ADS1148: ADS1148 noisy conversions and minimum / maximum clock frequencies

Hi T Slater,

Can you provide more info about the noise you are seeing? Plots of the data showing the noise, frequency of occurrence, etc. - basically anything else you can share with us to help us understand what is going on

You can also provide a schematic and your register settings for reference

There should be no issue running the SPI clock at ~1MHz

-Bryan

Hi Bryan. Whilst I'll have to check with colleagues regarding disclosure of schematics, I have a screenshot of the problem.

So, the noise is around 0.5C which in our code ends up being around 10 counts in the ADC. These are measuring PT100 RTD devices. 250uA excitation current. PGA set to 16.

Edit: I should say that the problem exhibits on the green trace in the top half. This then follows the change of module position, and continues as the red trace in the lower half.

Hi T Slater,

What does "change of module position" mean? What are you changing?

What are the plots in your screenshots meant to represent? The y-axis isn't labeled (I'm guessing degrees C?), and I cannot see the x-axis. 

Why are there 4 waveforms in your screenshot? What do they represent?

Can you send me a high resolution image of the topside markings of the device where this issue occurs?

Remember that I don't know anything about your system except what you've shared here. The more detailed and explicit you can be, the better we are able to help

-Bryan

My apologies Bryan, I'm the software engineer investigating the issue which has been highlighted from our Chinese colleagues.

System description:

An LPC812 talks to the ADS1148 on a small "module" about 30mm x 40mm in size. 4 of these modules sit on an IO carrier board, which is in a din rail mounted case.

The graph data:

The 4 plots represent a measurement rom each of 4 modules on a single IO carrier. X axis I can't help with, based on < 10 measurements per second it looks to be a few minute's data.

Further error info:

The error reporter has swapped two module's positions over, so the fault has then moved with one of the modules.

The error is reported to occur as the module ambient temperature is raised to 55 degrees C. Some modules exhibit the problem at room temperature. the errors are occurring in about 2% of modules produced.

Please bear with me while I try and sanitise some code to share. There are 3 possible areas - when we reset the ADC, when we initialise the ADC, and when we update the MUX to trigger a new measurement.

I will post back with further code / configuration information.

Hi T Slater,

Thanks for the additional information. It seems unlikely to me that the code is causing this issue, as then it should affect all of the modules, not just one

Can you send me high resolution images of the ADC topside markings i.e. the info on the top of the actual device? Please send me that info for all 4 modules, and clearly denote to me which one is "bad". I'll look into the history of these devices to understand if there is a lot / wafer issue or something like that

-Bryan

These are from two "bad" units. I'll post back with a known good unit - which should be from an older batch than these two posted above.

Hi T Slater

Thanks for providing this information, let me review and get back to you

-Bryan

Here's a known good one, date and batch codes are from a unit that is perhaps 18 months older than the two units above.

Hi T Slater,

Thanks for the additional info, I am still waiting on feedback from one of our internal teams. I will respond back to you as soon as I know more

-Bryan

Hi Bryan,

I am still trying to get to the root cause of the noise we're seeing on the ADCs. All I can say as of now is that reading a "raw" value from the ADC still shows the noise. In our signal processing chain, that is to say:

1. ADC value read over SPI

2. ADC value converted to float <--- This value read back as "raw" data

3. RTD lookup table + linear interpolation

4. End value written to output register