OP AMP Latch up

[morizabal]

Greetings,

I've been experiencing sparodic op amp failure (shorting the power supplies is one failure) in one of my designs. The op-amp (OPA 2277) is being used as a buffer. I was made aware that it is possible to that the inputs are getting a voltage before the power supply pins are. What is the best approach at handling this. Are there other methods besides using Schottky clamping diodes at the input and supplies?

 

[Lewish]

Is this a FET input OpAmp? If so, you need to clamp the input pins via schotky diodes to the power rails.

 

[Heydave]

Sometimes, op amps will oscillate when driving capacitive loads and appear to "latch up" loading the power supplies and getting very hot. This is most common when used at low gain such as a unity gain buffer. The frequency may appear very high compared to the signals you are normally dealing with and your test equipment may not see the oscillation. The capacitance load may be nothing more than an output cable. Note that a reflection can occure from down line due to a change in impedance at the far end. If this is the case, you may need to increase the impedance of the load by putting a resistor in series with the output and/or dampening the response of the amp. Sometimes, adding bypassing caps on the power supply pins will help as well. If it is truly a latchup, bypassing may help as well because the problem may be that the load capacitance may cause the op amp to pull the power pins below an input or feedback voltage just long enough to inject current into the chip's substrate. If this is the case, you may need to bypass with ceramic and low ESR bulk caps to get the power supply high frequency impedance, low enough to handle the dynamic load of the amp driving its load. In either case, try bypassing first with something like .1uF monolithic caps in parallel with 68uF tant caps and see if this cures the problem.
Good luck

 

[morizabal]

I am using 0.1uF ceramic capacitors on the op-amp. And there is an output cable provided with our product.

My problem right now is that the failures are random. Some of the op-amps fail immediately, some after we ship them out, and some none at all.

What would be the best approach to test and fix this problem.

 

[Lewish]

Do you have a series termination resistor between the Opamp output pin and the cable?

 

[IRstuff]

Are the inputs being driven by something that's powered by a separate and unrelated supply? If so, it's very likely that the other system may start at a different ground potential and the common-mode is being corrected and absorbed through the inputs on the op-amp.

you can certainly add stuff to the inputs to protect them, but it would seem to be a bad thing to do to a precision op amp whose claim to fame is low bias and leakage currents .

You could and should make sure that there is a solid signal ground connection between the source and the op amp circuitry. TTFN

 

[Heydave]

I would recommend output termination resistors in series with the output cable that match the impedance of the output cable. You may wish to increase gain to offset the added resistance if the cable is terminated on the far end. The amp should be bypassed with monolithic caps in parallel with some higher value, low ESR caps such as tant. On the inputs of the op amp, try to match the impedance between the two inputs. If you are using the amp as a simple unity gain buffer, try putting resistors in series with the feedback and the non-inverting input. You can get some idea of the bandwidth effect by using the input capacitances and the resistors as filter elements for analysis. The input resistor may be enough to limit power-up currents to safe levels while the feedback resistor should balance the bias currents and dampen the response that could cause oscillation.

 

[morizabal]

Do you have a series termination resistor between the Opamp output pin and the cable?

No, the output of the op-amp is directely connected to the output cable connector.


Are the inputs being driven by something that's powered by a separate and unrelated supply?

The inputs are being driven by a INA118 instrumentation amplifier which is powered by the same power circuitry as the op-amp. The output of the INA118 might be going higher before the op-amp supply pins are being driven.


I would recommend output termination resistors in series with the output cable that match the impedance of the output cable.

The only problem with this is that our customer may decide not to use our supplied cable and supply their own cable of different lenght.

I am also using 100 ohm resistor for my feedback resistor on the unity gain buffer.

Would using a signal line protector be of any benefit, I am thinking of using a device such as the MAX4505

Thanks All, M.

 

[morizabal]

Below is my current design that is giving me problems.

Below is the old design which I based my current design on.
Notice I ommited C7,R8 on my particular design. Can someone shed light on the purpose of R8,C7 on the below design. BTW the old circuit as far as I was concerned did not have the issue of latching op-amps. The reason I ommited C7 and R8 was to conserve board space since I didn't see the purpose of those components. But I may have assumed wrong.

Thanks.

 

[IRstuff]

I couldn't lay my hands on my active filter text from school, but the circuit topology looks awfully familiar.

It looks like you blew away a major portion of the active filter, as well as blowing away R10, which should be some sort of output damping resistance.

You've turned a multipole active filter into a passive RC followed by a buffer TTFN

 

[morizabal]

IRSTuff,

What would be good source to figure (on the net) out this topology. My electronics book from college didn't have that topology listed. Frankly I could probably calculate everything, but I haven't done that in a while

 

[IRstuff]

It looks very close to a Class 1B active filter; that's from a different book than the one I was looking for and it's essentially gibberish after 20+ years. The one I was looking for is Introduction to Filter Theory by ??.

As for the filter itself:

http://www.analog.com/Analog_Root/static/techSupport/designTools/interactiveTools/filter/filter.html

Anyway, it would appear that not having R10 would probably cause output oscillation.

TTFN