Sunday, February 9, 2020

S-Core 1.0, a somewhat-preliminary single-board blaster manager.

The Google Drive directory


This is my first crack at getting rid of Arduinos, perfboards, DRV8825 carrier boards, hand wiring and off the shelf 5V converter modules and replacing all that noise with a single PCB. Done around Aug-2019 and since then I have been running a couple of these.

This has an Atmel ATmega328P MCU, an AOZ1282 5V supply with the input filter from a recent post, an onboard tournament-lock (speed limit) trimpot, and inputs for a SPDT center-off selector, an analog pot knob, the bolt limit switch and the SPDT trigger like any other T19 and all of these contact input lines have the usual proven 1k pullup to vcc and 100k protection resistor going into the MCU pin. But the elephant in the room is the TI DRV8825 stepper driver and all its supporting componentry including a Vref trimpot, Vref testpoints, DC link cap, and current shunts. There is also an open-drain "single ender" solenoid driver that takes a DPAK (designed at the same time as the ACE LC1...) mosfet on this board - though I haven't used that for anything I would put this in yet. 2 layer 1oz copper, pretty undemanding stuff.

Gerbers

BOM (generated by Digikey)

Firmware v0.8

A Reddit thread about this

These are solid and do what they ought to. They have great thermal performance for the DRV8825 and run it cooler than Pololu boards do.

Misgivings are numerous!


  • I used a nonstandard motor drive signal connector pinout with 5 pins. Back when I designed this, I was not settled on where logic power supplies were even going to be physically located in a blaster - so these are: ground, NC (where BEC output is on old RC ESCs), throttle, tach, 5V (so far always NC on ESCs). Of course only 3 are necessary. I have settled on 3 pin with ground, tach, throttle as a pinout now after modding a few Spiders for tach with cables wired that way.
  • The ceramic reso and AOZ1282 inductor footprints are default ones too small and a massive pain to solder. The reso's alright, but the inductor is a bitch. Ok for reflow, but dumb layout for a hand-solderer. Needs a bit more area on the pad edge to heat from and it's OK. Lesson learned.
  • I discovered machined-pin headers: 2 pin connectors are secure if you just use good headers and not cheap ebay ones for the female side. The bolt limit switch can be a 2 pin.
  • There are no mounting holes since I opted to have a very fast cheap to print board bracket in the 19 instead. But for futurestuff that might be slimmer than a 19, I want either holes in the board or provision for slide-in mounting.
  • SMD electrolytics - again. Another pain in the ass. Lesson learned.
  • Looking back, my MCU decoupling was a bit meh (there's a via in one of them and the other has some trace length), but better than most commercial stuff and these never reset or glitch out.
  • The solenoid driver just needs to GTFO. There should be ONE OF a noid driver OR a stepper driver OR a third throttle channel on a given board. I actually put that in because I was expecting to use these in a HIR project where there would be a small noid for the feed gate (like in a Zeus or my old hopper loaded thing) and the DRV8825 would be running a feed roller motor. But that's probably best left off, and made external/piecemeal for such a very special rare case. It's deadweight in a 19 and in a solenoid-driven blaster, the 8825 is deadweight. Those apps need separate boards. Plus I think I would use a halfbridge in a dedicated hi-po solenoid drive board today instead of a single-ended powerstage so that decay mode could be fast if that is worth anything
  • Layout could be tighter.
  • Those Bourns TC33 trimpots. These are same the tiny, tiny, stamped sheetmetal/ceramic, things found on DRV8825 carrier/Stepstick boards for Vref. Small footprint, not hard to solder. Also found on Narfduino Brushless Micro (etc.). The one for Vref on the 8825 is fine - you use it once in a blue moon to set motor current just like the one on a stepstick carrier. But the one for the tournament lock - That really needs to be something larger, beefier, easier to turn with a small screwdriver and yet difficult to get accidentally turned from things brushing against it, accurate, and hard to break or contaminate. Like... a regular blue through hole Bourns trimpot.

Time for a brief component shopping session and a re-lay.

Other than that, MCU pinouts and hardware all good, works great, no problems. Running one of these in this:


"Crystal Patriot": Yoyi translucent red PETG, Yoyi clear (translucent whitish in a thick part) PETG, Yoyi translucent orange PETG for the flash hider and auxiliary controls, Inland/Esun trans. cobalt blue PETG.


Of course to go along with this there is a selector and an analog knob added as controls on the blaster. I put up the modded grip base and the knobs as well as the S-Core board bracket (glue the board in). This is the potentiometer and the rotary switch used (wire as SPDT center off). Can't beef with either, particularly the rotary switch, which gives a very solid and satisfyingly clicky selector.

No comments:

Post a Comment