Skip to content

Board Revision Notes

Jump to bottom
Jay Kominek edited this page Dec 18, 2020 · 26 revisions

ADC Board

Rev0

  • Misunderstood JTAG header and omitted the VTRef.
    • Workaround is an air wire.
    • Connected +3.3V to JTAG pins 1 & 2 in commit 27f18846
  • Omitted thermal relief on a lot of SMT pads, making them a PITA to solder.
    • Most (all?) fixed in 8e7716eb
  • Should include a power indicator LED.
    • Added in 8e7716eb
  • Relative connections of USB connector, FB1, and the USB ESD protection chip is probably a bad plan.
    • ESD protection moved inbetween USB connector and FB1 in 8e7716eb.
  • Footprint for the quad op amps was changed from SOIC to TSSOP in 157656b.
  • The Z/Y pins between the RS485 transceiver and the RJ45 jack were swapped.
    • Corrected in 0fce4a6
  • We were using alternate pins for USART1, because USB_OTG_FS_VBUS wants one of the standard pins. Unfortunately the bootloader doesn't seem to like the alternates. So we've moved the RS485 transceiver to USART3 in c76639c

★ Rev0 is partially brought up. Power regulation is fine, and the microcontroller is programmable, and can drive the LEDs. Analog design seems OK. Still need to actually test the USART/RS485 arrangement, but confidence is high.

LED Power Board

Rev0

✗✗✗ SHOULD NOT BE USED

✗✗✗ L78L05 has the wrong pinout in the schematic, installation as depicted causes current to flow "backwards" through the regulator, applying ~13V to the MCP1501 voltage reference, causing immediate catastrophic failure of the part.

✗✗✗ I used a SOIC-8 footprint for the opamps, but the intended part (MCP6V51) is MSOP-8. Boards must be respun.

  • Meant to switch the MOSFET footprint to TO-220, to use more common FETs.

    • Fixed in 5f9b64f4

  • J9 footprint is much smaller than the EB147A-02-D screw terminals I bought. C10 should be moved further away.

    • Moved in 5f9b64f4

  • Diodes D1-D8 are too close to J1-8 when screw terminals are used. A little bit of scooting towards C1-8 is appropriate.

    • Moved in 5f9b64f4

  • Add a 1Mohm resistor across +48V/GND.

    • Added in 5f9b64f4

  • Op amp footprint is MSOP-8 not SOIC-8.

    • Fixed in 5f9b64f4

  • Meant to put on some sort of control lines which could turn off the LEDs, for automatic calibration purposes.

    • df3742ef added an I2C I/O expander for controlling the voltage reference, and monitoring the status of the LED strings.

  • Forgot to round the corners of the board slightly, like the other two boards.

    • Fixed in df3742ef

  • Needs some general polarity protection.

    • Diodes added in df3742ef

  • ? Inductors on the +48V incoming power, to help with current in-rush?

Link Adapter

Rev0

  • Ground pins of "To ADC" port are isolated from the ground plane.
    • Workaround is an air wire.
    • Fixed in commit e903d9f

★★★ With the air wire in place, Rev0 works. The RS485 link has been tested to 3Mbps on a (very short) ethernet cable. The GPIO, pull up/down array, and MOSFET for detecting power over the link haven't been checked, but they're not critical. Absent further necessary/very useful changes to the design, I don't expect to designate and/or fabricate a Rev1.

  • Nice to have? P-Channel FET controlling the outgoing power on the To ADC port, tied to one of the GPIO lines, which is reprogrammed for PWREN? Not entirely sure about the PWREN behavior on the FT230. Basically want to ensure we don't send power to the ADC board until the USB upstream has agreed to give us 500mA.
    • Next step with that would be to add high-side current measurement, in such a fashion that the attached PC could get the information. (Maybe also just as a voltage which can be picked up on an oscilloscope.)
Clone this wiki locally