ADS1672: ADS1672 read out non-linearity

Sorry I copied and paste incorrectly.  Pleae ignore data in my last post. The following is corrected data.

Hello Hai,

Welcome to the TI E2E Community.

The data appears to suggest a gain error.  This could be caused by the Reference drooping or the input filter is not fully settled.

Please send a schematic showing the ADC, input amplifier, Reference, and the supply voltages used and I can take a closer look.

Thanks!

Regards,
Keith Nicholas
Precision ADC Applications

Hi Keith:

Thanks for the quick response. Several notes on the schematics:

1) The issue is there whether the VCM of op-amp is connected or not connected to ADS1672.

2) The issue is there whether CLK is 8 MHz or 20 MHz. At first I have 8 MHz so I change to 20 MHz.

Hai

Hello Hai,

The input amplifier configuration looks O.K.  However, I suspect the reference circuit is the problem.  The REF5030 output buffer does not have enough bandwidth to directly drive the input of the ADS1672.  As a result, the voltage will droop, which will appear as a gain error.

You mentioned the error is there whether CLK is 8MHz or 20MHz, but if this is the cause, there should be less gain error at 8MHz verses 20MHz.

You can use the REF5030 with an additional reference buffer amplifier.  Here is how the evaluation board is configured:

Another option is to use the REF6030.  This device has similar performance to the REF5030, but includes a high speed buffer, capable of directly driving the reference inputs of most ADC's.  I have not tried this part with the ADS1672, but based upon use with other similar ADC's, it should provide good performance.

Regards,
Keith

Hi Keith:

Yes, the gain error is slightly less at 8 MHz (@1.322V AINP and 3.621V AINN, the difference is -44,428 @ 8 MHz and  -48,905 @ 20MHz).

I will wire in the REF6030 and let you know the results.

Many thanks for your help.

Hai

Hi Keith:

The gain errors are actually worst with REF6030 as shown in the enclose Table. I have the REF6030 circuit wired per your picture.

Also with the data I sent to you, I have C99 and C98 connected to  AINP and AINN inputs respectively since I thought the 10 Ohm R68 and R69 might cause the input to droop at higher voltage.

What should I do next ?

Thanks

Hai

Hello Hai,

I just realized your gain error is around 0.8%.  If the above readings are the raw data from the ADC, no adjustments for offset or gain, then I think you are just seeing the expected gain error.

The ADS1672 has a max gain error of 1%.  The resistor network around the THS4551 will also have error.  Assuming you are using 0.1% resistors, this will add another 0.2% of gain error.  If you are using 1% resistors, then this could add up to 2% gain error.

Also, looking at your schematic, the ADC VREFN pin is connected to AGND2, the Reference ground is connected to AGND.  If there is any voltage difference between these pins, then this will show up as a reading error.

Regards,
Keith

Hi Keith:

Which one is better to use REF5030 or the REF6030 ? 

Would the REF5030 with its driver circuit OPA211and components would introduce more gain error than the REF6030 with less components ?

Also my final design will have a Ground Plane that might help with this issue.

Many thanks for your help.

Hai

Hi Keith:

If the reading accuracy is 1%, how do I calibrate to give me better accuracy ?

Thanks,

Hai.

Hello Hai,

You can get better overall DC performance with the REF5030 using an external amplifier, but it would require a dual stage amplifier to get this performance, similar to what is suggested in Figure 93 of the ADS9110 datasheet.

In most applications, the REF6030 performs almost as well and uses fewer parts.

Using a single ground plane should help with these issues as well.

Regarding your question on calibration, take a look at this TI Precision Lab that discusses the topic.

https://training.ti.com/ti-precision-labs-adcs-understanding-and-calibrating-offset-and-gain-adc-systems?context=1139747-1128375-1139104-1134080

Basically, you take two measurements, one near -FS (or ground) and the other near +FS, and then calculate the offset (b) and gain (m)  using the straight line formula y=mx+b.  Once you have these parameters, x is the raw reading, and y is the corrected reading.

Regards,
Keith