Saturday, April 28, 2018

Project T19 Part 9: Serial Number One

This needs to go up front:

More after the break.

Here's where we left off, with the main drive section parts coming off the printer as I worked on the last set of models: grip frame, grip panels, trigger, grip baseplate, crank web, bolt.

Assembling and sanity checking in cyberspace with those models.

Mockup existing parts.

The drive housing incorporates the bolt pattern for a stock Vector brand perfboard available on Digikey and my answer is going to be posting a schematic and build guide for hand-wiring the Core board. I opted to avoid even thinking about having any PCBs of the DZ Core backplane made right now, as a function of the experimental nature and the aggressive motor technology and integrated control strands in the project. Thus, strong likelihood of I/O expansion and hardware feature additions - for example, bolt control based on flywheel drive speed feedback instead of scheduled delays (that's in the pipe, and would definitely be 5x5).

It is entirely possible that any PCBs would be obsolete as soon as they arrived.

It's also on my mind to design an open-source ESC style, AVR-powered, SimonK-running motor controller targeted specifically at nerf flywheel drive applications to fill the unfortunate void left by the Afro line. That project is more important than getting bogged down in PCB Core boards.

"Bolt kit" printjob: Bolt, crank web, and crank pin bushing. I opted at this point to replace the COTS nylon bushing found in the Model Pandora drivetrain with a printable to remove one more annoying little vitamin. Thinking forward, I will probably integrate the crank pin into the crank web as one part on next rev. Can't hurt to add that layer weld to the cap screw's strength (the screw, or just a steel pin Devconned in, still needs to be there structurally to make me sleep at night) and remove the ability to misplace the tiny bushing during assembly.

Drivetrain internals being mocked up.

Bolt limit switch installed. The mounting pattern allows a large range of adjustment and accommodates any subminiature microswitch with a flat, roller or simulated roller lever.

Messy Site B workbench shot in which I caught a (very recognizable) printjob in progress.

Grip frame, including the left trigger cavity cover panel. I was not lying when I mentioned that I would "[approximately] convert the Model Pandora grip frame design to CAD models". The panel gets Devconned in there, since the trigger cavity would otherwise be some very nasty bridge work and kill any hopes of keeping the wonderful trigger pivot tolerances.

The structural design of this frame is inherently a bit unnerving; but I sliced this with (IIRC) 5 perimeters, 40% hexagonal, 10 tops and 10 bottoms at 0.2mm yielding a very solid part in which most of the thinnest sections of the trigger guard, area near the microswitch cutout and frame sides at the trigger cavity (3mm thickness) are void-free. Also, bolting the grip panels on forms a closed-section structure, and this is Makeshaper PETG which is extremely ductile and extremely durable stuff, run hot with excellent fusion. Trust me, this is not going to break. A hollow PLA FDL grip would break long before my frame would.

Frame, grip panels, and trigger.

Things come together.

2 center bolts mount microswitch. Switch case holes are tapped for 6-32, a methodology from the Prometheus and Model Pandora grip setups that may be a bit unorthodox but works well.

I may have forgot to mention that T19 is a dimensional halfbreed, unfortunately. Metric SHCSes are not as available over the counter here for the development process and I am very familiar with and have all the taps for 6-32, 8-32, 10-24 and 1/4"-20 fasteners, so T19 is designed and dimensioned entirely in metric except for the use of old imperial fastener lines.

There is also a theme of nonstandard screw lengths that must be zipped to length at build time to get both optimal engagement and clearance for other parts or blind holes, and cases (trigger, mag release, sling mount) where a SHCS is called for with a long unthreaded shank length to be used as a pivot/bearing surface and thus requires cutting a long screw down.

Another new part: the grip baseplate.

Fairly self-explanatory - this mates the grip frame and drive housing. See also the continuing textural unification of aesthetics on the Production. The "DZ Linear" texture is 1mm radius circular grooves with the centerline aligned with the surface and spaced 4mm on centers.

Back to the drivetrain, a crank installation detail. 10-24 setscrew. The crank web model contains a D-shaped shaft hole and locates itself properly on the shaft when installed.

Note counterweight. This is an attempt to offset the bolt reciprocating mass and yield a smoother shooting gun. So far, even at 13.8rps*, it is fairly smooth. Not totally balanced but decent to the point the recoil force of the accelerating dart is more tangible than the bolt cycling.

* (Wait, what? 13.8rps? Stepper? Uh... Yes. Turbo mode! More on that later.)

Battery harness: The main power switch on these is Bulgin C1300AA, high inrush rated (~150A).

Accumulating parts moving toward assembly.

Electronics compartment with everything in place.

This particular unit has an off-board logic power supply (Pololu S18V20F5) wired and wrapped into a harness module, and a 4 pin connector on the backplane which supplies both raw DC bus (for the bolt drive) and 5V logic power. The details of harnesses, connector locations and types, logic power DC/DC converter modules and such are still being optimized and experimented with, but I like this style for the larger modules which I am using for the high input voltage rating to be safer against DC bus transients. If I wind up recommending a smaller and cheaper module that can be direct-plugged into the backplane more readily like the original 3S-powered Model Pandora had, then I will revert to that for sure.

A detail not captured in any image: The mag release spring perch in the Production is in the drive housing. It is part of the front flange extension where the drive housing meets the breech.

This flange extension's existence is a fortunate mistake! I goofed when designing the breech/magwell flange pattern, and a very nice print of the breech already existed by the time I realized. However, the overly-low bottom bolts and flange edge - in line with the lower surface of the grip baseplate edges - and the resulting added extension on the drive housing to fix that problem permitted integrating the spring perch into the latter. Thus, removing the grip frame baseplate to access the electronics tray does not free the mag release spring and cause it to possibly launch off into oblivion.

Now for some more images:

That pretty much covers the physical aspects of the build. Next part coming tomorrow [today] or so will cover the developmental release of Production T19 part meshes that just went up, the list of bugs in the aforementioned that needed the file, knife or Dremel during this first build, and the first revision of files that is coming very soon to fix all those. Don't start printing just yet!

Also, the software front. Core "25.0" (developmental as well, actually) is posted on the Google Drive with some other Production-related tweaks, and one of the features added is the aforementioned "turbo mode". With the advent of closed-loop flywheel drives and corresponding switchover to 14.8V DC bus comes great new speed possibilities for the stepper bolt drive as well. At the same time, substantially more than 10rps isn't generally optimal all around, and yet I do not want to add any more controls or knobs. Thus, your "manual" for Core 25.0 reads something like this...

Turbo mode increases bolt travel speed for higher cyclic rate of fire. Caution: Using turbo mode decreases bolt force. Use is not advised with low quality ammunition or under adverse weather conditions.

To enter turbo mode:

* Power the blaster up with the trigger held down.

* After the flywheel drives have emitted the final long beep, release the trigger. (The trigger-down state is trapped when enabling turbo mode to avoid firing unintended shots. Boot will not commence until the trigger is released.)

* The bolt motor will emit 2 growls during the selftest routine, indicating turbo mode is active.

* Turbo mode will remain active until the blaster is powered down.

ROF with the posted preliminary implementation firmware is about 13.8rps with turbo mode active, which so far hasn't skipped any steps even with stuffed full Workermags. Also, base feed delay and recent shot comp have separate setpoints used when turbo mode is on. This is up to the user how to tune these - you can run them a tad longer to compensate for the higher bolt speed, as I do, or you can run them snappier, since high ROF is usually close-quarters spam where perfectly consistent velocity doesn't matter.

The commutate() function has been made much speedier by removing the silly old Arduino API digitalWrite() calls that were somehow still there instead of register writes despite all the other non-abstracted chip specific things in my codebase. Thus, non-turbo mode is also slightly faster now at about 11.3rps. For that reason, plus the slightly different Production drivetrain geometry and some startup oddities with these particular Afros and motors, stock base feed delay has been increased.

No comments:

Post a Comment