Ferrite inductor puzzle

[mole42uk]

Hi,

I am building a valve stereo decoder for FM broadcasts based on the circuit shown here.

Both of the coils are built on Vinkor LA2505 ferrite cores.

The 38kc/s auto-transformer worked just fine, although I had to increase the capacitance to about 490pF bring the oscillator to dead on 38kc/s.

I have tried 3 times to wind the 19kc/s filter. It calls for 1525 turns of 45swg enamelled copper wire, and although I have a coil winder for my lathe, I managed to snap the wire every time. I have now wound the filter coil with some 44swg which has been successful.

My question is this: the coil measures 3.63H on my tester, which gives reliable results on marked commercial coils. For this to resonate at 19kc/s I will need a 19pF capacitor, but the circuit calls for 180pF. Have I changed the impedance so much by a slight change in the wire diameter?

Hoping for some help, I remain slightly puzzled,

Richard

[GMB]

No, the wire diameter change didn't do it.

And 3.63H is way too high for that number of turns according to the spec. sheet.

Is it something like the core is meant to have a gap and somehow you have assembled it without a gap?

Are you pulling the wire off the end of the reel or unwinding it from the reel?

[Roy]

I have built a valve decoder using LA2505 cores, but using a different circuit.

Do you have the data for the LA2505? The "turns for 1mH" figure is 83.6, so 1525 turns should give you around 333mH.

To get 3.63H you would need to have wound over 5000 turns!

I think something may be wrong with your tester ...

Roy

[mole42uk]

I reckon 1525 turns at 83.6 turns per mH should give me 18.24mH, but maybe I'm using the wrong calculation?

Originally Posted by merlinmaxwell

Are you pulling the wire off the end of the reel or unwinding it from the reel?

Unwinding of course!

Richard

[Radio Wrangler]

Inductance is proportional to 'number of turns squared' no wonder you got a lot of inductance!

David

[mole42uk]

Originally Posted by Roy

Do you have the data for the LA2505? The "turns for 1mH" figure is 83.6, so 1525 turns should give you around 333mH.

With the specified capacitor at 180pF, I should need about 390mH for 19kc/s but I can't see from the data sheet how 83.6 turns for 1mH relates to the specified 1525 turns.

Richard

[Roy]

(1525/83.6)^2 = 333mH

with 180pF this would resonate at around 20.5 kHz, but the coil will have some self-capacitance which will lower the resonant frequency.

Hope I've got that right ...

Roy

[Radio Wrangler]

If 83.6 turns is 1mH, the 1 turn inductance is 1mH/(18.6 squared) which is 143nH

1525 turns will give you 143nH times 1525 squared which is 332.7milliHenries

With such a large winding, you will have appreciable stray capacitance in the turns and this will lead to a very distorted reading on an inductance tester, depending on the frequency you measure at. The self resonance frequency is going to be a problem.

If you want 18.24mH, then you need SQRT(18.24mH/143nH) which is 357 turns. and a lot easier to wind.

The stray C will be lower and the self resonant frequency will be dramatically higher, and it might just work with your intended capacitor value.

Pot cores are usually described by their 'Al' value in nanoHenries per turn squared, but not all early makers followed this. The inductance you got was Al (A subscript lower case L) times the square of the number of turns, in nanoHenries. If I buy a Siemens/Epcos pot core with Al = 630, and I wind on 123 turns, I get
630*123*123 = 9531270 nH which is 9.53mH in more comfortable numbers.

David

[GMB]

I also think it should be about 333mH.
Could 3.63H have actually been a misread for 0.363H ?

I suggest you add the capacitor and then try seeing what it does around the 19kHz region.

If it really is over 3H then there is something wrong with the core assembly, which I note may well be possible as I have seen notes about assembly issues with these cores (but I don't know what the problems were).

[kalee20]

Originally Posted by mole42uk

Unwinding of course!

That could be the problem - if unwinding, you need a jerk-absorber device between the reel and the winding. Unless, the reel is near empty and thus lightweight.

Pulling off the end is a bit inelegant, but avoids this problem.

[mole42uk]

Well, it was actually unwound off an old relay coil so there really wasn't any inertia.

Just clumsiness I think!
Richard

[Roy]

I'm sure you will find your coil is perfectly OK - try it in circuit!

Something that caught me out was forgetting to slip insulating sleeving over the leadout wires before terminating one of the transformers. I thought I might get away with it, but the core has edges sharp enough to penetrate the enamel on that very thin wire.

Good luck with your project.

Roy

[mole42uk]

Thanks!

I got it to peak at 19kHz with a 180pF capacitor by gapping the core. There's no guidance in the Vinkor data sheets, but I thought I'd try a gap. With a piece of thin rubber in there, I can tune the core by squeezing the cores together with the centre nylon screw, so once I find a gap size close enough I'll use a piece of plastic the right thickness and finally tune it to 19k with a trimmer capacitor.

I'm still a bit puzzled, but thanks to everyone's help at least I know a bit more about coils than I did last night!

Richard

[Roy]

Originally Posted by mole42uk

I can tune the core by squeezing the cores together with the centre nylon screw

I didn't know that was possible with the LA2505. The type I used had a tiny adjuster plug for tuning. Perhaps you are using the ferrite cores alone, without the container which clamps the assembly together? Without the adjuster, the inductance will be slightly lower than expected.

[mole42uk]

You've got it! I'm using just the cores and the bobbin, with a nylon screw to hold the cores together and to the aluminium chassis.

I didn't use the trim adjusters because the write up for the circuit I'm using (from WW, 1962) suggested that the cores were used without the metal containers. Apparently, the metal containers introduce an element of capacitance to earth. So, no containers, no trimmers either.

I found that, injecting 19kc/s across the coil & capacitor and looking at the signal on the 'scope, I can get a nice peak by setting the core halves apart by a small amount. I still need to discover by experiment how thick the spacer needs to be before I fit the coil to the chassis, and then select the best capacitor & trimmer combination to set the frequency dead on.

Richard

[Ed_Dinning]

Hi Richard, a lot of these cores, when used in their assemblies were supplies as 2 different halves, one of which had a threaded pin in the bottom of the central "hole".

The "top" half had a through hole and the small screwed slug bridged the gap and gave a measure of adjustment. It sounds as if you have 2 top halves on your coil.

I remember building this circuit many years ago and it set up just fine .

Ed

[Radio Wrangler]

Originally, some ferrite cores were gapless, and required a spacer to create a gap. Spacer material had to be physically stable under pressure. The problem was that the gap existed around the edge as well, and a lot of field was radiated from here.

To improve the self-shielding, ferrite cores were made with the centre post ground down to create an air gap, and half the gap was ground on each half so the parts were identical. There was no gap around the outside.

Later ferrite cores went to grinding the centre post on one half only. It was cheaper. The two halves weren't the same, but the base halves were all the same, irrespective of the Al value they were going to become when a gapped top was fitted. The bases had little nylon nuts glued into their centre holes to take adjusters, the top half got the extra grind, and had the Al value printed on the top.

When winding gapped pot cores, you can often get better Q by spacing the winding out so that wire isn't close to the strong fringing flux from the central gap. The art was to fill the bobbin part way with insulting tape, and then put the winding on top. The improvement from flux fringing offset the losses from longer wire.

David

[mole42uk]

I think that, somewhere in here, lies the answer.

I have taken the glued little adjuster studs out of the core halves, and I think that, if only one half carries the core gap, then I might have assembled the core with two 'non-gapped' halves. This would give me the error in the resonance frequency and explain why adding a gap will bring the frequency back into line.

The two halves carry the same identification code, AFAIR, but I'll physically measure them as soon as I get into the workshop again and find out.

Thanks,
Richard

[Roy]

I have just had a closer look at a spare core. Sure enough, only the upper half has the gap. I have wound quite a few of these, but never noticed that before!