Winding 30W Amp Transformers

[katopan]

This will be a while start to finish with the limited time I have at the moment. But winding transformers pops up from time to time and I thought I'd share the process I'm going through.

I'm crunching the numbers on transformers suitable for an EL34 based 30W amp (gee, let me think, maybe an Express? ). The OT won't be an Express clone, with its unused windings etc. (Shhh, details are secret!) Therefore it won't have the identical tone. But it should be suitable to use in the amp. Things are less picky with the PT so it just needs enough iron and copper for the power required and the performance should be similar enough.

I'm using old iron I have in a junk box - no lamination specs or any way of really measuring B-H curves, etc. So I'm relying on rule of thumb calculations I've been taught elsewhere and 'reality checks' against advertised transformers. You can calculate flux and everything in detail, but all that goes out the window if you don't have lam curves and know that what you've got will actually perform to them. So I've got donor iron and one spool of wire of the right size already. Probably need 3 other wire sizes - for all up a bit over $100 (and quite a few hours) for both PT & OT I think I can live with very good but not clone identical tone.

So for some general specs to get the ball rolling...
OT: 6k6 p-p primary, 4/8/16 secondaries, 30W rating, and the Express OTs I can find listed on other sites have US Ei-125 lams.
PT: I'm in Oz so primary will be 240V. 3 preamp valves and 2 EL34s needs 250mA, and if I want to use the iron for something 6L6 based in the future it won't draw as much as the EL34s, so secondary 1 will be 300-0-300V at 250mA. Likewise EL34s are more heater hungry, and with 3 preamp valves the total heater current is just under 4A. I'll design for at least 4.5A depending on wire size, so secondary 2 is 6.3V at 4.5A+. Other sites have US Ei-125 lams for the PT too but with a really big stack height.

[katopan]

Basic Design Calcs....

PT:
Secondary 1 load = 400Vdc x 250mA = 100W
Secondary 2 load = 6.3Vac x 4.5A = 28W
Primary load = 1.2 x secondary = 1.2 x 128W = 154W
(This is based on 80% efficiency but in reality will be around 90% so there is some headroom / safety factor designed in.)
Core size centre leg area for a PT is sqrt(Power) = sqrt(154) = 12.4cm2
US Ei-137 lams give 12.2cm2 with a stack height square with the centre leg width. I'll have to check my box of old transformers and see if I've got something suitable.
Turns per volt = 45/S (S = centre leg area in cm2) = 45/12.2 = 3.7 turns/volt
Secondary voltages will drop under load (called regulation) and so you design the turns ratios to include a boost factor to make up for this. I've only wound transformers once before and that time I used an extra 10%. Voltages were a few % high so this time I'm going to use 7%.
We start with Secondary 2 - 1.07 x 6.3V x 3.7 turns/volt = 25 turns
Primary = 25 turns x 240V/(1.07x6.3V) = 890 turns
Secondary 1 = 25 turns x (1.07x300V) / (1.07x6.3V) = 1190 turns x 2
Wire diameter = sqrt(I/2) (I = current)
Primary - sqrt((154W/240V) / 2) = sqrt(0.64A/2) = 0.57mm -> 0.56mm is readily available.
Secondary 1 - sqrt(0.25A/2) = 0.35mm -> use 0.4mm nearest size.
Secondary 2 - sqrt(4.5A/2) = 1.5mm -> Mmmmm.....
I can only get up to 1.25mm in the very small quantity that is needed for this winding (only 25 turns). I can either spend 3x more on a big spool I'll never use, or wind parallel windings for the heaters.
Secondary 2 - sqrt(4.5A/4) = 1.06mm -> That's better. I'll use 1.0mm as two parallel windings to supply the required current.
Next thing is to find out if I've got suitable iron and check if that will fit on the bobbin.

OT:
We want a 6K6 primary and 4/8/16 secondary taps.
Turns ratios -
6K6 : 4 ohm = 1 / 40.6202
6K6 : 8 ohm = 1 / 28.7228
6K6 : 16 ohm = 1 / 20.3101
Wire size we use the formula as used above for the PT. But when I reality check against OTs that this forum and other places have data on you quickly find that secondary wire size is commonly 'correct' for only the 8 ohm winding. The 4 ohm winding is overloaded compared to the calculated rated current for the wire size, but it is a small number of turns (relative to the primary) and it seems to work for everyone else, so that's what we'll do too.
Now we could work out primary current based on output power, but what about the bias current? Three different people will give three different answers for how to work this out. I'm going to use the anode current listed under max signal off the valve datasheets. Mullard EL34 datasheet has 2x110.5mA for 3K5 p-p load. 6K6 will be less but we're allowing 250mA from the PT, so it sounds good enough.
Primary - sqrt(0.11A/2) = 0.23mm -> use 0.25mm nearest size.
For the secondary we do calculate current from output power only for the 8 ohm tap as I said above.
Secondary - current = sqrt(30W/8ohms) = 1.94A
sqrt(1.94A/2) = 0.98mm -> use 1.0mm nearest size.
I could drop to sqrt(sqrt(30W/16ohms)/2) = sqrt(1.37A/2) = 0.83mm (0.8mm would probably be fine) for just the 16 ohm section of the winding if I wanted to.
Working out turns is for another day.
Oh, and you don't calculate iron core size for power with OTs. Any decent amp will have OT iron way oversized for power to ensure enough inductance for adequate bass response. Because of this the core should run pretty cool in any amp unless the OT is way undersized. As I said before I don't have lam data and so inductance is going to be pretty hard to calculate. So this is something you need to base purely on known commercial OTs. The Express iron I've found listed uses Ei-125 so that's what I'm using. Luckily I already know I have old iron of this size that is going to be perfect to use here.

[JazzGuitarGimp]

Good stuff - anxiously awaiting your next installment.....

[kdmay]

Watching with interest too.

BTW - where in Melbourne are you? Always good to know there's a local with a similar obsession! (I'm in Box Hill South).

[katopan]

Thanks guys!

kdmay - I'm in Cheltenham. Seems to be plenty of people around that have been bitten by the amp building bug. And an obsession it is!

So I had another look in the junk box last night. Gee sometimes you can be lucky. I went in hoping for an Ei-137 core thinking if I did have another Ei-125 it would be fairly square in stack height, where the Express PT has a much larger stack height. Out of the bigger iron first up I found I have 3 Ei-150 transformers, which is overkill but I would use one if I had to. No Ei-137s. But then I noticed another Ei-125 with a huge stack height. I didn't get a chance to compare it to the Express PT dimensions but by eye it looks possibly bigger than the Express PT photos I've looked at. I'll have to double check but it seems like I've got exactly what I need.

The iron I'm using for the OT has a nylon bobbin which I'll reuse. Will just have to cut out the centre divider. But the PT iron I found last night is wound the old way. My preference is to use a bobbin so if I can't buy one the right size I might have to make one for the PT. Last time I wound my own transformers I reused the bakerlite bobbin on the PT and made a new cardboard one for the OT, both without sides. I had a real problem with end turns falling off into the winding margins. Extra tape and I got through OK, but I don't want to have that problem again so this time I'm keen to use bobbins with sides. That way you can lay the paper against the sides and winding margins aren't so critical. But then last time I used greaseproof baking paper un-treated until the end when I vacuum pot varnished the windings. The baking paper isn't stiff at all. Maybe if I did it properly and used paper treated with shellac it would be stiffer and hold up the end turns better. Mmm... I'll have to think about that.

[katopan]

So the next thing to think about is interleaving. The smaller Fender Tweeds supposedly had none. Some guitar amp OTs have a bit. Other guitar amps, like the 18 Watt, used an off the shelf hi-fi OT which had a bit more. Some modern hi-fi OTs have a ridiculous amount. The number of layers and interleave pattern, as well as primary winding inductance are what mostly set the OT's frequency response. Add the harmonics from iron curve and how hard you drive up that curve and you probably have the three biggest factors which dictate how the OT will sound. Everything else just needs to hold up to 'reality checks' against other transformers that work well.

A bit of searching here and elsewhere and you find out the Stancor A-3801 was a hi-fi OT. Probably not 11 layers of interleave, but it would have some for sure. I'm no expert in transformers and only know the interleave patterns for a few Marshall amp OTs plus the generic patterns shown on transformer design websites. At the end of the day the interleave not only sets the high end response, but it also sets the balance between the two halves of the primary due to how they are split and differing lengths of wire between turns deep in the winding vs. near the outside. It's a pure guess where I go from here for what sort of tone I want....

I don't think a 30W hi-fi transformer is going to have the simpicity and imbalance of the bigger Marshall OTs. But going 7 layers like an 18 Watt OT seems overboard. So lets pick something from this site:
http://www.mif.pg.gda.pl/homepages/tom/transfor.htm
I have waded through Google Translate's mess of that site numerous times and think I've pieced together a lot of it. Three layer is not looking like enough for what was a hi-fi OT. 7's too much, so 5 layers sounds good. But which one?

Figure 7 shows a few options. 7b is listed as optimised for equal resistance and good for Class AB (by the way 7b is a Drake 50W OT used in Marshalls). But in the favourite progressions of fig 8, 9 & 10 I'm liking the look of 9. It's slightly more interleaved than 7b (both halves of the primary split) without adding more layers. I'd expect a slightly better performance over 7b. For this amp it seems like the appropriate level of complexity, hi-fi-ish without going overboard. Also I used this pattern in the 10W OT I wound for a 2 x 6P1P push-pull amp and that sounded great, so hopefully it will here too.

Now we have an interleave, we can start working out a winding pattern.....

[Colossal]

Great thread Katopan. Pulling up a chair...

[Structo]

Interesting about the Varistors on the Primary of the OT.

Do you think these are thermistors like, Inrush Current Limiters?

[cpollack]

Nice thread, Katopan.

Is the wire size as expressed in mm the diameter or the cross-ectional area? In the automotive business where I'm an application engineer for a connector company, we use ISO wires sizes expressed in mm^2 for cross-sectional area. A .5mm^2 wire is close in size to 20AWG.

Also, 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 .

-Chuck

[katopan]

Good to have you on board Colossal.

Tom - Sounds like you know more than me buddy. Are there varistors built into the OT primary windings or are they external? They could be anything depending on if they're PTC or NTC. Inrush current limiting or thermal protection? I have no idea without more info. Wasn't planning on putting anything but enamelled copper in my windings!

Chuck - Everything for me at work is mm2 cross sectional area too! I'm an elec eng working in an industrial process plant. But with winding wire it's specified as diameter, or often AWG for you U.S. guys.

Cheers, Craig.

[katopan]

Something that's had me going for a bit is PT lam stack height. I've been using the ClassicTone dimensions as a reference because their site has the dimensional drawing and clear photos. But that is a stupidly large stack height, and unless that's how the original was surely you'd just go up a lam size and make the thing squarer. My assumption was maybe the Stancor or Pacific trannies are just like that. It's been bugging me so back into the TW Files section to look at some photos of originals from this sticky: http://ampgarage.com/forum/viewtopic.php?t=19

Straight away it's apparent that Francesa doesn't have anywhere near the stack height they've used in the ClassicTones. But all the photos are angled so you can't determine the measurements. Then I found photo DSC00704.jpg in the zip file of the unknown Express. Another photo shows these to be Pacifics, and it's a nice overhead shot allowing some measurements to be made. We know an Ei-125 core is 3-1/8" along the shorter side. So I get the photo into a graphic program, resize until the core top edge measures 3-1/8" on my screen, and then measure the stack height. It's cool because this is the only photo that is square enough to measure like this and it's got a known reference (the core top edge) in both X and Y directions so we can account for any non-square pixel display for the monitor resolution.

The results: The PT stack height measures 2-1/4" and the OT stack 1-1/4". It might be a little rough but it's a close enough validation. Also for the PT it's still more than enough iron for the required power, and OT size checks out OK vs. other known OTs. And here's where I get lucky with my old iron. The big Ei-125 I've got (photo attached) has a stack height of 3-1/2" so that's exactly enough lams to make both the PT and OT for this project. I'll just have to make up my own bobbins for both. The smaller Ei-125 which has a 1-3/8" stack can go back in the junk box for another day.

[VacuumVoodoo]

Are there varistors built into the OT primary windings or are they external? They could be anything depending on if they're PTC or NTC. Inrush current limiting or thermal protection? I have no idea without more info. Wasn't planning on putting anything but enamelled copper in my windings!

Varistors eat voltage transients, they go short circuit when voltage transient exceeds specified limit and dissipate that energy as heat. Specs in data sheets are for threshold voltage and dissipated energy in joules.
Info on Toms page is generally solid but there are few booby traps, whether intentional - I don't know.
Re: core stacks, those oblong EI stacks are not optimal. Total transformer performance drops when central column (the "-" in E) cross section departs too far from a square and becomes a long rectangle. I wouldn't recommend higher ratio of length to width than 1,5 but opinions certainly vary. From transformer factory point of view it's cheaper to stock fewer lamination sizes.

[katopan]

Thanks Aleksander. That makes sense about varistors protecting against large voltage transients. I figure if the transformer is well made and both insulation and margins are adequate then there is no need for additional overvoltage protection.

I also agree about keeping the centre stack ratio between 1 - 1.5. It's got me puzzled why ClassicTone would make their Trainwreck PT with such a large stack height. Even at the height suggested by the original Express photos there is plenty of iron for the power needed by the amp. With very long rectangular centre legs not only does the transformer performance drop, but also you are generally trying to fit more copper into a limited winding window. Going up a lam size keeps the centre leg more square but also gives you more winding window to fit the correct size wire for the level of power the core will support.

[Structo]

I have also wondered about Mercury Magnetics transformers they advertise as Fat Stacks or Super Stacks.

Sounds like it is more of a gimmick than a good thing.

FatStack™ (SuperStack™): FatStack transformers are custom-designed Mercury Magnetics transformers that have extra iron. In an output transformer it extends the bass frequency response. In the power transformer it can minimize tone drift giving your amp more stability and headroom. Our SuperStack is similar but built higher to accommodate smaller mounting footprints.

[katopan]

So up until now it's been a case of applying the formula and getting a number. I've got core sizes from photos, the PT is bigger than my minimum calculated for the power, and the OT seems right compared to other amp OTs for the power. Wire size is set, target turns on the PT and ratios for the OT are all worked out. Next I really need to work out what I'm doing with the bobbin. Then comes the 'art' bit of working out how many turns and layers we want vs. what will fit on the bobbin.

The donor transformer doesn't have one that's going to be any use because it's the wrong size. I'll have to make them, but have to make a final choice do I want sides or not. Last time I didn't but this time I'm leaning toward having them. But that does have an impact on varnishing as you go vs. being able to dunk the whole thing in a vacuum tank. With sides and tight tolerances on insulation against the sides, vacuum varnishing won't be very effective because it won't get into all the air gaps. Without sides the varnish can easily get in to all the layers. Without sides can be made up with cardboard (varnished of course) but sides need to be a bit more sturdy.

Anyway, this is just a quicky post. I'll have to go away and make some choices. But finalising the bobbin will finalise the winding window I am left with. Also got to choose my layer and interwinding insulation (which I think I'll do the same as last time) which along with wire diameter will set the thickness of everything and how many layers I can fit within the winding window. This part of the design process becomes interative where you work out roughly what you want, see if it fits, and then refine things until it all looks right....