My input voltage may exceed the supplies. Is there a way to protect the inputs of my AD8250?
When the voltages at the input of many of our inamps, the AD8250 included, exceeds the supply, the ESD diodes become forward biased and begin to draw current. This can feature can be used to clamp your input voltages, and protect the inamp from damage.
The ESD diodes can only sustain about 5 mA of current, so you will need to put a resistor in series with the input to make this work for you properly. For example:
You have a supply voltage of 15 V, and may have input voltages as high as 30 V. You would need 3kOhm series resistor because with a 15V drop (30V input - 15V supply) across the 3k protection resistor gives the max sustainable current of 5 mA. If the resistor were any smaller, you would be putting too much current through the ESD diode.
I always had this doubt in mind !
I remember we used to say diode reverse saturation current doubles around every 10 deg C rise in Temperature.
Most of the apllication time I think these diodes will be reverse biased.
I wonder how this affects the working ! especially when we have precise data converter after the INAMP.
20 uA current rise on overall temp range and a huge 3K current limiting resistance will make a 0.06 V !
With ADC it is always a danger , usually an INAMP will have 12 V supply and ADC will have 5 V.
So wrong input or inband stray input can cause INAMP to swing to positive rail.
Some INAMPs at power ON do have a tendency to Swing to rail for short duration.
We usually put similar diodes externally to Single ended ADC and a resistor or a resistor of Antialiasing filter to protect the diodes and forward currents.
Don't you think because of such a huge resistor, we will never make ADC and overall signal processing system ingenral, truly above 16 bit?
Diode ingenral is temperature sensitive,many compensation and other techniques use this property of the diode.
Is there any other way to protect the devices easily without giving Gain and offset errors and drifts?
You are correct in pointing out that
1. Diode reverse leakage doubles every 10 degrees
2. This leakage creates temperature-dependent errors due to protection resistors
However, I believe Paul's comment refers to the protection suggested at the input of instrumentation amplifiers, leveraging the IC's internal ESD diodes. In this case, their leakage will be part of the input bias current spec over temperature, which is always specified in precision amplifiers. Then you have the advantage of being able to protect with single resistors and having a guarantee for the maximum input bias expected over temperature.
In general, there's always a tradeoff between protection level and performance. Finding the correct balance between protection level, system performance and complexity (and ultimately cost!) is part of the art of good analog design.
In the example provided by Paul, 3kOhm on each input will protect the amplifier only up to 30V with a 15V supply. During normal operation and if the protection is applied to both inputs, the offset current variation over the entire temeperature range won't exceed 30nA. In this case 30nA x 3kOhm = 90uV or in average you will get 90uV/125C = 0.72uV/C.
If you wish to increase the protection level using this method, you will certainly further degrade the precision of the system. As long as it is within your error budget to hit the target specs, you should be OK. If you need that level of protection but the performance hit does not fit your budget, you need to use a different protection scheme, one that will be more complex. You can start by adding external diodes, for example.
In order to avoid generalizations, I'd like to mention a few details that might be helpful.
Not all diodes are created equal. There is a large variety of processes that focus on optimizing different diode parameters, to make them suitable for just as large variety of applications. If you look at diodes such as BAV199, the leakage at high temperatures is typically very low (almost 3 orders of magnitude smaller than 20uA within the temp range of concern) compared to a schottky diode (especially power rectifiers) or even a more traditional general purpose diode such as 1n4148. The trade-off is that low leakage diodes also have a higher forward drop, which you have to take into account when designing a protection circuit. Additionaly, notice that the leakage is dependent on other factors such as reverse voltage. This means that you have a couple parameters that you could tweak to achieve the desired performance.
For your particular example, where you're worried about overdriving the ADC, 3k will be typically too high to be practical for several reasons that I'm going to skip for the sake of keeping this post short. All the situations you mention are very common and valid concerns. In this case, you have a few options: you can clamp with a low-leakage solution that can withstand the overdrive current with a reasonably small resistor or you can add an ADC driver that could take care of never overdriving the input. You may find that the follwing circuit note explains this in more detail:
I hope this helps,
Retrieving data ...