ADS1248 resolution

Hi Hamid,

High resolution and precision alone does not guarantee success. As you didn't give any details about the registers settings you are using or the amount of fluctuation in readings that you are seeing it is a little hard to tell if you are just seeing noise or if you have some other issue.

Most likely you are trying to resolve to some voltage in the nV region.  These are really small measurements.  Air currents, lights, cell phones, etc., all effect the measurement.  If you are using a prototyping system (as opposed to an actual PCB) it will be really hard to achieve the level of performance that you are probably seeking.  If you send me your schematic/layout I may be able to point to some possible issues.

On to the code snippet.  Is there something missing, such as brackets?  For the 'if' statement at the beginning, what gets executed?  Is it all of the remaining code?  How are you capturning DRDY?  Are you polling the DRDY pin?

As you are attending to strip out the most significant bit anything that could be close to zero volts differential and could potentially slip negative will appear as a near full-scale reading if you do not properly account for the sign bit.  The output result is binary two's complement.  You cannot just strip the most significant bit.  For example, 0xffffff is a -1 code, but if you strip the msb it becomes 0x7fffff which is positive full scale.

You must also be careful about making single ended measurements.  As the ADS1248 has an integrated input buffer it cannot accurately represent voltages close to the rails.  There is a common mode input restriction.

As I only see a snippet of your code I can't tell if you are calling functions, or if your variables are volatile.  You must also make sure that the variable V24b is initialized properly.  You must make sure that the contents is empty  before you '|' any contents.  If there is anything in the variable prior to 'OR' the result will be incorrect.

If you are trying to get rid of the msb, then your acutal code is wrong.  You should not use a shift, especially right shift as that gets rid of the lsb and not the msb which is the sign bit.  You must make sure you handle negative values appropriately.  Also, the total number of available codes in the positive direction is from 0 to 8388607 so to more accurately represent the value you should alter your equation so that the divisor is 8388607.0.

You should not need a delay in your code.  I also think there may be a slight misunderstanding as to the use of SDATAC and RDATAC.  The ADS1248 initializes in RDATAC mode.  This means that the result is automatically posted in the output register so that it can be directly read out of the device when DRDY transitions from high to low.  No command is necessary for the contents to be read.  If you issue a SDATAC command, the automatic posting of results to the output register will discontinue after the current conversion completes.  That means it will post one more time.  If you intend to read other register contents, you must make sure you wait for the current conversion cycle to complete before attempting to read data or the read will be corrupted by the automatic posting of the conversion results.

If you look at your code again, you will see that the issuing of RDATAC and SDATAC really doesn't make much sense.  I would suggest that you try to get things working in SDATAC mode only.  In this case you only need to issue SDATAC once when you first initialize your program.  Then in your code instead of issuing RDATAC, issue RDATA and read out the results.

One other thing I'm not clear about is why you are multiplying your results by two?  Is this to regain your bit you shifted out?  This is not a good way to handle your data, and the results will always be incorrect if the result is negative.

Best regards,

Bob B

Hi Bob,

I made more changes in the code based on your feedback and the result improved to some extent (2mV fluctuation now). There is no delay at the end of conversion function. I know this readings can be much more stable tough. The numbers are looking up but I am still trying to get a better understanding of the chip. There are 2 major issues as it is right now.

• The most significant byte (DATA1reads in my function) is always 0 with this configuration, so I am not fully utilizing the a2d's capacity.
• The second if statment is not really doing anying. I was just trying to see when the DRDY goes high. The problem is that, when I remove this statment the last 8 bits will be 0 this time and the conversion does not work.

Thanks again for your feedback.

Regards,

Hamid

Hi Bob,

What is the definition for DRDY0? Is this where you are doing the polling?  Yes we are polling DRDY as I mentioned earlier. The schematic also shows, pin 25 (DRTDY) of the A2D  hooked up to an I/O pin. We have called this pin  DRDY0 in the definitions at the top of my code. (The image couldn't capture the entier code.)

DigitalIn DRDY0(p9);

The link was very helpful. What A2D chip would you recommend that provides us with high resolution (24bit) and does not need differential input. The voltage range that I normally need to capture is in 0-5 V range. It would be ideal to have 7 or 8 high resolution a2d channels with low response time.

Thank you again  for your response.

Kind regards,

Hamid