2
\$\begingroup\$

First the background.
I’ve got a customer who wants to drive the inputs of an instrument from a separate signal generator. One of the inputs I designed correctly (series resistor and diodes to the supply rails.) but the other is just straight into an opamp input. (I didn’t expect someone would want to use an external source… but also the design is ~10 years old and I didn’t know as much back then.)

So when the instrument is powered up there is no problem. But if someone was to apply an external signal and the instrument was not turned on it would overdrive the opamp input. The opamp in question is an opa227, but I hope this question can be of broader scope. The typical opamp spec sheet says you can over drive by 0.7 volts over the supply rails. I set up a little test circuit (yeah a piece of white protoboard.) and sent in square waves from my sig. gen. (1- 20 Vp-p). Now exactly what happens depends on the power supply and the rest of the circuit. (The built in opamp protection circuit can power the supply rails.) But I think the worst case will be a dead short on the supply. So that’s what I did. It then looks like the built in protection circuit is a diode to the rail with a 10-15 ohm resistor in series. And at 20 Vp-p (10 V peak) I was seeing 5.2 Vp-p (2.6 V peak) at the opamp input. I let it run for several hours this way, and though I didn’t do any exhaustive testing, the opamp still seemed to work fine afterwards. (2.6V at 140 mA ~360mW of power dissipated)

So here’s the question: Can you over drive opamp inputs above the absolute maximum rating on the spec sheet? (I’m pretty sure we have all been prototyping something and turned off the power supply with the sig. gen. still connected.)

\$\endgroup\$
  • \$\begingroup\$ Can't you recommend that the customer add some protection circuitry between the generator and your unprotected input? At least a series resistor ... \$\endgroup\$ – Dave Tweed Dec 3 '14 at 17:03
  • \$\begingroup\$ I hope this is not an application with a concern for reliability. Abs max is abs max -- if you violate it no IC manufacturer will make any guarantees about the stressed chip. \$\endgroup\$ – Null Dec 3 '14 at 17:03
  • \$\begingroup\$ @DaveTweed, Sorry that's exactly what I suggested. (1k ohm series resistor in a pomona box.) And make sure the instrument is powered. But there is no guarantee they will pay attention. (The problem started when they hooked up a magnetic field coil to a heater output circuit. (clearly labeled "Heater Output".) And somehow manged to fry the LM395 output stage. (Don't ask me how I thought that IC was bullet proof.) \$\endgroup\$ – George Herold Dec 3 '14 at 17:35
  • \$\begingroup\$ The LM395? "These devices, which act as high gain power transistors, have included on the chip, current limiting, power limiting, and thermal overload protection making them virtually impossible to destroy from any type of overload." LOL. Customers.. \$\endgroup\$ – Spehro Pefhany Dec 3 '14 at 17:50
  • \$\begingroup\$ @SpehroPefhany, Yeah they are my expensive "go to" power output stage. My only guess is that because they hooked up a big coil to the output, when the thermal or current overload shut it down the overvoltage from the coil killed it.. or repeated blows from same.. goodness knows how long they left it on for. \$\endgroup\$ – George Herold Dec 3 '14 at 18:13
3
\$\begingroup\$

"Wish I didn't know now what I didn't know then.. " (an Ohrwurm for you)

The OPA227 is a bipolar circuit so latchup is not an issue. Usually, the voltage limits are not as important as the current limits, but there is no spec on those.

The datasheet specifically says:

The inputs of the OPA277 series are protected with 1kΩ series input resistors and diode clamps. The inputs can withstand ±30V differential inputs without damage. The protection diodes will, of course, conduct current when the inputs are over-driven. This may disturb the slewing behavior of unity-gain follower applications, but will not damage the op amp

So the differential voltage is not really a concern- meaning that subtle shifts in Vos or whatever are less likely, but the current you're seeing indicates that something else (like an isolation tub for the 1K?) is conducting.

Can you measure the input current? If it's just a few mA peak I'd not be too concerned, but if it's 50-100mA I'd worry.

\$\endgroup\$
  • \$\begingroup\$ Ear worm.. thanks. Yeah the current is 140 mA with 10 V on the input and the power rails shorted. I'm confused by the 1k ohm input protection. Why am I seeing so much current? (Did I fry the input protection without knowing it?) I'll power this up with +/- 5 V rails, drive it at +/-10 and see what happens. (then try a new opamp.) \$\endgroup\$ – George Herold Dec 3 '14 at 18:01
  • \$\begingroup\$ @GeorgeHerold Nominally you have a 5V overdrive, which amounts to ~5mA input current with a 1k series resistor to maintain overvoltage limits. Something's probably not right if you're getting 140mA input current... \$\endgroup\$ – helloworld922 Dec 3 '14 at 18:41
  • \$\begingroup\$ Fie on those who don't at least provide some diagram of the input stage. I think there are back-to-back diodes across the inputs (to protect the super-beta transistors from reverse breakdown of the BE junction) and 1K resistors in series with that. But the pins may have deliberate or parasitic structures between themselves and the supply rails. How rugged they are is anybody's guess from outside the mfr, but we might guess 'not very' since the input leakage is fairly low. How about hanging a couple diodes on there and be done with it? \$\endgroup\$ – Spehro Pefhany Dec 3 '14 at 18:59
  • \$\begingroup\$ Thanks Spehro, Yeah just forget it. I've wasted too much time over driving opamps today. Adding the +/- 5V supply didn't (significantly) change anything. I still get lots of current. For future boards I can easily add a 1k series resistor, and rest easy tonight. (diodes would mean a board redo) \$\endgroup\$ – George Herold Dec 3 '14 at 19:46
  • \$\begingroup\$ @helloworld922, Well after too much futzing around I can't make heads or tails of it. It appears that the 1k series resistors are after the diodes to supply rail, or there are no diodes to the rails??? (Weird though because in over-voltage I don't see any current till it's ~0.5V higher..just what you'd expect with diodes.) I could ask TI but it's not worth the time. \$\endgroup\$ – George Herold Dec 3 '14 at 19:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.