Roll your own Output Transformers

[CJ]

if you are on a battleship and bombs and machine gun shells are going off, you do not care too much about s/n ratios and frequency response,

so a lot of reduced figures are for military products and have nothing to do with their actual response unless they are marked for 20 to 20Kc,

so there are a lot of good buys out there if you know what the actual response is,

now when you put DC on an inductor, that funny curve you see on the above graph i posted turns into a straight line so you can not see much difference between it and the same coil around a core stacked differently

here is a graph of the same inductor stacked for mid-range with 1 ma DC,

everybody thinks that DC affects only the low end so that is why SE transformers are bigger, to make up for the bass, well  that is true but the high end also takes a hit, there is little inductance from the core after 2 k hz, so what ever is left gets wiped out by the DC bias on the core,

 

[All Ears]

Aha - so I guess that means in those 'radio' or milit stuff that those freq specs might be the specs measured AFTER DC has been applied to one of the other coils - that could cause a chop off the top and bottom as you say and could band-limit it....i think that is what you were getting at

It is certainly weird to think also that some of these 60's- 70's trafos I have (made by Freed or Triad for the military) might have actually 'seen combat' in the Vietnam War...

Have you got a graph which shows the effect of 'past experience/first ownership' on 2nd hand transformer performance ?  ;D

Anyway - my laminations and bobbin are on their way to me for my bifilar output jobs...I went one bigger than the usual 625EI...I chose 75EI instead

...and for lams I also went with 0.011 instead of 0.014...probably will use a stack of something like 25mm 0.011 per bobbin...that should have pretty decent bass response I would think.

I chose 28 gauge wire...but I was thinking I could have gone with 27 gauge or something a lot thicker but for this first test I am going to make an output trafo with double the number of turns of the 2503 and I wanted to make sure I could get them in - I am not familiar with 'window fill' or packing factor so this will just be a test. The wire was cheap and the lams are not expensive so I may have a few shots at this. If I have the time I will refine wire gauge and number of turns depending on how I like the sound. I am thinking that thicker wire has a lot of benefits in an output transformer - thicker is better as long you get the turns in ???

Also I'm wondering if there a rough and ready formula to determine turns available according to window fill or packing ? I can find the cross-sectional area of the wire but that does not tell me how the round wire may or may not pack in. I have been going with a rough idea of 40% available for wire out of the total window space...but perhaps that applies more to power transformers where there is lots of insulation and multiple windings etc. 40% sounds pretty minimal .

Looking forward to my 'weekend winding' soon (must be the equivalent of Sunday driving)

 

[CJ]

just figure out how many sq inches you have for wire on the bobbin, measure the winding width by height.

eg:  a bobbin for 75 EI has a winding width and height of 1 inch by 5/16th inch.

1 * .3125 is .3125 sq inches,

#27 wire gives you 4110 turns per sq in.

so .3125 * 4110 = 1284 turns,

divide by half and pri turns can be 642 minus room for tape, etc.

turns per sq inch are found on some of the wire charts but not all,

those xfmr specs you are talking about are not caused by DC, i just posted that chart as additional ino not related to the question, 

so expect 20-20k performance form a lot of the stuff you see marked down to show that they meet the basic requirements for the buyer, which in a lot of cases back then, was the military,

 

[All Ears]

Well cool...nice to know my calcs weren't far out

So now...

Using this...
http://www.magmet.com/lamination/eiindex.php
(I guess you meant 0.375" for window height for #75)

...plus this:
http://www.vias.org/eltransformers/lee_electronic_transformers_03_04.html

...and garnishing with this...
http://136.159.225.44/AlliedData2.pdf

I can get a much better estimate - cheers !...  8)  invaluable

Must say...i hate having to convert everything to imperial though (duck for cover) glad the Magmet site had the metric version as well

 

[CJ]

i measured the actual bobbin width and height, so if you use the window area, it will tell you that you can get a few more turns on there if winding right over the core leg,

so you will have some paper for a core tube, or a slightly thicker bobbin, which can be sanded down to match the paper former,

nice links!

 

[bruce0]

If you calculate the areas of the "circles" that will fit (circle stacking theorum or something), that area will be 90.69% of the available area, but in practice I wind and things cross and I get 85% yield, or a little better.

 

[All Ears]

ha ! I didn't even think of measuring the actual bobbin window...  :-[

and yes Bruce, table VI here (http://www.vias.org/eltransformers/lee_electronic_transformers_03_04.html) seems to show the space factor of 85% - 90% as you suggest.

Obviously external lead insulation at the winding start may cause a bulge but I can use spaghetti tubing instead of normal lead insulation

 

[All Ears]

Cores, eh ?

Anyone know for an output trafo, or any form of audio transformer for that matter, if a strictly square core is better than a rectangular core ?? - i can make either depending on the bobbin i choose.

I guess it is time to have a smoke and count laminations...

 

[bruce0]

Mr Ears:

I really appreciate that link.  I have been perusing it since you sent it.  Fantastic breadth and clarity on transformers.  I highly recommend it to anyone that hasn't seen the Reuben Lee site

ha ! I didn't even think of measuring the actual bobbin window...  :-[

and yes Bruce, table VI here (http://www.vias.org/eltransformers/lee_electronic_transformers_03_04.html) seems to show the space factor of 85% - 90% as you suggest.

Obviously external lead insulation at the winding start may cause a bulge but I can use spaghetti tubing instead of normal lead insulation

 

[All Ears]

Yeah - I also dug the attitude of whoever founded that whole VIAS website as well...
http://www.vias.org/about.html

And i will also get into some of this:
http://www.vias.org/electronics.html

it is incredible the kind of information you can get these days...pretty cool time for DIYers, inventors, mad backyard scientists or anyone else who likes self-teaching etc

 

[PRR]

> square core is better than a rectangular core ??

You mean Square Stack?

You want a maximum of iron with a minimum of wire length. Cylindrical core is best (and sometimes seen in HIGH power iron; approximated with 5 or 10 tongue-widths). Square is always a good bet.

Alternative is to read and digest Grossner. Do your math. A year later, you realize that for a one-off, you coulda just tossed it off with any reasonable yet generous proportions and been ahead of the game. (Exact optimization is vital when you make a million of them. One-offs can become try-again two-offs and still be quicker than heavy thinking.)

 

[All Ears]

Hi PRR,



I am digesting little by little...it is just a learning process for me - so it is all good education. You are right that it is not a good idea to let all the math swamp you. I am combining theory in small doses with practical experimentation. 

The 0.011 lams looked so so well made and were cheap so I will be winding a at least couple of pairs of different versions of o/p trafos with different wire gauges (and thus turns). I will probably make a gapped pair as well using some 0.08mm kraft paper at some point also. Then I will sit back and compare the versions with a good listening test in the right circuits which will probably be more informative than using the signal generator.

So far all good fun - no tears yet.  8)

 

[All Ears]

Just thought for completeness I would close off this thread with some pics of the transformers I wound back in May (haven't been here for a while)

Thanks to all who helped.

I wound 3 bifilar pairs on the same 75EI, 0.011 lams - each with a core stack of 25mm.

Firstly - Pair A...
- 312 turns 26 gauge wire
- DCR 5.25 ohms per coil (measured before lams inserted)
- 6.0nF capacitance raw, 6.8nf finished (post varnish)
- 530mH per coil when first assembled, falling to closer to 400mH when fully assembled and finished.

I can guess why the capacitance goes up after being soaked in varnish...but any ideas as to why the inductance would fall from the time of initital assembly to the final product ? Crappy LC meter ?

 

[All Ears]

Pair B

- 786 turns 30 gauge wire
- Capacitance 14nF initially, rising to 16nF finished.
- DCR 31.8 ohms per coil
- Inductance 2.4H

These had quite a lot of bass due to inductance I guess, but also had a bit more sizzle at the top - capacitance I guess.

 

[All Ears]

Pair C - Gapped Pair

- 366 turns of 28 gauge wire
- Gapped EI
- Capacitance 7.36nF initially, rising to 7.5nF post varnish
- DCR 8.5 ohms per coil
- Inductance 210mH per coil initially, falling to 200mH finished product.

These gapped ones will have to wait for a Neve project I guess. I just did them as a learning exercise.

Currently, Pair A are being used as the outputs in a 2 channel DOA preamp.

I guess I learned a lot in the process...probably realising that machines can wind these things a lot better and tighter than humans can (or at least DIY amateurs). These are probably not the most beautiful transformers in the world and probably their specs are not even that great, but as I said...all part of the learning process.

Cheers,
All Ears

 

[joaquins]

Nice job...

...
I can guess why the capacitance goes up after being soaked in varnish...but any ideas as to why the inductance would fall from the time of initital assembly to the final product ? Crappy LC meter ?

When I did mine, in order to get the I lams in place, I pull them with a clamp so in I look for the moment when inductance takes max value and stop, when I take them out of there it cames a little bit down, when I put the squad (I don't know how you call the metal sheet around the tx) I use the clamp again and it raise up again but never the max value I get first... I don't know how you build yours...

In order to get the flux inside of the core and don't interfere with close input TX you could put a cooper sheet around the TX (outside windows) and get it shorted... But I don't know if it affects inductance in any way...

JS

PS: I use 3903 3M tape to fisnish them nicely...

 

[All Ears]

Ah yes - that could well be it...the clamping must affect inductance

Now that you mention it, I remember after I first assembled them (see photo below) and they were tight enough to hold their shape nicely, but nevertheless still a bit loose, the inductance of Pair was reading 530mH. After I put the clamp shell on, I saw how it tightened up the lams a lot...and at that point I guess the inductance fell. I wouldn't have instinctively thought this would happen, but transformers are full of laws I know not very much about.

Pair A was the best sounding to my ears for my current preamp project...but it was interesting hearing the more typical scooped bass response of the gapped pair, Pair C...these I compared to a pair of UTC A-22's and actually liked mine better than the UTC's (which were a bit too light on the bass and a bit clinical). But parents are always biased.

Yes - i have some copper for a shielding strap...I see the API preamps with those, but don't quite know the theory of how those work. Should the copper strap short itself or not ? Would a shorted single turn act as a kind of winding and suck power and affect other parameters or quite the opposite ?

 

[CJ]

when you clamp the lams, sometimes the inductance goes down a bit, one possible reason is that you are moving the core away from the windings by just a bit, this will increase leakage inductance a bit, which is why you see a lower reading,

you can see this by connecting the LC meter and squeezing the lams by hand and watching the meter,

that copper strap can go all the away around without being a short, it would be a short if wound around the center leg, but outside is ok,

 

[All Ears]

It all makes perfect sense.

You wonder if a tiny bit more space between the lams might be a good thing if you can't squeeze one more lam in.

and yes - I'll roll them in copper - at least it will make them look like a surplus NASA part from Viking or Voyager if nothing else.

... "Transformers: More than Meets the Eye" by CJ

 

[zebra50]

>You wonder if a tiny bit more space between the lams might be a good thing if you can't squeeze one more lam in.

Further to that observation, if you can measure L at more than one frequency, you may find if you squeeze in an extra lam that the inductance increases at some frequencies and decreases at others. There is a science project all of its own!

Have you found a good reliable believable way to measure winding capacitance? Certainly on guitar pickups I have found it very difficult to get measurements that tally well with real-world sound. For ribbon mic transformers, which is most what I play with, it is not a big issue.