• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 82 subscribers
  • Views 1023 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.

TMS320F280041: ADC input leakage / sampling current question

Mark Pearson
Genius 12705 points
Part Number: TMS320F280041
Other Parts Discussed in Thread: TMS320F28054

Currently, we use the TMS320F28054FPNT in our design, and we noticed that some PCB’s had some vibration/oscillation in the motor torque.

This small vibration was found to be caused by the control loop but originated in the ADC measurements on the phase voltages.

Each phase is sampled through a voltage divider from the Bus Voltage via a 150k and 4.70k 1% resistor divider string, with a 100nF capacitor, 5%, directly on the ADC input.

We changed the 1% resistors to 0.1% types and this almost solved the issue.

The used ADC inputs are : ADCINA7, ADCINB3, ADCINB0 (Used because we don't want the PGA in the design)

We noticed that one of the ADC inputs, ADCINB0, had a different leakage current then the other 2. (also on other boards)

The other 2 where exactly the same, (I measured 0.00xxx mV between these other two inputs) but +/-2 mV from ADCINB0 to either of the other 2.

Yes, in the datasheet is stated that there can be +/- 2uA leakage current on these pins. 

I would like to know why is there a difference.

We plan to use the TMS320F280041CPZS on our next design, What re the leakage currents on the ADC input pins for the F280041?

Can I assume the same input specifications on the ADC input pins and use the same resistor-divider structure / leakage currents?

Thanks!

  • 0
    TI__Guru 60865 points

    Hi Mark,

    Is the ADC actively sampling and is there a difference in sample rate between the channels?  Is the delta in voltage the same even if you permute the order that the channels are sampling?

    Overall, driving an ADC input with a high impedance (you impedance in this case is 4.7k || 150k) can be challenging.  Since you have a large capacitor on the pin, it seems like you are using a "charge-sharing" input design, which can be sensitive to sampling too fast (this creates an effective input or output current, similar to the leakage current you are already looking at).  You might want to have a look at the following app. notes, whose simulation and analysis methods should also be able to be applied to F28054:

    https://www.ti.com/lit/an/spracv0/spracv0.pdf

    https://www.ti.com/lit/an/spract6/spract6.pdf

  • 0 in reply to Devin Cottier
    TI__Genius 12705 points

    Hi Devin,

    My initial response to the customer (in regards to the high input impedance) was as follows:

    Is there any buffer amplifier between the resistor string (150k & 4.7k) and the ADC input? The reason I ask is that the impedance that the string presents to the ADC input is ~4.557k. With a 2uA leakage current, this translates to ~9.11mV.  (It’s possible to have this, but it doesn’t mean you will – I’m just crunching some numbers). Customer answer: no. There is no amplifier/buffer in the design.

    Do you perform any calibration on the TMS320F28054 ADC inputs? The reason I ask is that the VOS error with a single-calibration (Vs. a periodic self-calibration) can be as high as +/-20LSB (or 16mV), and only +/-4LSB with a periodic self-cal. Customer answer: Calibration is performed only once - at power-up.

    Regarding the sampling questions: I don't exactly know, but I infer that the sampling rate is the same. Order of sampling - also unknown.

    I will see what I can find out.

    In the mean time - can you tell me about the ADC input leakage on the F28041 and let me know if the ADC input structures are the same or different between the F28054 and F28041?

  • 0 in reply to Mark Pearson
    TI__Guru 60865 points

    Hi Mark,

    There isn't any difference between these inputs on the F28054 device as far as I can tell.  Is this only occurring on a single system/unit, or on multiple?  ESD or EOS damage can result in increased leakage, and we would actually consider leakage higher than other pins and especially leakage higher than the DS limits to be a strong indicator that ESD or EOS has occurred on that pin.

    F28004x family devices use a different ADC and are fabricated in a different technology vs. F2805x family devices, so these will be significantly different.  One thing to note is that the pin leakage for F28004x devices is 0.1uA (vs 2uA on F2805x devices), which is a significant improvement the customer can look forward too when migrating.

    In addition to your comment about the offset self-calibration (which applies globally to the whole ADC), the customer should also be aware that there is a "Channel-to-channel offset variation"  error specification.  This error will result in ADC channels having a little bit of mismatch even if exactly the same voltage is applied.  In this case, it seems like you are seeing the error directly on the pin via DMM measurement through? The ch-to-ch offset error would only be expected in the actual conversion results. 

  • 0 in reply to Devin Cottier
    TI__Genius 12705 points

    Customer feedback on the sampling speed question:

    "We considered the sampling speed and sequence. All 3 channels are sampled at the same speed and we changed the value of the signal that is sampled before B0 to see if we have a crosstalk problem, but we did not see any difference. This left the leakage current as possible cause."  

  • 0 in reply to Mark Pearson
    TI__Guru 60865 points

    Hi Mark,

    I think mostly what is left to do is for them to investigate (if they haven't already) if this occurs on all systems or just one system (indicating the possibility of ESD or EOS).  Otherwise, like you previously communicated, this appears to be well within the DS leakage current specification.   

  • 0 in reply to Devin Cottier
    TI__Genius 12705 points

    Devin,

    In the original post - it is noted that the customer does see the issue "(also on other boards)".

    So, in summary,

    1) With respect to the question - is the leakage on each of the F28054 inputs, ADCINA7, ADCINB3, ADCINB0, the same?, the answer is 'yes' (with process variation) and these inputs have the same input structures.

    2) The ADC Inputs of the F28054 Vs. the F28041 are indeed different and the F28041 has much lower leakage current, 0.1uA (vs 2uA on F2805x devices).

    If the summary is accurate, then I think we can close this e2e post.

    Mark