ADS124S08: ADS124S08 always sends zeroes back during data reads, but registers read fine.

Hi Zisimos,

Please note that we cannot access Google drive files due to our internal firewall, can you please post the logic analyzer pictures in a subsequent post? I am not sure what you have shared so far, but it will be helpful to see the state of SCLK, CS, DOUT and DIN pins when you write a register and read back results, when you start conversions initially (you seem to see DOUT/DRDY toggling) and when you start conversions after selecting the temp sensor.

Also if you have a schematic to share that will be helpful.

Please note that with all 3x CS pins tied together you will not be able to configure each ADC individually i.e. they will all share the same configuration settings. I just wanted to confirm that this is the intended behavior.

-Bryan

Good catch!

I've corrected the two issues that you picked up on, now using SCLK=5MHz and correctly using the FSC registers (not changing from default values).

Now the results of the conversions are nonzero but distributed strangely.  Tried using on die temperature sensor source and also DVDD/4 test source.  Both behave in the same way: Conversion results are either very  close to 0x000000 or very close to 0xFFFFFF.


Some labeled logic analyzer captures are attached.  Here is a dump of the SPI traffic with some annotations.  For ease of implementation, 32bit transfers are always used and unneeded values are filled with NOP (0x00).

Will share a partial schematic as needed, but the connections are as shown in the first post, and all analog inputs are floating (connected to analog switches which are all open), using DVDD/4 for input source.

Hi Zisimos,

How are you determining when new conversion data is ready? Since you have not connected the DRDY pin, you must be polling DOUT/DRDY I assume? I did not see any of these transactions in the data you sent (maybe in the zoomed out images, but I cannot see the details of the SPI lines).

Also, from the table you have provided above, there seems to be a ~38ms delay from the START command being issued to data being read out. But you have set the data rate to 1000 SPS, which should give you a response within ~3 ms using the sinc3 filter and single-shot conversion mode. Can you clarify what is going on here?

-Bryan

Hi Zisimos,

After reviewing this some more internally, I see now that you are using the maximum delay setting (4096*tMOD) as well as global chop mode, so the 38.3ms makes sense. Please note that this reduces your effective data rate from 1000 SPS to 26 SPS.

Have you checked that the external reference voltage is valid into your system? You said you were using a 2.5V external VREF, have you measured this voltage at the REFx0 pins on the ADC to make sure you are actually applying 2.5V to these pins?

Also, make sure that the VREF buffer on REFP0 is disabled, since it is identical to the supply (VREF = AVDD = 2.5V). The buffer will not be able to drive all the way to the supply rail (AVDD), so the VREF voltage might droop even if there is 2.5V at the pin.

You should also understand that the coding scheme on the ADS124S08 is bipolar using binary 2's complement. So the values you are seeing e.g. FFFFEDh are not close to full-scale but a very low voltage negative value (-5.6uV). So the values you are seeing are all very close to 0V, both positive and negative.

-Bryan

I've figured out the issue.  G_CHOP being enabled was resulting in a near-zero value. 

Is the chop function averaging x and -x instead of x and -(-x)?  Or maybe chopping doesn't work with these internal test sources?

After disabling G_CHOP, both DVDD/4 and internal temperature sensor sources result in roughly the expected values. 

Not a big deal, I can implement chopping on the host side.

Thanks for your many good suggestions!