0
\$\begingroup\$

We are using this photo MOS relay in one of our designs. This design is in production and we test every relay in QC. Recently I came across a failed unit with one relay not turning on. enter image description here

I checked the input LED side and make sure that ~10mA of current is flowing through R49. According to datasheet (https://toshiba.semicon-storage.com/info/TLP3556A_datasheet_en_20200325.pdf?did=60315&prodName=TLP3556A) 0.3-4mA is the trigger current. So 10mA should be enough to trigger the output. However the LED Vfw is slightly less as I can see, about 1V. (should be 1.5V-1.8V according to data sheet, @ 10mA)

I then measured U9 pin 4 and it was ~24Vac and pin 3 was ~13Vac (floating). I turned on Q1 using a test command and still pin 3 is ~13Vac. If I connect a 1k load from COM to SW_OUTPUT, this voltage is 0V, which means the MOS relay is off.

I removed the U9 and tested it separately by biasing the input through a 220R resistor to 3V3 and I can see the MOS relay turns on. I have verified the gate drive from uP (FROM uP) is steady 3V3 when turned on. I cant think of any reason why it fails on the board. Any ideas?

Thanks,

\$\endgroup\$
5
  • \$\begingroup\$ You’ve not determined the root cause, so why try to guess? Have you tried another device? From what you’ve mentioned so far the fault could be elsewhere. \$\endgroup\$
    – Kartman
    Commented Sep 29, 2022 at 2:47
  • \$\begingroup\$ Thanks Kartman, I am looking at the MOS switch because here is where the signal breaks. I have tried swapping for a one I tried working external to the board. It doesn't seem like SW_OUTPUT is shorted to anything. May be I will probe for resistance from SW_OUTPUT to other nets to be 100% that there are no short ccts. I will order brand new device and try it on. Honestly the cct is rather simple and I have run out of ideas for any failure mechanisms external to the device. May be someone can shine a light here. Thanks again, \$\endgroup\$
    – Kaush
    Commented Sep 30, 2022 at 11:45
  • \$\begingroup\$ Yes, the circuit is simple, but you’ve not tested for the obvious. First test is for opens and shorts along with a good visual. You’d be amazed at how many faults can be detected this way. \$\endgroup\$
    – Kartman
    Commented Sep 30, 2022 at 11:59
  • \$\begingroup\$ Did you verify the value of R49? Everything makes sense if the value is much larger. \$\endgroup\$
    – Jens
    Commented Oct 1, 2022 at 0:30
  • \$\begingroup\$ Hi Jens, yes I have verified that. I also thought this could be a classic case of manufacturing failure :) But no luck. \$\endgroup\$
    – Kaush
    Commented Oct 1, 2022 at 23:04

0

Reset to default