Due to the U.S. Thanksgiving holiday, please expect delayed responses during the week of 11/22.

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.

AMC7832: Internal temp sensor repeatability

Part Number: AMC7832
Other Parts Discussed in Thread: LMZ12003,

I have 3 of these parts down on one board in different locations.  I re-read the temp sensor values from any of the parts and I see digital values that change dramatically.  For instance min/max over 10 reads (in hex)... 0x94 - 0xAC .  This corresponds to 37-degC to 43-degC.  The chips are powered from 5V with a 3.3V IOVDD.  5V comes from LMZ12003 (DC-DC converter), 3.3V comes from a linear regulator.  The internal ADC reference is used.  I have the recommended 4.7uF on REF_CMP.  I tried changing the conversion time (from min of 00 to max of 11) but no effect.

Any thoughts?  I could see a dithering of a couple LSBs, but this is huge.  Could the act of reading cause the variation?  Any suggestions to make this more stable?

Thanks,  Tom

  • Hi Tom,

    Thank you for your query. The engineer responsible for this part is currently unavailable due to the Thanks Giving holidays. He will pick up the thread once he returns.

    Regards,
    Uttam Sahu
    Applications Engineer, Precision DACs
  • Hi Tom,

    That is strange. Do you think you share your schematic with me? What is the time period that you measure the 10 temperatures? Do you have any dynamic loads on the DAC outputs? I ask because the internal tempurature sensor is measuring the die temperature, not necessarily the ambient temperature. If you load on the DACs is changing often, you could be seeing a real die temperature change, even by a few degrees in a just a second or two.

    Thanks!
    Paul
  • Paul,

    Thanks for the note.  Strange to me to.  I have found that not only is there dither in the temperature reading, but I see dither in all the ADC readings (7 to 9 counts or so).  I have a 4.7uF cap from ref_cmp pin to fat ground plane via and I have AGND2 to fat ground plane via, but I see that I did not connect AGND2 to 4.7uF cap directly.  I will rework this to see if this makes any difference.

    All my DACs go to op-amp inputs so current draw is minimal.  My ADC lines are basically static (e.g. power supply monitoring).  My +5V input is actually 5.1V.  This might vary a little because of variable loads, but I expect it never dips below 5V (I'll check again).

    Do you have experience with these parts where you can read and re-read the ADC and see it dither on one count?

    Here are the three instances of the part on the board....  (again, thanks for your inputs)

  • Sorry, one more thing. In rework, I lifted the REF_OUT1/REF_OUT2 pins and shorted the VRANGE_A/B/C/D pins to ground on all three instances (I had my ranges wrong).
  • Hi Tom,

    A few LSB of noise can be expected, but if you look at the conversions from a time domain standpoint are you seeing a ripple at a certain frequency? Maybe 60Hz? I think your decoupling scheme for the REF_CMP pin is probably okay. Do you see any ripple when you measure your supplies?
  • Thanks again for the note.  There will certainly be some ripple on the power supplies.  The main 5V supply comes from a DC-DC converter so the ripple is up in the 10's or 100's of KHz rather than 60Hz.  I have filtered these voltages and used .1uF decoupling at the power supply pins as recommended.  Still, the comparison is done against the reference rather than the supply voltage, so I would hope there is some built-in resilience to supply variation.

    I'm going to try improving the 4.7uF reference supply cap wiring and I'll take another look at the power supplies themselves.  You are saying that a few LSB is to be expected.  I'm seeing 2-3 times that--even with temperature sensing which is an internally synthesized signal.  I'd be curious what the eval board does for this part.

    I'll hunt this.  My goal in contacting you was to see if there were some known issues or practical experiences that others have seen with this part.

  • Hi Tom,

    I understand. The level of noise you are seeing is abnormal and I have not seen this issue with the AMC7832. I think you are on the right track with investing the REF_CMP configuration. Please keep me updated if you solve the issue, and feel free to ask more questions if you have any issues.

    Thanks!
    Paul