CNCCookbook Customer Spotlight Extended Cut

My friend Bob Warfield at CNCCookbook has a series where he is shinning spotlights on customers of his GWizard Calculator and GWizard Editor software products.  A very cool idea! Along with my buddy Brad Martin of TacticalKeychains getting the spotlight I was fortunate enough to be one of the ones to be featured in a Q&A style write up about War Machine LLC and how it got its start.

You can check it out here by clicking on the image below

customer spotlight

However there is much more to the story than that so I have a bit more of a detailed write up in a more prose format that I would like to share with you below.


In 2005 I had no idea what it took to make a product from scratch, manufacture it, and bring it to market, but I figured it couldn’t be that hard. Well, I was about to find out hard, how hard it could be, the hard way. Being an avid firearms enthusiast since the age of 9, hunter, and shooting competitor, I began some serious new AR15 builds when I came across a curious problem involving the rifle’s gas block. In short I recognized a problem with the existing front sight base and the advent of the increasingly popular optics and red dot sights. I thought about it for a while and searched high and low for a solution but found none. Every product on the market that allowed for a flip down front sight for a clear sight picture was either too low to accommodate the vast majority of sights, made of comparatively soft aluminum, made of steel but too heavy and clunky, looked like it was hacksawed from a railroad tie, did not have a sling attachment point, too wide to fit inside of a modified forearm rail, did not have a functional bayonet lug, or was not pinned to the barrel. Nothing out there satisfied all of the ideal criteria for me as somewhat of a functional perfectionist.


I thought about how hard it would be to make one in my garage from scratch and soon concluded that it would probably be really really easy…ahem. Over the next few months I couldn’t shake the idea and began trying to make prototypes out of paper, then plastic, then plaster of Paris. I kept searching for the perfect product during this time hoping it would turn up so I could forget about it and spend my evenings doing something else. I started drawing up plans on graph paper to maybe give to a machinist somewhere and hope he wouldn’t use it wipe his spindle bore or something. At some point I heard about CAD and tried coming up with a design using but that never really went anywhere, especially when I saw how much a prototype would cost. Later I somehow got ahold of some old, outdated, decrepit copy of AutoCAD and a few books from the library on how to use it. Here is where I modeled the basic design over the course of a few months. With a thumb drive full of binary code I consulted with my alma mater university engineering dept. to have a 3D printed prototype made, but they seemed put off that it was a gun part and were reluctant to get involved. Next I searched some sites like for a shop that would make it with an imposed NDA. Ding I found one that was in America and didn’t require a kidney as payment, unlike most of the quotes I got. I had the first prototype made, or rather the first thing made. Apparently when I sent the shop the file they had machined the wrong part, it was still a gas block but it was a goofy design that I was playing with in a hidden layer and not my final design…and that’s what I got for many hundreds of dollars and many months wait. I tried it anyway and it didn’t work at all, the rifle would not even function. Which was weird because I had used more than enough duct tape to secure the rifle to the broken pallet while pulling the trigger from behind a hay bale with about 30ft of high grade bailing twine. I took what I could learn from that expensive chunk of garbage and made some design tweaks and tried to get another one made. That one turned out right to my design but not so good either, still I redesigned, tweaked dimensions, and upgraded the tape and twine to a bench clamp and kite string. Finally I got the one you see in much of my web art mounted on the carbine. It turned out rough but much better and seemed to fit OK. I Parkerized it in my garage with some chemicals I got from Walmart and pinned it to my barrel with a drill press that I bought on sale from Sears just for this occasion. Oh, and I did all of this the night before leaving for a professional carbine training course where I would be firing about 3k rounds at high intervals and taking only this rifle. It was a brilliant plan and nothing that I could think of could possibly go wrong.


Well unbelievably, shockingly, impossibly, I was right, nothing went wrong! I almost forgot that I was firing a science experiment until about the 4th or 5th magazine down range. During the course I had some instructors come up to me in the live fire house and ask me what kind of gas block I was using or where they could get one, being especially intrigued that it was pinned to the barrel and not clamped or set screwed as is typically found. They thought I was joking when I said “I made it and you can’t get one.”

IMG_1232 IMG_1230zoom IMG_1243

During the next few years I went to numerous rifle courses like this one, participated in numerous 3-gun competitions, and did lots of shooting on my own with no real hiccups related to the gas block design! Since I modeled the gas block from scratch with some old calipers it did however always have a slight gas leak around the gas tube because the original model I made specified a gas tube bore that was about 0.002” over the size it should have been. As a result it sometimes malfunctioned with extremely under-powered ammo. Otherwise when using good ammo it functions just fine and in fact it is still on my rifle and going strong to this day. Upon positively identifying the source of this occurrence I corrected that lone error and now thought it was time to level up and have more made to sell. I tried to get the original shop that made the prototype to make more but they declined. Apparently it was waaay too hard to machine and I was told that they were piling up too many broken endmills on it. I bounced around trying to find a shop to make them with no success, again mainly because the design was too complicated to machine which would make them too costly. Too costly was an understatement. Trying to find a manufacturer was proving futile. Most wouldn’t get back with me, the ones that did were very negative on trying to make them, and some were straight up rude.

IMG_1199War Machine LLC's WarBlock High T-4 AR15 High reciever level same plane Railed Gasblock War Machine LLC's WarBlock High T-4 AR15 High reciever level same plane Railed Gasblock

During this time I was still looking for a product that had all of these features so I could finally not feel “obligated” to create it. Not finding any I began to wonder if I was missing something, maybe there was some reason no one made it, after all it seemed to me to be a no-brainer so maybe I really didn’t have a brain. I really couldn’t think of any and that perplexed me even more. I finally said “F it, I’m going to make it myself and see what happens.” “Now where do I start…there must be some sort of doohickey that makes things like this,” I thought. My only experience with milling machines happened when I worked part-time at Toyota during college servicing welding robots. We were replacing a large copper cable on a robot that was malfunctioning. Our test equipment showed the cable was OK except for the terminal being pitted from electric arcing. Instead of scraping the extremely expensive cable we walked over to an area that had a HUGE drill press, at least that’s what I thought it was at the time. It was actually a CNC converted Bridgeport. We gave it to some machinists and told them what we wanted done. They whipped up a quick program, put the cable terminal in a vise, and then the machine came alive for about 10sec. facing off the terminal to a smooth flat surface. I remember I was highly intrigued staring at the mirror like copper surface as we walked it back to the robot thinking of all the possibilities that this new-fangled witchcraft represented. My first passion was a desire to make silencers but that never materialized. I was lost even after frequenting the DIY section on silencer forums and just being bewildered by the tools of the trade.

Back to the present story I started a frenzied research endeavor on-line to discover just what sort of wizardry was required to sculpt metal in such a manner. I inquired about taking some college courses in machining but I worked full time and it just wasn’t in the cards. That’s when I stumbled on CNCCookbook. Here I spent many hours reading everything I possibly could about this mysterious and fascinating field that I had never given much serious thought about except always thinking it was likely unattainable to the common man. While reading through all of the blog articles, something clicked and I was finally starting to get it. All of the reading there was paying off. Suddenly the prospects of making these crazy gas blocks that were otherwise beyond the realm of mere mortal machinists didn’t seem so distant and enigmatic. My first real money-on-the-barrel-head commitment was buying a $60 set of Gibraltar strap clamps on sale from Enco. I thought “Well it’s too late to go back now,” but in reality I had already passed the point of no return.

After a ton of research I took the plunge in 2010 and ordered a brand new CNC mill, a 3hp Mikini 1610L, for fear of the complexity of converting one or buying a used lemon. The thought of buying a NEW lemon, with no real support, from a company that would promptly go out of business after my purchase, had never occurred to me, especially after all of the glowing reviews I had read online. When I ordered it I was still lost as to how to operate it, what programs I would need, or what tooling or tool holders to use. I figured now was the time to act, not over analyze it, I would figure out the details at a running pace along the way. In addition to buying a mind-bogglingly expensive CAD/CAM package I had also purchased Gwizard CNC Calculator and it has been invaluable to getting the correct cutting parameters without burning through piles of endmills, which I did anyway but that’s a different story.  Months later when the machine was delivered I ran into a whole separate saga that I won’t get into here but it is chronicled in an article here at Three years later I was finally ready to use the now disdainfully nicknamed war machine (if that isn’t an attention getter). The first project I decided to tackle was unintentional, but I had stumbled on some very nice pieces of grade 5 titanium for a good price because they were odd sized scraps. I made some novelty precision dice with a decorative box, and that project, code named Luxor, has been written about right here on CNCCookbook.


After getting a feel for the new motor and electronics I was ready to attempt what every other machine shop I had contacted refused to do. It was either going to be an epic success or a complete failure, mediocrity was not an option. I didn’t care, I was ready to give it my best, if for nothing else to just rid myself of this burning desire to make the best most perfect taper pinned receiver height same plane Picatinney railed AR15 gas block with bayonet lug and QD sling swivel socket, ever made. Project WarBlock was a go. It was a daunting prospect especially for a full time father of a little boy and an infant to care for, with no time, and even less money. Again I won’t go over it in detail, as I already did that on my blog here, and my Youtube channel here, my Facebook page here, and my most active venue Instagram here, but I think I gave it a run for its money. You can let me know what you think! After I made the first run of WarBlocks, it dawned on me that I have to pin these onto barrels myself. I can’t expect the customer to do this highly technical and potentially perilous operation! Not an easy task to say the least. I think the first few barrels I did had me rethinking the whole predicament I had gotten myself into. I lost more money on broken drills and reamers per barrel than I was grossing in revenue from the sale itself! I however knew it had to be done so I chalked it up to tuition and I worked at it. Now I’ve turned barrel pinning into a science and in fact I might be the only person that actually CNCs this process, a process traditionally done with a manual Bridgeport.

WarBlock standard configuration Carbine AR15 receiver level same plane railed gas block with bayonet lug pinned with taper pins

There may be a million ways to skin a cat but there are far fewer ways to do it efficiently and with superior results…and with less angry meowing. I’m still working feverishly refining the manufacturing techniques for the WarBlock and its variations hoping to one day take the design to a new level and make a bullet-proof adjustable version that would be good enough to be considered standard operating equipment on any AR15. Although judging by the gushing feedback I get from virtually every customer I have, I think the WarBlock is already there. Though the WarBlock is about as perfect as it can be at this point, refining the process to be faster, more precise, easier on tooling, tweaking tolerances, getting the perfect nitride finish, seeking dealerships with component manufacturers, sourcing raw materials, and finding the best tooling, etc. has been the name of the game since the beginning and is definitely an ongoing process and not a destination. I’ve had great experiences with Speedy Metals, Maritool, Bob at CNCCookbook, Tormach, and especially Carl from Lakeshore Carbide who really took an interest is seeing this project succeed. It’s not all about machining though, penetrating the AR15 market without lots of capital is a tough road. Website development, SEO, marketing strategies, etc. are all areas way outside my field of expertise but things I work on a little everyday trying to get better and better. I have other product ideas in mind and am excited to dive into them but they will have to wait their turn, I am still up against the clock more than ever and still trying to make due with a small slow mill, limited tooling, and limited space. The important take away is that even with both hands tied behind your back it is still possible, you just have to hop a little. One thing is for sure, the future will require bigger, badder, and faster cat skinning machines…and a bigger shop!

WarBlocks AR15 receiver level same plane railed gas block with bayonet lug.

The WarBlock High T-4 Receiver Height AR15 Gas Block is now in stock!


The WarBlock High T-4 is now in stock and orders are open! They are available as a stand-alone unit to pin yourself (highly ill-advised unless you know what you are doing) primarily for gunsmiths, armorers, manufacturers, and highly technically proficient DIY with the tools and the know how. They are also available already professionally pinned to the CMMG line of barrels using standard taper pins. In addition there is also an option to send in your unpinned barrel for WarBlock pinning.

WarBlock flyer website

This video details a lot of the main points behind why the WarBlock is superior to mounting your BUIS to the forearm rail as well as many of the advantages it has.

Rounded out with the maximum versatility of rail mounted flip-up sights without sacrificing the proven robustness of the standard front sight base. In fact the WarBlock is the best of both worlds offering a hard mounted platform for the plethora of flip-up front sights while maintaining the lightweight and ruggedness of the standard front sight base.


-Receiver level 1913 Picatinney Rail
-Reinforced standard bayonet lug
-Intergral QD sling swivel socket
-Weighs only 4oz., less than the standard FSB at 5.5oz.
-Pins securely to barrel using standard taper pins
-Fully CNC machined from hardened 4140 billet
-Salt Bath Nitrided finish

The pre-release sale price is only $87 + free shipping! That is a steal! This sale will not last long as that price is a loss leader to promote earning an early customer base while we’re working out the kinks of production.

S4450007 S4450001







Please support us and help us take this product mainstream!

WarBlock Receiver Height AR15 Gas Block CNC Machining (17): soft jaws II squaring stock and barrel pinning

Here we are with the second installment of the soft jaw series within a series. In this episode we explore the other two primary functions that the soft jaws were engineered to serve. The first being the final machining off of the back tab. We start with a problem encountered when using the traditional vise jaws, actually a two-fold dilemma. The first being that the standard jaws only had a length of 6″ so the maximum number of blocks I could hold at once was 6. However all of the other operations have a throughput count of 8, leaving us with a bottleneck to production. The new soft jaws are 8.5″ in length so that we can now hold 8 blocks. The second dilemma was that minor variations in the tolerances of the blocks could leave some that are unclamped and uncaptivated by the tightening of the vise jaws in a conventional manner. Just as with the upper row of bolts in the moveable jaw we have a lower row that are spaced 1″ apart and they will serve a similar function here. The blocks are machined to be 1″ thick and when stacked together in the vise for machining operations they can individually be clamped by virtue of 8 bolts eliminating any possibility for a loose block or one that is not secured against the fixed vice jaw. The lower row of bolts are also centered on the vertical height of the blocks so that when tightened they apply a balanced load acting against the block forcing it tight into the fixed jaw and ensuring a square cut.


In practice this worked well and some unexpected benefits were revealed by this design. The first of which was discovered when facing the 12″ long bars of saw cut steel square and to size. In the past I had been using a long aircraft drill bit to act as the round stock used when squaring stock, and while this worked it was always tedious trying to wedge the shaft in between the moveable jaw and the bar of steel without the bar falling while simultaneously trying to tighten the vise. A much better method I found was to place a length of non-springing stainless steel wire to take the place of the drill. The advantage posed by the soft jaws is that the wire can be wrapped around the the far bolt on one side, ran tightly across the face of the jaw roughly centered on the thickness of the bar, and then wrapped around the last bolt on the opposite side. This created a round stock that would stay put between re-orienting and changing the bar and ending up being a surprisingly significant time saver. The stainless steel wire was also hard enough to resist flattening even with significant vise pressure.







The second benefit discovered was pertaining to squaring the saw cut sides after the 12″ bar was sawed into 5 individual 2.25″ blocks after facing operations. The problem is that the saw cannot cut nearly square enough compared to the mill and the saw cut side is what needs to be indexed to square it to the rest of the already machined sides. It would be very difficult to align 8 blocks at one time to be square to some reference point other than the vise jaws. However with the bolts that can be individually tightened we now have an easy solution. The first step is to open the jaws just enough for blocks to be easily placed between them, then one block is placed on the far left side. A 1-2-3 block is placed beside it sitting flat on the parallel and since the saw cut block cannot be trusted to be square to the parallel it is canted in the X direction to butt up tightly to the very square side of the 1-2-3 block which is square to the parallel. While this arraignment is held securely by hand the appropriate bolt is tightened to clamp the block pinning it against the fixed vice jaw and securing its orientation. Then all of the subsequent 7 blocks can be added and clamped together in the X direction with a large C-clamp thus butting all of the machined sides of the blocks against each other and to the same squareness as the first captivated block. The top can them be faced smooth and square. When flipping the stack to machine the other side this process can be omitted as not necessary since the freshly faced side is now presumed square.

Now for the final designed function of the soft jaws and to many what will be the main event. If you recall in the last video there was a shallow slot and blind threaded holes that were milled into the back of the fixed vise jaw. These features are for another fixture that will be used in conjunction with the fixed jaw itself. When drilling and pinning a gasblock into an AR15 barrel they must adhere to certain criteria with regards to spacing and alignment. First the obvious is that the gasblock must be aligned rotationally such that the gas port outlet of the barrel matches up with the gas port inlet of the gasblock. Although there is a certain amount of fudge factor built into the stringency of this alignment given that the barrel gas port is on the order of 0.063″ in diameter (varies with barrel length and gas system) whereas the gasblock inlet port which must align to it is 0.1405″, a pretty generous misalignment safeguard. However a standard front sight base, gasblock with integrated front sight, or railed gasblock must be much better aligned than this or else it will affect the front and rear sight alignment. In this case with a railed gasblock we have a registration feature in the top flight deck portion of the rail since it is milled perfectly square with the rest of the gasblock as well as the bore. So we must simply devise a way to rotationally index the top rail of the gasblock to the gas port of the barrel but more precisely to the top rail of the receiver. How do we do this without a receiver? An AR15 barrel extension has a 0.125″ dowel pin called an indexing pin drilled and pressed into the the exact 12 o’clock position in relation not only to the bolt lugs and feed ramps, but also the gas port. Then in later assembly this indexing pin aligns the barrel to the receiver buy slipping into a milled slot in the exact 12 o’clock position of the receiver which is also square to the top rail. Armed with this knowledge we can then design a holding and aligning fixture based on these features.

Doing a cursory review of some preexisting barrel pinning fixtures on the market they seem to index the barrel in a similar way to how the barrel in secured to a receiver. There is basically a metal plate with a 1″ hole for the barrel extension and milled slot to accept the indexing pin. Fearing that this method would not be as precise due to any slop or play in either the extension hole or the indexing slot I set out to make something better, something with a more positive dead reckoning. I pondered this and settled on the only method that would always ensure zero play in the alignment no matter what the tolerances of the barrel and that only required 3 points of registration, one being adjustable to fit. The mighty 90 deg. angle also known as the V block. I began thinking of ways to mill a V-block by canting the angles in the vise but that lasted about 5 sec. when it dawned on my that I have a CNC machine I can make any angle I want, we don’t need no stinking sine vises. Then it also dawned on my that since we are dealing with right angles offset 45 degrees I already have a tool to do this, the 45deg. chamfer mill. From here I simply machined the main “V-block” with endmills being careful to fully support the thin “ears” within the vice jaws, then in the same setup machined the other “V-block” with the chamfer mill to essentially make two V-blocks in one.

20150118_153118 20150118_153226






In making this fixture I made sure to key all the mating pieces so that they can positively align and remain square with only one bolt holding them together. The long bar simply rides in the snug fitting slot in the back of the fixed vise jaw so that it is also keyed square. When thinking about this fixture it only made sense to make it adjustable for different barrel lengths so the long bar can be pulled out to accommodate the different length barrels. I also made it reversible for no reason other than it was easy to do at the time and would really help in the off chance I needed to secure a barrel from the other side.







So how did it work? The fixture worked perfectly but there with a few tooling hiccups. The first time I tried to use it I went out to drill and pin a customer barrel and realized the large drill chuck I have will collide with the hand guard cap behind the gasblock. The reason being is that I have to bottom out the drill bit and the reamer in the chuck so I can measure their tool offset and it will be repeatable from then on. I also do this with common drill sizes and spot drills. With no good way to extend the length of the drill and reamer downward and maintain repeatable tool offset I would either have to do the drilling and reaming in separate operations measuring the tool offset each time or simply find another way to extend the stickout of the tools from the chuck. I searched for awhile and the best the solution seemed to be to use an ER collet tool extension. The widely available set screw type extensions are only found in standard nominal sizes (usually 0.25″) so they wouldn’t work without introducing a huge and unusable amount of run out. Settling on a colleted extension I also needed a somewhat small diameter to clear the hand guard cap. I ended up having to order the tools off of Ebay from China since I couldn’t find an equivalent tool extension domestically. This meant a long shipping time. I went with the smallest I could find which was ER8m. The m just meant they used a slimmer nut that required its own specialty wrench. These arrived and worked great when installing in the drill chuck, just like a 3/8″ drill bit would, but they are long enough to bottom out in the drill chuck for repeatability. The only problem encountered with this set up was that the the collets weren’t tight enough to hold the tools. In the case of the reamer it just simply wasn’t torqued down enough but in the case of the #31 drills the 3.5mm collet had actually reached its maximum compression and would not grip the drill bit any tighter. That of course broke both of my carbide drills trying to drill through the very hard nitrided surface of the barrel. I replaced that collet with a 3mm and it fits the drill much better. Since the proper way to pin a gasblock is to ride the edge of the barrel, meaning the center of the hole should be right on the horizon of the barrel leaving the hole exactly half way into the barrel, this means that the drill is essentially cutting through the hard nitrided surface the entire way through the barrel. This is very hard on the tool and in my case when switching to the #31 TiN coated cobalt drill and the drill chuck, even a rigidly held tool in an otherwise good cut was burned up. I need to stick to TiAlN carbide and perhaps step up the RPM and reduce the feed.


I did a lot of testing to hopefully find the perfect depth to ream to leaving only the hammering in of the pin remaining for completion. What I found after testing many dozens of pins is that the pins are not of a consistent enough size to do it this way and get the perfect depth pin every time. The pins really must be hand fit with the reamer to get a good equidistant protrusion on each side. Given that, I thought I could still use automation to my advantage and ream most of the way down and then simply finish it by hand, a strategy that worked very well despite the reamer slipping and not reaming the hole deep enough.


Fortunately the spot drilling portion of the operation went well thanks to the 140 deg. 3/8″ spot drill from Lakeshore Carbide which apparently now only come in double ended versions (just be careful seating them in the drill chuck).

Lakeshore Carbide 140deg. spot drill

Lakeshore Carbide 140deg. spot drill

The other issue was with the rigidity of the barrel during drilling. When drilling the forward leg that is unsupported the barrel tended to flex downward from the force. This was no doubt exacerbated by the fact that the drill responsible for that was the cobalt drill with the rapidly disintegrating cutting edge but it is still an issue that must not be allowed to persist. Simply placing a vertical support to hold the barrel from any downward Z movement will eliminate this problem entirely.


While waiting for the tool extensions from China to arrive I didn’t want to waste any time and had started production on the gasblocks on Dec. 16 2014 which coincidentally was the 241st anniversary of the Boston Tea Party, and just as the colonists did in 1773 we will once again find ourselves brewing up an ocean of High “T” before you know it.

If you didn’t get that reference I guess you’ll just have to stay tuned for the next installment.





WarBlock Receiver Height AR15 Gas Block CNC Machining (16): soft jaws and milling off back tab

After the excitement of the first production run we need to float back down to earth for a little bit and face the reality that these gasblocks still aren’t done. We need to remove the back tab that has been securing into the fixtures up to this point. While it sounds easy enough it was actually quite a complicated little thing to tackle. First off I had to design a set of “soft” jaws to hold the things now that they are a nice irregular shape with very few flats and a tendency to crush or deform if clamped too tightly. I say soft jaws in quotes because they are not actually very soft, rather than the soft aluminum they are traditionally made from I decided to make them out of hardened 4140 steel just like the gasblocks. In fact I actually made them from the very same bars I buy to make the gasblocks with, so out of two bars I get two vise jaws and two gasblocks. Here at War Machine we are always looking for economical ways to do things so we can pass the savings on to you.

With that said, let us begin with part one of the multi-purpose soft jaw series within a series.

First off these jaws had better be able to serve numerous roles to justify the time, expense, and design work tied up in making them so for part one we are just going to focus on the primary objective that these jaws accomplish and that is milling off the back tab of the gasblocks. As with all of our other fixtures we are going to try and harness some economies of scale and the power of CNC by creating jaws that can hold as many gasblocks as there are coming down the pike to eliminate any potential bottlenecks to production. Both fixtures 1 and 2 contain eight stations so we have a target number for the jaws. The only way to hold eight gasblocks in a vice is vertically with the fantail pointed skyward. From this orientation there is no flat for the movable jaw to seat against, in fact the only flat practically available is the one inside the sling swivel socket…I knew I put that there for a reason. So a boss on the movable vise jaw could fit the bill but lets think about that for a moment. The added complexity and CAMing time notwithstanding consider that with fixed bosses any tiny deviation in the gasblocks dimension could compromise the clamping force when the vise in now clamping the largest gasblock and not clamping any others. It’s true the boss would prevent the gasblock from being ripped up towards the endmill but it would not be rigidly held by any stretch when confronted with the whirling helical flutes of a cutting tool. We must devise a way to clamp each part individually. Well, the socket hole is 0.377 in diameter so coincidentally a 0.375 socket head cap screw fits right in there no problem. We have our clamping mechanism. Drill and tap holes to run bolts through the movable jaw but make sure they are spaced such that any out of square tabs are tolerated. I left a tolerance of 0.05″ between the actual tab dimensions of 1.0″ for a spacing of 1.05″. The bolt also need a little bit of work so I squared the faces of them flat with an endmill.

resized youtube thumbnail 16

Since I don’t have 8″ parallels and the ones I do have are for the standard height vise jaws I designed in fixed parallels which would support any workpiece where the jaws overhand the vise bed and can be used in conjunction with normal parallels to raise the work to a similar degree in relation to the taller jaws. Of course the main design feature is their height, leaving just enough clearance above the jaws for the back tab to be milled away and they stickout from the face of the jaws only enough to register squareness across the broadest part of the forecastle but yet just shy of the logo so as not to encounter burrs.

upper bolts on gasblock soft jawsparallels on gasblock soft jaws







For milling I initially wanted to use the big EM90-750A 0.75″ indexable endmill from Glacern Machine Tool using Mitsubishi APMT1135-VP15TF inserts.

glacern 075 indexable endmill

Initial testing suggested this was not feasible. At the depth of cut required (~0.25″) that endmill exerts so much force on the workpiece that is just smacks it out of the way even when clamped traditionally in a vise. Knowing that the gasblocks would be clamped even less rigidly than that I knew it would mean big trouble. I then decided the size endmill that I’ve had great luck with even when the part is not in the most secure setup possible, that’s right, the good ole’ 0.375 rougher. This size is good for not transferring too much force to the work. I’m still using the “Fireplug” from Lakeshore Carbide but it has taken quite a beating with numerous chipped teeth on the cutting edge. Up to this point it is still working so I’m still using it.

Even with this smaller endmill the initial cut testing did not go well. First the part pushed over in the vise and became canted so the cut wasn’t square. Then there were overhangs of steel that broke free from the tab and became loose canons that could damage the fragile carbide of the endmill much like a loose canon that breaks free of its lashings and roles around the ship damaging the hull.

First the toolpath. I chose that Y axis toolpath for a few reasons. One reason was to confine the cut to one part at a time so that if it were to break during the cut at least I could have some finished ones instead of all of them being in a unfinished state, that was probably the least compelling reason. The secondary reason was that I incorrectly assumed that the force exerted on the part was in the axis of motion so I thought by going up and down in the Y axis meant that the part was always being pushed into a vise jaw. However the biggest reason was that there would be fewer interrupted cuts endured by the endmill. When the cutter goes in and out of material it is stressed more so than when remaining in a cut. Despite there being unavoidable holes that the fireplug has to plow through at least it wasn’t also engaging and re-engaging with every part as it went through all eight gasblocks as it would if it were moving horizontally in the X axis.

Well here is why I was wrong. First of all the parts were not rigidly secure right off the bat but here is why my toolpath appears to have exacerbated the problem. The force imposed on the work piece is actually quite lateral to the direction of travel. It’s as if you are mowing your grass and expecting the clippings to be thrown forward in front of you because that is the direction that you are walking. Not so, the clippings and in this case the chips but more importantly the force exerted on the work is in the direction more or less tangential to the rotation of the cutter.  So by going up and down in the Y axis I’m actually pushing the part over in the X axis where it is least supported against opposing forces.

Now the problem with the overhangs. That is more of a unique problem per part and in my case this toolpath also made the problem much worse. The overhanging tab, once cleared from connecting material, will simply fall away. However if there is no place for it to completely clear the cutting plane it becomes a hazard to the cutter and possibly to the work as well. My Y axis toolpath ensured that the hanging tab would not fall away.

What I did to remedy this was to change the toolpath to have X axis travel. This pushes the work into the vise jaw and eliminates hanging tabs that can fall into the cutting path. The one hanging tab that does fall away does so right at the end of the cut and I made sure the endmill just barely overhangs the part to ensure the hazardous free tab is thrown clear by the endmill and if it wasn’t at least it wouldn’t fall into the its path. In the words of Brodie from Mallrats, “You face forward or you face the possibility of shock and damage”.









So addressing the simple fact that this setup lacked rigidity I tried to design something quick and dirty that could fix this problem. After a test cut using a bolt for a T-slot I ended up moving forward with a bracket of sorts that has a threaded bar so that the gasblocks can be secured downwards towards the parallel keeping everything square. I used flanged sleeve bearings to serve as washers because they could slip down in the bore of the gasblocks maintaining their alignment without slipping off. I used 3/8-16 x 1″ socket head cap screws since the heads just about fit in the 14mm hole on the gasblock tabs, I had to grind the diameters down a little so they would fall through. I also milled a 60 degree chamfer on the near side of the bar to match up with the chamfer of the bayonet lug. This not only allowed me to center the threaded holes on the bar which make for easier machining since I did most of it manually but it also help to support the bayonet lug and prevent it from flexing when clamping down. The flange bearings also need a little flat and a radius to clear the radius of the forward leg and underside of the gasblock. Nooge.

By just bolting on cross members to the threaded bar this allows them to just slip under the vise jaw overhangs forcing the threaded bar to remain flat and tight to the vise bed. It also can move somewhat freely to reposition for other purposes.

support jigsupport bar for gasblock soft jaws








We are finally finished with the machining phase of the gasblock although this is not to say that improvements will not be made, different tools used, or different techniques will be tried to improve the product. These will all certainly happen and of course I will update this blog with them when they do. The next stop is to send these initial space monkeys off for finishing. I was originally going to Manganese Phosphate Parkerize them but have now tentatively opted to have them Salt Bath Nitrided. That is how I wanted to finish them many years ago when shopping around for a machine shop to make them for me but the price was prohibitive. It seems the price has fallen, quality has improved, and is more widely available. don’t you just love the free market. Nitriding is a far superior finish to Parkerizing and very very hard, after all it is the same “coating” that is applied to my carbide endmills to cut hardened steel, and it is more corrosion resistant.








So join me next time for part 2 of the soft jaw mini-series where we will explore the secondary objective of these jaws. Snoochie boochies

WarBlock Receiver Height AR15 Gas Block CNC Machining (15): first test production run

Finally after many months of work preparing for an actual production run we are now here. Before that we first knock out a little side project that I had done some work on months ago for a customer. I had worked closely with this customer on a prior successful project so I thought this would be an easy “reproduction” run. With the previous part there was a problem that required some further machining but it is good now. The material used is a carbon based alloy that is soft but difficult to work with due to being gummy. Nevertheless it went fine and with the exception a couple of sort of bumpy patches and some areas that left some fuzz had an unbelievably smooth surface finish especially on the back side. The level of detail was simply amazing. There was a lot of chatter in areas, especially at night, but it quieted down somewhat eventually. It held decent tolerances at 21.5 inches +/- 1.oooo” or so for being such a hefty part at 7lbs 6.2oz. Unfortunately I noticed that even though the tool was cool to the touch the work was quite warm and had moved a little. The customer also complained a little about residual thermal expansion as this piece grows a little dimensionally everyday and may have some stresses in it as it continues to warp and change shape. I also had a slight problem with coolant drainage. Apparently I had quite a bit of tramp collect in the bottom of the part which must have became rancid. The total op time needs to be improved on a little as it is taking a little longer than the first one at 38 weeks vs 37 with the first but is still under the estimated program run time of 40 weeks. There was a another problem with the machine overheating and puking on me but that was early on in the program and she eventually settled down. The program run finally ended at 4:49 am Sept 12, 2014.

Previous part
Previous part
New part
New part








Moving along, I was never really satisfied with the way the blanks mated to the fixture plates. It seemed that the tolerances required to keep the part aligned were so tight that hitting them seemed like a matter of chance. I’m specifically talking about the main drilled bore hole and the gastube hole, those are features that align with the bosses on the fixtures and need to be snug with no play. The way I had been doing it before was basically just luck that they would be the exact right distance from one another after I flip the stack of blocks over from one op to the next. After getting a particularly bad fit on two of the test blocks I decided to rethink it. I first greatly increased the depth of the gastube hole chamfer thinking this would allow some flex in the boss by engaging it more towards its tip and make for a better fit. After trying it on a few the fit was marginally better but still not good. Let’s rethink it again, we need a repeatable bullet proof solution. The large boss is the reference point of all the gasblock geometry and machining operations so it needs to be fairly snug, and it is. The boss that goes into the gastube hole is simply for rotational alignment nothing more. I remember someone telling me to use diamond shaped bosses for rotational alignment because they are a little more forgiving of anything short of an exact match. I thought about how they work and decided that instead of modifying the fixture plates to either accept diamond boss pins, or to machine the round bosses into diamond shaped pins, I would alter the blocks slightly. If the only real purpose of these bosses is to rotationally align the part and the only real problem with them now is that the holes that mate with them are not always the exact same distance from each other as the bosses are, well then really the only important dimension is the side-to-side captivation of the boss. And thus I milled a small rectangle pocket right over the gastube hole. Now the vertical boss/hole misalignment can be as much as 0.1″ in either direction while the long sides of the rectangle pocket snugly secures the boss and prevents any rotational play. Perfect. This also had the added benefit of making donning and doffing the parts on and off the fixtures much easier.

All of the Fixture 1 operations went almost perfectly. The only issues were just CAM errors where I left a plunge feed too slow or something. The two endmills in these operations that are doing the bulk of the grunt work are the long 0.5 Lakeshore carbide rougher and the 0.375 “Fireplug” rougher of the same make. They display some great performance but have taken a beating in just one run of eight. The 0.5″ rougher sustained numerous chipped teeth and I wonder if it will make it through another run. I will try it and see but the cuts starts to get louder and the deflection climbs. The Fireplug wasn’t completely new when I started this run but it also has a lot of chipped teeth. Both of these endmills are still cutting so I will just see how long they actually last until failure. The 0.5″ rougher needs to last more than one run of 8, lasting 16 gasblocks would be acceptable, 32 to 48 would be ideal but probably unlikely. It’s one of the most expensive cutters in these ops and the longer it lasts the lower I can keep the price of the gasblock. All of my other endmills are holding up remarkably well. I did however replace the 0.1875″ endmill because the chipped flute tips from the Mach brain fart are now causing the endmill to deflect more than usual.

I replaced it with  this one from Lakeshore carbide.

Lakeshore carbide corner radius endmill
Lakeshore carbide corner radius endmill

I think this one will last much longer given radiused flute tips. I got the least amount of radius at 0.01 to start with and even that should make a big difference.

fixture 1 x8 toolpathS4100002



















I understand CNC parts are expensive but this project came to fruition from the perspective of me being a consumer a customer and a user. I like parts that above all work great, are reliable, over built, elegant, and also affordable. Seeing parts grossly over priced for the sake of their rarity is a decision the maker ultimately has to make, but as for myself, a believer in the unfettered free market, sees an opportunity to streamline the manufacturing and/or increase the supply to meet the demand and reduce the economic impact of scarcity thus keeping the prices in check. Otherwise you are inviting the competition to undercut you. To me exorbitant prices out of league with reasonable manufacturing costs are off putting. I have a naive vision that this gasblock will become the gold standard gasblock for the A3 and A4 flattop AR15 much as the flattop AR15 has all but replaced the A1 and A2 style fixed carry handle AR15. Making it extremely costly is a perfect way to remain obscure and, while a perfectly legitimate market practice, ultimately short sighted from a business perspective in my view. This is why if given the opportunity to invest in larger equipment and automation my ultimate intention is to increase production and make these gasblocks ubiquitous features. I’d like to see them as an evolutionary time point in the history of the AR15, casting off the vestigial traits of the previous and increasingly maladaptive generation.

Transferring the gasblocks from fixture 1 to fixture 2 was a breeze and totally solid with the new way I was milling the rectangular boss captivation pocket in the back. Again as with fixture 1 the Fireplug is doing an inordinate amount of the material removal in fact the bulk of the material removal here on fixture 2. While it has its battle scars it still seems to cut fine. The especially concerning aspect of the long 0.5″ rougher sustaining damage in one run it that it is not used anywhere else but on fixture 1 front side. It doesn’t pull double duty anywhere else like the 0.375″ Fireplug does so its economy is somewhat fixed.







I was apprehensive about the milling on fixture 2 because so much depends on perfect alignment to get everything to cut just right and not slightly gouge the part here or leave a little too much there. The main reason for the apprehension was the fact that my Haimer 3D Taster indicator has some X and Y backlash as mentioned before and I wasn’t sure if I could completely trust it to indicate correctly. When machining one of the practice parts I noticed a somewhat significant misalignment error in the 3D contouring causing some gouging on one side of the fixture and too much material left behind on the other side between flips. I narrowed down the cause and attributed it to the Haimer backlash. When indicating the Y axis I approached the fixture from two different directions from each side of the flip. This resulted in the backlash showing up in the part. However in this production run I made sure to always approach in the same direction so as to cancel out or negate the backlash from having an effect between parts.

The program run times took a little longer than hoped but were also much better than I had originally anticipated. Much of this is due to the deflection caused by using those long 2.5″ stickout endmills on fixture 1 resulting in a lot of deflection and thus requiring multiple spring passes. Perhaps a more rigid machine that can handle larger diameter cutters can alleviate that. The next most time costly operations are the 3D contouring. The main issue with those operations being as long as they are is due to my spindle speed limitation. Only having 5000 available RPM really limits the feedrate that you can push small ballmill cutters. The 10,000 to 30,000 RPM spindle of a larger machine with servos could really speed that up. There are some things that I could potentially do to reduce cycle times once I get a handle on the price to performance ratio of pushing everything harder but right now I will leave it alone and just get some cooking and out on the market. I will update the cycle times of each op here when I get a more accurate time of the final procedure dialed in. I’ll mention when I do this in a future blog post so you can refer back to here if interested.

Cycle times x8 Production Run

Raw material cost x8: $72

Stock preparation (facing, squaring, etc): 58 min average per x8

Main bore and gastube hole: 51 min cycle time

Fixture 1 Front Side:

Fixture 1 Top Side:

Fixture 1 Bottom Side:

Fixture 2 Side 1:

Fixture 2 Side 2:

Back tab Milling:








You probably noticed the new vise jaws in the video. I made those for milling the final operation but I designed them to serve numerous roles. That will be detailed in the next installment.

I ran into another major problem between machining side 1 of fixture 2 and the flip to side 2. The machine computer would not boot up. Through some investigative research I discovered that the harddrive had likely failed. I have a drive rescue kit which is comprised of a device that powers any harddrive and converts it to USB, and when I powered it I could hear is was spinning slowly and clicking, not good. I did the old trick of putting the drive in the freezer, letting it get nice and cold, then powering it up. It actually sounded better but didn’t spin up fast enough. I tried toggling the power on and off to give the disc enough momentum so that it might take off. To my surprise this worked! It took about a minute of power on power off cycles before it just took off. I was able to make a disc image of it using imaging software but I also simply copied the contents of the drive over to my desktop computer. I restored the image to a new SSD I went out and bought thinking it would be more immune to the machine vibrations but it would not boot either. I spent quite a while trying to fix the master boot record and repair the boot sector using the recovery console in Windows XP. Nothing worked, the data on the drive was no doubt corrupted from the crash. Fortunately I made a disc image when I first got the machine and restored this to the SSD and everything worked fine. Of course I had to go and update all the drivers for everything, touch screen monitor, chipset, LAN, onboard graphics, etc. I also had to reinstall the latest versions of Mach3 and the ethernet smoothstepper plugin. This was all no big deal really. What was really bothering me was all my Mach settings. Well I scoured the files I was able to recover from the failed harddrive only to discover many of the files that had been open right before I shutdown the previous day were gone including my .xml settings file and my tool3.dat file containing my tool table. The crash must had deleted them. I recovered my custom screenset but some of the .jpg images were corrupted or gone. I downloaded new ones and was able to recover the backup of the current .xml I had been using in the xmlbackups folder (go figure). What confused me was that they did not have the .xml file extension but something like .16b11 instead. Turns out all you have to do is change the extension to .xml and it is the same current .xml you were using before. Wow that saved my ass from having to redo all that stuff, half of which I don’t even remember. The only thing I lost was the tool table and some hand written Gcode programs that I used frequently. I remeasured all my tool offsets, rewrote two of the Gcode programs but the 3rd was too complicated to rewrite on the fly. Then I remembered that I posted the code to the CNCzone forum a long time ago when I was trying to get help with writing it. Luckily I was able to search CNCzone and find it. Back in business. Of course I made a new drive image once I did all that in case that ever happens again.

Well I must say that everything up to this point has been an enormous undertaking and a long and difficult journey for just one guy with more on his plate than any 10 people should have to deal with but I am getting very excited with the progress. We are still not done yet. There is still much more to do so keep watching, keep reading, and I’ll keep working. Stay tuned!


WarBlock Receiver Height AR15 Gas Block CNC Machining (14): finalizing side ops

In this installment I just wanted to finish up cut testing on the side ops mainly to eliminate chatter and to improve surface finish. There is not a lot to comment on here so this will be a short blog post. However here are some things I found useful to know when troubleshooting and developing tool paths.

Raise your 3D toolpaths above the model:

This has the benefit of tailoring the 3D tool path to your actual part geometry instead of a theoretical model. Backlash, tool wear, and mostly (in my case) tool deflection all contribute to a complex workpiece that does not exactly match the computer model. The work generally ends up being slightly larger than the model I’ve found, especially if you are careful to use mainly climb cutting where the tools deflect more away from the part being cut. This means that if you generate a 3D toolpath it will cut true to the model and on an oversized part leave artifacts, tool marks, and abrupt changes in surface geometry. By simply experimenting with lifting the tool path up off of the model you can eliminate these issues and make for a better more aesthetic part. In the case of this gasblock I only needed to lift the 3D contours 0.001″ above the model to get it to blend well due to the spring passes on previous ops. On the other hand the 3D paths on the lightning window needed to be raised a full 0.005″ to blend. This is no doubt from the concave geometry causing endmill deflection on that 2.5″ long cutter despite the 2 additional spring passes.

Create operations that negate the requirement of absolute accuracy of previous ops from a different setup:

If you machine something in one fixture or setup, then move it to another fixture or setup and expect the geometry to match the new tool paths based on that geometry, you’re going to have a bad time. It can be very very close but you are walking a tight rope. As the endmill from the previous ops wears, the parts geometry will change and this change will be evident when later paths are now cutting “too deep” relative to the part even though they are cutting true to the model. Consistency suffers. Also unless the positioning of the part on the new fixture is absolutely accurate i.e. no play whatsoever this can induce minor surface inconsistencies. What I had to do was run another pass over the areas that had previously been machined from a prior setup to make sure the geometry was “aligned” or matched when in the new setup. The more independent you can make operations in a new setup from those in a previous setup the better. With this knowledge in mind you can even think about this when designing a new model in CAD.

Don’t be afraid of spring passes:

As was mentioned before endmill deflection plays a huge role in part accuracy and I have found that nearly any milling will result in some amount of deflection. What is acceptable is up to the tolerances of the part. Areas that either need to be very accurately machined or whose dimensions are critical for a future op such as 3d milling could certainly benefit from a spring pass or two, and yes even with carbide tools. Yes it takes a little longer, yes it adds a little more wear to the tool, but your parts will be more accurate, more aesthetic, and more consistent. Plus it is probably ultimately less wear and less time than if you took many more lighter less aggressive cuts to eliminate the deflection. I would say the longer the tool and the heavier the cut the more useful a spring pass is going to be. Adding 2 spring passes on top of the finish pass in the main profiling op using the long 2.5″ endmill really tightened up the geometry proper to the model and eliminated potential problems down stream.

Turn down the RPM to eliminate chatter:

This one is a no-brainer but I was surprised just how far I had to reduce RPM to get the squealing to stop, by about half in some cases. In other cases I was right on the edge of chatter. For example with the 0.25″ finish passes after the side gutting I went from 3145 to 2800 to 2450 to 1850 and it still chattered. Then going from 1850 to 1815 completely eliminated the chatter like a switch and the finish is amazing. Like Carl at Lakeshore Carbide says, once you drop the RPM to the point it stops chattering you have found the happy chipload for the cut, then you can increase the RPM and the feed accordingly, maintaining that same happy chip load. While I have found the happy chip loads, I have not speed things up yet. Perhaps I will do that later to even further optimize the process but for now I just need to make them without breaking any more tools.

Use surface planes to modify 3D ops:

If your CAD/CAM combo can do this (it should), you can add planes or surfaces to specific areas around the model to fool the program into thinking they are part of the model and thus generate tool paths around them. Simply lifting the tool path above the model by increasing the stock material value is a very two dimensional exercise. Suppose you wanted to lift the path a little in one area but not in another, copy a surface of the model then lift or tilt it upwards in the Z. Generating a 3D op around this new geometry will alter the tool path. Case in point notice how I created some planes in the image on the right to not only modify the behavior of the ballmill while still cutting the same geometry but also to create more of a gradual entry into the work piece. You can also use this technique in conjunction with containment regions to control where the tool changes direction so that it does so out side of a cut so as not to leave tool mark artifacts in the surface of the work.

arc fitting 3D tool paths
3D tool paths to model with arc fitting
3D containment surfaces
3D tool paths using containment surfaces









Use plunge milling to relieve tight corners before a profile pass:

When confronted with a corner radius that is equal to the radius of the endmill obviously the best action is to drop to a smaller radius endmill. Due to the fact that the radial engagement spikes to 50{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} almost instantaneously this causes a very loud chirp and will often chip the flutes of your fine finishing endmill. If you are already slotting or engaged 50{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} or more radially taking a hard turn wont hurt anything as the endmill is already at maximum deflection. However in some cases dropping to a smaller endmill may be impractical because the next size down may not have a long enough flute length as was the case with me forcing you to use an endmill of the same radius as the profile cut. Slowing the feed way down didn’t really help, chipping away at the corners first in very light stepovers didn’t really help either. Absent a trochoidal or high speed tool path the best solution seemed to be simply plunging the corner out. This actually worked surprisingly well and has yet to hurt the flute tips of my endmill.

Use as few Z levels as you can get away with:

Do it all in one Z level if you can. This engages more of the cutting flutes of the endmill and minimizes the amount of time the flute tips are engaged. The flute tips are the most fragile and most vulnerable part of the tool and often when they get chipped or damaged, as they typically do long before any other part of the flute, the tool stops cutting well and is trashed. Imagine cutting to a depth of 1 inch for 1 inch of travel. Taking a 0.1″ DOC results in 10 light passes or in other words 10 linear inches of cutting is imposed on the flute tips and bottom 0.1″ of the flutes. However taking a 1.0″ DOC engages the flute tips for only 1 linear inch and uses much more of the tool you paid for. Both cut geometry’s are the same but the gentle and light 0.1″ DOC is 10 times harder on the tool than the aggressive hogging at 1.0″ DOC and will potentially offer a 10x shorter tool life. Sometimes tool deflection and chatter will limit the amount of depth you can take so you may have to divide it up into 2 or more Z level passes but the goal is as few as possible. If you must sacrifice something to pull it off you should usually sacrifice width of cut to get a greater depth of cut. Again in my case I wanted to do the side gutting in one Z level but simply couldn’t due to the tool deflection of my 0.375 rougher (may it rest in peace). Now with the Lakeshore Carbide Fireplug the limiting factor is not deflection but flute length so I kept it at 2.

Don’t be afraid to slot:

Much has been said about avoiding slotting ops or where you have 100{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} radial engagement. While sound advice, in my experiences here sometimes I was doing far more damage to the tool by trying to avoid slotting. I was creating all kinds of funky tool paths to take multiple passes simply to avoid the dreaded slot. I was also chipping a lot of tools doing it this way. Imagine my surprise when I just said “screw it” I’m slotting and the slots ended up being the quietest and gentlest cuts I’ve ever done. The caveat is that you really need adequate coolant pressure to really clear those chips out.

Don’t be afraid to helix:

Like the slot the other dreaded op is the plunge. So logically since a helix employs both it should be feared right? Well, what I’ve found out here is that provided the helix angle is shallow enough the helix is a great way to create an accurate odd sized hole for which there is no drill size or for some other reason that precludes drilling or circular interpolation.

Always try to maintain a climb cutting condition where practical:

This just leaves a nicer finish, is easier on the tool, is more predictable as far as deflection goes, is much less likely to gouge or undercut the part, and it’s easier to clear chips from. The few times where I tried conventional milling (out of necessity) I was quite disappointed with the finish. This also should impact how you arrange the part on your fixtures. If I had known better I would have arrayed the gasblocks on the far side around the other way instead of mirroring them so I wouldn’t have to fiddle with reversing each tool path but instead just mirror and flip the whole thing and be done with it. The way I have it now the tool paths on the far side are opposite the near side so that involved quite a bit of tweaking and uncertainty to get them back to climb milling. Luckily there were no major issues. Live and learn.

Here is a parting shot of all the side tool paths for 2 gasblocks in stations 1 and 5 of fixture plate 2

side ops toolpaths x2

Now with all that done we are ready to begin expanding the instances of these ops and switch gears into production testing. There are very exciting things coming up next so say tuned!





WarBlock Receiver Height AR15 Gas Block CNC Machining (13): double down and fixture fixing

After I machined fixture plate 1 to have the correct sized bosses I decided to jump right in and machine two blanks simultaneously, the first time I’ve done more than one on the fixture. The only problem was the the gastube hole was drilled slightly off location in reference to the main 14mm bore hole (slightly too far away). This caused the blanks to require way too much force to seat to the fixture, much more than usual and ended up raising a small ridge of deformation on the back of the blank ultimately preventing it from seating flat to the plate. After I fixed this by filing the ridge down and chamfering the holes deeper I realized two problems. One; I think I changed the origin point of the spot drill operation and this resulted in it not drilling deep enough so there was nearly no chamfer around the hole. The chamfer is important because it relieves the contact area around the boss and facilitates the blank seating flush to the plate. It also lessens the necessity of perfect alignment since if the hole engagement is only near the top of the boss then some amount of flex is possible to ensure fully mating to the surface of the fixture. Lastly if there is some deformation of the blank caused by excessive seating force it will not affect the ability of the blank seating flush to the plate. Secondly; when I was re-indicating the blanks after I flipped them to drill the gastube hole I noticed the Y was off 0.001″ from where it was before. Consequently this is also where the error was. In the future I need to simply indicated the Y off of the fixed vice jaw and keep that value after the flip. Doing it this way will also require that the blanks either be perfectly square (which these weren’t) or I need to add a strip of tape as Brad at Tactical Keychains does or perhaps some aluminum welding wire as Tom Lipton at OxtoolCo suggests in his book Sink or Swim to the movable vise jaw to allow the side against the fixed jaw to sit flush and square. Other than that small issue everything cut perfectly.

Double Trouble

There was a slight code problem when cutting the bayonet with the 0.1875″ endmill. Somehow I managed to change the tool in the CAM to a 0.375″ so I had to go back and fix that.

The spiral cuts with the 0.5″ rougher went fine and I was pleased that it cut the two blocks simultaneously just fine. I had never tried it before it was all just theory in my head. The cuts did sound a little rougher than usual because that endmill is getting dull and has a few chipped teeth. I may retire that one rather than risk breaking something. Hell, maybe I’ll just break it. We’re testing here right?

The finish pass and two spring passes cut just fine and reduced the top to bottom draft from 0.003″ to 0.001″. The only issue here was the long thin chips building up between the parts like steel wool. I blew them out with air but perhaps will one day search for one of those serrated finishing endmills. Those leave a fine finish but break up the chips much smaller so that they can be more easily washed away by coolant. In fact I might reposition the nozzles a little so they blast more coolant between the parts without one part occluding coolant flow to the other.

The 0.109″ roundover worked great after I increased the birth to 0.006″ and this left no ledge. Preventing a ledge is tricky with these cutters and seemed to just require trial and error. The chamfering worked great too but I had to go and do it twice because for some reason the Logo didn’t cut deep enough. I remeasured the tool and determined it was slightly off a couple of thousandths, I then set the new offset and ran the op again. The logo turned out great at 0.002″ deep. It is subtle but boldly emblazoned in the most conspicuous place on the gasblock, the butress, just forward of the flight deck. I am fiercely proud of my logo, my work, and my reputation. At the same time I don’t believe a serious part like this needs to have a billboard plastered all over it. Plus it only takes 16 seconds per part to engrave. In the future I may add more text somewhere with a laser engraver but until then the logo will suffice and will probably always be machined in as a matter of syle.

0.002" deep War Machine logo
0.002″ deep War Machine logo

I made some minor changes to minimize unnecessary machining on the recoil grooves of the Picatinney rail. Other than that the top side ops all went perfectly. I’ll probably go back and increase the plunge feed to rapid to further speed things up now that I feel more comfortable with it.

The main changes I made to rectify prior issues in the bottom ops was to widen the 0.047″ roundover to take a 0.005″ wider pass to eliminate the ledge it was always leaving behind. This worked perfectly and left no ledge at all. I also changed the entry point of the Woodruff cutter to initially engage in a thicker area as well as to raise the RPM slightly to take less of a chipload and have less deflection. The most significant change I made to eliminate the bump it was pushing out on the side walls was to leave 0.003″ of material per side up from 0.001″. I actually tried 0.004″ at first but the sling socket swivel button did not fully retract after being depressed so I bumped it to 0.003″ which worked great and left no bump at all. A new path I added and you probably noticed it from the last video is widening the amount of the material the 0.375″ rougher takes from the bottom. I did this when I realized the side gutting entry would partially engage the cutter on the conventional milling side and I wanted to avoid that for the sake of tool life.

Speaking of tool life I improved the entries of the rougher so it doesn’t abruptly enter the material over and over again. I did this by introducing radial engagements, cut rounding, and mixing climb and conventional milling to some of these toolpaths. One of of these radial entries I accidentally left as a rapid. This claimed my favorite endmill, the mighty 0.375″ rougher. I would like to give this tool a 21 gun salute and taps. Its death was a violent as its life and it will not soon be forgotten.

maritool 0375 rougher
Maritool 0.375 rougher

Its successor is the “Fireplug” from Lakeshore Carbide that I’ve talked about before. I’ve not had a chance to use it much at all yet but what little that I have done with it suggests it will give be even better performance. The verdict is still out but I do have some concerns about tool life with the fireplug. I used it to machine down the face of fixture plate 2, an operations that I have done twice before with the Maritool rougher not to mention all of the other cuts I’ve made with it and it seemed to appear like new for a long long time. In fact it wasn’t until very recently I even noticed any damage to this endmill. When using the Fireplug I dialed down the aggressiveness of the cut a little bit to better extend tool life because speed was not an issue here. Disappointingly and quite surprisingly that one op left several small chips in the flutes of the endmill. I’ll just continue to use it and gauge the total life of the tool before making any determination about which one to stick with.

LakeShore Carbide 0.375 “fireplug” stub rougher

The problem. So after I machined the two blanks I put them on fixture 2 and started taking some measurements. Curiously it appeared that one side was still higher than the other side despite fixing the side to side play issue of the bosses on fixture plate 1. In fact it was a very consistent measurement even between the two different in-process gasblocks, one side was always 0.004″ higher that the other. I first thought that it was caused by the index point being slightly off from the first boss. After investigating this I discovered that this was partially the case. The numbers I was using for the offset from the sides of the plate where I probe and the first boss was off in the X by about 0.002″, I could even barely see the misalignment with my eye in the spot where the large roundover intersects with the bore. Great, simply change the offset. Wait, this didn’t account for the other 0.002″. I took both of my fixtures to the surface plate and then to the mill for some very accurate and precise final measurements. What I found was that the bosses on either fixture plate were not parallel to the sides where I probe from and thus were not centered as they should be to the body of the plate in the case of fixture 2 resulting in Z height errors. In the case of fixture 1 it would result in increasing Y axis errors going from left to right in the X, which would ultimately result in all manner of location errors when in-process gasblocks are loaded onto fixture 2. This error was 0.002″ at its maximum deviation. Bingo there it is.

The cause. How could this be. I always carefully align the vise when mounting it to the table but deductively this was the only way this error could have occurred. I had recently removed the vise and disassembled it for a thorough cleaning so I had oiled it, reassembled it, and adjusted the movable jaw lift set screw. I dressed both the vise bottom and the table surface with a file to make sure everything was as perfect as it could be before mounting it. First I aligned it as I usually do, with the Haimer 3D Taster across the fixed vice jaw. Having a suspicion about the validity of this measurement given the backlash discoveries I made about the Haimer I re-indicated the fixed jaw with the dial test indicator. Whoala! it was off 0.001″ over the 6 inches of the jaw. This translates into an error of 0.002″ over 12 inches, the length of the fixtures, and consequently the exact amount of maximum error I was measuring on the plates. The bottom line is that the amount of backlash in the Haimer makes it unsuitable to indicate and square a vise to less than 0.001″. Unacceptable. I mentioned before that I had contacted Haimer USA to see about having my 3D Taster inspected and perhaps repaired if it was indeed found to have a backlash problem outside of normal specs. I was told the inspection would be as high as the cost of a brand new unit. Ok that is seriously effed up right there. Scratch that idea. I guess I will abandon using the very expensive Haimer for aligning anything accurately in the XY plane and instead use the more sensitive and much cheaper Mitutoyo DTI instead.

The fix. Well fixture 1 was a relatively easy fix. I simply aligned the bosses instead of the sides in the Z axis using a 0.001″ shim, adjusting the shim until the bosses were perfectly parallel with the X travel in the Z axis. and re-faced all of the sides square. I then very carefully measured the offsets from the new side surfaces to the bosses using the dial test indicator and Noga magnetic base attached to the spindle. I did this several times and it was always within a few tenths. Easy Peasy.

Fixture 2 was more tricky. I first repeated the procedure as before with fixture 1 to square up the two sides relative to the bosses. This was somewhat more complicated because unlike with fixture 1 I had to machine a precise amount of material from each side to ensure that the bosses were perfectly centered between these two sides. Once I did that and verified it I then had to buzz down the bosses on the other side and re-machine the whole face again using the new offset numbers from the other side. I then verified that everything was right with the world on all sides in every possible way to the extent of the precision of my meteorology tools to verify everything so that it could no longer being a lingering variable in my mind.

The aftermath. I had several stiff drinks and fell asleep.








WarBlock Receiver Height AR15 Gas Block CNC Machining (12): 3D machining side ops revisited

Welcome back. I’ve been having a difficult time keeping up with getting my blog posts out on time but rest assured I have been machining every second I can to get this process down and ready for production. I have a bit of a back log of video and writing so while the power is down for the day I get to step away from the machine and work on catching up on my laptop. This episode will be similar to the last one but with some new tweaks to try out. I also discover a problem with fixture plate 1 and have to re-machine it.

So lets dig in. I started out by modifying some of the toolpaths from the bottom to create extra clearance for when the part is moved to fixture 2 and the endmill engages from the side. I also changed the side gutting ops to use a stub length endmill rather than the standard length one I’m actually using. You’ve heard me talk about the Maritool 0.375″ rougher I’ve been using before and it has been a real trooper but it was starting to show its battle scars and I knew it would need to be replaced soon. Even though I decided to just keep on using it for testing until it broke I went ahead and re-programmed these cuts for two Z levels passes because the “Fireplug” stub I plan to replace it with only has a 0.625″ flute length instead of a 1.0″ like the standard one. My deepest depth of cut is 0.89″ but there is only one op like that, all my other cuts with this endmill are 0.4″ or less. Maybe in the future I’ll just get more tool holders and use the longer endmill here if it saves time. I was able to further eliminate some chatter areas like the slotting in the bottom op with the 0.1875″ endmill by simply increasing the feedrate a little while taking it in two passes. Widening the roundover a little more helped with the ledge issues but perhaps not enough. Also the Woodruff cuts were modified to try and eliminate the bump it pushes out of the thin side walls but again perhaps not enough.

Now that I have pretty well demonstrated the power of the spindle and for the most part am able to base my depth of cut on the maximum allowed by the part geometry, one new thing that I’m trying out is fiddling with the optimum width of cut. Before I was defaulting to 70{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} max width based on some things I had heard from Sandvik Coromont but I’ve always been curious about this chart posted on CNCCookbook in an article about deflection.

0 radial deflection when climb milling at 20{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} and 60{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} WOC

If you look you can see that conventional milling always results in radial deflection of the tool either away from the part with cuts below 50{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} engagement and into the part with cuts greater than 50{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a}. However the different forces involved with climb milling result in some curious idiosyncrasies. With climb milling the tool is never deflecting into the part and is generally back driving the axis (which is why you shouldn’t climb mill on a machine with an Acme lead screw due to the backlash causing the tool to sort of violently jump back and forth during the cut). However at 2 points in the radial engagement spectrum the forces are balanced between pushing the tool away and forward and pulling the tool into the part and backwards. According to the chart these points occur and 20{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} and 60{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} radial engagement. I’m not sure if programming cuts at these widths is good or bad in terms of chatter. On one hand the deflection being absorbed rotationally or axially instead of radially may minimize chatter due to less sideways endmill deflection. On the other hand the chatter may be exacerbated by this very fact and could potentially result in the equivalent of milling right on the center line at 50{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} (which is to be avoided). So far the latter does not seem to be an issue. However if there truly is no radial deflection this has the potential to increase the accuracy of the cut as well as make subsequent cuts with finishing endmills easier.

The slight ringing of chatter heard in the roughing ops when side gutting are definitely from part deflection rather than tool deflection as it happens on the front leg of the gasblock that is sticking out unsupported once the center bridging material is milled away. It’s not bad and I’m not worried about it since there isn’t a whole lot I can do about it without cutting productivity.

The finishing profile using the 0.25″ endmill has never really gone well without loud screeching in those tight 0.125″ radius corners. I only want to change the model to widen those radii as a last resort so I’m trying to come up with a slick way to make those turns without chatter, preferably with the same 0.25″ endmill. I started by first trying to do it in two slow passes. That didn’t work so then I tried relieving the corners first with a series of five 0.005″ passes. That didn’t work either. So I’m either forced to use a smaller tool or plunge mill it. Since my next size smaller tool won’t cut deep enough axially I’m going to try plunge milling the little crescent of corner out before going back and profiling it. Then later in the profile cut I’ll add a small radius to the turn so that the tool avoids the very tightest part of the corner and won’t engage. The finish on the linear portions is fine but I run a spring pass anyway to make sure the dimension is as accurate as possible so that the 3D ops match up better with the geometry. I may need to lift the 3D ops off of the model a little more in future testing and if I do I could probably drop the spring pass. I also have another tool from Lakeshore Carbide I want to try after this one wears out. It is a 4fl variable flute corner radius long length 0.25″ endmill with a 1.125″ flute length. I don’t have a lot of experience with variable flute endmills but when I do use them they work phenomenally well. Usually much better than standard flute as far as chatter goes. The small radius on the tips will also enhance tool life and potentially improve the surface finish when plunging.

Lakeshore Carbide long length variable flute finishing endmill

Next I came back to the issues of the 3D toolpaths cutting too deep. After some examination I have determined that fixture plate 1 has a flaw. I mentioned before about how when I first machined it I tapped the holes after I machined the bosses causing the bosses to slightly increase in size. This forced me to re-machine them resulting in me cutting them too small. Now the fact that the bosses are too small the drilled blank fails to index properly due to some side to side play. I measured the total play to be 0.014″, in other words HUGE and completely unacceptable. So when I bolt down a blank to fixture plate 1 it maybe as much as 0.007″ off to either side. It is then machined correctly but yet cockeyed to the boss. Later when it is transferred to fixture plate 2 where is fits tight the part is now between 0.007″ too high or too low rotationally around the boss depending on it’s orientation to how it was secured on fixture plate 1.  The only way to fix this issue is too re-machine the face of the fixture plate. Fortunately I could reuse the fixture plate by just buzzing off the bosses and sinking the face down 0.1″ and re-profiling new bosses to the correct size as on fixture plate 2. This worked out great and now everything fits tight. Actually a little too tight but I think increasing the depth of the hole chamfers on the blank should take care of that.

Before I made the modifications to fixture plate 1 tried the 3D machining again on a blank that happened to fit on fixture plate 1 more snugly than the first one did. This greatly improved the problem with the cuts being too deep but they are still slightly too deep. I determined that the mostly likely explanation for this other than fixture plate 1 misalignment is draft due to deflection. When doing the profile cuts with those long endmills on fixture plate 1 endmill deflection makes the part right on dimension at the top but slightly wider at the bottom. In fact with one finish pass and one spring pass still results in about 0.0025″ or so of draft from top to bottom. This consequently is about the amount that the 3D paths are cutting too deep on the back of the gasblock. The front being right on. I’ve gone back and added another spring pass to the profile resulting in a draft of only 0.001″. Not perfect but much better. In addition to the two spring passes I will also lift the 3D path up a little more to 0.001″ off the model. I may go more but we’ll see if it’s necessary.

Next I lifted the 3D path for the lightening window up off the model from 0.002″ to 0.005″ which turned out great and was even perhaps a little too much so I later turned it back down some”. Lifting the path off the model 0.004″ was perfect but with the two spring passes I might even be able to lower that to 0.003″ or maybe even less. The original toolpath here left a small divot and a slightly chattered vertical wall which doesn’t matter or hurt anything but I’m going to try and get rid of them as a matter of principle.  Re-programming the toolpath a little actually made it worse so I contemplated going back to the way it was before but I can’t leave well enough alone. My motto seems to be “if it ain’t fixed, break it until it is”.

Perhaps the biggest improvement gained in this installment is the addition of arc fitting to the 3D toolpaths. When the natively generated 3D toolpaths are simply and huge series of straight lines this causes the machine to seriously slow down around curves which extends the estimated cut time by about 3x and makes the ground jitter and shake like one of those old fashion exercise machines that would wrap around your gut and shake the fat off your ass. When I was able to go in and add arc fitting (a feature of RhinoCAM Pro) I was able to speed up the whole operation and get it to cut so much more smoothly. It also greatly minimized the chatter. Those areas where it slows down reduces the chipload and increases the likelihood for chatter, by keeping it moving along at the programmed feedrate keeps a healthy chipload and make for a quieter more peaceful cut. Playing with the parameters such as tolerance, minimum arc length, and max radius arc this basically replaces the series of lines segments that make up a curvature into one or more simple arcs. It also has the added benefit of drastically shortening and simplifying your code. For example in the case of the lightening pocket the native code generated was just over 13,000 lines. With arc fitting this is reduced to about 3300.

linear 3D tool paths
Natively generated linear toolpath
arc fitting 3D tool paths
Toolpath with arc fitting (arcs in dark blue)









We end this episode with a 15x video of re-machining fixture plate 1 to fix the errors with the boss sizes. By cutting the bosses to 0.0005″ under per side (0.001″ smaller diameter) than the hole size of 0.5512″ resulted in what previously had 0.007″ of play from side to side down to 0. Unfortunately this was not the only problem with the fixtures as new ones were uncovered and rectified but I’ll save that for next time.



WarBlock Receiver Height AR15 Gas Block CNC Machining (11): 3D machining lightening pocket and sling swivel socket

As hoped I was able to get some machining done this week. In this episode I wanted to begin operations on the lightening pocket which always seemed somewhat daunting to me. When I first bought my CAM software a few years ago the first thing I did was try to figure out how to do this feature using 3D milling and it left me feeling a bit intimidated. However it is one of the few remaining major operations so I figure just jump in and see what floats. Well I don’t know what is floating but I certainly know what sinks, may bank account, in the form of carbide endmills. The first attempt at this pocket left me with more carbide to add to the broken pile.

My thinking was that since a square endmill is prone to chipping and a radius corner endmill less so, then a ball endmill must be even more so. This combined with the prospect of how long 3D milling can take led me to turn up the feed on these cuts. The first op I did was simply to pocket out the main flat bottom area. I tried a ramp angle of 10 deg. which chattered but the actual cut sounded OK for the most part except for the chirps resulting from the ballmill engaging too much material in a 90 deg. or tighter turn. Right at the end of that op the turns start to become more and more acute and the radial engagement starts to become more abrupt causing tool flex. You can actually start to see the ballmill flex a little on the few passes before it just snaps. The flutes never chipped though. On the bright side it did not take too long and the finish was incredible with a 2{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} stepover. Even though the floor of that pocket was simply done with successive passes of a ballmill it appeared to have been face milled with a flat bottom cutter.

After some deliberation with my CAM I decided to first relieve the area where those tight turns were to be made using a back and forth ramp. In another post I mentioned needing to modify the model to be better suited to using a 0.1875″ ballmill. What I did was change all the radii of that pocket from 0.1″ to 0.09375″. This means I could cut the majority of the curves of that pocket with a single pass of a 0.1875″ ballmill instead of many passes.

After breaking my only 0.1875 ballmill on the initial profile op I ordered a few more from Lakeshore Carbide but this time instead of getting a standard length I got the stub length for added rigidity, I don’t use much DOC anyway with these. The stub length ballmill has a reduced flute length so that the shank portion (the strongest part) comprises more of the tool.

Lakeshore Carbide stub length ballmill

You can see the difference in flute length between the stubby and broken standard length ballmills here

broken Maritool vs Lakeshore Carbide stub length ballmill
broken Maritool vs Lakeshore Carbide stub length ballmill


So going back and making a clearance slot involved a back and forth 3 deg. ramp which did not turn out overly well due to chatter. I’m not immediately sure how to deal with this other than to slow the rpm down or decrease the ramp angle even more. Nevertheless it made the cut and cleared out the material albeit leaving a chattery finish on the vertical wall.

Since the first pocketing op resulted in some sharp corners I went back and massaged it a little with arc fitting and cut rounding, both features of RhinoCAM. This simply fits arcs anywhere they can be fit depending on what tolerance threshold you set and also rounds any change in direction so that these transitions happen much smoother with less shocks to the tool.

no cut rounding
2{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} stepover with no cut rounding
cut rounding
4{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} stepover with 0.015″ cut rounding







Ultimately the latter cut with the rounding will be smoother but I think I will change back to the 2{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} stepover and see what that looks like. When dry running the code many of the blue arcs resulted in full circles instead of half circles in Mach so I found out that you must check the “limit arcs to 180 deg.” box in the post processor so they will path correctly.

Running the simple profile pass around the edges of the pocket resulted in a beautiful finish without the need for numerous passes.

Now onto the 3D machining. I played with numerous 3D techniques but ultimately parallel finishing made the most sense so as not to require any hand manipulation of the code. I lifted the path off of the model by 0.002″ to be on the safe side for getting a nice natural organic blend of the contours but it still cut too deep. I attribute this to the actual part being slightly over sized from deflection. I have two options here. The simplest is to just raise the toolpath a little more off of the model to maybe 0.004-0.005″ and see what that looks like. I’m betting it will be fine but if not this 3D toolpath has a feature called radial loops. This causes the tool to gently retract radially away from the model, do a loop in the air to more over the step, then gently radially re-enter the cut. In fact this is such a cool feature that I might use it anyway and of course document the details when I revisit this operation in the near future.

Next I completely rethought the bottom ops and programmed the sling socket swivel feature. I decided that I wanted to do even less machining on this side than I originally thought. Just enough to get the sling swivel socket profiled, rounded over, and then the drilling and bore done. This feature had me a little perplexed about the best way to approach it. I wanted to minimize tool changes so if I could reuse one of the tools I was already using I would. Normally the bore would be drilled then finish machined but the fact that I needed to drill the gasport hole off center made me think the drill would wander badly. I contemplated plunge milling with the 0.375 rougher since I was already using it and it was the right size but was dissuaded of that notion. The next option was helical interpolation using the remaining available tool, the good ole 0.1875″ endmill. Carl at Lakeshore Carbide suggested this would be the better option as well. Fearful of chipping flutes tips as I always am I figured “hey they are already a little chipped and this endmill is a heck of a lot cheaper to replace than the 0.375″ rougher” so I ventured forth.

Again I was amazed at the sound (or lack thereof) of the 0.375″ rougher just quietly pushing it’s way though those heavy cuts, and this endmill is starting to show some wear on it. The slotting with the 0.1875″ endmill is still a little irritating to me because even cutting the DOC in half didn’t seem to change anything about the chatter. Next time I may reduce the RPM to see if that helps. The roundover is still ever so slightly too tight so I will widen the berth a tad more to avoid making a ledge.

The interesting part in that I’ve never done it successfully yet is the helical interpolation. I keep hearing about how it is so much better than anything else out there from places like Sandvik Coromont I thought this better be good. The last time I tried it years ago I ruined a $98 endmill within 30 sec. The very idea doesn’t sit well with me what with the flutes tip engaged from both the side and the bottom. It’s like they are just begging to chip. However people like Brad from Tactical Keychains does it all the time like in this video of machining the notorious grade 5 titanium and he’s getting good tool life. So I gave it a shot.

I went with the often recommended 3 deg. helix angle instead of the 7 to 9 deg. angle recommended by GWizard’s interpolation mini-calc. but I used its recommended RPM and feedrate. It made the hole without breaking the endmill however the flute tips were predictable (by me) chipped. They were already a tiny bit chipped from the Mach3 brain shart in the previous video but they were much worse now so I can’t necessarily lay blame to the cut itself, which sounded good. I know when a tool get chipped it begins on a rapid course to failure so this could be the case here and not the helical cut. I had tons of coolant forcefully shooting right in the hole and there was even the drilled through hole for chips to flow out of. I didn’t even really hear much chip re-cutting if any. Nevertheless I have opted to replace this endmill with again one from Lakeshore Carbide that has a 0.010″ corner radius thinking it might resist chipping better as well as perhaps making better cuts.

Lakeshore carbide corner radius endmill
Lakeshore carbide corner radius endmill

I chose the 0.010″ radius but may be inclined to try a larger radius in the future depending on tool life.

Now the crazy part was after the helical interpolation was done a coding error caused by some idiot who shall remain unnamed resulted in the finish pass to run at only 1300 RPM, much slower than any recommendation from GWizard that I could get. To my shock the cut was silent and the finish looked like a precision ground and lapped bore, it was that perfect. It’s so reflective I can’t even get a decently convincing picture of it, and it all happened by accident. The bore size was also spot on as the sling swivel stud fit it perfectly without rattling. This diameter dimension going from 0.375″ to 0.377″ was another change to the model I made recently and it definitely paid off, 0.375″ would have been too tight.

1300rpm 4ipm 0.4 DOC 0.001 WOC
amazing finish: 1300rpm 4ipm 0.4 DOC 0.001 WOC

Onto the sling swivel stud capture groove. I was able to use the Woodruff cutter for the first time and it cut just fine. If you look closely in the bore you will see the unique clover leaf like geometry of the groove. This is called a rotation limiter and it prevents the sling stud from being able to fully rotate around and get the sling twisted. The only issue with this op is that the sidewalls bulge just a little where the groove is cut because the stock material is too thin here. I can either change the model to tighten in the groove since I made it slightly too big by about 0.0025″ per side or widen the whole feature. Or perhaps both since I don’t want to widen it too much because it’s designed to be thin enough to fit inside of a forearm rail but yet don’t want to tighten the groove too much risking not working with slightly over sized stud bearings. Maybe just a little with each.

I just realized that all of the features of this gasblock have been cut one way or another (except the back which is just milled off). Now it is just a matter of refining the cuts to eliminate chatter, preserve tool life, and make accurate dimensions as efficiently as possible with the best possible surface finish.

Thanks for reading and see you in the next installment.


WarBlock Receiver Height AR15 Gas Block CNC Machining (10): chatter and interpolation

It’s been awhile since my last video or blog post due to too many extenuating circumstances to mention but I’m back with another one. This one will be yet again of particular interest to my machinist friends and perhaps no one else. From the last installment I was playing around with trying to plunge mill the bore to save time and tooling. While that might still yet pan out to be a viable option in the future for now I’ve decided to continue to interpolate the bore with the roughing endmill and then ream the bore to size. If you recall I was having a hard time with chatter when roughing the bore but not when milling the same cut linearly on a piece of scrap, so I decided to roll up my sleeves and get to the bottom of it. I was aware of the phenomenon of increased radial engagement when milling a bore but I was not aware of the scale of the problem. Let me explain. When you make a cut linearly your programmed step over is the same as the radial engagement so these kinds of cuts are the easiest to program predictably in terms of deflection. For example when cutting linearly if you have a 0.5″ endmill and your step over is 0.25 your radial engagement is 50{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a}, very simple. You can then use an excellent program like Bob Warfield’s GWizard Calculator to calculate the tool deflection and see if it is prone to chatter or stressing the endmill.

However when circular interpolating (or spiraling out to a certain diameter) the radial engagement is actually higher for a given step over than for that of a linear cut because of the way the work piece sort of partially wraps itself around the endmill. Even though the actual width of the material being removed may still be shallow the fact that more flutes are engaged at any given time or at least engaged for longer still translates into increased radial load on the tool, and thus deflection. This is intuitive but the real question is how much of a factor is it?

The problem I was having was that I could easily cut 0.8″ DOC, 0.07″ WOC linearly with my 0.5″ endmill no problem. It sounded good and cut just fine. Then I would drop the endmill down into a 0.5512″ drilled hole and interpolate out to a diameter of roughly 0.75″ using the exact same parameters and it had more chatter then an episode of The View. How could this be affecting the cut so drastically?

Well if we do a little analysis in CAD the magnitude of the problem immediately becomes apparent. The radial engagement drastically climbs when spiral milling a bore even though the step over is still only 0.07″. Interpolating using a 0.5″ endmill in a 0.5512″ bore effectively triples the radial engagement so that what we thought was a 0.07″ radial engagement producing 0.0039″ deflection is actually over 0.2″ of radial engagement producing 0.0086″ deflection. That’s over 40{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} engagement! No wonder it’s chattery. Move over Whoopi you’ve got some competition. Of course as the bore enlarges the radial engagement is not static, it changes and fortunately it declines as the endmill spirals out since the work is progressively not wrapped so tightly around the tool. So all one must do to calculate the appropriate step over for interpolation is to calculate what I call the magnum of arc. This is the maximum engagement that occurs at the end of the first fully engaged revolution of the cut just before the spiral arc takes the tool around to cutting previously relieved material. Then simply limit the step over to this, the maximum that the tool will ever radially engage in this spiral, as your desired radial engagement. What a mouthful.

And that’s what I did. I reprogrammed the step over distance of the spiral to be something more commiserate to ensuring an actual radial load of 0.07″ or 14{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} of tool radius. This step over value actually ended up being only 0.0089 or almost 8 times less than what we thought and what had been previously programmed.

Both cuts have a 0.07″ step over but the circular interpolations results in radial engagement approaching 1/4″.

spiral cut width amplification 5

I started to write an equation that would define all of these parameters but decided that is probably better suited for someone who can actually plug it in to a nice mini-calc in an already existing program…ahem Bob. Perhaps input parameters for initial hole size if there is a pre-drilled hole or if plunge milling a hole size equal to the tool diameter (which would always bring back an initial 100{bf1535d22b332f69d7a2d53a4e01e5867624aab1b456f60c0cd75d6084bc173a} radial load because any radial motion away from a plunge will always initially be full engagement), and desired actual radial engagement. I envision the mini-calc simply grabbing the tool diameter from whatever active tool is and then once the adjusted step over is calculated saving it to the main cut width box. One can then simply input this new step over value into their CAM software when programming interpolation or spiral operations and not run the risk of grossly over engaging the tool. This feature would be especially useful in hard unforgiving materials.

While doing these cuts the deflection was also noted in the bore and the code was modified to leave approximately 0.004-0.005″ per side for the reamer.

On this cut I also lowered the rpm a little bit to eliminate some of the polar harmonics that was occurring. Now overall the cut sounds much better. Here is the endmill I’m using from Lakeshore Carbide:

Lakeshore Carbide extended length 1/2″ roughing endmill

I did finally notice a couple of tiny chips on the teeth of this endmill but it still runs great and could just be a cumulative effect from all the screeching cuts I’ve made with it this far. Time will tell with a new tool and better cuts.

Moving on, I couldn’t leave well enough alone. I decided to go ahead and try a full depth 0.4″ full slot cut on the bayonet lug with the 0.375″ rougher with the idea that I’m using more of the flutes which therefore will result in fewer overall cuts with the flute tips engaged (typically the first part to chip and ruin an otherwise good endmill). It turned out amazingly good. I’m always surprised at what this endmill does and it seems to like slotting better that anything else, it just gets dead quiet when doing its thing with full radial engagement.

I’ve made an order recently for more endmills and am going to try a similar 0.375″ rougher from Lakeshore Carbide they call the “fireplug”. It is a stub length, variable flute, variable helix, corner chamfer, thick core, Viagra coated, roughing endmill. So basically all the tricks of the trade to make the most rigid endmill possible. Ok maybe not Viagra coated but certainly prescription strength stiffness and with less vibrations than well never mind. I went with it because I analyzed all the cuts I do with this size endmill and determined that nothing is really in excess of 0.4 DOC. There is one cut that does exceed the flute length and that is the cleanup pass when gutting from the side but I can easily program around that. Anyway here is that endmill:

Lakeshore Carbide fireplug roughing endmill

The spiral roughing for the profile suffers from the same phenomenon of radial overloading so had to use the same technique to tweak these cuts as well.

You can see the difference the 0.07″ step over and then adjusting it to a 0.0411″ step over has on the cut width albeit much less drastically due to the larger diameter circle. I have since modified the the current step over distance again but don’t feel like grabbing another screen shot but you get the idea.

spiral cut width amplification 1spiral cut width amplification 2

So when experimenting with the profile pass again I took note of the part dimensions to asses the severity of deflection. My goal was leave 0.004″ per side for finishing but when I programmed it that way it actually left 0.007″ per side on top of that due to deflection. So I ran a spring pass reducing it to 0.002 on top of that. Then a second spring pass leaving 0.0005″ over that. Hmmm two spring passes is no good so I altered the code to only leave 0.002″ for finishing and one spring pass. This ended up leaving 0.00375″ per side for finishing…just about perfect.

Using the same finishing parameters as the previous videos but right to the surface dimension of the part this got to 0.00025″ over. Probably good enough but a spring pass got it dead nuts on target. This kind of accuracy is important because of the front roundover step. As shown in prior videos if the profile dimensions are slightly oversized (due to deflection) then the roundover tool will leave an unwanted ledge. Even so I still modified the round over tool path to take a wider berth leaving 0.005″ of stock and eliminating the finish pass to ensure reliability in production and withstand any minor variation. I also added back the finish passes on the chamfering ops.

There was also a strange problem when cutting the bayonet lug recesses. I zeroed everything just fine, made some cuts, then when I switched to the 0.1875″ endmill and resumed the program the machine moved maybe 0.1″ too far in the X direction and it plunged my endmill straight down into the part (which damaged the flute tips and destroyed the part). I paused it quickly and studied the situation. Finding nothing out of the ordinary except the position of my tool I didn’t do anything but re-run the program and it then went to the correct location and did the cut accurately. I then simply loaded another part, did all the cuts up to that point and it cut everything correctly. The only thing I can think of is an anomalous Mach 3 brain fart.

Bear with my boring narration in all my videos and perhaps especially this one. When I make these videos it is usually late at night (early in the morning) and I am almost invariably dead tired. However I have found that my videos and blog posts serve as almost like a public notebook or journal that I have found myself referring to all the time when programming.

Hmmm what did that cut with similar parameters sound like?

What kind of deflection did I get with that endmill and that DOC?

What RPM or feed seemed to chatter with that cut?

What feed gave me a better reamer finish?

Was coolant able to get in there and clear chips OK on that cut?

These are all questions that pop into my head while programming and while my actual notebook is quite useful nothing beats being able to actually go back and see and hear that cut again organized with the cut data right on the screen to go along with it. That’s why I continue to post redundant and what I can only imagine to be boring video to non-machine freaks and lately I have even been really recognizing the value of adding as much data and information as I possibly can to each installment. It helps me tremendously and I only hope you all find it interesting or helpful to you as well.

With the foresight of having a small backlog of footage I don’t need a crystal ball to tell you that the next installment will be 3D machining the main lightening pocket and breaking more tools. Yay.

Until next time, and hopefully not so far out.