ADS124S08: Issue with noise performance between REF0 and REF1 when reading RTDs

Hi Berrie Dirks,

Can you try disabling (opening) the powerdown switch that is connected to REFN0? It serves no purpose to be closed since you have REFN0 connected directly to ground anyway. This is the only difference I see in your circuit compared to REFx1

-Bryan

Hello Bryan,

We've ran some tests with the powerdown switch open but there seem to be no change in the results so this does not solve the issue.

Is there any more information we can provide that would help troubleshooting this issue?

Kind regards,

Berrie

Hi Berrie Dirks,

It is a holiday here in the US, we will get back to you next week. Thanks for your patience

-Bryan

Hi Berrie Dirks,

How are the IDACs connected to the RTDs? You show AIN0 and AINCOM in your register map settings, but these are not connected to anything in your schematic (maybe you purposely cut these off from the image you sent?). You also mentioned muxes. Can you send the complete schematic?

If you bypass the mux and measure an RTD directly, while also using REFP0/REFN0 as the reference source, do you see a difference in performance?

If you reduce the IDAC current magnitude, do you see any difference?

What is the voltage magnitude of the error you are seeing? You only show the plot with respect to temperature and resistance, but it would be interesting to know what the actual voltage error is

Also what is approximate resistance of the RTD you are measuring? It that 200ohm as shown in your plots?

What is the plot actually showing? Is each signal a separate RTD? If so, how many measurements are being taken per channel in that image? It would help to explain in detail what info you are actually providing

-Bryan

Hello Bryan,

Bryan Lizon86 said:

How are the IDACs connected to the RTDs? You show AIN0 and AINCOM in your register map settings, but these are not connected to anything in your schematic (maybe you purposely cut these off from the image you sent?). You also mentioned muxes. Can you send the complete schematic?

I did indeed remove some parts from the schematic because I thought they would be irrelevant to the problem. I think I cropped out too much. Here's the complete circuit, AINCOM and AIN0 are configured as current source.

Bryan Lizon86 said:

If you bypass the mux and measure an RTD directly, while also using REFP0/REFN0 as the reference source, do you see a difference in performance?

We have 8x the circuit above on a single board and we have one separate ADS124S08 with two PT1000 sensors connected directly. All pins used are equal in this circuit but no muxes are present.

On this 9th ADC we see the exact same problem where noise is higher when REFP/N0 is used, so I do not believe that the problem lies within the muxes or in the switching between sensors. Here's a short measurement of 2 sensors connected directly to the ADC. The sensors are hanging in free air.  On sensor 1 you can clearly see if someone is standing close the temperature goes up but in the 2nd picture the noise is dominant.

Bryan Lizon86 said:

Also what is approximate resistance of the RTD you are measuring? It that 200ohm as shown in your plots?

Sensors are all PT1000's with a base resistance of 1000 Ohms but we intend to use it between 20C and 350C (~1100-2200 Ohms)

Bryan Lizon86 said:

What is the plot actually showing? Is each signal a separate RTD? If so, how many measurements are being taken per channel in that image? It would help to explain in detail what info you are actually providing

I think you've misinterpreted the plot from the first post (I should have labelled the axes). It is showing temperature over time and not temperature vs resistance. Our temperature sensor array is on a different board and we interface between the two with flexcables. Each cable has enough traces for 8 sensors. The image is showing the 8 sensors on the left and 8 on the right, each set having its own cable. The figure was just to illustrate there's a pattern. We tried swapping the cables but the noisy channels are always linked to REFP/N0 channel on all ADCs on the board.

So far we've only had the hardware to heat up to 120C but over this entire temperature range (20-120C) we see no change in noise performance.

Bryan Lizon86 said:

If you reduce the IDAC current magnitude, do you see any difference?

What is the voltage magnitude of the error you are seeing? You only show the plot with respect to temperature and resistance, but it would be interesting to know what the actual voltage error is

Looking at the included figure the top picture the error sits around 20uVpp and the bottom picture at the same time sits at 460uVpp. We did not run a test yet with reduced current on the IDAC so that will be our next test which is currently planned for Tuesday.

I hope this additional information clarifies a little.

Looking forward to your answer.

Kind regards,

Berrie

Hi Berrie Dirks,

Thanks for providing all of this additional information

How are you taking the data that you show in the plots? For example, on the 9th ADC that does not use any external muxes where you have two RTDs connected, are you taking one measurement on Ch1, then one measurement on Ch2, then Ch1, then, Ch2, and so on, for 5+ minutes? Or are you sampling Ch1 for many samples in a row, then Ch2? Since you show you are single-shot mode I am guessing the first method, but I just want to confirm. If you are ping-ponging between channels, can you just take data on one channel for a few minutes, then take data on the other channel for a few minutes, and see if the behavior changes?

In the "Heater PT1000" plots you sent in your last post, are those RTDs supposed to be measuring the same temperature? You said they were hanging in free air and they are on the same board, so I assume that means they are very close to each other, but It looks like #1 is ~20C, while #2 is ~23C. Can you confirm? If so, is this within the tolerance of the RTD e.g. +/-3C, or is there also an offset associated with the measurements themselves (in addition to the noise)?

Can you try similar tests using a 1kohm resistor instead of a real RTD, to see if you still get noise on the VREF inputs?

Have you measured the ADC input and reference voltages with a scope to see if the fluctuations you are measuring on the output are actually present on the input?

Is there any possibility that the REFP0/REFN0 circuit on the layout is physically close to something that would give off heat e.g. an LDO, or something with switching noise e.g. charge pump, that might influence that specific reference resistor and not the one at REFP1/REFN1?

There is no fundamental difference between the two REFx circuits internal to the ADC, so there is no obvious reason why this would occur. I am just suggesting some troubleshooting steps to see if we can understand or change the behavior in any way

-Bryan