[SOLVED] input magnitude violation of ADC

[talking]

I have an ADC circuit in our lab, and I need to measure the ADC output by applying sine signal to ADC.
Due to some experimental purpose, the sine input signal should be higher than the ADC input range, and so the clipped sine signal needs to be measured from the ADC output.
My worry is the ADC can break down by applying high voltage input signal. How much high level signal can be applied to the ADC, to avoid this worry in general?
For example, if the ADC input is from -1V to +1v, then how much level input signal can be applied? If you have any theory of this (input signal with xx% of ADC input range can be safely applied, or etc.), it would be better.

 

[betwixt]

Up to 100% of the input range is acceptable and will not cause damage. For best accuracy you should aim for close to 100% so the proportion of voltage per step is smallest. Usually, you can go SLIGHTLY under lower voltage limit and SLIGHTLY over the high limit without damage but it is not desirable to do so and in any case never exceed more than about 0.5V either way.

If you are using a single supply ADC (0V ground, 5V VDD for example) you will have to offset the input voltage by adding a fixed positive voltage to it so the result never goes below 0V. If you do that you must also be careful that the positive peak of the signal plus the offset is still within the maximum input voltage range. For example, if you are using 5V FSD and you have a sine input of -1V to +1V (2V p-p) you should consider adding +2.5v to the input voltage so it stays well within the allowed range. With 2.5V added, the voltage will go from +1.5V to +3.5V. Depending on your signal source, adding the voltage may be as simple as using a potential divider across the supply with the center tap connected to the ADC input. If that isn't feasible, you can add it with an op-amp summing circuit.

If the peak to peak input voltage is actually higher than the ADC input range, either reduce it with a potential divider or protect the ADC input by adding a series resistor in the signal path and a pair of small signal Schottky diodes, one from ground to the ADC input and the other from the ADC input to the positive supply. Wire them cathodes toward the positive supply so they do not normally conduct. If the input goes below ground or higher than supply, they will become forward biased and dump excess curent to ground or supply to protect the ADC.

Brian.

 

[talking]

Thank you so much!

 

[talking]

Brian, may I double-check more clear meaning from some of your comments? Sorry for follow-up questions.

>Up to 100% of the input range is acceptable and will not cause damage.
As given an example, if ADC input range is from -1v to +1v, then "100% of the input range" in your comments means voltages from -2V to +2v? or -1v to +1v?

>in any case never exceed more than about 0.5V either way.
This means the input signal should be from -1.5v to +1.5v?

 

[betwixt]

By 100% I mean from the lowest voltage the ADC can measure to the highest voltage it can measure. Keeping the signal close to the full measurement range gives best accuracy. But be careful, almost all ADC can only accept positive voltages with respect to ground. Check if your ADC can handle negative voltages at it's input. If it can only handle positive voltages, you have to shift the signal so it's most negative point does not go below 0V and by adding the shift, double check it's most positive peak is still below the maximum voltage it can handle.

Brian.

 

[talking]

Thanks for additional explanation, Brian.
I think my previous description was not clear. Sorry about that. I need to little more clarify what I'm trying to do.

I need to first apply a small sine input signal to an ADC. In this state, the upper/lower peak values of the input signal is sufficiently small not to be clipped by an ADC at all. Next, as you also mentioned before, I have to shift up this small sine input, in order to purposely make an ADC to clip some values at the upper peak of the input signal. So clipping by the ADC is the purpose of my experiment, due to some philosophy. (The reason of this would be wordy, and so please just understand my situation.)
I want to shift up the input as high as possible, not to give damage to the ADC.

Here is my original question. how much can I shift it up, assuming the ADC input range is -1v ~ +1v? If possible, could you please give (approximate) value based on +/-1 range?

 

[betwixt]

It will clip at maximum ADC reading, clearly it will not be able to produce numbers higher than it's maximum. The maximum ADC reading depends on the ADC type and your circuit, sometimes it is the supply voltage, sometimes it is set by an externally supplied voltage (+Vref). To deliberately overdrive the input is dangerous though and I would not advise you to do it. The maximum voltage will be mentioned in the ADC datasheet under "Absolute maximum ratings" but these are the "point of no return" values, beyond which the ADC may be damaged. If you want clipping, a far safer solution is to use the ADC within it's normal safe range and mathematically clip the results beyond the threshold you want.

If you absolutely must overdrive the input, look at the data sheet to see the internal components around the input pin. They will give some clues as to how much current it's protection components can handle and you could limit the signal current to a value lower than would damage them. The protection components are not really designed for continuous use like that though and in any case, adding resistance in the input line could have other consequences to your measurements.

Brian.

 

[talking]

Thank you so much for following up the discussion, Brian. Let me go check the ESD circuitry in the datasheet. Actually, I expect I need ~5% of input Vpp, e.g., 100mV of 2Vpp more or less.