Variable Inductor using resistor

I came up with this idea a few months ago, but simulations were not conclusive so I left it for later.. I returned to it when I had problems sourcing some variable inductors I needed, and have just prototyped an oscillator using this scheme -

In theory, I think this scheme allows any variable inductance to be created - I have only done one small prototype, and was able to achieve what I needed - I intend to do more tests over the next few days, and would welcome any input from people with better understanding and/or more expierience with inductors.

link to image (http://www.therasynth.com/assets/images/resistive_var_inductor.JPG)

[R1 = 15k, VR1 = 50k .. R1 should not be <1k ever!]
Bifilar windings (http://books.google.com/books?id=HQCr2wBclOMC&pg=PA216&lpg=PA216&dq=bifilar+winding&source=bl&ots=_xlddTcw3m&sig=Lmza94dkIxdVa-9SzLos2_NT8Tk&hl=en&ei=hdjBSY2uOKTJjAeNoe2WCw&sa=X&oi=book_result&resnum=9&ct=result#PPA216,M1) are used because they seem to work better - at first I tried a common mode choke - it worked, but not well over a large range. There is some loss of power due to dissipation (microwatts) through the resistors - and I think that a common mode choke has greater losses from eddy currents - the bifilar winding scheme, I think, cancells the field at (or near to) its origin, leading to lower eddy losses.

I used a small ferrite torroid core (Coated ring core, 10.8mm dia 4.75mm H.. RS 647-9907 EPCOS B64290L38X830)
and 30SWG En Cu wire, and had a bifilar winding right 'round the core (no overlap).. this gave me 1.3mH per winding, and I was able, with the circuit above (using wider range of resistance values), to achieve adjustment from 100uH to 1mH.. The values in the circuit above allow the same adjustment as a typical 'red' IF transformer (About 250uH to 480uH) and I see no reason why this should not work in EW oscillators.
The resistors should be at the least sensitive side of the circuit (the +Ve end in the EW).
I will post more details and photos ASAP.

Alas, I doubt that this scheme will work for equalizing coils.. Best performance is when the 'feedback' current is lowest ('feedback' resistance is highest).

Beware - If the current through both windings is equal, there will be no inductance, and only the wire resistance to limit current - you could easily blow up your circuit!

Fred, just my spontaneous observations:

a) The current through both windings will never be equal since one has the resistor in series which attenuates and de-phases the current.

b) It seems like there is the second inductance partially added (or subtracted) via the resistor. That makes me think that the resulting inductance will be frequency depending because of the high pass effect of the R - L2 combination.

So I guess that this may work in a small specified frequency range where R has to be chosen in the same range as 2*pi*f*L2.

Thanks for these observations, Thierry.

[i]>>"a) The current through both windings will never be equal since one has the resistor in series which attenuates and de-phases the current."[/i]

True - That is the reason for R1.. The warning was because reduction of R1 can cause the total inductance seen by a circuit connected to this "variable inductance" to be extremely low.

Wirewound resistors are constructed using bifilar windings - the inductance of the windings are cancelled out, leaving only the wire resistance.. Reducing R1+VR1 to 0 effectively turns the circuit above into a wirewound resistor. (well - not exactly.. a WWR would be constructed by connecting 1B to 2B, the resistance would be seen across 1A to 2A, and the inductance from 1A to 2A would be close to 0.. All other components / connections removed)

[i]>>"b) "It seems like there is the second inductance partially added (or subtracted) via the resistor. That makes me think that the resulting inductance will be frequency depending because of the high pass effect of the R - L2 combination."[/i]

I have not done much testing, and cannot verify practically or theoretically what you say here - but it makes sense.. The amplitude of the oscillator waveform does change significantly as the value of VR1 is changed.

I was not going to mention this.. but I am making a pitch oscillator using a tri-filar 'transformer', and using the extra winding to deliberately couple frequency dependent current (replacing R1+VR1 with a small C and VR1, or perhaps a variable capacitor trimmer).. I believe I can get this to linearize the response.. As the oscillator frequency increases, greater 'feedback' level is coupled by the C, and this increased 'feedback' acts to reduce the circuit inductance, increasing the oscillator frequency and giving a non-linear response which, I believe (hope) will be the inverse of the sensed capacitance non-linearity.


>>[i]"So I guess that this may work in a small specified frequency range where R has to be chosen in the same range as 2*pi*f*L2.[/i]

My first crude prototype's frequency was adjustable from about 300kHz to 600kHz by adjusting VR1..

I have not yet worked out what this circuit will do with regard to Q. At present I only know with certainty that I can use this circuit in my oscillator. I suspect that trying to use this circuit as part of a filter (for example, a BPF for use in a volume circuit) may give unsatisfactory results - on the other hand, it may be possible to configure the circuit sand tailor its response / frequency dependancy, to make it a great 'component' for a BPF.. Time will tell.

This is the full circuit of my test oscillator:

link to image (http://www.therasynth.com/assets/images/bifosc1sm.GIF)

And a link to the full size images (http://www.therasynth.com/html/scrapbook.html)