• State TI Thinks Resolved
  • Locked Locked
  • Replies 3 replies
  • Answers 1 answer
  • Subscribers 47 subscribers
  • Views 109 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

ADS8900B: High speed low noise current loop to voltage conversion

Chris van der Aar
Intellectual 510 points
Part Number: ADS8900B
Other Parts Discussed in Thread: ADS127L11, ADS131A04, ADS1271, ADS127L01, ADS127L21, TPS7A20

Tool/software:

For a new project we are using capacitive sensors for very accurate height measurements (3mm range with nanometer resolution). The cap sensor amplifiers have a current loop output (4..20mA) related to the height (0..3mm) and we want to use that for height measurements by our motion controller

I have to design a current loop to voltage converter because we want to read the sensor info with a 20 or 24 bit SPI ADC with an update rate of 20kHz. I already got some info from the Amplifiers forum wrt the current loop to voltage converter but for the ADC choice they referred to the Data Converters forum. Besides the exact implementation for the current loop to voltage converter also depends on the ADC choice (inpput range).

Do you have any suggestion which ADC devices are suitable ( 20-or 24 bit ADC with SPI interface and fast conversion time)? And maybe an evaluation board or a design example that could work for this application? With these we could already start some testing together with our software engineers.

When I look in the ADC selector for 20 bit or higher ADCs with fast data rate (>250ks/s) and SPI interface  I can only find the 24 bit ADS127L11 or the 20 bit ADS8900B. With 20kHz we have 50 usec cycle time but within this time that data has to be sampled and converted and ready for the next conversion cycle. .Besides we want to read the data via BISS interface so we have to transfer the SPI data to BISS protocol which will also take some time (we already found a chip that can do this task). So therefore I think we need an ADC that will take not too much time for sampling and conversion?  Or do you think we can use an ADC with a lower sampling rate?

Best Regards

Chris

  • 0
    TI__Expert 5480 points

    Hi Chris,

    Thanks for your question. Approximately how long does the SPI to BISS transfer take per cycle? That may help decide the margin of minimum sampling speed you need.

    ADS127L11 is capable of ~1MSPS data rate, similar to the ADS8900B, when using low-latency filter mode. Do you need the wideband filter to be enabled for your application?

    ADS127L11 is a delta-sigma ADC, while the ADS8900B is a SAR ADC. Delta-sigma ADCs are typically more suitable for applications that require higher resolution, like yours. We do have an evaluation module and example code for the L11 that may be useful to you. I also found this related thread where someone is using the ADS131A04 for a similar application.

    I am looping in Keith, who supports the ADS127L11, for his insight.

    Best regards,

    Samiha

  • 0 in reply to Samiha Sharif
    Intellectual 510 points

    Hi Samiha!

    Here some response of the manufacturer of the SPI to BISS converter chip that we planned to use for this application wrt speed:

    >> As soon as a BiSS frame is started, our chip will start clocking at the fast sensor interface (max. clock frequency = 20 MHz).

    >> Your ADC can provide the latest data directly.

    >> Assuming 24 bit, they can be received by our chip way below 2 us.

    >> Our chip will then (optionally) calculate a CRC and output the data via BiSS.

    >> BiSS cycles of 50 us should not be a big deal.

    >> We have customers who have implemented cycle times < 20 us.

     So for this chip the 20 kHz loop frequency is not a problem. For the ADC a single channel version is OK because I have to make a  few of these converter boxes but each box gets its own PCB with 1 current loop input (4..20mA ). Wrt expected noise I don't know because the current loop is generated by a demodulator module that we buy from a company specialized in these types of measurements and I expect not too noisy at the current loop output. These modules are placed on an DIN rail and our PCB will be also placed into a DIN rail box so the distance wrt cabling will be not too long. Besides we can do some hardware filtering (RC  filters) at the input stage where we have to convert the current loop to a voltage matching the input range of the selected ADC. Furthermore I think we have to see in practice how much noise we get and then we can always put some extra filtering in software. I van ask our software team if they can make something that we are able to change the fillering easily during testing.  FinalIy I can ask your amplifier team for a good solution for the current loop to voltage conversion when I know the right type of ADC.

    You suggested a delta sigma ADC for this application. When I look to single channel, SPI ADCs with min100ks/s sampling rate I find :

    - ADS1271 ( looks more an older Burr Brown type)

    - ADS127L01

    - ADS127L11

    - ADS127L21

    The final one has the fastest sample rate (and very low gain drift as the ADS127L11) however maybe our SPI frequency of max 20MHz can give some limitations here wrt SPI transfer data rate?  How much time will it take from start conversion untill data ready (so the BISS converter can convert the ADC data for these devices (ADS127Lxx) ? Do you think the ADS127L01 would also be sufficient for this? It llos that that one is a little bit accurate? We do not measure absolute positions but a relative displacement and of course we will have some delay between the actual position and the actual data that will be read but that should not be a big deal for us. 

    Which device would you suggest? So I can ask to the Amplifier team for a good solution for the current loop to voltage conversion. And do you have an example schematic that I can use for our PCB so we are sure that everything will work with our design?

    For the analog power supply we normally get +24V from an external rack supply but these types of supplies will certainly have some times. However nowadays there are DC/DC converters for +24V to +3V3 with very low noise figures. I assume this will be also an important factor here for accuracy?

    Best Regards 

    Chris

  • 0 in reply to Chris van der Aar
    TI__Mastermind 38320 points

    Hello Chris,

    In your system, you need to measure an input current, and then transfer that result in a total time of 50us.  Let's assume we need 20us to transfer data from the ADC to the BISS and to process and then transmit over the system bus.  The actual time may be much less than this, but since 20us was mentioned, I think it is a conservative estimate.  This leaves 30us for the ADC to acquire and convert the data.

    If using a SAR ADC such as the ADS8900B, the total time for a fully settled reading will simply be the conversion period.  ADS8900B operates up to 1Msps, so this total time is 1us.

    The Delta-sigma ADCs will take longer for a fully settled reading due to the internal digital filtering necessary in this topology.  However, the ADS127L01, ADS127L11, or ADS127L21 can easily meet the 30us requirement.

    The total amount of time required for a fully settled reading depends on the filter option selected.  The ADS127L11 is probably the best choice in this case.  Using the SINC4+SINC1 filter setting with OSR=32x4=128 will result in a total conversion time of 18us and an input noise floor of 4uVrms, resulting in an effective resolution of 21b.  If using the OSR=32x10=320, total conversion time will be 33us with an input noise floor of 2.8uVrms, resulting in an effective resolution of 21.5b.

    The ADS127L01 could also be used, but the ADS127L11 has more filter options which allow a good tradeoff between noise (effective resolution) and speed.  In addition, the ADS127L11 as internal buffers which are easier to drive using external amplifiers, so my suggestion is to use the ADS127L11 in this case.

    As a side note, you could also use the ADS8900B, but the noise will be higher due to the higher bandwidth of the device.  You could get similar noise to the delta-sigma by averaging several samples (similar to the internal filter of the delta-sigma), but this would require additional digital processing of data.

    If you want best performance from a high resolution ADC, a good high PSRR low noise linear regulator such as TPS7A20 is recommended for the supplies.  However, there are good low-noise DC/DC converters that can also work, but it is difficult to determine based on datasheet specifications and usually requires verification in the lab.

    Regards,
    Keith Nicholas
    Precision ADC Applications