search

sinara-hw / Kasli

Kasli is a powerful FPGA carrier, capable of controlling 12 Eurocard extension modules.
Other
16 stars 1 forks source link

Kasli 2.0 12V input short circuit #87

Closed sbourdeauducq closed 1 year ago

sbourdeauducq commented 3 years ago

I have two three four Kasli 2.0 from TS and Creotech that died in exactly the same way: after a few weeks, a short circuit developed on the 12V input. No burned components or other visible damage. Both failures occurred while boards were inside metallic chassis (ESD is very unlikely).

I'm opening this issue here to collect information about this problem.

kaolpr commented 3 years ago

Have you tried disconnecting the fan? As problems with these have already been reported (#76) maybe it's generating not only vibrations...

sbourdeauducq commented 3 years ago

This is on Technosystem boards. They came without fan, and we fitted them with another fan model that we used many times before on Kasli 1 without such an issue. Also, this is a hard short that can be measured with a DMM, without power applied to the board. The MOSFET that controls the fan cannot have voltage above threshold on its gate.

gkasprow commented 3 years ago

This could be related to the input protection diode. If you use a power supply with an output voltage that exceeds 12V significantly, you may heat the D1. It may burn after some time and get shorted. Just remove D1 and see what happenes.

sbourdeauducq commented 3 years ago

The power supply is the same AKM65US12 that we have been using before on many Kasli 1.1 without problems, and every time I measured their output it was well below the 13.3V breakdown voltage of the SMBJ12A used on all Kaslis. So this scenario does not sound likely.

sbourdeauducq commented 3 years ago

One of the boards was actually from Creotech.

Just remove D1 and see what happenes.

Still the same.

sbourdeauducq commented 3 years ago

Have you tried disconnecting the fan?

For good measure, I also tried that on the affected Creotech board - still the same, as expected.

maciejprzybysz commented 3 years ago

We have just checked Kasli v2.0 sent to Technosystem. We found that the cause of the short circuit is C3 capacitor on the TLV62150 DCDC converter input. In the project 10uF / 6.3V capacitor was used. This converter is powered by 12V, so the capacitor should be rated at least 16V. This problem affects all of the Kasli boards manufactured in accordance with v2.0rc2 and v2.0.1 release documentation.

C3_kasli

We removed the capacitor from the board. It looks like the short circuit is gone and the board is working fine. Unfortunately, there are no suitable 10uF capacitors in 0402 case on the market, so a design change is required.

dtcallcock commented 3 years ago

Would stacking 2x 4.7uF on top of each other be a reasonable way of patching existing boards?

sbourdeauducq commented 3 years ago

Unfortunately, there are no suitable 10uF capacitors in 0402 case on the market

What makes a capacitor suitable here besides the obvious - X5R dielectric?

Would stacking 2x 4.7uF on top of each other be a reasonable way of patching existing boards?

Or maybe try to fit a 0603 if the footprint happens to be large enough, or flipped 90 degrees and with solder bridges.

sbourdeauducq commented 3 years ago

Patch with 0805 capacitor (Murata GRM21BR61C106KE15K) and solder bridges image

Board works again!

akaminska commented 3 years ago

Thank you all very much for your help with understanding this issue! Attached is a description of a recommended patch Kasli_v2_patch_210122.pdf