tuning an LC

"Klaus Kragelund" John Larkin wrote: John Fields

Electronic organs made in the early 1960s ( with Ge PNP transistors ) had 12 master oscillators providing the tempered scale. Each oscillator used a ferrite pot core with adjustable slugs followed by multiple stages of octave division using flip flops.

The inductors were tapped and polystyrene caps set the 12 frequencies, which be from about 8kHz to 16 kHz.

Even after 10 years or more use, the master oscillators would require little or no retuning.

Absolute accuracy was about 0.2% ( +/-0.5 Hz in 440Hz ) and temp drift over the room temp range even less.

This kind of set up ruled until the late 1970s when single "master oscillator" ICs arrived that provided all 12 frequencies derived from a crystal reference.

.... Phil

Well I figured the input signal... a bit harder if it's not a BP filter.

Getting enough C in a variable cap is hard. The piston ones are very nice, but low C. The squishy mylar/ mica compression ones are.. 'squishy' (I t hink that's the technical term) There are some big rotating plate things b y (Murata??.. I'd have to go search again.) Expensive and I've never playe d with them... As I recall the maximum C is less than 1nF.

George H.

Well someone already mentioned this, but have you thought about a HV opamp.

You'd have to add some 'power booster' to get the current drive. But at least the inductor goes away.

The customer specified the L and the resonsnt frequency. I'm guessing that some modeat tolerance on L would be OK. I'd rather not ask them, lest it become an issue.

Again, I'm not sure, but something around 50 is the ballpark.

The current is forced by the voltage (60 v p-p) and the reactance. If you add enough R to limit the current, you'd get a very low Q.

A slug-tuned pot core would work (some are very big) but calculating/buying/winding/assembling/mounting would be a hassle. I can make a

Indeed. Often, the oscillators were tube, and the dividers Ge: saving cost where they can, using better when they have to.

Tim

What is the end object of this exercise? Just a 60V, 107kHz sine wave? What loading? ...Jim Thompson

Do inductors drift much over time? TCs are generally positive, numbers like +100 PPM or so.

We stock several tiny surface-mount variable caps. The Muratas are around 50 cents each, but are in the 10 pF range.

If I step up 4:1 maybe, I'd get a 16x c multiplier. So I'd need a 150 volt, several hundred pF variable cap. They exist, but would cost ballpark $15.

We have so many parts in stock that the shelves are getting crowded, so we tey to avoid creating new stock numbers if we can use what's already there.

How much power do you think to put through the circuit?

Please remember that the power going back-and-forth in the resonant circuit is Q times the power through.

The dimensioning starts to feel like a LF transmitter with some quite heavy Litz wire in the coil with a sturdy core.

As noted, the LC runs at 60 volts p-p. That defines the current in the L and in the C.

Please remember that the current in an inductor is E/Z.

The L can be a small surface-mount Coilcraft part. Got samples yesterday.

The currents are pretty high, 20% of 0.5A is 100mA, but maybe consider a trimmable capacitance multiplier?

The switches and caps solution is very low risk and simple.. maybe coarse + fine adjust. Set the fine to midscale, tune the coarse, then the fine.

You could buy better 2% inductors, if cost isn't too important:-

Still need +/-600pF of adjustment range.

Best regards,

--sp

Those Ls look good; I'll contact them. Thanks.

If I use the 20% Coilcraft parts, I'd probably use one plugin cap, like the Wima radials, one of four possible values, and span the rest with a rotary hex switch

I asked Coilcraft if they would select better tolerance parts, but no answer yet. I'll try MPS.

You did not specify the load impedance, especially its real part. You also left out whether this is a series or parallel resonent circuit.

If you run a parallel resonant circuit with pre-defined voltage, you are constricting the Q. A pure voltage source is a short-circuit as a load impedance.

Tha's right, but you Q is lost.

A trick used years and years ago: BEVEL the inside part of the cup core. Rotating it changes the inductance by changing the effective air gap.

In early modem days I used pot cores that had an adjustable gap (threaded slug). ...Jim Thompson

I had to deal with similar problems. This is what actually works:

1. Sort out all inductors by L. Keep stock of sets of capacitors to make capacitance in desired range. Put right combination of capacitors on the boards.
2. Modify your schematics so it won't have to be resonant.
3. Make (semi)custom inductors with tighter tolerance.

Those variants are trivial and they get the job done. No need to be fancy.

Vladimir Vassilevsky DSP and Mixed Signal Designs

No, all you care about is the overall capacitance. A Y5V cap will swing as much as a hyperabrupt varactor, so say 470 nF can drop down below 50 nF at high voltage.

The required cap is nominally 1/((2 pi 107kHz)**2 * 75 uH) = 29.5 nF.

Say we start with 36 nF in series with a couple of 2200 nF Y5V caps. 36 nF in series with 1.1 uF is 34.9 nF, and 33 nF in series with 100 nF is

However, if John doesn't want to have to supervise the frequency with the MCU, probably the switch and binary-weighted caps approach is a good one.

Cheers

Phil Hobbs

The tank circuit itself is the load.

I suppose if you connect an L to a C, it's both series and parallel. But it's more "parallel" in this situation.

I described the drive in another post: +30 DC on one end of the tank. A mosfet grounds the other end through a 100 ohm resistor. Gate drive is pulses, 107 KHz, 25% duty cycle. So you get about 60 volts p-p across the LC (a little more, in fact!)

???