Actions
Forum Thread Details
  • State Suggested Answer
  • Replies 6 replies
  • Answers 1 answer
  • Subscribers 79 subscribers
  • Views 1119 views
  • Users 0 members are here
Related

Max11300 sample phases

davegadgeteer

The Max11300 has about 20 channels of ADC, but I need to know the time difference between the channels' samples.

 

I want to digitize multiple signals and study their phase differences.

 

Ideally I'd sample and hold all channels at the same time, either muxing the ADC or having one ADC per channel.

 

If I use multiple ADCs, they need to be sync so they sample at the same time.

 

Anybody dealt alt with a problem like that?

Parents
  • 0

    I get involved with lots of simultaneous sampling data acquisition at high(ish) speeds and resolutions.

     

    The MAX1100 is a 12 bit converter with an ability to mix/match ADCs and DACs on the same set of pins - unless you really need the DACs there are better ways to go.

     

    Several people (AD and TI spring to mind) make simultaneous sampling ADCs.

    If you don't really need simultaneous sampling but can live with knowing the time offsets then using the 12 bit ADC on a micro is probably the simplest and cheapest bet. I'm very familiar with ST's range of ARM based micros which offer quite good 12 bit ADC facilities but only 2 x 12 bit DACs.  If you could let us know the resolution, sampling rate and number of channels of ADC and DAC you need I could be more specific. Using on chip ADCs will usually make it easier to control precise timing.

     

    MK

  • 0 in reply to michaelkellett

    I'd really prefer simultaneous sampling, and I don't need DtoAs.

    I think 10 bits might do, but if 12 doesn't cost a lot more I'd use 12.

    I need sampling fast enough for decent audio, so somewhere around 48ksps. I could perhaps use audio SAR digitizers (24 bit??) if they don't cause me to lose the benefit of simultaneous sampling.

    I need at least 4 channels, but would like to be able to expand to dozens.

    I'm trying to use a 3D array of microphones or accelerometers initially to find directions to noise sources and eventually to use the sound wave curvature info to determine the position of nearby sources. Lots of computation.

Reply
  • 0 in reply to michaelkellett

    I'd really prefer simultaneous sampling, and I don't need DtoAs.

    I think 10 bits might do, but if 12 doesn't cost a lot more I'd use 12.

    I need sampling fast enough for decent audio, so somewhere around 48ksps. I could perhaps use audio SAR digitizers (24 bit??) if they don't cause me to lose the benefit of simultaneous sampling.

    I need at least 4 channels, but would like to be able to expand to dozens.

    I'm trying to use a 3D array of microphones or accelerometers initially to find directions to noise sources and eventually to use the sound wave curvature info to determine the position of nearby sources. Lots of computation.

Children
  • 0 in reply to davegadgeteer

    Your cheapest route to simultaneous at chip level is probably to use an 8 channel audio chip like the Cirrus CS5368.

    There are other suitable more industrial chips which might have advantages if you build more than 8 channels (the Cirrus chips can be hard to synch across chips) you might do better with ADS8568, ADS1178, AD7768.

    To actually use these chips you'll have trouble getting the data into the processor (unless it's a DSP) - typical micros don't have the right kind of interfaces.

    To process the data for multiple chips for perhaps 32+ channels you'll need an FPGA.

    Designing data acquisition systems like this is not trivial.

    My involvement is mainly on the FPGA side in a commercial activity and a new 8 channel ADC board (starting from scratch) would represent a few man years of engineering work, at least 3 board re-spins (8 layers, fine pitch BGA et etc).

     

    So, you might consider just buying an off the shelf audio converter, - you'll need to check if an 8 channel board truly samples simultaneously - but I can assure you that it doesn't matter - the maths to compensate for non simultaneous sampling is trivial.

     

    Suppose instead you use an STM32F4xx type processor which can do 12 bit data (12 bits is not really very much with audio !) but easily 1MHz sampling rate - on 16 channels you have 16uS delay to compensate for or about 0.5mm position error in the microphones. The nice thing about doing it this way is that the basic acquisition system will cost you between £10 and £30.

     

    In summary, 12 bit non simultaneous on a micro - dead cheap.

    24 bit good audio quality on a bought in sound card or box - quite cheap (<<£500) and easy to get data into a PC for processing.

    Full welly 128 channels, 24 bit 48k sampling, sub uS channel to channel phase error,  - think in terms of £40k+ if you buy it ready made -  or a lot of work.

     

    Hope this helps - how are you going to process the data, do you need real time results ?

     

     

     

     

    MK