Winding own oscillator coil on a ferrite former

[regenfreak]

Hi
Anybody here has tried to wind their own coil on a commercial blank former with ferrite slug using Litz wire for DIY built? I am not talking about replacing an existing broken superhet oscillator that you can count the number of turns .

The challenge is working out the number of turns for given wire diameter to produce the desired inductance from my design calculations. I have seen online calculators for single layer coil on a ferrite former and multi-layer coils on an air core former. I have no luck in finding equations for predicting inductance of a multi-layer coil on a ferrite former.

Any one knows any literature with the appropriate or parametric equations?

What is the approximate diameter of the Litz wire used in vintage superhet oscillator coil?

The alternative method would be trial and error, using enamel wire in place of Litz wire, scraping and re-sealing the enamel, measuring the inductance during the winding process. But it will a Q killer unless a thicker enamel wire is used for MW oscillator coil.

[Radio Wrangler]

The slug will more likely be dust-iron or 'Carbonyl Iron' of which there are a few grades and several sizes. Ferrites give more inductance per turn at the price of poorer stability and greater uncertainty.

There are formulae for single-layer solenoidal windings without a core. Add a slug or a second layer and you won't find anything other than empirical data if you're lucky. Manufacturer's data sheets tell you the materials and their frequency ranges, their magnetic properties and loss factors versus frequency. Coil former data gives you mechanical dimensions. Nowhere will there be any mention of how to determine the inductance!

You'd have thought the headline parameter of an inductor would have had a starring role.

Back in the day it was usual to wind a large test inductor and measure it. remove a turn (or few) at a time, clip the surplus wire and re-solder, measure again.

This way you built a graph of inductance versus turns appropriate for the pattern your coil winder gave.

You need to use the wire you intend for the final product. Inter-winding capacitance distorts the effective inductance result, so you must test in the intended frequency range.

For a one-off, wind a coil you're confident is too big, then start removing turns and measuring.

Adding turns in a controlled pattern is too difficult and adding wire is problematical. Removing turns solves these problems.

David

[kalee20]

It's as Dave RW says. But might not be all that bad!

If you know the dimensions then you can use the formulae that you mention to calculate inductance, with no core. Adding a core might increase inductance by 30% (ie inductance increases by a factor of 1.3). So, use your formulae to calculate the turns you will need for an inductance of 1/1.3.

Then you'll have to wind it and tweak it. But you will be in the right ball-park. To tweak it, remember that due to the square law, 1% change of turns gives 2% change in inductance - which then results in 1% change in frequency in a resonant LC circuit.

Be prepared for winding at least one trial coil, before going for the real thing.

Wire gauge for a MW coil? Probably around 30SWG or 0.3mm. But on a miniature former it would be smaller; on a large air-cored jobbie it could be 26SWG or 0.45mm

[regenfreak]

David, kalee20

Thanks for the inputs. I have a miniature, NOS blank 7mm ferrite former with an adjustable slug. I was thinking along the line same of calculating the multi-layer air core coil first and multiplying the inductance by a guess factor...perhaps popping out of an existing ferrite core oscillator coil to estimate the scaling factor.

The permeability of unknown ferrite core can be measured by winding a single layer dummy coil and use the ballpark value for ferrite inductance per turn squared AL (in nH), something like this

http://sklaic.info/forum/index.php?topic=175.0

But I agree it is difficult to know what AL to use if the ferrite former is vintage

In any case, miniature MW oscillator seems a pain to wind. Winding the wire in criss cross pattern by hand is possible but can be difficult if we are talking about 500-1000s turns. It seems something that not many people would try..

I am inclined to replace ferrite slug former with a large toroid core where manufacturer data and calculators are widely available. But I hatre winding toroid core because you have to work out the approximate wire length first and their inductance can only adjustable by tappings.

[kalee20]

Knowing the permeability of the slug is not particularly important. As long as it is greater than 30 or so, inductance hardly changes much. You could replace the slug with another, of permeability 100,000 and as long as it is the same size and in the same position, inductance will be only a few percent greater.

That's because the inductance is dominated by the lines of flux being mostly in air. The effects of this dominate. If you had a toroidal core, then you would indeed find a vast difference.

[regenfreak]

This is an interesting point that i didn't know about.

I find it surprising that there is no empirical equation for the inductance of multi-layer ferrite slug coils as if the inductance change per layer is completely non-linear

[Mike. Watterson]

It's approximately N squared. So four layers is 10 to 16x (at most) more inductance. The slug increases it from minimum and increases it somewhat. Transistor sets might have ferrite, the valve sets are likely dust iron or related.

As someone wrote, guess (look at similar inductance coils) and wind too much and take off till just less than required without a core.

[regenfreak]

Cheers. Some rough guide is better than messing around in the dark.

[G6Tanuki]

In the back of the old RSGB handbooks of the 60s, there are graphs showing inductance/number-of-turns for the once-popular "Aladdin" coil-formers; I've used these with success in the past.

I really don't think that the actual inductance will vary that much between using "Litz" and an equal-cross-section solid wire: certainly not to an extent that won't be adjustable-out by tweaking the core-slug a bit.

[regenfreak]

I may poke around old valve radio design books to see if they have those charts

[G4XWDJim]

Look up Professor Coyles calculator. You may find it useful.

Jim

[regenfreak]

yes i saw that when i built spider web coils. The table for Litz and enamel wire is handy in Prof. Coyles calculator.

[Ed_Dinning]

Hi, let me know if you need any Litz wire as I have a range of sizes here.
If you wind with enameled copper first that should get you quite close.

Ed

[regenfreak]

Thanks Ed! It seems the main challenges are patience and finger dexterity in winding fine wires.

I have a phobia of handing very fine wires as you can imagine there are lots of cursing and frustration when they break. But hey they often sell Denco oscillator coils on ebay around £25 and they often don't match the tracking requirements of whatever combo you intend to use. DIY coils can offer a way out.

[regenfreak]

I have this idea in my mind suddenly that has kept me awake with my insomnia. There is a simple way to minimize stray capacitance like basket coils without the hassle of complex winding technique using metal pins in bigger formers.

I would try to wind the coil in a clean, criss-cross pattern without messy windings or scrambling. The winding layers can be restricted by two circular cardboards (act like a bobbins or spool) with four slots cut at 90 degrees. The slots stop the winding unravelling on its own and also acts as a criss-cross guard. Each layer must be at 90 degrees to each other so that the windings are at 45 degrees to the circumference of the former. Masking tape is needed to hold the first turn. I could try 3 slots or 5 slots using the same idea.

Using simple geometry (if my maths is correct):

Total winding length = no. of turns x wire diameter + diameter of the former

The criss-cross pattern is repeated every two layers by rotating it to the next slot at 90 degrees. It does not have to be 90 degrees, it could be 72 degrees, 60 degrees etc. Ideally no layers of windings are parallel to each other.

I have played around the air core multi-layer calculator. The number of turns required is not that great for MW valve oscillator coils which are typically around 100 to 280 micro-Henry. I can get away with moderately thin wire and maybe 4-6 layers. I think I can manage that. With transistor MW oscillator, it requires finer and more turns, having typically between 350 to 630 microHenry.

[regenfreak]

i think four slots will end up parallel wires but they are "precessed" in layers (like the precession of a satellite orbit around the Earth due to its auto rotation). Maybe 5 slots work better. It is very difficult to visualise it in my head, it is one of those things that to be tried and experimented

[Mike. Watterson]

You can make a wave winder for small coils with Lego Technic. However a random wound version works nearly as well. Use a couple of plastic disc, glued on outside to former, to stop the wire falling off the ends.
The point of wave (basket) winding is to reduce self capacitance which is a problem with multiple layers.

Just move wire feed point back and forth as you turn the former.

[regenfreak]

Mike, I will have to look up Lego Technic as I have no idea what it is.
If I understand you correctly, I would do the figure eight hand motion in each two cycles of winding but I thought it would be difficult to keep track the number of layers as it looks messy.

I have looked my vintage old collection of MW valve oscillators that they all have similar form factor like a pancake with layers at about 30 degrees to each other repeating themselves every 180 degrees.

I have played with a roll of toilet paper with PVC wire and the 30 degree winding is kind of doable. Initially I thought about 90 degrees layer alignment but then the mutual inductance is minimal which means more windings..which is not so clever.

I could try to wrap the former with paper with rough texture to stop the winding sliding (Ever try to wind a conical tesla coil?)

[regenfreak]

in the multi layer inductance calculator i would have to use the former diameter divided by cos30.

[Mike. Watterson]

It's the total turns that counts, not the layers.