OT: a photo essay on DIY output transformers

[brett]

Hi
I'm no expert about electical engineering or output transformers, but I've made a few of these and found that it's a lot easier than I though it would be, and is very economical for small output transformers (e.g. 1 to 20W).  In my case, I couldn't get the transformer I wanted (24:8 ohms) anywhere, so it was the only option.

These comments apply to coupling transformers as well.
Let's say I wanted an output transformer for coupling a 24 ohm push-pull pair of output transistors to an 8 ohm speaker (coz I do).  It needs to handle at least 5W, and possibly 10W if ever I use a 4 ohm speaker.  So I started with this 15VA output transformer (DSE M2860 Oz$5).  Larger transformers are good because they need fewer turns of wire on them, and have space for heavy guage wire, which reduces losses.  So choose one with a VA rating of 2 to 5 times the wattage you require.



This transformer was held to gether with a metal binding around the outside, which was easy to remove.  Inside, the laminations were held together with a very weak glue or oil or something.  Most wall-wart transformers are glued together with stronger stuff.  I use pliers to remove the first lamination (which is often ruined in the process, but that doesn't matter), then unglue and remove the rest by separating them with a utility knife and pulling them out by hand. 

Here's the laminations and bobbin.



Sometimes, the laminations are bolted together.  In this case the laminations usually seem to be oiled rather than glued, which makes things easy.  Make sure that you keep the insulating washers with the bolts.  Not insulating the bolts causes problems of the kind we really, really, don't want.



I wanted the wire from the secondary winding because it was a handy size (about 0.4mm diameter).  I wound it onto a tube of silicon sealant).  This transformer had the equivalent of $5 worth of  0.4mm enamelled wire.  (I'm so cheap it's ridiculous)



I didn't want the primary.  It is usually very thin, and only useful for winding high-resistance, low-current primaries for valve amps.  Output transformers for valve amps have primaries of  5000 ohms or more - very different to the 24 ohms I wanted for my transistor amp.



Once the primary and secondaries are off the bobbin, I cut off the divider that is between them.  It's there for safety with mains voltages, but no use to us (leave it there if you want).



That hole through the middle is where the "core" of the transformer sits.  Measure the area of the hole (in square centimetres), because it is important in some calculations that come later.



We also want to measure the width (in centimetres) of the "tongue" of the "E" laminations.  Multiply that distance by 6 to get what is known as the "Magnetic path length (cm)".  Another way of estimating this is measuring the "average" distance around one "loop" of the two "loops" that make up the transformer.



Decide the lowest frequency (in Hertz) to be transmitted by the transformer at full power.  This is usually low E, equal to 80 Hz.  From this frequency, calculate the required inductance (L, in Henries), from L = resistance/(2*pi*f)
where resistance is the resistance of the primary.

From L, we calculate the number of turns on the primary (Tp), which is equal to:

sqrt((path length/10*L)/(4*pi*0.0004*core area/10000))

The number of turns on the secondary (Ts) is equal to:

Tp  * sqrt(secondary resistance / primary resistance)

For this transformer, the path length was 11.4 cm, the area was 3.6 square centimetres and the required inductance was .048 Henries.  This gave 175 turns for the primary and 100 for the seconday.

Now, choose some wire that won't overfill the bobbin.  I have a spreadsheet for calculating "window fill" and "copper losses", but some general rules seem to work, too.  For low-impedance windings (under 50 ohms) use 0.4 mm or heavier wire.  For medium impedance (100 to 300 ohms) use wire of about 0.2mm.  For valve transformers, with impedances of several kohms and several hundred turns on the primary, use "angel hair" wire, as shown in the photo of the primary , above.

You'll often want a "centre tap" on the primary.  If so, wind half of the primary and tie it off, then wind the secondary, then finish with the last half of the primary, started at the wire that was the end of the first half.

Here's a typical start, and fifteen quite ugly windings.  It doesn't have to be too neat, but it should be better than this.





Here's the finished bobbin.  Winding it took about 15 to 20 minutes.   I put some solder onto the starts, finishes and centre tap.  A hot iron (400 C) helps to  melt through the insulation (and the bobbin too if applied too long).  Next, measure the DC resistances of the primary and secondary to check that everything is connected.  The resistance should be 2 to 8% of the target AC impedance.  If higher, consider using heavier guage wire.



If there's an open circuit, then one of the wires might be broken or you haven't melted through the insulation at the terminations.

Re-insert the laminations, in a similar way to which they came out.  First the Es from alternate sides, then the Is, starting in the middle and working out.  You probably won't get them all back in.  In this transformer I got 38 out of 39 back!



This is the finished transformer.  I measured the inductance at approx 50 milliHenries (the target was 48).  After shooting this photo, I put some craft glue around the edges and put the metal case back on it.



All up this took a couple of hours, but at least half of that was the photography.  The transformer work too about one hour, including the maths.  My frst one took about two hours.

cheers

[MKB]

Very nice!!  It really isn't as hard as people think it is.  Also, there are some great sources for finding info on the internet to wind the OT's.  Here's a link to probably the best out there: http://www.pmillett.com/Books/FTR_ref_data.pdf  This book is Reference Data for Radio Engineers; in it is a chapter that gives a by-the-numbers method for designing and winding an OT.  But indeed your method is much simpler.

One thing to be careful of is to make sure you have some insulation between your primary and secondary windings, especially if you will have a large voltage differential between them in say a tube power amp circuit.  Another thing would be the lamination types; most inexpensive power transformers may not have the optimum lamination formulations, although they will work well.  The source for those would be old blown audio transformers, if they are from high quality amps such as Marshalls or Fenders the lams should be very high quality.  Unfortunately you can't tell if they are audio quality laminations other than performing heavy metallurgy on them (or of course listening).  The lamination type can make a big difference in tone, as the transition into core saturation is different (less sharp) and they in general sound smoother.

It's also very easy to wind custom power transformers; many power transformers have the 120V primary wound near the core, which you don't need to change, so you're already halfway there.  It's easy to disassemble a cheap Radio Shack transformer, remove the secondary, recalculate the secondary for the voltage and current you need, then rewind the secondary and reassemble the thing.  It can be done in a very short amount of time, especially if your secondary voltage is low (less # of windings).  I rewound a 6VAC secondary transformer to a dual secondary isolated 2.5VAC (for a tube preamp project using very old 1930's direct heated triodes) in less than an hour.  The biggest pain is finding a transformer that can be easily disassembled; some of them are vacuum varnished that makes them nearly impossible to disassemble.

[JimRayden]

Thanks brett! I'm off to buy some small PT's now.

---------
Jimbo

[zjokka]

another great, true DIY thread, brett!

thanks
zj

[JonFrum]

Nice work Brett. I saved this page a while ago:



http://geek.scorpiorising.ca/windingchokes.html


There's another page at this site for winding power transformers.

[markm]

Ha Ha!!
That fantastic Brett!
Great job! 

[John Lyons]

Very nice! I need to be doing this or at least getting my head around it.
I have build a few 1-2 watt amps and the Output transformers are small and somewhat expensive. ($30)
Now just to add some time to the day...
Thanks for writing this up and inspiring us.

john

[brett]

Hi
thanks everyone for the positive feedback and the links.  I found it a bit difficult to get information at the start and to understand the maths.  Geofex.com has a page in the tube amp section that was my introduction to the language and concepts.

Another thing would be the lamination types; most inexpensive power transformers may not have the optimum lamination formulations, although they will work well.  The source for those would be old blown audio transformers, if they are from high quality amps such as Marshalls or Fenders the lams should be very high quality.

Yes, high quality laminations (eg M6) are found in expensive output transformers, and their characteristics are better than regular transformer "iron".  For example, their permeability might be 4200 instead of 4000.  But these DIY transformers can be operated well below saturation, where a 10% difference in permeability won't affect the results.  That's one reason why I suggest re-winding a transformer with at least 2 times the VA rating of the unit required (the other is that you get a big bobbin that can take lots of heavy-guage wire).  In summary: Transformers with low quality iron will have lower VA ratings.  Not exceeding 30 to 50% of the original VA rating should prevent distortion due to saturation of a low quality core. 

Basicaudio mentioned $30 for a small transformer.  Here's what I'm paying in Oz:
Model Type Rating $
125A  Universal  3 watts 50.00 (US$40)
125B  Universal  5 watts 53.60
125C  Universal  8 watts 56.00
125D  Universal  10 watts 59.40
125E  Universal  15 watts   69.80 (US$56)

After paying fifty bucks for a 125A transformer for my Firefly, I figured that there had to be a better way.

cheers

[brett]

That link is one of the best I've seen.
It's got lots of good info.