PAY-SSM v3 001 Manufacturing

[vogeldylan]

First of the v3 PAY SSM boards, pretty exciting. This issue will track manufacturing notes and rework/repair checklists while the board is being assembled.

Missing Parts

• 1 FDC637AN N-MOS

TOP SIDE:

• [x] C9 and C10 on CAN are shorted
• [x] Pins 7 & 8 are shorted on U9 (PEX)
• [x] U10 (PHT2) has two pins shorted to vias
• [x] U1 (DAC) has pins 1 and 2 shorted
• [x] U13 also has a via short on the right side
• [x] U18 has a via short
• [x] Pin 19 and 20 on U17 are shorted
• [x] Pin 18 and 19 on motor 2 IC are shorted
• [x] Lots of solder balls around humidity sensor footprint

[vogeldylan]

Updates from Bruno's inspection, and Dylan's rework

• C9 and C10 are a defect. They have been reworked once.
• Pin 7 & 8 on U9 are part of the same net, not a defect.
• U10 pins are on the same net, not a defect.
• U1 (DAC) pins are a defect. They have been reworked once.
• U13 pins are on the same net, not a defect.
• U18 is not a defect, same net.
• Pin 19 and 20 on U17 is not a defect, same net
• Pin 18 & 19 on motor 2 IC is not a defect, same net
• Solder balls on footprint is a process indicator. Balls have been wiped off by hand and inspected.

TODO:

• [x] Inspect C9 and C10 rework
• [x] Inspect U1 (DAC) pin 1 & 2 rework

[vogeldylan]

• [x] Pads around U1 (DAC) need to have caps attached
• [x] Big solder glob on S3
• [x] C63 100pF on DAC was not added

[vogeldylan]

Jaden and I finished up everything but the connectors and the pressure sensor on the bottom of SSM. Still missing N-MOS

• [x] U3 looks shorted on pins 11 and 12 (buffer)
• [x] U19 has some solder paste which hasn't re-flown on pin 13
• [x] C45 has un-flown solder paste, other side is very poorly connected
• [x] R17 has un-flowed solder paste
• [x] Take R54 (resistor array) off and redo it
• [x] R43 shorts power and ground on pin 2
• [x] Take U19 off and redo - also unshort the vias beside 17 and 18 while at it
• [x] Check all of the resistor arrays (4x) for unflowed solder paste spare R41 (hit em again)
• [x] Spike on pins 2 & 3 of U21
• [x] Recheck R35
• [x] Reflow pins 3, 4, 5, 8 & 9 on U3

[brunofalmeida]

Programming headers: PAY-SSM (right side) - notch facing down PAY-Optical (left side) - notch facing down

[brunofalmeida]

PAY-SSM 32M1:

• fuse programming worked
• program upload worked
• UART RX and TX worked
• UART headers (from 32M1's perspective):
  • top 2 - TX
  • bottom 2 - RX
• PROG <-> RUN header
  • need to be in PROG mode to upload programs
  • need to be in RUN mode for SSM_RST button and UART to work

[brunofalmeida]

PAY-Optical 32M1:

• fuse programming worked
• program upload worked
• UART RX and TX worked
• UART headers (from 32M1's perspective):
  • DI - TX
  • CLK - RX
• need to disconnect RX when programming

[vogeldylan]

• Soldered on the missing 100pf
• [x] Still missing that N-Mos, was never ordered
• [x] missing thermistors to test with
• [x] missing cheap heaters to test with

[brunofalmeida]

• PWR/RST LED lights up when RST button is pressed
  • next iteration should use an open-drain inverting hex buffer - LED will light up when the button is not pressed

[brunofalmeida]

• ADC numbers channels from 0-11, but schematic numbers thermistors from 1-12
  • future schematics should number thermistors and other ADC components from 0-11

[brunofalmeida]

Motor testing updates:

• possibly some problem because the CONFIG value is latched when exiting SLEEP, but CONFIG is tied to VINT which is only valid (high) when SLEEP is disabled - possibly a race condition
• cut CONFIG wire on MOTOR 2 and connected green wire so we can choose 3.3V (indexer mode) or 0V (phase/enable mode)
• motor seems to only work in phase/enable mode (CONFIG low) - will edit library to use that mode
• need fast SPI transactions on port expander for switching - SPI clock is currently around 125kHz (division of 64 from clock frequency of 8 MHz), should increase SPI clock speed in lib-common

[brunofalmeida]

Motor testing:

• broke the green wire on the motor 2 driver CONFIG pin, very little of pin left so difficult to solder
  • CONFIG is pulled down internally so did not connect anything to it - defaults to phase/enable mode
• implemented phase/enable mode software
  • works for motor 2 driver
  • nothing happens when connecting the motor to motor 1 driver, so it is assumed to properly be in indexer mode as per the original PCB design
  • in the future, should control CONFIG pin in software
  • in the future, should account for phase/enable mode pin setup so 32M1 vs. PEX pin assignments are balanced (e.g. AENBL and BENBL should both be on the 32M1 or both be on the PEX)
• period of motor cycling is much longer than expected (for expected ~2s, got ~12s)
  • probably because of slow SPI clock speed, need to increase

[siddMahen]

@brunofalmeida SPI clock speed is currently 1/64th F_CPU, so 125 kHz. Does this fit the approx. times?

[brunofalmeida]

For what we observed, yes

[brunofalmeida]

Motor testing:

• desoldered the CONFIG pin on the motor 1 driver, left open (default to phase/enable mode)
• fixed shorting the nFAULT pin on motor 1 driver
• tested both motors actuating the platform simultaneously - works
• whichever motor is connected to the motor 2 driver has better torque and is harder to stall
  • tried switching the motors - consistent result that it depends on the motor driver, not the motor
• hypotheses:
  • APHASE is tied together for both motor drivers, but BPHASE is separate between motor drivers - BPHASE requires 2 SPI operations on the PEX compared to 1 operation for APHASE which messes up timing and changes the phase offset
  • SPI too slow - try increasing clock speed
  • eliminate/cache PEX read result and just do write operation when toggling output pins

Updates:

• found a 10ms delay on every PEX read and write (was using old version of library for faster prototyping without having to change it to use the new interface)
• motors much faster after removing PEX delays, but were much easier to stall (required less force)
• increased motor cycling period from 20ms to 200ms, much better torque and more difficult to stall
• probed motor driver outputs - lots of noise, want to have more consistent current for motor inductors to improve torque
• still going into constant current mode on power supply
• might redesign next SSM with dual H-bridge

[vogeldylan]

Soldered on the last remaining MOSFET. Have some prototype heaters which need to be tested with. Closing this issue.

Still testing to be done. Can be re-opened if new issues arise.