L oscillator stability

I don't quite understand as the setup is the same from one reading to the next. Even with few uH coils, it is stable. Wiring should be less of a problem with higher inductance values. Shouldn't be...?

Yes. Have you looked at the waveform with an oscilloscope? Maybe the distributed capacitance of that relatively high value is causing some problems-- you might see high-frequency hash during parts of the oscillator wavform, for example, so you could have an oscillator on top of an oscillator.

Best regards, Spehro Pefhany

That's a good point. I'm gonna check this very carefully. If this is the case, maybe the small RC I'm adding in series lowers the oscillator gain.

J. Hunter

Good Answer. I wanted to say "Parasitics," and I wanted to suggest hanging a couple of feet (1/2-1 meter) of wire off one or the other end of the DUT and wave your hands around, to see if what you actually have is a theremin. :-)

Good Luck! Rich

If it works, leave it. There's nothing wrong with empirical design. The PHDs want to wrap higher-order equations around everything, but if it works, it works. :-)

Especially if it's repatable, and especially especially if you're getting the "right" reading while poking around the circuit with your fingers. :-)

Cheers! Rich

Not really, it is simple. Th inductance is involving an area. The connecting wires are part of the enclosed area.

Rene

"J. Hunter" wrote in message news:cEmke.24531$ snipped-for-privacy@weber.videotron.net...

The way I read it, you have the inductor(s) wired in the same position as the oscillator crystal would normally sit. I was initially going to reply that it could not possibly oscillate at low frequencies due to loss of phase shift in the fixed caps. Then realised I was assuming a cap in // with the inductor. Then realised there wasn't one. Then realised series LC resonance would be needed anyway . Then realised stray C around the L's would give a number of additional oscillation modes. Then figured the inverter output impedance and input limiting diodes will also give trouble. Then gave up and simmed it :-). Surprisingly it can work all the way upto those 1H inductors but you now have an oscillator that readily insists on oscillating but for all manner of awkward reasons. AC voltage levels vary all over the place as the frequency varies and as the other guys mention, stray capacitance can have a sizeable frequency effect even with the big inductors (the circuit will "squegg"). The circuit functions but not really well enough as a measuring meter. It's far better to resonate the Ls against a known, fixed, decent sized, C value. Use the arrangement in the original AADE meter and it's clones (LM317?) and -then- feed into the counter chip for subsequent counting. Even then, there will be problems trying to get 'low Q' coils to resonate. A final solution needs more complex oscillator electronics, resulting in an even bigger headache than wading through those resonant frequency sums inside the micro :-) regards john

In message , J. Hunter writes

Your system has more than one frequency which meets the zero phase+ loop gain ^1 criteria. The oscillator has a varying gain with amplitude as it goes into limiting so once its going it will lock also oscillation builds per cycle so that the higher frequency if the gains the same will build first. All this means reduce capacitive coupling due to the self capacity on the inductance best done by reducing the termination resistance both sides on the inductance, this probably means a better oscillator. Finally You want most or nearly all the oscillation current through the inductance so it could usefully be placed in one arm of an AC bridge with a compensating capacitor and resistor in the other arm and set this into the oscillator circuit.