strange OPA problem, latch-up?

ManAtWork

Hello,

I use a general purpose OP amplifier (TLC271) to drive the input of a frequency inverter. I have an RC network at the·non-inverting input to filter a PWM signal and a resistor divider at the·inverting input to set the gain. Supply voltage is 12V and the output should swing between 0 and 10V. Most of the time it works, but sometimes it doesn't. I have the impression that it has something to do with how the FI is grounded. But it is not reproducible. The exact same setup works one day and fails on another. In this case the output is locked at 0V while the + input is 2,5V, for example, and the - input is also 0V. The only way to get out of this state is to cycle power, then everything is working normally, again.

Is there such a thing as latch-up for CMOS OPAs? I've heared something about polarity inversal for bipolar OPAs when an input is going negative (below ground for single supply). But this is not happening.

I also tried to provoke the effect intentionally by heavily "abusing" the circuit, such as applying negative or positive (>12V) spikes to the output, switching on power of the FI and my circuit in different order. But without luck, the fault occurs only when·*it* wants and not when I do. I tried several different boards and different FIs. The possibility of occurence changes but it happens with all devices.

Can somebody explain what is happening? Or can somebody recommend a latch-up immune OPA? It has to be single supply type that tolerates 12V and can handle input and output voltages down to zero. Rail-to-rail is not necessary.

Thanks a lot


·

desiko

Hi, ManAtWork

Take a look at this application note from Microchip.

http://ww1.microchip.com/downloads/en/AppNotes/00763b.pdf

May be helpful. Good luck.

desiko

Christof Eb.

Hi ManAtWork,

have· you had a chance to look at the signals with a oscilloscope?

I am thinking of high frequency swinging of the circuit. This might be damped by a capacitor in parallel to the feedback resistor.

Good luck, Christof

ManAtWork

Hello Desiko and Christof,

thanks for the hints. Yesterday I had a chance to analyse the problem a bit further, again. It seems that it's not a classical latch up ike that described in the microchip app note. The OPA does not draw much supply current and doesn't get hot when in the latched/locked state. I checked it with the scope. The + input is at +0.4V, for example,·with neglible noise. The output is in fact oscillating but with no dicrete frequency but rather a strange series of spikes and chaotic waves with cycles in the range of several microseconds and up to 1V amplitude above and below ground. The inverting input shows the same oscillations with half the amplitude (gain is set to 2). Ground and supply are stable with no remarkable noise.

·This is really weird because

1. There is no negative supply. Any swing below GND must be undershot or the result of some "magic" charge pump effect.

2. If it was a classical oscillation the OPA should still cancel out any DC (average) offset between the inverting and non-inverting input except for the offset voltage of the OPA itself which is only several mV and not 0.4V. Even when taking clipping into account, the oscillation should be around the expected average output and not completely out of range.

Fact is, I still don't understand what is happening. But I found a cure against it. I replaced the TLC271 with a TLV2371. This is also a general purpose low power OPA but with true rail-to-rail characteristics. Now, it works without problems.

Cheers

whicker

Yep. welcome to analog electronics, where your amplifiers oscillate and your oscillators don't.

(please forgive my silliness in dragging up an old saying).