A/D input voltage limit?

First, apologies for a late response.  To your first point on the recommended amplifiers - Most of the amplifiers also have an option of a 3.3V supply as well, which will limit the signal swing at the outputs to around 2.5V or less.  The max rating on the ADC input pins is about 2.1V which is more of a DC type input meaning the part could potentially accept a non-continuous signal of slightly greater amplitude.  However, there has not been any test data collected under that type of condition because there can be a number of different types of signals.  A DC signal is much less complex to evaluate, test, and specify.  Even if you were to use the parts on a 5V supply with a potentially higher output swing from the amplifier, this should really only be an issue if you were providing a large signal to the input of the amplifier to cause the output swing to go higher.  If there isn't a large signal swing at the inputs, you shouldn't see one at the outputs and therefore have no problems with operation from the ADC side.  There isn't any "protection" circuitry for an overvoltage condition on the inputs other than the ESD protection on the input pins.  Given the statements above, so long as the input signal to the amplifier is not large and you do not drive the amplifier to its maximum signal swing, there should not be an issue from the ADC.  Hope this helps to answer your questions.

Hmm, not really.  That's an awfully hand-wavy way to design an A/D input circuit.  The reason I posted this question is that it's always a possiblity of overdriving the input amplifier, during transient, power up, fault conditions, or even normal operation such as the user turning the gain too high.

I wish Analog Devices would provide specs on the actual max signal swing of the drivers for various supplies and loads.  Op amp specs often have this, but not these differential A/D drivers.  Without this, all I have is "around 2.5 V", which still exceeds the A/D specs.  I may have low frequencies or transients which saturate the driver output for a few milliseconds.  That seems close enough to DC to cause damage.

I think one recommendation if you are concerned with these transient type overvoltage conditions is to use some diodes on the inputs to clamp any voltages that go above the maximum ratings of the ADC analog inputs.  If repeated overvoltage conditions do occur, over time there will be damage to the ADC analog inputs.  This would be the best way to try and ensure that the maximum voltage levels specified in the data sheet are not exceeded.

Thank you.  I have designed a TI diff amp in with rail-to-rail outputs and drive levels shown in the data sheet.  That in combination with an anti-alias filter with known attenuation will guarantee that I won't exceed the A/D max input.

I'd like to address a few points you've raised in the postings here and offer some recommendations.

Yes overdriving an ADC front end is something the designer always needs to pay attention to, external clamping circuitry can degrade performance, but there are other options. I hope that the below posted thoughts may help.

1. The ADA4937 as you mentioned, actually has a min supply voltage of 3.3V and not 5V as you stated and so would help reduce a potential overdrive condition.
   
2. Our new ADA4930-1/2 is designed to drive 1.8V ADC's and would be a good choice for you, it operates off 3.3V.
   
3. Our AD8036/37 is targeted at driving ADCs, and provides the user the ability to set high and low voltage limits at the output via a Vhl and Vll pins where the designer sets the output clamp levels via an external reference and has only a 1ns of over drive recovery, something you wont get when a RRO saturates. Its a very nice "clamp amplifier or "clamplifier"
   
4. Very few of our diff amps have RRO, there is a built in headroom at the output, (this provides better distortion performance) this also provides some built in limiting, this along with power supply selection will help reduce output swing in an input over drive condition.
   
5. Our amplifiers do not produce any kind of overdrive or spiking at power up.
   
6. It sounds like you might be using the rail to rail feature of the diff amp to limit the output swing, this can introduce harmonic distortion and recovery problems when the transistors are coming out of saturation, typically RRO also have lower distortion performance.
   

I hope that you might find some of these suggestions helpful or at least thought provoking.

Jack Ardizzoni

High Speed Amplifiers

Product Manager

Thank you for the excellent and complete reply, Jack.  Unfortunately that design has already gone to layout so there is little chance of incorporating your suggestions at this point.  I will keep them in mind for the future however.

Which brings up a sore point: it takes a really long time for Analog Devices tech support to respond.  Since starting the system design I'm on now, I've had about five questions which I've posed to this forum and directly to tech support.  In every case it's typically a week before I get any response at all.  It's been a month since I first posted this question, which I also sent directly to tech support email.  I still haven't had a response to that.

You guys must be overworked. It's leading to frustration and design losses out here though.  Analog Devices is a great company with great products, but your tech support makes it hard sometimes to design you in.  Forward this on to your manager, and hound him till he doubles the size of your tech support group.  :-)

Gerrit

Hi Gerrit-

Thanks for your feedback and being an Analog Devices customer.  I apologize for the delays that you have experienced.  We strive to provide timely technical support on questions that you pose on EngineerZone or to our direct technical support.  You mentioned that you are still waiting to hear from technical support on an issue you entered.  If you contact me via private message with some more details on this, I can follow up on it for you.  Just click on my username to access my profile and then click the "Send Private Message" link.

Thanks,

AndyR

EngineerZone Community Manager