• State
  • Locked Locked
  • Replies 3 replies
  • Subscribers 32 subscribers
  • Views 148 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DS250DF410: Heating issue

Bence Reucsán
Prodigy 25 points
Part Number: DS250DF410

Hi,

We are using the DS250DF410 in multiple designs, and we've noticed a strange heating issue with some of the chips. For example, on one of our PCB, there are 4 instances of the DS250DF410. We manufactured 3 pieces of this PCB, so we have a total number of 12 DS250DF410.

Out of these only 9 were working correctly, the other 3 were not working properly. (One PCB was 100% fine, the other two PCB had 1 or 2 faulty chips.) (Abnormally hot die temperature, not responding on I2C bus, pulling down constantly the I2C bus.)  

We thought it was a manufacturing error, so we've replaced the 3 chips with new ones. The new chips are working fine indeed, but now the other chips (which were working fine) seems to be faulty. (The same issue: heating up, I2C error.) These chips were not heated during the repairment process, so it's definitely not soldering issue.

Do you have any suggestion what can cause the chip failure? Could this be an ESD issue? If so, do you have any recommendation?

Here I attach photo taken with thermal camera. You can see that the two middle DS250DF410 are heating up to about 70°C, while the leftmost and rightmost one are about 30-40 °C. (This photo was taken without any network traffic, and the default state of the retimers. Also usually we use heatsink, but we removed it now to see the heating issue more clearly.)

Thank you in advance.
Best regards,
Bence

  • 0
    TI__Mastermind 20782 points

    Hi Bence,

    This is not an issue I have observed before.  I have a couple quick thoughts on issues to look into.  However, in general, this does not match behavior we expect from the device.

    • Is the power supply rail voltage stable and matching your expectation?  On power cycling, do you observe any strange transients?
    • Are your RX/TX signals AC coupled?

    Beyond this, I can look over your schematic/layout to see if anything unusual stands out.  You can share this via e2e DM so that it is not in the public forum.

    Thanks,
    Drew

  • 0 in reply to Drew Miller1
    Prodigy 25 points

    Hi Drew,

    Sorry for the late reply.

    We'll check the power supply for any strange transients, though we think it's fine, because we have a board out of the three that has been working perfectly for several weeks without any problem.

    The RX and TX signals are AC coupled with 100 nF capacitors. However they are on another board, so you will not see it on the schematic. 

    Basically, there is a backplane board (I cannot share its schematic unfortunately) with QSFP connectors and multiple slots for CPU cards. (The backplane <-> CPU board connection is made with ExaMAX connectors.) 

    On the backplane there is another slot for the board that has the four retimers we are talking about. This board is actually a kind of loopback card, mapping each CPU slot to a channel of one of the QSFP channel. (The retimer board is also connected with ExaMAX connectors.)  The AC coupling is in the QSFP modules and on the CPU cards, not on the retimer board.

    I can share you the schematic and layout files of the retimer board in private. We'd be really thankful if you could have a look on them.

    Currently we are thinking that the issue is either ESD related or a soldering issue on the BGA package.

    Best regards,
    Bence

  • 0 in reply to Bence Reucsán
    TI__Mastermind 20782 points

    Hi Bence,

    Thanks for looking into power supply and for confirming AC coupling caps in HS signal chain.

    I'll review the files you sent and get back to you.  Assuming nothing is out of the ordinary with schematic/layout, I think ESD or manufacturing issue are good hypotheses for this behavior.

    Thanks,
    Drew