Hi,

I have a task to design an input protection circuit for a differential amplifier. Requirements are that the signal can only come in the range of -2 to 2V, everything beyond that must be somehow pulled down to that range. Another requirement is that I need to achieve the minimal possible signal distortion ( I have a sensitive 50 ohm terminated voltage spike-looking (2-6 us duration) signal that I am digitizing after the analog frontend.) Usually I would use something like a regular zener diode, but this time I need something more sophisticated (considering dual rail requirement). Any ideas or suggestions? Thank you in advance.

 

Is it the differential signal than cannot exceed ±2V, or the common mode signal?What is the frequency of the signal?

 

Make +1.4V a -1.4V power sources and clip the excessive signal to them using fast diodes (low capacitance).

 

Is it the differential signal than cannot exceed ±2V, or the common mode signal?What is the frequency of the signal?

It is a commond mode signal that I have to account in -2V to +2V with reference to ground voltage range. There is no specific frequency as I am digitizing the pulses. Pulse width is around 2-6 us, am I expecting anywhere from 0 to 50 000 of these pulses in a second

 

Establish a couple of references at +1.4V and -1.4V using two diodes to set the reference, then a resistor to the positive and negative supplies.
They can be any sort of diode: 1N4001 etc.
Then clamp each signal to the reference using two fast diodes. It will clamp at about 3x V diode which should be near enough 2V.
If the signal has a limited amount of energy you could use 1N4148. If it can cause more damage use a UF4001, or a Schottky, but a Schottky in conjunction with two normal diodes would clamp at about 1.5V.

 

One nice tiny but powerful option is TL431 (reciproke Zener)

 

The required circuit would be called a "Diode Clamp"circuit which all of the design considerations should be in a good engineering reference book. As stated, it uses diodes arranged so that as the voltage passes some limit the diodes start to conduct. So you also need to have some series resistance in the signal line prior to the diodes, to limit the current. For a differential input you would need four diodes, two for each line.
Similar arrangements are used to protect balanced microphone inputs on some PA amplifiers, if you want to see similar examples of the protection scheme.

 

What is the maximum signal frequency and/or rise-time?

 

Establish a couple of references at +1.4V and -1.4V using two diodes to set the reference, then a resistor to the positive and negative supplies.
They can be any sort of diode: 1N4001 etc.
Then clamp each signal to the reference using two fast diodes. It will clamp at about 3x V diode which should be near enough 2V.
If the signal has a limited amount of energy you could use 1N4148. If it can cause more damage use a UF4001, or a Schottky, but a Schottky in conjunction with two normal diodes would clamp at about 1.5V.

May I ask you to provide a circuit diagram of your mentioned protection? (very simple one drawn by hand or paint, just want to make sure I get your idea right.) Typically diodes introduce non linearity to the signal, and the signal shape itself is the most important thing to preserve here. Wont the diodes cause signal distortion? Thank you in advance.

 

May I ask you to provide a circuit diagram of your mentioned protection? (very simple one drawn by hand or paint, just want to make sure I get your idea right.) Typically diodes introduce non linearity to the signal, and the signal shape itself is the most important thing to preserve here. Wont the diodes cause signal distortion? Thank you in advance.

I’ve drawn the fast diodes as Schottkies even though they are probably not, so you know which ones have to be fast and which ones don’t.

 

I’ve drawn the fast diodes as Schottkies even though they are probably not, so you know which ones have to be fast and which ones don’t.

thank you very much, I will test this circuit in real conditions.

 

Hi,

I have a task to design an input protection circuit for a differential amplifier. Requirements are that the signal can only come in the range of -2 to 2V, everything beyond that must be somehow pulled down to that range. Another requirement is that I need to achieve the minimal possible signal distortion ( I have a sensitive 50 ohm terminated voltage spike-looking (2-6 us duration) signal that I am digitizing after the analog frontend.) Usually I would use something like a regular zener diode, but this time I need something more sophisticated (considering dual rail requirement). Any ideas or suggestions? Thank you in advance.

clamp circuit regardless if its zener with anodes tied back to back or clamping circuit to a regulated voltage rail will work, but will chop the top of the signal.

 

This is a schematic of a clamping circuit for balanced/differential signal with regulated +/- 2V supply derived from a voltage dividers W/ zener. diodes from a +/- 15VDC supply. Input resistors should be used to prevent overcurrent conditions with large signal voltages. In this schematic, I'm applying 8V of signal. 50 Kc pulse should be fine with standard PIN diodes like 1N914A since its below 200 Kc

 

Why do you want to protect the inputs? Are you protecting the diff. amp inputs or are you protecting the inputs of the next IC? If you want to limit the output of the amp, then that is probably a different circuit.
How big is the input voltage you want to clamp? Can you get a 50-volt spike on the input? or is it more like 5V?
Please provide the schematic of the input amplifier you are using. There is a good chance it has input protection inside the IC.

 

Requirements are that the signal can only come in the range of -2 to 2V, everything beyond that must be somehow pulled down to that range. Another requirement is that I need to achieve the minimal possible signal distortion

If you try to preserve the signal peaks, like you would do with an audio signal, you would use a look ahead limiter circuit.

 

Why do you want to protect the inputs? Are you protecting the diff. amp inputs or are you protecting the inputs of the next IC? If you want to limit the output of the amp, then that is probably a different circuit.
How big is the input voltage you want to clamp? Can you get a 50-volt spike on the input? or is it more like 5V?
Please provide the schematic of the input amplifier you are using. There is a good chance it has input protection inside the IC.

That circuit I posted is a standard ESD protection circuit used in balanced audio that ties to the +/- supply because the standard compensates are rail to rail input instrumentation amp or op amp. clamping of the signal into an ADC is optional, but is commonly used in instrumentation applications. Of course, they have specialized instrumentation amps like the INA 851 that will clamp the output to the lower voltage rails of ADC ic.

 

This is my exact application. This IC converts the single ended signal to a differential one and feeds it to the ADC. My ADC can only take 0-4V with reference to ground. So I am expecting to limit my single ended signal to -2/+2V with reference to ground, apply a 2V common mode voltage to the LMH6550 and feed that differential signal to the ADC.

The only relevant part is what happens before the LMH6550. The diff amp will have +-5V supply, but I want to limit the signal itself to +-2V to prevent distortion.

In other words, all the diff amp and ADC stuff is not really important, I just want to preserve the signal content as much as possible in the +-2V range and clip everything thats beyond. The single ended signal will be coming from another OPAMP, so it wont carry very large energy or current source capability, but during transients and other stuff, I expect my signal to shoot up to +- 8 or 9V lets say, and at normal large signal conditions, maybe +-3 or 4V

 

Is there a difference between limiting the input OR limiting the ADC input to+ 4V?
What I am thinking is that as the diodes start to conduct there will be some distortion. By limiting after the input amp the signal is 2x higher and the diode circuit will have 1/2 the diode problems.

 

Is there a difference between limiting the input OR limiting the ADC input to+ 4V?
What I am thinking is that as the diodes start to conduct there will be some distortion. By limiting after the input amp the signal is 2x higher and the diode circuit will have 1/2 the diode problems.

I can not put limiting circuit after my diff amp or just supply the diff amp with 0-4V supply as it would prevent the ADC from overvoltage but clip all the signal that is in the 0 to -2V range with reference to ground. Please ignore the resistor values in my last picture, this is a picture from a datasheet, I will make the unity gain of the diff amp. So regular diode clamp may not be the best solution preserving the signal form. Are there any more advanced approaches? What about tunnel diodes or other simpler forms of protection that does not distort the signal? Maybe I can use additional high speed opamp buffer with +2 to -2V rails? I am not very experienced in the analog design so I am hoping for any advices. Thank you once again.

 

I can not put limiting circuit after my diff amp or just supply the diff amp with 0-4V supply as it would prevent the ADC from overvoltage but clip all the signal that is in the 0 to -2V range with reference to ground. Please ignore the resistor values in my last picture, this is a picture from a datasheet, I will make the unity gain of the diff amp. So regular diode clamp may not be the best solution preserving the signal form. Are there any more advanced approaches? What about tunnel diodes or other simpler forms of protection that does not distort the signal? Maybe I can use additional high speed opamp buffer with +2 to -2V rails? I am not very experienced in the analog design so I am hoping for any advices. Thank you once again.

The whole point of clamps is to prevent damage to input circuitry if the voltages get too high. By then linearity is the least of your concerns.

If you really want to eliminate common-mode voltages, then you need a transformer.