• State TI Thinks Resolved
  • Locked Locked
  • Replies 15 replies
  • Answers 6 answers
  • Subscribers 46 subscribers
  • Views 4115 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

ADS1248: Reference voltage

Satheesh C R
Prodigy 90 points
Part Number: ADS1248
Other Parts Discussed in Thread: INA333, , ADS1220, ADS124S06, ADS124S08

I am planning to use ADS1248 for strain measurement. Schematic is given below. Can I use 5V dc as my reference voltage? INA333 out put will 2.5 +-2 V , so my ADC input ANI0 will swing from 0.5 to 4.5V.

  • 0
    TI__Guru* 98215 points

    Satheesh,


    The ADS1248 is specifically limited in the maximum reference voltage. If you look in the Recommended Operating Conditions table on page 8 of the datsheet, the Voltage Reference Inputs are limited to (AVDD-AVSS)-1V. With a supply of 5V, that means the reference is limited to 4V.

    If you need to run a bridge measurement, you should do it with a source/reference 1V less than the supply. Alternately you could use an ADS1220 or an ADS124S06. Both of these other devices do not have the same reference limitation.


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Thank you Joseph. Is there any 8 channel ADC having high reference input capability(5V)? Actually I need to measure 6 strain gage bridges outputs, if possible with single ADC.

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points

    Satheesh,


    I hadn't realized that you needed to measure quite so many channels. In that case, you should use the ADS124S08. This device has 12 analog inputs plus a common. The reference input voltage extends to the rails so the maximum reference voltage is AVDD-AVSS.

    Read through the datasheet for details, but the ADS124S08 should work for you. Once you're finished with the schematic, we'd be happy to review it for you.


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Thank you Joseph for your quick response. I have modified my schematic with ADS124S08 and same is given below. Kindly have a look at it for any corrections or improvements. I have a doubt, if I am using REFP0 and REFN0 for my reference, do i still need to give 5V at REFOUT?

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points
    Satheesh,


    I think it would be better to remove the INA333 entirely and just use the PGA on the ADS124S08. The problem is that the noise for the INA333 is still rather high. The noise of the INA333 for the range of 0.1 to 10Hz is about 1uV peak to peak referred to the input. In a gain of 1000, this means that the noise seen at the ADC pin will be 1mVpp. In comparison, the noise of ADS124S08 PGA will be significantly less. You can look through the noise tables (Tables 1-8) in the datasheet for an approximate metric. Even though the gain is less (max 128V/V) you'll still get a better resolution because the noise performance is better.

    If you are measuring 6 bridges, you would use 6 differential input channels from the 12 analog inputs of the ADS124S08. Again, I think this is the correct device for your application.

    I would note that REFOUT is the output for the internal voltage reference. You do not tie it to another voltage. If you are using internal reference voltage (either for a measurement, or for IDAC currents, you would only need to attach a capacitance to this pin).


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Joesph,

    Though I have put my maximum output from strain gage bridge as 2.5mV, most of the time its less than 1mV. 2.5mV is at the highest load, but most of the time I need to measure half of the highest load. So wont it be better to put an amplifier which can give higher amplification than 128V/V.

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points
    Satheesh,


    The noise that the INA333 will still be a problem. However, the noise that you see from the ADS124S08 PGA will depend on the operation that you choose. I'll put up an example with some numbers to help explain.

    Your input range is ±2.5mV, which would be a full scale total of 5mV of range. If you use the INA333, then it's like having 1uV of noise at the input of the PGA. This will reduce your resolution to about 5000 counts (where the ADC will be able to read 5mV/1uV - or about 1 part in 5000 noise free).

    Look at this from the side of the ADC, with a 2.5V reference for example. If you have ±2.5mV as an input, with a INA333 in a gain of 1000, you have an input range of ±2.5V. Here you use your entire full scale input range, using all 2^24 bits to represent input codes. However, the INA333 imparts 1uV of noise at the input, so the ADC sees 1mV of peak to peak noise. That 1mV of uncertainty becomes (1mV/5V)*2^24 codes or 3355 codes of noise. Out of 2^24 codes, you have 1 out of 5000 parts of noise.

    In comparison, the noise for the on-board PGA for the ADS124S08 is significantly lower depending on the mode of operation. I don't have specific numbers for an external reference, but running with the sinc3 filter, internal 2.5V reference, and global chop disabled (Table 1 in the datasheet) the noise is significantly lower at low data rates. With a data rate of 10SPS using a gain of 128, the equivalent input noise is going to be about 0.23uV (peak to peak). This will give an equivalent resolution of 21700 (from 5mV/0.23uV).

    As in the previous example, you can look at this from the ADC side. The input will be smaller, no longer using the entire full scale range. The input is now ±2.5mV*128 or ±320mV. Because of this, you'll only use (640mV/5V)*2^24 = 2.147e6 codes, which is 1/8 of the full scale range. However, the noise is significantly smaller. With a noise of 0.23uV at the given configuration, you'll see a noise of 0.23uV*128/5V = 29.4uV. This is the equivalent of (29uV/5V)*2^24 = 98.8 codes of noise. The resolution is again 1 part in 21700 (from 98.8 codes of noise, to the full scale of 2.147e6 codes for the full scale).

    Using the INA333, the noise is larger and the noise is amplified to an even larger value. Using the on-board PGA, the noise is smaller, and the noise is amplified less. Even though I don't use the extent of the full scale range, the noise is less of a factor.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Ok Joseph, then I will not use additional amplifiers..i'll reduce the reference voltage to get maximum resolution with pga gain of 128. giving a reference of 1 or 1.25v is it ok?

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points
    Satheesh,


    I think that would be fine. However, you should remember to use a precision voltage reference. Any error of the reference voltage appears as a gain error. If there is absolute error in the reference, this is reflected in the ADC measurement (a 1% error in the reference, becomes a -1% error in measurement). Also, noise is also a large factor. In the same way that there is reference error the noise is seen as a changing reference. You'll want to consider this noise as a factor in your measurement.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Joseph,

    Can you suggest me a voltage reference IC part number which will give me minimum error?

    One more doubt is, I will be connecting my bridge outputs ( differential) across AIN0 & 1( bridge-1)  or AIN2 & 3  ( bridge-2) ..AIN10 & 11( bridge-6). In this case, my bridge outputs can be either +ve or -ve  ( +-2.5mV max), whether differential amplifiers can take negative inputs? OR do i need to use AINCOM and convert each leg of these bridges separately and then take difference to get differential reading?

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points
    Satheesh,


    I'm sorry I missed your last post. I've been looking over this thread again and I have a few more comments and questions.

    In my last post, I spoke to soon about using a smaller reference to increase the resolution. In your application, you have a bridge which is a ratiometric measurement. That means that the reference would be the source driving the bridge and the output would be measured by the ADC. In that case you'd already have a low noise because changes in the reference are correlated with changes in the input. In that case, generally you wouldn't want to use a reference that is separate from the bridge excitation. You'd have to make the measurement, but the exception is if both the bridge excitation and the reference are very low noise.

    In the end, I think it's still better to use a ratiometric measurement, using the supply as the reference, and using the PGA from the ADC for amplification. Adding in more circuitry usually just adds on more noise. Out of curiosity, what bridge or strain gauge are you using? Do you have exact specifications for it?

    As for your other question, generally, I would use a differential reading for each bridge. For the ADS124S08, that would allow for 6 bridges of measurement.


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Joseph,

    Daily I was checking this thread for a reply from you.. Once again thank you for replying.

    My doubt of differential measurement is still not clear, request you to clear two point regarding this. First one is, if i apply a -2.5mV as differential input across say ANI1 and ANI2, with a gain of 128, my PGA output will be -320mV, whether an amplifier without -ve power supply will work? ( we are not giving any -ve supply to ADS124S08, right?).  Second one is whether ADC built in is a bipolar ADC? I assume its a unipolar ADC.

    If I use Ratio-metric measurements, what can be maximum bridge excitation voltage? Can I have a 5V excitation? In ratio-metric measurement I have to give excitation voltage itself as reference voltage, right?

    Actually I prefer a higher excitation, so that my bridge outputs can be maximum.

    My measurement requirement  is,  6 component strain gage balance measurement. I have a force balance which can measure 6 outputs ( 3 forces and 3 moments). These are basically full bridge strain gages. These outputs can be either positive or negative ( depending on my direction of loading)

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points
    Satheesh,


    Like most of the other precision ADCs of this type, the input is bipolar, meaning that the ADC can measure both positive and negative signals. However, both AINP and AINN must be within the supply range to be valid. A negative signal means that the AINN is higher than AINP (measuring AINP-AINN). As an example, if the ADC AVDD supply is +5V and the AVSS is ground, then both AINN and AINP must be within those voltages. If AINN is +3V and AINP is +2V, then the ADC would measure -1V. This would give a negative measurement, a negative supply and a negative reference is not used. I think this is the question you were asking about. If I've misinterpreted it, let me know and ask again.

    A bridge measurement is ratiometric uses the bridge excitation as the reference. For the ADS124S08, this reference may be as high as AVDD-AVSS. If the ADC supply is 5V, then you can use a reference of 5V.

    I would note that the ADS124S08 Evaluation Module has a set of connections that you can use for a bridge measurement. You could get one of these boards and test the application. At the very minimum, the ADS1x4S08 Evaluation Module User's Guide has an explanation for the bridge measurement circuit. It is worth downloading the guide for more details. You can find the User's Guide at the following link:

    www.ti.com/.../sbau272a.pdf


    Joseph Wu
  • 0 in reply to Joseph Wu
    Prodigy 90 points

    Ok Joseph,

    I will by one ADS1x4S08 Evaluation Module and will try and test  different features of this IC. In case of any doubt I'll write to you. Thank yo very much for all your help.

  • 0 in reply to Satheesh C R
    TI__Guru* 98215 points
    Satheesh,


    I think using the EVM is going to help you test your setup. As I mentioned, there's a setting for a bridge measurement, so you should be able to test at least a single channel with the EVM.

    In your original post, you mentioned using the INA333. I still think the extra INA will put far more noise in your setup, but you should just get one and look at that, just to see it. Even though the INA333 could amplify your input signal to 1000, I believe that you'll get better noise performance from the ADS124S08 PGA, even if you use far less of the full scale range.

    Good luck with your evaluation. If you have further questions, come back and start another post.


    Joseph Wu