Shop OBEX P1 Docs P2 Docs Learn Events
ADC input over-voltage — Parallax Forums

ADC input over-voltage

parskoparsko Posts: 501
edited 2008-03-19 19:23 in General Discussion
Hi all,

I have a MCP3002 ADC running at 3.3V reference. My input is from an op-amp (LM334). The op-amp is running off 5V, with the output going to 3.5-3.7V max (from 0V).

Is it a problem placing a higher voltage (3.7V) into the ADC running with a Vss/Vref of 3.3V? Meaning, I have an overvoltage of up-to 0.4V. The output is max'd out at 255 (I'm only using 8 bits of the 10) most of the time.

I do know I can adjust this so I don't get this scenario using my op-amp feedback loop, but I would loose a bit of resolution, which I would like to maintain, if possible.

I'm not sure I can find (or, better yet, don't know how to interpret the correct info) from the datasheet. Is there a rule-of-thumb I can go by??? Should this be avoided? I am pretty familiar with the datasheet, having read it many (many (many)) times...

Thanks,

-Parsko

Comments

  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2008-03-17 20:56
    The datasheet's Absolute Maximum Ratings states that the analog inputs cannot go higher than Vdd/Vref + 0.6V. But also read the caveat, where it says this is a stress rating only, which does not reflect recommended operating conditions.

    You may need to switch to the MCP3004, which has separate Vdd and Vref pins. That way you can power the chip from 5V and set Vref to 3.3V.

    -Phil
  • Tracy AllenTracy Allen Posts: 6,666
    edited 2008-03-18 06:01
    One thing to watch out for in relation to overvoltage is spillover from one channel to others on the same chip. By that I mean that overvoltage applied on one channel affects not only the reading on that channel, but also on all other channels of the device, even if the current into the substrate diodes is not above the stress rating that Phil referred to. The MCP3002 is a two channel device. I am not sure if this applies to the MCP3002, and I don't know if 0.4 volts above Vdd would be a problem. That sort of thing is seldom mentioned explicitly in data sheets.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com
  • parskoparsko Posts: 501
    edited 2008-03-18 13:29
    Phil and Tracy,

    Thanks for the input!

    Phil, where, specifically, did you find the Vdd/Vreg + 0.6V spec??? I just checked the datasheet and couldn't find it anywhere. EDIT: I just looked again, and did you gather it from Figure 4.1 "Analog Input Model", as the threshhold voltage? If so, does that imply that the lower threshold is 0.6V, what does that mean??

    Tracy, interesting concept. I would have never thought that spillover could occur, of even exist. Is this a major concern? Or is it one of those concepts that can happen "theoretically", but in reality only affects a few poorly (or specifically) designed chips?

    My instinct told me that the slight over-voltage probably would not be an issue. It appears to be the case. I certainly would not have blasted it with much more (for instance 5V or higher).

    -Parsko

    Post Edited (parsko) : 3/18/2008 2:32:30 PM GMT
  • Phil Pilgrim (PhiPi)Phil Pilgrim (PhiPi) Posts: 23,514
    edited 2008-03-18 16:40
    Parsko,

    The 0.6V figure comes from the "Maximum Ratings" section (1.1) and applies to all pins, both analog and digital. The lower limit is Vss - 0.6V.

    -Phil
  • parskoparsko Posts: 501
    edited 2008-03-18 17:14
    Phil Pilgrim (PhiPi) said...
    Parsko,

    The 0.6V figure comes from the "Maximum Ratings" section (1.1) and applies to all pins, both analog and digital. The lower limit is Vss - 0.6V.

    -Phil

    ...DUH... Wow, that is obvious, isn't it! I love it when the (little) lightbulb goes off above my head. Thanks for flicking the switch!

    -Luke
  • Tracy AllenTracy Allen Posts: 6,666
    edited 2008-03-19 01:10
    Parsko,

    Spillover between channels is just something to check for in a design review. I look for it because I have seen it happen. The main thing is that the data sheets specify a range for Vin+ from Vss to Vdd, and as you know from poring through the specs, it does not elaborate much on what might happen outside those limits, except that it will peg the reading and max or min. Outside the limits, it is necessary to run your own tests and not assume nothing! Double negative for emphasis!

    Another option might be to use an op amp with rail to rail output, and run it from the 3.3 volt supply. Then there would be no problem and very little loss of resolution. The LM334 is a temperature sensor. Did you mean LM324? National and other companies make quite a few nice rail to rail output op amps that will operate at 3.3 volts.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com
  • parskoparsko Posts: 501
    edited 2008-03-19 15:07
    Tracy,

    Yes, I meant LM324. Bought it at the 'Shack. I'm having a hard time determining which are rail-to-rail components for op-amps. There are so many to choose from. Is there an "industry standard" you would recommend to use with microcontrollers?

    I flailed around trying to get a TL082 working, only to find out that it doesn't work well under the conditions I was trying. I think there are a few rules of thumb associated with Op-amps that I learned in the process. This is one reason why I'm asking questions... I'll try to post what I've learned, so others won't go through what I did.

    -Luke
  • Tracy AllenTracy Allen Posts: 6,666
    edited 2008-03-19 19:23
    The r-r io op-amp I use the most is probably the LT1490A. I have lots of others on hand though. The thing about op-amps is that there is no single one that combines all possible desirable features. That is why there are so many available. You have to become familiar with a range of op-ams that fit the projects that you do.

    Texas instruments does make a many rail to rail op amps in the TLV series. I have the TLV2464 which is 4 channel like the LM324, or the TLV2774, which I really like because of its combination of pico-amp input bias currents and relatively high bandwidth. Something in the TLV family would probably be good for your project.

    I still like the venerable LM10, an op-amp combined with a 0.2 volt reference in the same package. It was the first rail to rail, low voltage capable, op amp, and it was and is an amazing accomplishment, designed by pioneer Robert Widlar. National application note AN-211, written by Widlar himself to introduce the LM10, should be required reading.

    Other National Semi op-amps I have on hand are LM6062, LM6482 and LM6582.

    I like Linear Tech products especially because they have a high voltage bipolar process (used in the LT1490), whereas a lot of other manufacturers limit their r-r products to 6 volts or less. Linear Tech takes particular care for fault protection.

    ▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
    Tracy Allen
    www.emesystems.com
Sign In or Register to comment.