Audio trafo winding help

[CJ]

here is a pic of the transformer with the weird parallel primary

Stancor  HF-40

 

[mitsos]

I was confused thinking it was an EI, but ít says dual coil, so that's some sort of DU or UI core?  I don't think it's on the paper, unless I missed it.  Interesting to note the wire gauges, and DCRs of primary and secondary.  I was told that paralleling 2 wires gives something equal to a wire 3 gauges larger (lower), which is what they were going for. 

But why didn't they do the same parallel tricks on the secondary?

When would you choose P-S-P over P-S-P-S-P or over P-S-S-P? or any other variation??

thanks!

 

[CJ]

coil geometry example: 600:600 transformer with P-S-P coil geometry.

turns ratio is 1:1 for our 600:600 xfmr,

so we will have the same amount of turns on the pri as the sec,

when you keep the copper wire length of the pri and sec the same, you get better coupling, and thus, less leakage inductance,

so when you wind 1/2 pri sec 1/2 pri, the average turn length for both pri and sec come out about the same,

the sec has turns close to average length,

the pri has below and above average turn lengths,

so it averages out to the same  turn length as the sec.

asymmetry causes leakage inductance in a coil, so if you make the windings the same length, the leakage goes down.

next, we have a formula that states that

F/L = 1/N^2

N is the number of sections in the coil, for our coil, N=3, so N^2 = 9,

F/L = 1/9

F/L is Leakage Factor, the lower the better,

so for a 4 section coil like P-S-S-P, you get

1/4^2 = 1/16 F/L which is better still.

Peerless used a 4-5 Geometry, S-P-S-P-S-P-S-P-S,

so the get 1/9^2 = 1/81 F/L, which is 9 times better F/L tha the P-S-P coil.

but labor goes up, so you pick a compromise in your marketing strategy.

one coil is cheaper than two, so if we wind an output transformer, we do not need massive CMMR, so we can use an EI core and get extra core volume for more power.

coil geometry for EI based transformers will have the odd/even type coil geometry like P-S-P-S-P, which would be hard to do on a dual coil xfmr.

dual coil transformers usually has even number of pri and sec windings,

rare to see more than 7 windings because you need insl between layers so the copper has no place left, and the capacitance goes up because of the increased surface area,

oh wow, more carnage, this one has pics

Western Electric 94E  Repeat Coil

http://www.groupdiy.com/index.php?topic=35859.msg440346#msg440346

 

[mitsos]

Thanks, CJ.

so when you wind 1/2 pri sec 1/2 pri, the average turn length for both pri and sec come out about the same,

Now that I read that, I remember you saying that some other place.

So, not with increased sections (P-S-P-S-P or SPSPSPSPS or whatever in between) we get lower leakage inductance but increased capacitance.

Any reason to go P-S-P-S-P instead of S-P-S-P-S?

Is it normal to wind layers back and forth (left to right, then back right to left)?  From what I've seen it's sometimes done in one direction only, ie left to right then all the way back to left in one swoop, to start the next layer, left to right.  What is the benefit of winding only left to right (is that what you call foldback?).  It would seem that this might make the coil look like a few capacitors in series, but then that's what pie-winding is for, right?  Is pie-winding the only way to "series-fy" winding capacitance?

All this Q/A is starting to look a bit like the 1st Edition of Audio Cyclopedia!

 

[PRR]

> a bit like the 1st Edition of Audio Cyclopedia!

You want Radiotron 4th.

 

[CJ]

<any reason to go P-S-P-S-P instead of S-P-S-P-S...>

mechanical:

if you have a transformer with a high turns ratio like a 1:10 mic input, you will have two different wire sizes, big and small,

from a mechanical standpoint, it is wiser to wind the big wire first,

this is because if you wind the small stuff first and then pull hard on the big wire, it will split the fine windings apart and this is not good.

big wire can collapse pressboard coil formers, but nowdays, everybody uses nylon bobbins, so this is no longer a problem unless you wind retro like Magnequest or some other botique shop.

so P-S or S-P might depend on wire size

voltage gradients:

innerstage and output transformers for tube equipment can have high voltage on the windings, you want to make sure this voltage does not short to the core because the core usually ends up getting grounded to the chassis when you mount the transformer due to internal shield and ground connections being soldered in place,

so you may want to wind the low voltage winding first, so that it acts as a spacer between the hi voltage winding and the core, but not always. you may wish to wind the HV section first if you need it to be coupled to the core first.

also, you may want to balance leakage C by placing the windings in a certain configuration,

this will change depending on the application, for example a single ended  output transformer will be wound different then a Push Pull output, a push pull xfmr will be balanced, an SE xfmr will be set up so that the high voltage terminals are in a certain place with respect to ground, balancing is not much of an issue as with push-pull.

here is a jpg with coils and stuff,

 

[mitsos]

> a bit like the 1st Edition of Audio Cyclopedia!

You want Radiotron 4th.

Of course. I was only referring to the question/answer format of this thread. I have pete millet's RDH pdf, chap 5 is great, the math made my head hurt at first, but it's getting better. Maybe I'm developing callouses!  ;D

CJ, thanks for the explanations and image, that helps a lot!

PRR's comment made me go and pull RDH off the shelf again (thanks for that!) and to answer my own question

Is it normal to wind layers back and forth (left to right, then back right to left)?  From what I've seen it's sometimes done in one direction only, ie left to right then all the way back to left in one swoop, to start the next layer, left to right.  What is the benefit of winding only left to right (is that what you call foldback?).  It would seem that this might make the coil look like a few capacitors in series, but then that's what pie-winding is for, right?  Is pie-winding the only way to "series-fy" winding capacitance? 

I think that winding layers in one direction only (L-R than going back to left to start again) would effect  capacitance between layers?  If we consider the start of the winding as 0V, and end of the layer as 1V, if you start the next layer at the end of 1st layer and wind back towards the beginning, you'll have 2V (end of 2nd layer) right near the zero (begin of 1st layer).  BUT if, after the 1st layer, you go all the way back to the beginning and start winding the 2nd layer in the same direction, you'll have 1V difference all along the layer, never more never less.  Thinking about it, the average V difference will end up being 1V anyway, but with the second method, it's always even.

Am I more or less on track there and is one better than the other?

 

[CJ]

you explained the fold-back winding better than i could,

bringing the wire back across the layer creates additional space between layers. which would mean less capacitance, but the main reason is to keep the voltage between the windings down which keeps leakage C down,

 

[mitsos]

you explained the fold-back winding better than i could,

bringing the wire back across the layer creates additional space between layers. which would mean less capacitance, but the main reason is to keep the voltage between the windings down which keeps leakage C down,

It came out of RDH4 Ch5, thanks to PRR for reminding me to reread it! Glad something from that book sunk in, it only took 4 years!  ;D

 

[CJ]

Partridge Articles

enjoy!

http://www.ig-transformatoren.com/core_distortions.html?file=tl_files/igtransformatoren/Partridge/Wireless%20World%20%22Transformer%20Distortion%22%20Seite%201.pdf

http://www.ig-transformatoren.com/core_distortions.html?file=tl_files/igtransformatoren/Partridge/Wireless%20World%20Teil%201%20Seite%201.pdf

http://www.ig-transformatoren.com/core_distortions.html?file=tl_files/igtransformatoren/Partridge/Wireless%20World%20Teil%202%20Seite%201.pdf

http://www.ig-transformatoren.com/core_distortions.html?file=tl_files/igtransformatoren/Partridge/Wireless%20World%20Teil%203%20Seite%201.pdf

http://www.ig-transformatoren.com/core_distortions.html?file=tl_files/igtransformatoren/Partridge/Wireless%20World%20Teil%204%20Seite%201.pdf

 

[mitsos]

hey CJ, thanks for the links. I got one of them to fully download but the rest I get only one page..

maybe it's my 'puter, gonna keep trying.

Can't wait till this weekend, gonna pick me up a new 10K10K and a 10K:150+150  8)

 

[CJ]

those articles are not complete but the main stuff is on the front page,

don't worry about the math and the details, just go for the overall concepts like flux density vs inductance, etc.

 

[mitsos]

I did some Fr response charts with a solid state pultec in another thread to show the response of a UTC A21.  In the same test I threw in the 10K:10K on the input, it's pretty flat, but is about 11-12 dB below an HS56 "remake" (wired 600:600) in the same position.  I think I have seen people have using 10K trafos for pultec inputs before (using the same 620R load resistor), but I'm sure they were more "modern" transformers, with much less DCR.  Is the level difference due strictly to DCR?  Just to refresh memories, DCR of the 10K is about 1K3 Ohms while the HS56 is about 50-60 Ohms.

thanks!

 

[CJ]

could be the core loss,

the dcr is not too bad for a 10K:10K

if you use a core with higher perm, the turns will come down and thus the dcr,

make sure you adjust the input level when you change transformers,

 

[mitsos]

The 10K:10K core is 625EI, half M4 CRGO steel (something like 10mils) and half 14mil nickel (I think 49%).  The particular 600:600 is M6 steel, EE lams like the HS56, just steel. 

hysterysis, eddy currents, I'm lost... gonna run the tests again when I get home tomorrow.

BTW, I picked up the new coils 10K:10K and 10K:150+150, they look nice, using 39AWG for the 10K windings (38 didn't fit with all the shields and paper insulation) and 33AWG for the 2x 150 Ohm coils. No measurements yet, have to wire them up, won't get to it until tomorrow night or monday.

 

[PRR]

> about 11-12 dB below .... DCR of the 10K is about 1K3 Ohms

Do math. You have a voltage divider with 1300 on top and 620 on bottom. 1:0.32 or -9.8dB.

> I think I have seen people

People shouldn't use 10K:10K to drive 600.

 

[mitsos]

You're awesome PRR! Thank you!  Picturing the 1K3 as a series R is thinking way outside the box for me.  I had a suspicion about that 620R, but like I said, I remember some posts about people using 10K:10K on pultec inputs in some old posts. Anyway, it was the only circuit I had to quickly test this thing, (I had originally stuck it on the output ,which is closer to where it will end up) so I'm glad to hear it's normal!

Have a good saturday night!

 

[ioaudio]

> about 11-12 dB below .... DCR of the 10K is about 1K3 Ohms

Do math. You have a voltage divider with 1300 on top and 620 on bottom. 1:0.32 or -9.8dB.

just nitpicking but is'nt it 1K3 for the primary plus 1K3 for the secondary - 1:0.19 or -14.4dB ?

 

[CJ]

<EE lams like the HS56, just steel...>

i think the triad hs 56 might have  50/50 lams, radiometal, whatever you want to call it, which is between M6 and super 80.

looks darker than 80 super, but if you scrape the gunk off it you can see a difference,

 

[mitsos]

Yes it kinda looks like steel but different, much softer, and reads different than M6, 49% or 80% lams.  Definitely nickel though, I didn't want to take chances and had some tested. I think mfg standards or alloys have changed since then.  Judging by the uneven winding job, maybe they also cheaped out on the lams back then? 

just nitpicking but is'nt it 1K3 for the primary plus 1K3 for the secondary - 1:0.19 or -14.4dB ?

had to check the trafo equivalent circuits, they show the secondary DCR on the primary side, but PRRs number makes more sense, cuz the trafo graphs a bit lower than that, which might mean that the trafo (quite possibly) has other losses as well. The coil is far from perfect, (but the next one will be better, I hope).