will this circuit fry the shottky diodes?

[danjel]

I recently had a pcb made with the following input path to one of the ADC pins of an atmega168:
(see attached)

Input jack (with normal to gnd)-->inverting opamp(gain 0.5)--->inverting opamp (gain 1)--->shottky diodes to +5V and Gnd---->ADC pin.

I totally forgot to put a current limiting resistor before the shottky diodes but I am wondering if it is needed?

The expected input is 0-10V (which gets reduced to 0-5v)

The max overvoltage possible (accidental at input) is 15V (so 2.5V over at the shottkys).

Do I need to hack my pcb to add something like a 4.7K resistor in the before the shottky or can I leave it as is?
I am worried about damaging the AVR ADC input (which afaik has an internal input resistor)
I guess also ruin the opamp/shottky if there is an overvoltage and nothing to limit the current flowing through them?

 

[MikeMl]

What is the TL072 powered from? The biggest problem I can see is the following: Suppose the output of the TL072 tries to go to 10V. Without a current limiting resistor between the output and the diodes, the amplifier is going to try to force current into the diode connected to +5V, which will likely force the 5V higher. How high? There is no max current output specified on the TL072 data sheet, but I'm guessing ~20mA, so that may force the 5v Vdd supply up...

I vote for a current limiting resistor. Why didn't you just use a single rail-to-rail amplifier powered from 5V and GND; that would have precluded this problem.

 

[danjel]

The TL072 is powered from +/-12V so it would probably result in the problem you described.

Using a railsafe unipolar opamp powered from +5V for this stage probably would have been a smarter thing to do :S

 

[danjel]

also if I had used a rail-to-rail opamp powered from 5V I would not have got a true 0-5V range since (afaik) the actual output range is always less. Usually rail-to-rail opamps allow you to go right to 0V but they don't let you go right to the power rail voltage (just closer than a normal op amp)

 

[MikeMl]

also if I had used a rail-to-rail opamp powered from 5V I would not have got a true 0-5V range since (afaik) the actual output range is always less. Usually rail-to-rail opamps allow you to go right to 0V but they don't let you go right to the power rail voltage (just closer than a normal op amp)

I have used 5V CMOS opamps that will pull to 4.95V and 0.01V. You can always scale your true signal so that it ranges from say 100mV to 4.9V, and take care of that in software downstream of the AD.

 

[danjel]

The opamp is just a plain tl072 powered by +/-12V.
The shottky diodes are bat54STA.

I looked at the datasheets for the schottky and it can handle a surge current of 600mA and a max continuous current of 200A (which seems wrong, must be 200mA).

Also the TL072 does have built in SC protection to either supply rail and it can be held in SC indefinitely. I am pretty sure the max current it can supply is 20mA as well.

So given all this I would guess that my design is actually safe from harm then? i.e. no need to mod but obviously change the design in future revisions.

 

[MrAl]

Hi,


I wouldnt worry about those Schottky diodes as much as the ones on the uC chip. If that particular uC chip has protective clamp diodes, those might conduct before the external ones do and thus might have to take the bulk of the current. It's much better to use a current limiting resistor from the output of the second op amp stage to the diodes. You'll also want to double check the max output current when the output goes negative as that will be worst case power dissipation, which will be 12 times whatever current it can output. If the power dissipation for the op amp chip goes too high it will burn up.
Another added advantage to using a current limiting resistor is that it will isolate the output of the op amp from the clamps, which means the op amp will stay in the linear mode at all times. With the clamp diodes connected directly to the output the output will not always have a chance to keep the op amp in the linear mode and that could cause some unusual problems too, so a well chosen current limit resistor is very wise.

 

[danjel]

The uC clamping diodes can handle about 1mA max so anything more than this must go through the external schottky clamp diodes.
I don't think it is possible for the OpAmp to burn out if it has SC protection but I have to see what the max power dissipation is given a +/-12v rail.

If I do put the current limiting resistor in I think it would have to be around 2.2-4.7k range.