ADS5272: Data read errors

Hi Joseph,

We are looking into this. Who is the new vendor you are using to procure the ADS5272?

Regards,

Rob

There is a new vendor supplying the analog signal which eventually drives the ADS5272.  Nothing to do with the part itself, other than we had no issues of this sort for ~1K parts, until we started using the new vendor's analog signal.

Hi Joseph,

Thank you for the update. Do you have a problem with all new devices you have? Or just a few?

Thanks,

Rob

At the moment, the failure is limited to just a few parts.  Peripheral parts outside of the 5272 have changed.  We want to understand from TI whether the ADS5272 can get stuck in a mode where it draws less current, and is either not communicating or sending FFF for voltage queries.  

Can the ADS5272 see an excursion at one or multiple analog inputs that may put it into this low current mode with incorrect response?

We are sending a reset signal to the 5272, but it may be prior to these excursions happening, and we are reasonably certain the 5272 is not performing properly when this happens.

We have had very good success with the part thus far, and we want to understand if and how our peripheral changes are upsetting the device performance.

Joseph,

Does your board power up sequence follow what is shown on page 8 of the data sheet? Does the excursion cause any of the ADC power rails to dip? 

Regards,

Jim

Hi Team,

On behalf of the customer, please see below comment:

"Note: I thought I put a reply in this forum to Jim S. and or Bob Reeder, but it disappeared, and there is no option to reply at the bottom of the thread now.  Can someone add the comments below and reply again.  I may have clicked the wrong button accidentally in the E2E forum.  The thread appears to be closed. 

ADS5272: Data read errors - Data converters forum - Data converters - TI E2E support forums

We believe the power supply is sequenced properly and settled.  We use the same 3.3V regulator thru separate chokes to the both analog and digital supplies to the 5272.  The reset is applied and held for .5 msec after the supply is settled.  The 5272 is not queried for seconds after the turn on.  The supply sequencing we use has worked for years and is still working, except in some cases where the indirect external source, prior to buffering, of the 5272 analog inputs have changed.

The external source of analog inputs are buffered thru opamps AD8042 and a differential AD buffer, AD8138, prior to reaching the 5272.  Both of these buffers are supplied by the same 3.3V supply as the 5272.  Historically, everything has worked properly, except with this new external source, which is buffered prior to reaching the 5272.

We believe this new external source of analog inputs has some large voltage excursions prior to the buffers, and the original source (which works fine) did not.  The 5272 will see excursions limited to the 3.3V of the buffers. We see this problem only at cold temperatures, the problem is indicated by consistent 5272 reads of greater a threshold on multiple channels of the A/D.  These reads are incorrect even after the entire product warms to hot temperature (over 60 minutes of time).  We are comfortable that the voltages are correct at the input to the A/D, once the excursions are over.   We also see a significant reduction in current (~60 mA less when the problem occurs) on the 3.3V supply, whose main user is the 5272. It is also possible that the reads are simply not working at all.  This problem corrects only at the next cycle of power.  Not clear why it does not happen every time, may be cold temperature related.

Can large excursions on the 5272 analog inputs limited to 0- 3.3V cause the 5272 to get into a mode where a reset is required, if the 5272 supplies are consistent after the reset?  The single reset we use now is after the supply is settled but is no longer active when these excursions occur.  The problem does not correct unless we cycle the supplies.  We never saw this problem with the original external supplier, but have two alternate external suppliers of the external part where this problem occurred. "

Best regards,
Gerald

Gerald,

What exactly is the amount of overload at ADC’s input? Can the customer draw input’s time domain profile and corresponding output data of ADC?

 Does a RESET fix the issue or only recycling the power supply?

Can they monitor the voltage rails to see if there is a drop at the ADC power pins when the excursion occurs?

Is it possible to use external power supply for the new external source? 

Regards,

Jim

Hi Jim,

On behalf of the customer, please see below comment:

"

1. Unfortunately our problem is only happening at cold temperature when the entire assembly is within an oven and we have no direct access under these conditions to measure anything.
2. We have strong evidence that when the issue occurs, the supply feeding the 5272 is drawing substantially less current, 60-90 mA less.  The issue only occurs with a new vendor's analog output.  Analog outputs from old and new vendors have always been buffered thru 3.3V opamps and a single ended to differential converter.
3. We have an FPGA which communicates with the 5272 which has had no issues with over 1000 units, which all used the previous vendor, whose part has less excursions on its analog output.
4. We are trying to qualify the new vendor, but the 5272 appears to be hanging up, possibly in need of a reset.  We think the 3.3 V supply is steady after the reset is sent, but can’t measure under failing conditions due to #1.
5. We also can’t communicate directly with the 5272, we can only send read requests to the FPGA, which in turn queries the 5272.  The FPGA evaluates whether the result is above a threshold.  The results we get indicate that all of the 5 channels we use going to the 5272 are higher than the threshold, even as the device warms to temperatures where we know everything works fine, and we are very confident the analog outputs are correct and settled.
6. This only corrects with a restart of the supply sequence and a 5272 reset.  We only reset the 5272 once after a supply cycle.

We would like to provide more information but we can’t due to #1.  As you are the experts on the 5272, we would like to know whether it is possible for the 5272 to get into a failure mode where it is either not communicating or sending readings that are above our threshold, possibly FFF on all channels.

We would like to know if this failure mode can be caused by excessive excursions at the analog output, but within 0 and 3.3 V, given that there are two buffers using single positive 3.3 V supplies, the same supplies as the 5272.

Is the device more sensitive to this at cold temperatures?  We believe the start at all temperatures is very similar, but we have only seen this occur cold.  It is also very repetitively failing.

We have already replaced the entire CCA with the 5272 and FPGA, and the problem repeats.  When we return to the original analog output vendor, the problem goes away.

We realize we are asking you to make a speculative call with limited information.  Have you seen the 5272 draw lower current, and be in some sort of a failure mode where reads are high or communication is not working?

Can this be caused by excursions between 0 and 3.3V at the analog input after the supply is settled?

We have been looking into our startup sequence as well."

Thank you and best regards,
Gerald

Gerald,

I have forward this to the design team. Hope to hear back soon from them.

Regards,

Jim

Hi Jim,

Thank you. We will wait.

Best regards,
Gerald

Hi Rob,

I would like to gently follow up on this. We hope to hear an update. Thank you.

Best regards,
Gerald

Hi Gerald,

I apologize for the delay. Jim is OOO and I am jumping back in.

I have an email out to the design team to see if they have any comments on this.

I hope to respond by tomorrow.

Thanks,

Rob

Hi Rob,

May I ask for updates? Thank you so much.

Best regards,
Gerald