Forums » General Electronics Discussion

This circuit is supposed to level shift the ouput of a band gap temperature sensor which produces 10 mv/°K. I need to produce a 0 to 60 mv ramp from -20°C to +40°C. When I step the RF input from 0v to 0.1v it works as expected, but with 0.2v or greater on RF it just oscillates at about 100KHz. Is this actually the opamp performance or is your simulator doing something wierd.

https://www.circuitlab.com/circuit/k9qeu9/ab1/

Op amps can get unstable when you run them at low gain.

Try putting a 1000pF capacitor across the output to input resistor.

No help. I added 3 1uf caps to the output and both inputs. The oscillation frequency dropped to 10 KHz.

I then tried adding a positive feedback resistor, as you would with a comparator, and stepping it's value from 100M down to 1M. It xtill oscillates when you step RF up from 100 mv to 200 mv. I need it to be 1000 mv.

Help!?!

You are loading the opamp output with a large capacitor. This can make the amp unstable.

You might like to have a look at:

Don't put a capacitor, ever, from an op-amp input or output to ground. Nothing good ever happens that way.

Put a 1000pF capacitor from op-amp output to the minus op-amp input. that will roll off the response and probably stop oscillations.

arduinohacker, you confused me with your first response. That's when I added the caps. I took them out and it works now.

The oscillations are caused by the startup conditions. At t=0 I was trying to put 0.8 v dc on one input and 0.8v dc PLUS the triangular waveform on the other.

Unfortunately, the triaangle was starting from a phase angle of 0°! That gave me an initial condition of 1.0 v on the second input which the opamp regarded as a VHF pulse, so it oscillated!

I changed the startup phase angle of the triangle to 90° and it starts cleanly without confusion.

Thanks a lot arduinohacker and signality.

BTW signality, that link to analog's in loop compensation is dead. Do you know where I can find a copy of the original article.

@Paul_L, the link:

http://www.analog.com/library/analogDialogue/archives/31-2/appleng.html

works for me.

There's also this one:

http://www.analog.com/library/analogDialogue/cd/vol31n2.pdf

where you'll find the article on p19.

Note that if the appearance of your oscillation depends on your initial conditions then you have a conditionally stable system.

That means it's conditionally unstable too.

That's bad.

Follow the advice in the apps note and check your sim with a number of different (shorter rather than longer) step times

:)

Signality, thanks. The first link is dead but the second brings up the complete issue of Analog Dialogue.

This entire approach to stability refers to driving capacitive loads with op amps not designed for stability under these conditions.

The instability I saw was caused by the step differential voltage applied by the simulation. The real world condition for this circuit will have a band gap temperature sensor and a reference voltage applied to the differential inputs. In continuous operation there will be no sudden steps. Normal temperature changes will produce very slowly changing voltages.

The only chance of step induced instability will be during power up, and, I must confess, I have little idea how to observe this. I suspect that it will be stable if the op amp, the reference divider and the temperature sensor are all powered by the same regulated power supply, but that will have to be checked in the actual hardware.

Thanks again.

Paul_L