This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS124S08: Slow response of signals.

Part Number: ADS124S08
Other Parts Discussed in Thread: ADS1235, ADS1261

I am currently using the ADS124S08 in a strain gauge design.

I am running at 800SPS and using a Gain of 64 and a reference of 1.25V

I am running a strain gauge that is a 350 ohm wheatstone bridge at EXC voltage of 5V.

I get poor response to impulse data.  it takes several seconds to decay back to a baseline.

I tried running a frequency generator square wave of +3V to -3V with a GAIN of 1 and a Reference of 5V.

I am not able to get a waveform definition for any thing over 1/5th Hz.  anything above that and the squarewave looks like a triangle wave at reduced amplitude.

I have tried using the Sync filter and the low latency filter with little or no change.  I have moved the SPS around and no improvement either.  I have taken all averaging out of my software and cannot get a frequency response over the 1/5th Hz.

Is this a charge dissipation problem?  not sure what I am missing here it seems like this chip should be easily capable of tracking a 10Hz signal.

Should there be resistors to ground to bleed the voltage off?  I tried a single 1K across the differential with no improvement.

  • Hi Greg,

    When running your tests with a 5V reference and the square wave of +/-3V, you need to make sure that the voltage as it varies is in the correct input range.  So how have you set up your generator to the ADC inputs and have you verified the inputs with a scope relative to AGND?  You cannot drive the input below AGND with a unipolar analog supply.  Do you have any analog RC input filters that can affect the frequency response?

    Do you have a schematic you can share of the bridge input configuration?

    Best regards,

    Bob B

  • thanks Bob.  I took all my hardware RC filtering out as well as all of my software averaging.  The bridge goes directly into the ADC on chan 1+ and chan 1-

    Looking at the signal, the NEG chan input to AGND varies from 2.2v to 1.5v  and the POS chan input to AGND varies from 3.1v to 2.4 volt.  The scope is across the differential inputs to the ADC and shows a square wave going from 0.25v to 1.7v.  I will try to attach a graph of my data, I could not get the attach feature to work last night.  The picture is of the square wave as the ADC reports it.  The frequency is 1/8Hz.  

  • Hi Greg,

    If you could send me the register configuration you are using and the raw data as codes I can more properly discern what you are seeing.  Keep in mind that you are using a Delta-Sigma oversampling ADC and not a SAR.  The Delta-Sigma is sampling the input at 256kHz and converting to a modulator data bit stream.  The oversampled/modulator data forces the quantization noise to the higher frequencies and the internal digital filter is essentially a low-pass filter to remove the quantization noise.  This makes the Delta-Sigma ADC very suitable for dc and slow moving signals to increase precision.  

    The fast edges of a square wave is not what the Delta-Sigma is great at capturing.  For example, the sinc3 filter is essentially averaging the step response and will take 3 conversion cycles for the digital filter to settle to the final value.  The low-latency filter should be able to show the step response a little sooner, but may still show some of the same behavior based on the FIR filter architecture. 

    Based on the input voltage and the codes shown on the plot it would appear that the output code never fully reaches either the high plateau or the low plateau.  Are you in continuous conversion mode?  And have you tried the 4ksps to see if you get an improvement in the waveform?

    If you need a faster response you could look at the ADS1235 or ADS1261.  However, keep in mind that as you increase output data rate, the cutoff frequency of the digital filter is raised so that noise from quantization will start to appear in the measurement.

    Best regards,

    Bob B

  • Yes, I read where any input hardware filtering should not have a Fc that is less than 1/2Fmod I suspect for the reasons you stated.

    As for the problem, I think I have found it, it was a combination of several things I was going, one being a had a filter in my software that was doing some heavy averaging of the data that I forgot was even there.  The price of doing a little here and a little there I guess.  I am getting a lot better response for quicker signals now.  As for registers this is what I am doing:

    after [power up:

    MUX set to Ch1

    PGA 0xE8

    RATE 0x2A

    REF 0x20

    Then I read CH 1 data and set up for CH 2

    Then read CH 2 and set up for CH 3

    Then read CH 3 and set up for CH 1

    I have read where you should not talk to the ADC during a conversion?  Should I not be setting up the next channel while I am reading the third data byte from the current channel?

  • Hi Greg,

    It is best to minimize the communication as much as possible and it is best to read the data as soon as possible following the end of the conversion period to cycle through the mux as quickly as possible.  The sharp edges in the digital communication can bleed into the conversion so it is suggested to use 50 Ohm resistors in series with the digital signal traces.

    As the SPI bus is capable of full-duplex operation you can read the data the same time as you transmit the next configuration.  The example is shown in Figure 92 on page 71 of the ADS124S08 datasheet.  If you read the data result at the same time you change the mux, there should be no degradation of the conversion result due to the communication.  The reason being is the conversion result clears the digital filter and restarts the conversion following decode of the mux channel change.  Of course you can also achieve the same thing by reading the data first and then changing the mux.

    Best regards,

    Bob B

  • Thanks Bob, yes I used the example on page 71 as my template.  Should the SPI lines have the 50 resistors as well?

  • Hi Greg,

    I can see that I was a bit confusing when I stated digital signal lines.  I should have more clearly stated the SPI digital signal lines.  The 50 Ohm resistors are not necessary on the more static pins such as START, RESET, etc..

    Best regards,

    Bob B