I have an application (a tuneable filter) that requires a variable inductor and the GIC (gyrator?) implementation had all the right parameters, so I decided to build a breadboard version to see how it performs. The breadboard version of the circuit showed some interesting behavioural problems which was a bit like opening a (small) can of worms, so I was wondering if you could share your knowledge and experience with me.

(In the actual filter, the sythesized inductor forms a parallel tuned circuit with a 1 nF capacitor so that the resonant frequency can be tuned between 5 kHz and 20 kHz. I won't show this part of the circuit because it's not very relevant to the problem.)

I started with version 1A of the circuit (see attachment). It worked pretty well with an LF347 and I was starting to feel all happy and relaxed about it, but then I decided to pluck the LF347 out of the breadboard socket and plug in a TL074 expecting to see the same performance (the specs for these two op-amps being very similar). It did the filtering OK but unfortunately it also oscillated when I tuned the filter below 10 kHz. The oscillation low level (a few 10s of mV) and it was at about 1.8 MHz, way out of the filter's normal operating frequency, so clearly it was due to some feedback loop phase lags all adding up and making the op-amps unstable. I noticed that the amplitude of the oscillation increased when I touched the inverting input of the op-amp and doing this also provoked the problem with the LF347, so I concluded that it was most likely the breadboard stray capacitance causing the problem. But still, it would be nice not to have any instability at all, so I had another think about it. This is what led to the opening of the can of worms.

I realised that theoretically there are four possible ways of implementing the filter (1A, 1B, 2A and 2B on the attached schematics), depending on where I put the capacitor and the inductance tuning pot. Versions 2A and 2B seemed more promising from a stability point of view because the capacitor forms an integrator with op-amp U2, so I rearranged the circuit. The result was a very stable filter but a far shallower peak when I tuned the circuit to the low inductance end. Version 2B produced a nice peak at the low inductance end but sadly (despite containing the integrator) was again prone to oscillations with the TL074.

So the question is: from a practical point of view, assuming a less-than-perfect layout, real op-amps and all, is there any advantage of one particular arrangment over the others? Admittedly, my breadboard prototype will have its own particular stray capacitance characteristics, but in general is there any winner amongst these circuit options?

I thought maybe some of you will have had far more experience with these GIC based sythesized inductors and would be happy to share your thoughts with me.

Thanks,
MajorMajor

 

This type of gyrator, based on a GIC, has 24 basic topologies.

I've attached an image showing them.

You are using the best known, the A1A topology.

As you have discovered, it has different properties depending on whether the capacitor is connected at the y2 or y4 position.

Most of the topologies are unstable, but you might try the A3A and B2B topologies. Some of them are DC unstable; that is, one or both opamp outputs latch up at the rail as soon as they are powered up.

Some of them are AC unstable; that is, they oscillate.

Some of them are AC unstable depending on which position the capacitor occupies, stable for one and unstable for the other.

But, try the two I mentioned above and, please, report back.