Winding 30W Amp Transformers

[VacuumVoodoo]

Make some 4-5mm holes radially in the sides of bobbins. Just like the holes for flying leads. That's all you need to get the impregnation in.

[katopan]

Great tip. Thanks!

[katopan]

Been a bit busy to post on this, but I'll get to working out turns soon.

Just a pic to keep things going. This is the beginnings of an OT bobbin. About 0.3mm cardboard, put together with superglue. What's shown is a strip glued in a wrap around fashion to 2 layers for the centre piece, plus two layers of end sides. I'll cut the 'H' lines drawn on the sides, bend the flaps out and glue to the centre piece. Couple more layers of sides to pad them out and then finally another wrap around the centre for 4 layers all round. Then I'll varnish it all once assembled.

[Colossal]

Very cool, enjoying this thread!

[Phil_S]

...one of your countrymen in Canberra has a rather thorough tube HiFi site and there are a few pages on transformer winding and theory at http://www.turneraudio.com.au/education+diy.html .

It may be a bit forward to contact him directly. People like their privacy. I've had a nice exchange with Mr. Turner at RAT (rec.audio.tubes). I just posted my transformer question and soon enough he got involved.

You can get to the group via a web browser
https://groups.google.com/forum/?fromgr ... udio.tubes (for unknown reasons the link is not carrying the stuff on the end; copy and paste to your broswer) but it used to be better with a news reader. I just looked and it seems there is now a decent threaded interface at Google. I just clicked the top post and there he was in the mix.

If you are really hard up for a newsreader, you can use Outlook Express. When you set it up, avoid the email part of it all together or it will take over every time you launch. IOW, don't enter your email address. Then you need to find a news provider. Free ones (at least they were free at one time) include news.eternal-september.org and aioe.org.

[katopan]

I'll be doing a few things different to what is on the Turner Audio site. But it is a great and thorough description of how he makes his transformers. Rare to find such detailed info about how someone winds their own. Great stuff!

[katopan]

Ok, I now have an OT bobbin and have started on the one for the PT. The superglued cardboard worked really well and feels just as stiff and sturdy as a moulded nylon bobbin if not more so, and it's almost half the thickness. I had to make a second set of sides because the first set weren't accurate enough. Hey, thin cardboard sheets are cheap and I've got to have something to do in my work lunchbreak, so nothing really lost. Cut the second set with a guillotine which helped get things right, straight and square. Cut outs are done with a sharp utility knife. The pics show a bit of how it was all glued together.

There are many ways to skin the cat when it comes to winding transformers so you need to really think through and visualise the whole process in your head before even making the bobbin. Even making the choice to have bobbin sides I still had to weigh up flying lead mounting options (taped within the windings, terminal strip, etc), do I want to run interlayer connections outside through holes or have them inside hidden within the bobbin (and how does that effect interlayer insulation and winding margins), etc, etc. I've decided to have hidden interconnections and make the winding window square all around such that my final external wires can just come out of the windings on the final layer like they would if I didn't have bobbin sides. I'm also leaning to varnishing as I go even though the vacuum varnished worked well last time. That way I don't need any holes in the bobbin sides and everything will be quite neat. Well, the picture in my head now is quite neat. We'll have to wait and see if it actually works out that way.

Now I've got definite final measurements of the winding window I can proceed with working out layers and turns....

[JazzGuitarGimp]

It's exciting seeing this come together - Thanks for sharing, Katopan.

[katopan]

So back on the 2nd post of this thread I worked out the OT will have wire sizes of 0.25mm (0.28mm OD) for the primary and 1.0mm (1.08mm OD) for the secondary, although the 16 ohm section could drop to 0.8mm (0.87mm OD) if needed to fit. My new bobbin has a winding window 43mm high and 13mm deep. The bobbin centre externally measures 39mm deep and 36.5mm wide. Without target inductance values or a way to measure the iron characteristic, to work out turns we need to introduce a bit of engineering judgement. Yes, this is rough. No, it won't produce a clone of a known OT. But yes, it will work and give me a perfectly reliable and toneful OT for the job.

The best starting point is this - ever seen an OT with only half the winding window full? No, me either. If it was going to be half full, the designer would use a smaller core size and go back to an almost full window for around the same inductance. In other words if we're referring to a known OT size for an amp size, you basically want to fit as much copper into the windings as possible without going over what will practically fit when you build it and try to put the lams together. This will give you the most inductance and best bass response out of the core size you are using for the lam characteristics you're stuck with from the donor iron.

Next we look at winding numbers of known transformers. I only know the details of Marshall 18W, 50W & 100W OTs. An 18W has an 8K p-p primary with 3030 turns. JMP/JCM 50W has 3.4K p-p with 1600 turns. 100W has 1.7K p-p with 1240 turns. We're designing for 30W and 6.6K p-p. This is where we set up a spreadsheet and play with the numbers to see what will fit and whether it looks like it's something in between the above numbers.

Traditional OTs use varnished paper insulation. 1 layer between each winding layer and often 3 layers between each actual winding. Modern transformers often don't use anything between layers and polyester tape between windings. I'm going half way between - varnished baking paper (greaseproof paper) between layers and polyester film sheet (overhead transparency) between windings. My baking paper is 0.0508mm thick and the overhead transparency is 0.102mm thick. Nothing beats the breakdown voltage vs. thickness of polyester film and it's got a great temperature rating too. For 0.102mm polyester sheet datasheets say it has a dielectric strength of around 3800Vac/mil = 15kVac breakdown. That just stomps on any other option for interwinding insulation and will make for a long lasting transformer. Not vintage though, so if you want that vibe in full use a few layers of varnished paper instead.

A few numbers are floating around for air breakdown voltage but the number I use is 20V per 0.001" which corresponds to 787V/mm. In Australia our standards require the insulation to accomodate 3kVac without breakdown for any mains connected transformer (the PT only here). 3kVac is 4243Vpeak and requires both insulation that can handle that voltage and a creep distance across surfaces between windings of at least 5.4mm. This creep distance is only required for old style PTs where the secondary is wound over top of the primary, and is implemented by leaving a minimum 3mm winding margin at the edges to give 6mm creep distance. Modern transformers have split/segmented bobbins and so don't require the same winding margins. An OT isn't split but also doesn't need to be built for 3kVac. By default the OT will be subjected to close to 2x the B+ voltage from signal swing, and I'm guessing up to another B+ worth from inductive load spikes when overdriven. If I leave a 2mm winding margin, that's 4mm creep distance giving over 3kVpeak breakdown. That's over 7x the B+ and I think will be more than adequate.

Next we work out a few basics.....

[katopan]

With a winding window width of 43mm, if I leave 2mm winding margin at each edge that reduces to 39mm useable space. It's common to reduce that by 90% to allow for small gaps in the turns because the wire won't sit exactly right up against the next turn. Also last time I found that to be adequate as long as I stayed on top of the turns being neatly against each other. 90% of 39mm gives us 35.1mm to work out turns per layer. For our primary wire 0.28mm OD that gives a max of 125 turns/layer. For our secondary wire sizes 1.08mm OD gives a max of 32 turns/layer and 0.87mm OD gives max 40 turns/layer.

I'm also using a 90% factor for the winding window height. Some people use 80% here but I found 90% was fine last time. 90% of 13mm gives us 11.7mm. We need to fit interwinding insulation at 0.102mm per layer, interlayer insulation at 0.0508mm per layer, and all our copper into that height. To work that out we need to work out how many layers we need for what turns. All of that then becomes interative until you work out something that fits.

Some time playing with the numbers and I've come up with the following winding plan (in layers):
Bobbin
Polyester film insulation
P1-1 of 122 turns + baking paper insulation
P1-2 of 122 turns + baking paper insulation
P1-3 of 122 turns + baking paper insulation
P1-4 of 122 turns + baking paper insulation
P1-5 of 122 turns + baking paper insulation
P1-6 of 122 turns + baking paper insulation
P1-7 of 122 turns
Polyester film insulation
4ohm-1 of 32 turns + baking paper insulation
4ohm-2 of 31 turns
Polyester film insulation
P2-1 of 122 turns + baking paper insulation
P2-2 of 122 turns + baking paper insulation
P2-3 of 122 turns + baking paper insulation
P2-4 of 122 turns + baking paper insulation
P2-5 of 122 turns + baking paper insulation
P2-6 of 122 turns + baking paper insulation
P2-7 of 122 turns
Polyester film insulation
8ohm-1 of 26 turns + baking paper insulation
16ohm-1 of 37 turns + baking paper insulation
Polyester film insulation
P2-8 of 122 turns + baking paper insulation
P2-9 of 122 turns + baking paper insulation
P2-10 of 122 turns + baking paper insulation
P2-11 & P1-8 of 60+60 turns + baking paper insulation
P1-9 of 122 turns + baking paper insulation
P1-10 of 122 turns + baking paper insulation
P1-11 of 122 turns + baking paper insulation
Polyester film insulation
Paper external label

This follows the interleave pattern I chose earlier. It gives me 2560 primary turns, 63 turns on the 4 ohm tap, 89 turns on the 8 ohm tap and 126 turns on the 16 ohm tap. Extremely close to 6.6K for all secondary taps. These numbers look pretty good in the context of the known OTs mentioned above. Layers will all fit within the winding margins with some to spare. And most importantly the total height is spot on for my 90% winding window height. That's made up of 21 primary wire layers, 18 primary baking paper layers, 3 secondary 1.0mm wire layers, 1 secondary 0.8mm wire layer, 2 secondary baking paper layers, 6 polyester film interwinding insulation layers, and the outer paper label.

So in the end I will have to use 0.8mm wire for the 16 ohm section instead of the 1.0mm I was hoping to fit. But it makes the difference in height and more dramatically turns per layer of what I can fit.

[JazzGuitarGimp]

Observations and questions:

I don't see a layer called P1-8, though this is in keeping with your summery that you will have 21 primary layers.

Do all of the P1 layers belong to one side of the primary (one output tube) while all P2 layers belong to the other output tube?

If so, I see an imbalance in turns on each side of the primary. I understand that the last set of P1 turns will be longer than the first set, meaning that P1 would likely have more DC resistance than P2 if there were 22 layers. But, a turn is a turn in terms of voltage, current and impedance ratios.

What am I missing?

[katopan]

Yes, P1 & P2 are each half of the primary. P2-11 & P1-8 are on the same layer and are 60 turns each. In the middle of that layer is the primary centre tap. All is balanced again when you account for the lost P1-8 unless I've stuffed up something I haven't noticed. Also that interleave pattern is one of the more balanced ones with the primary interleave order of 2/3rds of P1, 2/3rds of P2, 1/3rd of P2 and 1/3rd of P1. Having the two sections of P1 first and last, with P2 sections in the middle evens out the wire length of the two halves of the primary and results in a fairly balanced primary winding resistance. It's a feature of this interleave pattern.

[katopan]

I now have a PT bobbin. Same sort of assembly except with the extra separator to make it a divided bobbin. It only shows in the photos if you know it's there, but the centre divider also has a layer of overhead transparency on one side. This will add many times the breakdown voltage of the cardboard to between the primary and secondary to easily meet the 3kVac requirement.

Next I can paint both all over with varnish which will soak into the cardboard and seal the surfaces. I better order the additional wire I need too. At this rate I'll be winding soon enough.

The only huccup is last time I started using a calculator with the equals button wired to a reed switch - I'd read about other people using this, where you press 1 + 1 and then the number increments every time the magnet passes the reed switch effectively pressing the equals button. Well the calc was $2 and had conductive trace on polyester film for the buttons, going into a very small PCB with the chip covered in goop. I managed to solder wires to the very edge of the PCB and everything worked for about two layers. Then the calc just stopped working. I got another one and did the same thing. Another two layers and it did the same. I'm guessing the chips are static sensitive and don't like being wired up to the outside world. I actually finished off both OT and PT counting manually, marking on paper after each 10 turns. Very tedious! Anyway, after finishing those transformers I half bit the bullet and starting making a proper CMOS counter circuit into a box. I never finished the counter so that is also going to be needed before I get to actually making turns of wire.

[rock_mumbles]

Wow that's a lot of work!

[katopan]

It's a labour of love Rock. You wouldn't do it otherwise! But I'm excited about having a second go at winding my own trannies. The last ones worked well and of course I learnt a lot first time around to apply and plan these ones out better.