It took some time to get so far since I posted here about a CNC servo stepper development. Chip shortage added extra time for changing the design two times…
Now I’d like to share my thoughts and appreciate any comment or improvement proposal - PCBs are not yet ordered. That’s my first lesson learned - get your parts beforehand.
Some key facts:
- Separate boards for AS5147 Encoder and Driver - space & height restriction at NEMA17 in my CNC mill
- I will reflow solder myself -> all SMD parts on one side for stencil, only SMT (plugs) on other.
- For NEMA17 with 2 x max. 2.5A phase current , low resistance, 36V Power Supply
- 2 x TI DRV8876. Preferred 8874 not available at all, DRV8876 ordered from semi-reliable source…
- 2 layer, 1 ounce board. -> 2 oz. and/or 4 layers too expensive (checked PCBWAY, Seeed)
- Mitigate heat problems by mirroring the 1.1mm power tracks on backside and stitching with 0.8mm/0.4mm vias -> how can 3.8A (spec. max) flow through a 0.45mm pad connection?
- TI PowerPAD specs followed ,but with 0.3mm vias -> 0.2mm holes drive PCB costs
- Differential 100Ohm RS422 line drivers DS34LV86/DS26LV31 -> 1.5m distance through noisy energy chain. No stock for DS26LV32, equivalent should work
- use of 8-wire, energy chain qualified CAT5 cable with crimped Molex contacts -> no room for RJ45 plugs
- ABI encoder signals used. local ATTiny85 to initialize AS5147 over SPI. -> Minimize cabling. STM8: no stock
- Bluepill clone piggy back, so STM32F103C8T (don’t use USB and 5V supply concurrently - diode missing)! -> no stock for any intended STM32 48pin MCU.
- Isolated 5V/3.3V converter -> don’t trust the Bluepill converter. And I have a big 5V supply on mill.
- Isolated interface for STEP/DIR/ENA or SPI input from GRBL controller. Made with ISO7763DBQR
- non-isolated UART interface for SimpleFOC monitor
- Inline current measurement with 2 x INA253 to compare with DRV8876 internal measurement
Let me know what you think!
Regards
Peter
P.S. Testpoints not yet on board.