1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

How to extend the range of an ADC

Discussion in 'Electronic Projects Design/Ideas/Reviews' started by ADWSystems, Aug 5, 2017.

  1. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
  2. Nigel Goodwin

    Nigel Goodwin Super Moderator Most Helpful Member

    Joined:
    Nov 17, 2003
    Messages:
    39,218
    Likes:
    640
    Location:
    Derbyshire, UK
    Simple maths - for example if you multiply your result by 5 (to scale it) then the last digit can only be either 0 or 5. So it looks horribly messy, hence you truncate the last digit.

    The same applies to all scaling factors, some are worse than others - get your calculator out and give it a try.

    By doing the scaling in the attenuators, and using a final FSD reading of 102.3 (or any other variant of 1023), you get a perfect result.
     
  3. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    It's either really early, or I'm still not following. Where is 102.3V coming from (other than 1/10th of 1023)? What relevance does it have on the 0-25V input signal?

    I have never understood the reasonings put forth on ADC scaling to drop digits within the displayed value. I find it odd this topic only ever seems to come up on the forums. I have never had this discussion in the workplace or with any customers.

    WHy do you keep mentioning 1%? Where does the 1% come from?

    P.S. There are multibillion dollar companies that don't think 0 and 5 as LSD look horrible.
     
    Last edited: Aug 7, 2017
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. crutschow

    crutschow Well-Known Member Most Helpful Member

    Joined:
    Mar 14, 2008
    Messages:
    10,592
    Likes:
    477
    Location:
    L.A., USA Zulu -8

    Yes.
     
  6. Nigel Goodwin

    Nigel Goodwin Super Moderator Most Helpful Member

    Joined:
    Nov 17, 2003
    Messages:
    39,218
    Likes:
    640
    Location:
    Derbyshire, UK
    It's scaling the input to match the resolution - so make the maximum input to the analogue input equal to 102.3V at the input of the attenuator. This gives you a nice display, with no scaling required (other than inserting the decimal point in the correct place).

    When you input 25V the output of the ADC will be 250, you simply stick the decimal point between the 5 and the 0 to give the correct reading. No maths involved, and it gives you 0.1V resolution.

    When you use your multimeter does the last digit only display some values? - I thought not :D

    If you don't understand the basic principles it's unlikely you would be discussing it with customers, or indeed if you should be talking to customers at all?.

    Because it's difficult to manage 1% accuracy (and calibration), and my suggestion gives better than 1% accuracy (depending on the rest of the circuit). In fact for your 25V requirement it gives 0.4% accuracy.

    Pretty crappy companies then - and bear in mind that's only ONE example of poor scaling, and the most useable one - if you're scaling by other values it gets worse.

    As I said before, get your calculator out and do some sample readings - as we're talking about a 10 bit conversion, try the 13 values from 500 to 512 and apply your scaling maths to them, and see what results you get - perhaps then you'll understand the issues.
     
  7. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    Nigel.

    Here are your 500-520 values computed based on a 5V maximum and a 25V maximum.

    Count 5V 25V
    500 2.441 12.207
    501 2.446 12.231
    502 2.451 12.256
    503 2.456 12.280
    504 2.461 12.305
    505 2.466 12.329
    506 2.471 12.354
    507 2.476 12.378
    508 2.480 12.402
    509 2.485 12.427
    510 2.490 12.451
    511 2.495 12.476
    512 2.500 12.500
    513 2.505 12.524
    514 2.510 12.549
    515 2.515 12.573
    516 2.520 12.598
    517 2.524 12.622
    518 2.529 12.646
    519 2.534 12.671
    520 2.539 12.695

    I see exactly what I expect, very close to what I see on the bench (quick wiring, not neat, somewhat noisy). What are you talking about?

    How are you defining accuracy (what vs. what)?
     
  8. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    Why 20k and 20k for the unity inverting opamp? Why not 0 and 0? (I assume it has something to do with the 20k used in stage one, but don't see the relation or impact)
     
  9. ronsimpson

    ronsimpson Well-Known Member Most Helpful Member

    Joined:
    Apr 17, 2007
    Messages:
    7,253
    Likes:
    961
    Location:
    Loveland, CO USA
    0=short or a piece of wire.
    20k and 20k = gain of -1 in this case. (inverting) Not 1 but -1.
    The first amp inverts the signal and the second one also inverts. -1 X -1 = 1
    Use any two resistor of the same value. (ok not any...) but 200k or 100k or 51k.... It is a good idea to use the same parts all over the design.

    If you don like the last amp .... Do the inverting in software.
    (for 8 bits; 0=5 volts and 255 = 0 volts)
     
  10. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    Oh. I haven't a problem with the last opamp. Even understand that it is gain -1. I thought there was an opamp configuration (inverting buffer) that did not have any resistors in the feedback loop.
     
  11. ronsimpson

    ronsimpson Well-Known Member Most Helpful Member

    Joined:
    Apr 17, 2007
    Messages:
    7,253
    Likes:
    961
    Location:
    Loveland, CO USA
    You can get a gain of +1 with out resistors.
     
  12. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    Oops.
     
  13. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    Hey wait a minute. The voltage reference is -4.5V. Where I am I supposed to get that?!

    I just spent the entire afternoon putting together one huge simulation of the entire system. Unfortunately, the offset is also needed in the third stage but adding instead of subtracting. I'm not sure the project managers are going to like the design very much. Crutschow's tagline pretty much hit this on the head (inside every little problem there's a big problem trying to get out). I'm dumping this on the manager to find out how far the customer is willing to push out their limits and requirements.

    I also know see the "pile of resistors" that has been mentioned. I was only thinking of the resistors for creating the offset. I was not thinking of the resistors that form the differential opamp. Once upon a time, I ran into a differential opamp design and don't recall having to deal with "a pile of resistors". After nearly two hours of searching and review, I found out why. That circuit didn't use a differential opamp, is used an instrumentation opamp. No pile of resistors. Ok. Back to step 1.
     
    Last edited: Sep 29, 2017
  14. crutschow

    crutschow Well-Known Member Most Helpful Member

    Joined:
    Mar 14, 2008
    Messages:
    10,592
    Likes:
    477
    Location:
    L.A., USA Zulu -8
    I used -5V.
    You can get that from the negative supply and a LT1634-5 shunt reference along with a resistor, as shown here.
    upload_2017-8-7_20-31-24.png
    What does that stage do?
     
    Last edited: Aug 7, 2017
  15. Pommie

    Pommie Well-Known Member Most Helpful Member

    Joined:
    Mar 18, 2005
    Messages:
    10,016
    Likes:
    317
    Location:
    Brisbane Australia
    Is there a reason why a through hole 14 bit ADC can't be used?

    Mike.
     
  16. ADWSystems

    ADWSystems Member

    Joined:
    Mar 14, 2008
    Messages:
    711
    Likes:
    5
    Location:
    USA
    Sends control signal out to the control box of the lab equipment. I got scope locked worrying about the out-of-range reading signal from the equipment, I forgot about the control signal.
     

Share This Page