• State
  • Locked Locked
  • Replies 19 replies
  • Subscribers 31 subscribers
  • Views 1890 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

Ethernet of TM4C1294KCPDT dosen't work under minus30℃ ,why?

Jiaming Sun
Expert 1300 points

dear all

We test the TM4C1294 board ，at beginning the interface of ethernet works well under normal temperature, but when the temperature goes under -30℃，the ethernet fails. I want to know why , the datasheet shows that TM4C1294 can work under -40℃。

Hope to get reply soon.

  • 0
    TI__Guru**** 244400 points
    Hello Jiaming,

    Technically that should not happen. To help you better, I would need the following information

    1. What is the RBIAS resistor being used? Component name or link to the manufacturer part number is what we need to see for tolerances
    2. What is the Crystal being used? Again component name or link to the manufacturer part number.
    3. Layout information of the board.
    4. Signal quality at room temperature and at low temperature.
  • 0 in reply to Amit Ashara
    Expert 1300 points

    We did'nt use RBIAS resistor. We use liquid nitrogen  to drop the temperature of the  crystal , it works well .But when we spray the liquid nitrogen onto TM4C1294 , we see the wave of crystal disappear on the oscillograph ,after a while ,the wave appears again.So we think the crystal works well under -30℃, but TM4C1294 can't work well under that temperature.The signal quality at room temperatrue and at low temperature didn't  have obvious difference.

  • 0 in reply to Jiaming Sun
    TI__Guru**** 244400 points

    Hello Jiaming

    Jiaming Sun said:
    We did'nt use RBIAS resistor

    If you do not have RBIAS resistor installed for the on-chip Ethernet PHY, then there lies the issue. The RBIAS is a "MUST" for using the Ethernet PHY

  • 0 in reply to Amit Ashara
    Expert 1300 points

    sorry,I misunderstood your meaning.I thought that RBIAS means resistor for crystal .The RBIAS resistor for TM4C1294 is 4.87KΩ ,1%  accuracy.

  • 0 in reply to Jiaming Sun
    TI__Guru**** 244400 points
    Hello Jiaming,

    That clarifies RBIAS. Now you mentioned earlier when the cold spray is applied, the wave of crystal disappears. Were you monitoring the crystal output? Could it be that the crystal stops oscillating below -30C? Can you please list the full part number for the crystal being used.

    As another experiment, you can use a singled ended external clock generator to supply the clock instead of the crystal, apply the freeze spray and see if the device still stops or not?
  • 0 in reply to Amit Ashara
    Expert 1300 points
    I do the experiment you mentioned, I use oscillator ,connect it to osc0,but TM4C1294 dosn't work. Why?did you use oscillator for TM4C1294,dose it work?
  • 0 in reply to Jiaming Sun
    Guru 27665 points

    Jiaming Sun said:
    But when we spray the liquid nitrogen onto TM4C1294

    I can understand using this as a quick and dirty test but

    • Blasting an IC with a 77K liquid is quite a shock (more than a 200C differential)
    • It's uncontrolled, you don't know what temperature you are getting
    • There's likely a high differential temperature across the IC

    On the 'plus' side, it will likely show up bad solder joints. And bad solder joints or similar may be what you are highlighting, especially if this is only a single board.

    Robert

  • 0 in reply to Jiaming Sun
    TI__Guru**** 244400 points
    Hello Jiaming,

    I mentioned an external single ended clock generator, so that there is no other component except TM4C129x. Also I checked on the device characterization and ethernet PHY tests have been performed at -40C across devices and there has been not a single failure seen.

    I agree with Robert. May be the device temperature is lower than -40C.
  • 0 in reply to Amit Ashara
    Expert 1300 points

    We did the test in an  environmental testing chamber  at first ,so we can get the accurate number of temperature ,-30℃ under which ethernet dosn't work. We use ethernet to communicate with PC outside  environmental testing chamber,when temperature goes below -30℃,communication dropped.

    After that ,we use liquid nitrogen for convenience to do our experiment because we need to weld some wire on the board ,if we use environmental testing chamber ,it cost time .We do the test in order to find which part caused the failure.

  • 0 in reply to Amit Ashara
    Expert 1300 points
    yes,we use an external single ended clock generator,connect this clock signal (3.3V) to OSC0 of TM4C1294,but very stange ,TM4C1294 seems dosn't work ,do we need to set some registers to get the chip work?
  • 0 in reply to Jiaming Sun
    TI__Guru**** 244400 points
    Hello Jiaming

    OK. Now a couple of things to check

    1. Check the stability of the 3.3V supply and 1.2V core voltage when the freeze is applied. Is there a dip/power loss
    2. Is each board having the same issue. If not then, replace a working part on the failing board and then check.
  • 0 in reply to Amit Ashara
    Guru 27665 points
    As cb1 might put it "Be Ware the single board anomaly"

    Jiaming, that seems a reasonable use of local cooling but I'd still be a little wary of thermal shock.

    Assuming thermal shock is not an issue you may still be seeing an issue with the build of the PCB. Poor solder joint, cracked trace, cracked component (thermal shock again), contamination, etc.... Any of these might cause you to have a marginal signal that only fails under certain conditions.

    I have seen capacitors crack before in such a way that they were difficult to find although some exploded during testing.

    Robert
  • 0 in reply to Robert Adsett
    TI__Guru**** 244400 points
    Hello Robert,

    I too agree with you and cb1. If it is now a single board or multi board anomaly needs to be understood,
  • 0 in reply to Robert Adsett
    Expert 1300 points

    On this board ,there is not only TM4C1294,but also other TI mcu ,such as TMS570xxx,they work well under -30℃.So I think solder joint,cracked trace,and so on are not the reason why TM4C1294 fails.

  • 0 in reply to Jiaming Sun
    Guru 27665 points
    I would not take the fact that other processors do not show an issue as good evidence that there is no issue with solder, cracked traces etc... Those often show up in single locations. Either because of process issues, design issues or handling.

    For instance those cracked caps I mentioned above showed up on a single location on the board. The root cause was design/handling.

    Robert

    You do need to check whether this happens on multiple boards.
  • 0 in reply to Jiaming Sun
    TI__Guru**** 244400 points
    Hello Jiaming,

    I think based on the evidence provided so far, we would like to know if this issue is device endemic or board dependent? The experiment of testing multiple boards and device swap would help us follow up on the next course of action.
  • 0 in reply to Amit Ashara
    Expert 1300 points

    We do the test that downloading very simple porgram such as flipping the IO ,TM4C1294 could work well under -30℃.If we add ethernet code into program ,the chip failed under --30℃.From this phenomenon ,could you give some advice?

  • 0 in reply to Jiaming Sun
    Guru 27665 points
    Test on multiple boards

    Robert
  • 0 in reply to Jiaming Sun
    TI__Guru**** 244400 points
    Hello Jiaming,

    A single device test is not conclusive. As everyone else on the forum is suggesting the same, please try multiple boards.