6
\$\begingroup\$

I'm making a board with MCP2515T-I/ML due to space constraints and it won't start up.

(datasheet) https://ww1.microchip.com/downloads/en/DeviceDoc/MCP2515-Stand-Alone-CAN-Controller-with-SPI-20001801J.pdf

This is the pcb layout and schematic. pcb layout

The top layer copper pour is 3.3V and the second layer is a solid ground plane.

schematic

I have checked for shorts, there are no shorts. I have checked the MISO, MOSI, CLK and CS lines and they all behave appropriately excluding MISO. It seems not only does the MCP2515 not want to turn on but the 8MHz external clock is not oscillating. I checked using a tiny SA and a H field probe (which correctly detects oscillations of other crystals).

I have conducted both manual hardware reset and SPI reset on the device and attempted to communicate with it, it still doesn't communicate.

After replacing the MCP2515T-I/ML multiple times on different boards I'm beginning to wonder if the issue isn't only that the crystal never starts up and the MCP2515 remains in "oscillator startup timer reset mode".

The crystal is ABM3B-8.000MHZ-10-D-1-G-T.

(datasheet) https://4donline.ihs.com/images/VipMasterIC/IC/ABRC/ABRC-S-A0001211322/ABRC-S-A0001211322-1.pdf?hkey=6D3A4C79FDBF58556ACFDE234799DDF0

The crystal load capacitor is 10p and I've replaced the crystal as well, no dice. I don't currently have 0402 capacitors to vary the load capacitance but am wondering if the 3.3V top layer copper pour could be providing so much parasitic capacitance that the board needs to be redesigned. According to C1,2 = 2(CL-Cstray), I should have used C1,2 = 14p if Cstray was 3p. Supposing Cstray is much higher I may not have needed to use any load capacitors. In desperation I've removed the load capacitors one at a time and attempted to see if the crystal would oscillate, in case the parasitics were in excess. That didn't work either.

I don't feel comfortable sending off a redesign until I know exactly what went wrong here, since it could be multiple things. I'm aware I should have had a keep out area and guard ring in the top layer around the crystal. Would that be the only issue though?

If possible I'd like to measure the actual load capacitance but my MM doesn't seem up to the job. The capacitance value measured across one load capacitor is either hovering around 12p or 2n.

Any ideas?

\$\endgroup\$
0

3 Answers 3

8
\$\begingroup\$

In line with SpehroPefhany's answer, if you want to try connecting the ground pad you can do it as follows.

  1. Using a small razor knife, make four cuts (as shown) to isolate the plane area under the chip from the 3.3V top side plane.

  2. Scrape off the solder mask in the area shown, especially on the ground via.

  3. Apply some solder paste to the newly exposed area (including the via)

  4. Resolder the part. The ground pad should now be connected.

enter image description here

\$\endgroup\$
3
  • \$\begingroup\$ This is a good idea. I did it and made sure VCC and ground were not connected and that the solder mask scraped area was connected to ground with solder. The results were... zero changes. I tried it twice. \$\endgroup\$
    – Tony
    May 10, 2023 at 4:18
  • 2
    \$\begingroup\$ Actually, I didn't notice that the vcc pin is fed by the inner pour that runs under the pad, the same one I disconnected from VCC when making a make shift exposed pad. The modification needs one more cut to the left of the middle pin on the right or else a short will be created. I was able to reroute power from another vcc pin to this pin and that solved the problem, the oscillator is oscillating. It was the exposed pad! \$\endgroup\$
    – Tony
    May 10, 2023 at 6:33
  • \$\begingroup\$ Thank you user. \$\endgroup\$
    – Tony
    May 10, 2023 at 6:34
7
\$\begingroup\$

You're supposed to have this footprint:

enter image description here

But you've omitted the pad for the EP. Can't say for sure that's the cause of your issue, but it doesn't help at all.

An RC + diode is recommended on the /RESET pin. You've just got a resistor. If you do an SPI reset it may not be necessary to have the more complex circuit.

Another possibility is that you've got the wrong value for the xtal load capacitors by a huge margin (eg. 20nF rather than 20pF). Or an open on one of the pins, it's hard to inspect leadless packages. If the package was soldered by someone who doesn't do this sort of thing regularly with proper tools, my guess would be on that as the issue.

I don't think minutia with the layout or capacitance values is the problem, you've got something else.

\$\endgroup\$
9
  • 3
    \$\begingroup\$ You can usually check the crystal and reset pins with a DMM in diode test mode (red lead to Vss, black to pin) . Most likely they'll show as a diode to Vss with power off. It could be an assembly error, but less likely with a contract assembler. EP is electrical as well as thermal but it's not likely that would stop the oscillator. \$\endgroup\$ May 10, 2023 at 0:45
  • 3
    \$\begingroup\$ The data sheet does say that the exposed pad is to be connected to ground. Page 4 Table 1-1 . Last entry in the table. \$\endgroup\$
    – RussellH
    May 10, 2023 at 0:49
  • 2
    \$\begingroup\$ @Tony - Regarding the EP: Even connecting pin 8 to VSS does not mean the EP can be left floating. The datasheet p4 says for the EP "connect to VSS"; no other options (e.g. floating) are given by the manufacturer. I remember we've had a couple of questions - 1 recently, 1 long ago, where significant misbehaviour was caused by a floating EP. The one long ago that I remember was confirmed to be caused by the floating EP by soldering a very thin (32AWG?) wire to the EP. The IC was then soldered back to the original PCB (no EP footprint) and the wire connected to VSS. Then it worked correctly. \$\endgroup\$
    – SamGibson
    May 10, 2023 at 0:50
  • 1
    \$\begingroup\$ Not continuity, that won't work, diode mode. In diode mode most multimeters will read open for one polarity and something like 600 counts for the other. \$\endgroup\$ May 10, 2023 at 0:57
  • 3
    \$\begingroup\$ Thank you Spehro it was the exposed pad indeed. \$\endgroup\$
    – Tony
    May 10, 2023 at 6:34
3
\$\begingroup\$

Since you are primarily suspecting the oscillator, you can remove the crystal and directly feed a 50% duty cycle clock (0 - 3.3V level) directly on OSC1 pin. Leave OSC2 open with nothing connected to it. Then, put the reset pin low manually and release it. If the device starts responding, it could indicate clock issue.

Any ideas?

The +3.3V supply also needs to be probed to see if there are any dips in the supply causing chip reset.

\$\endgroup\$
1
  • \$\begingroup\$ That's a good idea. If I have clock problems in the future I'll try this method. \$\endgroup\$
    – Tony
    May 10, 2023 at 6:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.