So the last image showed what was about to happen, more or less. Well, it happened. But not after a bit of bug hunting.
Once I wired the cage, controllers and TIM up, I realized that the ESCs would die and reboot when attempting startups under certain conditions. Brownout resets. They especially didn't like my sub-C NiMH pack that I tested on at all. Throwing a better battery at it "solved" the problem but left a nagging concern. The intended battery was still only 7.4V and a multitude of conditions (like low charge, cold weather, ...) could result in low voltage or increased battery IR.
The ESCs (ZTW Spider 40A non-BEC) had two linear LDO regulators in series to power the ATmega8 microcontroller. Now, this is not exactly an awesome strategy for deriving 5V logic power from a 7.4V battery that is also being hammered with noisy, switchy high currents from a PWM motor drive. It gives basically no voltage headroom at all and does nothing to ride through low spikes because linears don't block reverse current like buck/boosts do. Any caps hopefully placed on the 5V rail are basically useless here.
So linear logic power probably works great for applications that use 3S lipo or higher where there is enough extra voltage to keep the noise and transients up out of the trouble zone, and squeaks past for 2S when all the factors are not stacked against it working reliably, like they were here. In the end, I installed a buck/boost regulator (Pololu S9V11F5) on the TIM board, desoldered the 5V linear reg from each ESC, and wired the ESCs to get their 5V rail externally. The TIM now powers both ESCs through the wiring harness. Sort of inverse BEC. It wasn't too hard a modification, works great, and gives a lot of peace of mind.
Lessons learned for brushless builders: One, try to avoid 2S. The only case I can think of that wants 2S is this one, i.e. 4500-5800kv inrunners on 34mm flywheels. Most outrunners and everything else will use 3S or 4S. Moar voltage is good! Two, ESCs that have a big voltage range like 2S-6S probably have stacked regs (double the dropout voltage...) to manage power dissipation on the regs at high battery voltage. Perhaps all ZTWs do. Try to avoid those on 2S. Three, use enough battery. And four, test the hell out of everything before HvZing this type of power system.
So without further ado let's get to the part where we shoot darts.
This is actually an old-ass project and concept of mine. It isn't something I would have done now or in the future, but this is Project T17. T as in Tacmod, and 17 as in 2017 and as in a reference to the Milsig M17 CQC, a gun that suggested the basic layout of a short, brutish, functional carbine way back when.
I was over Rapidstrike receivers. There are a few problems with them, but the most glaring is that they mount the cage straddling the main receiver clamshell and the barrel shrouds which just drop in there. If anything warps, and these cheap parts are likely to be warped, it all fucks itself. Badly. The Tacmod 3.1 was one Rapidstrike too many. While I love bodyworking the things and they have this nice ready-made SCARness, they ain't that great. Similarly, barrel length. I stuck to my well proportioned rifle builds for too long. Nowadays, inner barrels are just deadweight on the elemental and purposeful flywheel blaster. It was clear that shorter was the way to go.
So what else can I build on? Nowadays, post-Prometheus, I would say, "nothing, fuck stock blasters". But in 2016 it was more like "Hmm, that Stryfe looks fairly nice." But it isn't full auto. Can I use these XSW parts?
Back when these drop-in full auto kits came out, I bought one. Never bothered reviewing it because honestly, it's a piece of shit. Aside from things like the crappy stock wiring and coming with an AA holder, the molded parts are all warpy and flashy and covered with grind marks from manual rework at the factory. The gearbox mounting is all kinds of fucked. The stock mounting plate bolts down to the receiver with one screw and mounts the gearbox with one machine screw and nut. Otherwise, it just kind of sits there. The sector gear and rack are the wrong dimensions to fit correctly in the unhacked Stryfe, so the gearbox mount has to force the gearbox badly off-angle to get the sector to engage the rack and not slip. The design of the bolt is awful, using the stock Stryfe pusher mounting boss, which is what is used to take the impact when the thing returns open (and will eventually snap off). The rack is offset about 3/4" to the left of the bolt centerline. The thrust from the gear pressure angle is pushing down on that offset edge of the bolt creating a large torque which is only restrained by the same mounting boss. A lot of friction and potential wear results. The plastic shim collar supplied for the mounting boss is the wrong thickness and must be hand fitted for the thing to function.
Then we get to the marginal bolt stroke, perhaps too short for a dart on the low end of the tolerance. And most of all, the weak-ass gearbox. It is a 48:1 "TT" box, a cheap component that has been around forever and seen service in things like the VLocity paintball loader. But it is TINY inside. Miniscule plastic gears. It makes a Rapidstrike box look like a Muncie M22.
In the end I think most of these Chinese full auto conversion kits are total garbage and not worth anyone's time. I also have similar sentiments about stock blasters, but that is a later subject as to why this is my last mod.
I did buy some 63:1 20mm metal gearmotors hoping to use them with modded XSW parts for bolt drive. Never even tried that. They are even tinier inside than the TT gearboxes and packaging would have been horrid with the motor sticking way out the side. Instead, since at this point this build was looking like a one shot deal, this happened.
I had one sitting around and the fit was too perfect to not give it a shot.
Now it needs cycle control, so we need a limit switch, a follower and a switch mount:
There. Quick, dirty and functional like the rest of the build.
For a stock setup I am running a buffer tube and AR-15 compatible stock. I initially planned to just insert a simple fabricated piece to align the stock with the Stryfe's wonky low bore axis on one of these Worker/F10555 buffer tubes that have a Hasbro square mount on the other end, and leave the thing as a Hasbro square-mount stock that can be removed. I didn't do that in the end though. I am also running my battery in here so it needs to be gutted (it's clamshell and stock it has screw bosses inside) and Devconned together.
Here is what became of the Hasbro square mount part. RIP, good riddance. I cut my F10555/Worker stock base in half, Devconned it to each receiver half, and did a thick PVC buildup. There is a reason for the thickness, you will see later.
Also all switches hooked up.
Workbench shot and some more progress going on with bodywork including the flat top job and the magwell. Perhaps you see the sling mount built into the stock base now as well, and the finished stock assembly with its mating plate to bolt up on that flat surface I created.
And some internals now.
Detail of TIM location above breech, and motor controllers in old battery box area.
This was a few days before the TBNC war last weekend. Still missing a lot of shit.
Here's a clearer shot of the sling mount. Steel pin goes through that back plate. Designed for Magpul MS2 type setups.
And the flat top. Photobomb by HPA rig in the background.
Top profile.
This was the night before the game at probably 0200 right after I got off work, bolted everything up and test fired.
And there we have it. Still missing sharktooth delete panels, a left side rail for flashlights, magwell fences, a motor cover (which I think I will make clear on this one), and a flash hider.
Also shown is the 2S 2600mAh battery. Sony VTC5 cell. The energy density of these newer high current/power tool hybrid cells is wild! They are up to 3000mAh now. Gonna be building one of these exact buffer tube packs in 3000mAh for someone shortly.
This pack did eliminate the ESC brownouts and I would have probably been OK to just run it with stock ESCs. But, I would still ideally use more battery on guns with electronics if battery packaging weren't a concern. That's the unfortunate thing with the hybrid cylindricals, current density/IR has stayed constant. This pack is equivalent to a decent 1500mAh lipo. The burst ratings are higher on the power tool cells than for lipos, but IR is IR. I wish there were something like a 1200mAh, 4mOhm 18650 cell rated for 60A continuous. As it stands, these 1P 18650 hybrid packs, and ~1500mAh lipos that also fit in buffer tubes, have about the same IR and that's about as much punch as you can fit in there.
Oh, one other thing. These motors serve their purpose, but they are crap. The bearings are noisy (and nonreplaceable like many inrunners). I had to get a defective motor with huge endplay replaced to get 2 good ones. DO NOT buy the GWS GWBLMxxxx series motors.
If I work with this type of setup in the future (will likely brushless my Riot Cage and drop it into this blaster for its review run if it has the 20mm mount sockets, and later on if it doesn't and needs to get milled first) I will be trying out Turnigy XK2040.
No WIP images of either the electronics compartment cover or this side cover, but this is what is on the left side to cover the pusher motor and the cycle control switch. Note oddly appropriate bit of print line from the PVC pipe (3" D2729 drain, "3000# min crush") that I flattened into sheet to make this.
Finally, rather than a video of shooting at a wall, how about a video of shooting fools with it? This was why I was rushing to get it ready to go. Oh. by the way, Tampa Bay Nerf Club is back, with improved props, rules and gamemodes and soon to be a permanent private field to play on. Link in the video description for all Floridians.
Will you be scratch building flywheel blasters?
ReplyDeleteWill you be scratch building flywheel blasters? if so, will you be building your own cages or using aftermarket ones?
ReplyDeleteYes.
DeleteMy own.
That's some nice looking fabrication, damn. Beats the crap out of any other auto-Stryfe I've seen as a functional field machine.
ReplyDelete