Shop OBEX P1 Docs P2 Docs Learn Events
FINALLY! An ADC and DAC shield worth the time... — Parallax Forums

FINALLY! An ADC and DAC shield worth the time...

Read the latest issue of Nuts and Volts and this one highlights an Analog Shield for the Arduino, but with truly interesting specs: 4 channel A/D, 16-bit, 100 ks/sec, and 4 channel D/A, 16-bit, 100 ks/s. Digilent is making them- the board was jointly developed by Stanford and TI. Schematics on Digilent show a pretty straight forward design. Uses the TI converters: ADS8343 and DAC8564.

I have not yet acquired one, as the magazine just arrived today, but it appears that this shield is controlled via SPI. Advice in the article points the interested reader to github.com/wespo/analogShield to obtain drivers with a bit more operational stability than those Digilent evidently supplies.

Point is, we now have a reasonable analog board to play with, that can be used for truly interesting and useful things (100 KHz sample rates and 16 bit resolution will find immediate use in many higher end applications), but the driver software will require a port for the prop.

Comments

  • I'm guessing this the board you're talking about?

    MFG_410-309.jpg

    I agree it it looks interesting.

    It sure has a lot going on.

    While it looks interesting, I can't think of why I need one right myself (at least not right now).

    Do you have a particular application in mind?

  • jmgjmg Posts: 15,140
    ServoM wrote: »
    FINALLY! An ADC and DAC shield worth the time... ... Uses the TI converters: ADS8343 and DAC8564.
    hmm, nice parts, but not cheap, so you really have to need 16b
    A better solution for more general ADC/DAC use, where 14b/12b is fine, may be a EFM8LB1, which has 14b ADC, and 12b DACs & is sub $1 in moderate volumes.
  • Agreed... at $50 I'm not buying one unless there's a very strong need. Yikes.
  • This response is interesting to me.

    I have built many industrial control systems that require 16 bits and reasonably fast sampling ADCs and DACs (closed loop servo control), and these have traditionally been built from PCI daq cards (~$1000-$1200 per board) running in industrialized PC systems under an embeddable RTOS. These are WAY more expensive to build than something potentially contrived using these $50 ADC cards and a prop (or two or three) and a coordinating ARM system. And, packaging, power consumption/thermal management of the latter systems make this approach a no-brainer.

    I don't get the resistance to the price point of $50, but I suppose as a hobbyist, I could understand it. But in terms of shear value, hands down, it is fire-sale in terms of cost.

    I have been looking at how to do something like this for about 3-4 years, and have seen absolutely no activity in the prop world in this direction- quite a few 12 and even 14 bits, some 16's that are SLOWWWW, but nobody tackling something in this category until now.


    It'd be even better if this board could do +/- 10v input.

  • Mark_TMark_T Posts: 1,981
    edited 2017-05-25 15:18
    For high-spec audio I2S sigma-delta converters would be better/cheaper (and 24 bit), but they don't work at DC and are two
    channel only.
  • Heater.Heater. Posts: 21,230
    ServoM,

    I'm curious. What is it that you have been measuring in industry that requires an accuracy and/or resolution 0.0015% ?
  • I'll donate an ASC+ to anyone willing to go ham on writing kickbutt Propeller drivers for this board. You must be able to demonstrate the pair of boards doing something interesting.
  • I'll donate an ASC+ to anyone willing to go ham on writing kickbutt Propeller drivers for this board. You must be able to demonstrate the pair of boards doing something interesting.

    Do you have a language requirement? I'm certainly not opposed to adding another couple chips to PropWare's HAL, and I don't have an ASC+ yet.
  • Considering that the price of single unit ADS8343+DAC8564 as chips will set you back $32, the $49 price of the evaluation board (also available at Digikey etc.) is a steal. The interface to the ADS8343 looks to be straight-ahead SPI, clocked up to 2.4MHz, with 24 to 32 clock cycles per channel conversion, 100kHz maximum sample throughput. It has the option for either internal or external conversion clock.

    I regularly use slow sigma-delta 16 bit ADCs ( ADS1115) for quasi static measurements, like thermocouples. Even that would be adequate for the temperature test example that Eric Bogatin showed at the end of his N&V article. However, the board would really show its power for faster signals, say, evaluating the stimulus response of a laser+photodiode system. Up through audio bandwidth.



  • It worried me that the article author had to search long and hard (he almost gave up) for a library that worked. As he says "Initially, I tried the library that Digilent recommended - it never worked very well and gave me hours of frustration."

    He also says"In fact, of all the posted libraries, this was the ONLY library that worked and was stable for the Uno and Due ..." (the Wespro library as mentioned in the original post).
  • Just an FYI, as I discovered this unfortunate fact today: The AnalogShield appears to have been retired.

    I have several instruments that rely upon them, so I'm interested in hearing about viable replacements. The bipolar DACs, in particular, appear to be hard to replace in such a form factor.
  • I've used the PCI daq cards for years. National instruments, Measurement computing, Advantech, etc. But now these OEM's are dropping the PCI line more and more in favor of the PXI line, USB, and Network connection.
    So for a couple of years now I've been making my own USB data acquisition using the Prop, ADC's, and Dac's. The daq cards require good signal condition or else your reading will jitter all over the place. These cards just have an ADC with a multichannel MUX and a PCI bridge of course.
  • In my industrial servo controller I use an audio ADC to read position from a resolver. The PCM1803A has 24 bits resolution and around 100dB SNR which means you have 16 usable bits without noise at 96kS/s. Fortunatelly, absolute presision is not important when reading resolvers. Everything depends on the amplitude relation of two signals. So gain and offset errors are cancelled out as long as tracking beween the two channels is good. The best thing is the stereo ADC costs only ~$1. If you need bandwidth down to zero (DC) make sure to take the 1807 instead. The 1803 has a fixed high-pass filter at ~2Hz.
Sign In or Register to comment.