Oscillator question

A series RLC is purely ohmic at the LC resonant frequency. Maybe you can do something with that.

My Measurements model 59 'Megacycle Meter' is the best thing I have for that. Otherwise, an AC bridge would get my vote.

The active device in a self-limiting oscillator presents wildly varying impedances to the tank circuit. ALC helps a lot, but is more work.

Cheers

Phil Hobbs

A regular logic inverter with the inductor from output to input an a cap from input to ground pretty much always oscillates. Of course, a low Q would result in lousy phase noise but that's probably something you don't care about in this case.

If there is a chance that some of the tested coils are high Q you might want to add a resistor in front of the input, to avoid potential bzzzzt situations. Also, provide 1M or so to VCC to tie the input when no coil is inserted. Otherwise it might spew EMI.

Yay! Got one as well. Problem is, they don't make'em no mo'h.

I have to confess that I have become decadent and sometimes use a network analyzer or for stuff above 200MHz a Signalhound analyzer/gen combo. I know, I know, it always feels like to fire up the big truck to go buy a couple of rolls.

That's where the 74HCU04 or CD4007 comes in handy.

I know--I heard about them from you some years back. I now have a couple of my own, courtesy of eBay. (Well, one back end and two heads.)

Well, if you have a network analyzer, it's a natural tendency. What with all the optical stuff, my lab is failing the cat test fairly spectacularly at the moment, or I'd have one myself. I've been meaning to update the lab photos on my web site, but it hasn't been neat enough for some years now. ;)

Cheers

Phil Hobbs

It's not an official "oscillator" but a small ARM could (depending on the resonant freq) DDS a sine wave, and a dumb comparator or phase detector could measure the current. When the current is in phase with the voltage, you're at the LC resonance. Could be cute and cheap.

We're about to use an LPC1758 to software-DDS generate a sine wave at up to, maybe, 5 KHz, out its 10-bit DAC. Numbers like 50 KHz shouldn't be too hard.

At low Q there is some resonance frequency Q relation that falls on a circle. Brian Pippards "Physics of Vibration Vol 1." ? It's either here or at work.

George H.

See Equation (3), Page 2, Radiotron Designer's Handbook, Chapter 9

See the comment:

"This relationship shows that in practice there is little difference between these two frequencies. Q must, for example, be less than four to make f,, differ by 1% from f,. (equations omitted.) Q normally exceeds fifty, for which value the two frequencies differ by about one part in twenty thousand."

Main url:

I'm using an AC bridge with an SRS DSP lock-in but want to do better in terms of resolution on small changes in inductance.

Something like this with R 50x higher and a 1kHz update rate would be nice:

I guess the servoing the phase so it is resistive is what I want if I'm using an oscillator.

What's the frequency? If it's in opamp range, I can imagine a little circuit that will oscillate at exactly the LC resonance.

Getting the oscillation frequency right is the easy bit. Getting the loop gain right to sustain the oscillation without driving the circuit into clipping is a little trickier.

For a fairly well defined inductor it's easy enough. The Almost All Digital Electronics L/C Meter IIB does it for a fairly wide range of inductances.

Spehro presumably knows about that - Win Hill recommended it some years ago and several of us bought it and found it useful.

Hi, John:-

I think we'll get the best results somewhere in the high audio range to maybe 1MHz, so within op-amp range.

Best regards,

--sp

Have you seen the LDC1000 Inductive-to-digital interface from TI? It covers your frequency range you mentioned and has pretty fast response time. I used it for some inductive speed sensors. Their eval board comes with some neat software to measure inductance and impedance.

Thanks Tom, So the natural frequency is what you get if you ping it and let it respond. If you drive it then the resonance frequency may be different depending on how you measure it. Maximum amplitude or phase shift of 90 degrees.. Does one of those give the number Spehro is looking for? (I'm too lazy to do the math right now.)

George H. give the

How about this?

Assuming that Rl represents all the losses, Vb lags Va by 90 degrees at resonance, independent of the value of Rl.

So integrating and inverting Vb and driving Va makes an oscillator at the LC resonant frequency.

Yes, I've been playing with it- it's oscillator-based. Resolution on the display is pretty coarse (only 1nH) and I had to ballast the inductances up to get it to work at all (which introduces other issues). Handed it off last week to another person to do some more tests.

How does Keysight get **1pH** resolution in a 10uH inductance in the E4980A? It costs ~$20K so they can play some interesting tricks, I'm sure. Update rate is relatively slow.

Best regards,

-- Spehro Pefhany

Since most inductors have huge TCs (+100 PPM/K if you're lucky) that sort of resolution is meaningless.

Yes. You can measure voltage or current in the inductor or capacitor. I believe there are three different resonance points depending on which one you measure. In any event, the Barkhausen criterion shows the total phase shift around the loop must be zero or 360 degrees, including prop delay through the active elements. See

The phase rule overrides the amplitude criteria. You can add phase shift to an oscillator and it will adjust the frequency to bring the total phase back to zero even though it may be far from the peak amplitude.

Amplitude is not so significant as long as the loop gain exceeds 1 for startup. It will quickly hit some limit to make the gain equal to 1. The phase is zero or 360 degrees.

I don't know. He seems to be looking for a change in inductance, in which case probably any oscillator that can start might be suitable. I think a simple 74U04 Pierce with the inductor and capacitors in a pi configuration would give the best result. It would be very easy to try it in LTspice and see what the limits to Q are.

Those big tempcos are from the core material- if a non-magnetic structure with a low CTE is used then you can get very low tempco (but little inductance in a reasonable size). Ferrite cores make a decent temperature sensor.

Anyway, isn't that a bit like concluding that measuring better than a nanosecond is meaningless? ;-)