Advice please on winding an audio transformer.

[Radio Wrangler]

There are a couple of questions you haven't asked, and as they set the foundations of later things, it helps to have them out in the open. Let's take a couple of steps back, and look at things from a different direction:

What is the function of the output transformer. Why have one at all?

1) It acts as a safety element, allowing the loudspeaker connections to be at or near ground potential, isolated from the high voltages needed by valves - even at flea power.

2) The loudspeaker impedance is inconveniently low compared to the optimum impedance that your valve(s) should be loaded with. The transformer lives up to its name and transforms the impedance your speaker presents into the impedance your design work says you should load the valves with. Your valves could drive the speaker impedance directly, but the power you could get would be very low and the efficiency would be lousy.

3) The transformer allows push-pull amplifier operation without needing (or faking) complementary devices. You can make push-pull amplifiers with one valve pulling up and another pulling down, but they are very complicated. Having a push-pull transformer is the easy way out.

Point (2) above raises the question; 'what is the optimum impedance to load your valves with?'

To answer this, you get a plot of the anode characteristics of the valve you've chosen and decide what is the maximum anode current you want to run the valve to, and what is the maximum anode voltage.

Now it's time to pin the tail on the donkey... well, two tails, actually.

For point 'A' find the horizontal level of the max current you've chosen, and pick a point along it where the anode current has come out of the curvy bit at the low voltage end, and is looking flatter.

For point 'B' the current is zero for a class B design, (or a bit above zero for class AB).
Pitch the voltage of point 'B' halfway between the voltage of point 'A' and your chosen max anode voltage. The voltage of 'B' is also the HT voltage from your power supply. B here is the name of a point on a graph (Americans use B+ as their name for the high voltage supply rail where Brits say HT. This is just coincidence, I could have named those points Fred and Charlie).

Using a ruler draw a line through 'A' and 'B'. Extend it rightwards up to your chosen max voltage. This will go into negative anode current regions on your graph, which looks silly because valves don't conduct in reverse. But this is a push-pull amplifier and the 'negative' current is current from the other valve, reversed by the transformer.

The slope of the line 'A' to 'B' shows a linear drop in voltage versus current... A linear resistance from the power supply.

It is this resistance that the output transformer transforms the resistance of your speaker to make.

Have a general read up on 'load lines' now and it should make things a lot clearer about what is going on. You are NOT trying to match the anode impedance of your valve. If it comes close, it is only coincidence, nothing more. You are setting the impedance presented to the valve. Given an HT voltage, and a number for the amount of power you want, the presented load impedance comes out of the maths with relatively little influence of your choice of valve. At first, this looks surprising, but it doesn't half make life easier.

Once you've found the load line impedance, you get your transformer turns ratio from it and your speaker impedance.

Next, you need to know how much inductance you need to go as low as you want in frequency. Primary inductancetimes 2 times Pi times Frequency = Load line resistance will give you your FL -3dB point. Pick a -3dB frequency and plug in the numbers to get your primary inductance..

You now need to design a transformer for this inductance, with a core big enough to be clear of saturation with a primary current od point A on your load line.

It all fits together like a jigsaw. I was a bit concerned that you were looking too intently at the pieces and not at the picture on the lid

David

[AdrianH]

Thanks David points 1 and 2 I can fully understand, sorting out the load line without a resister load I can not follow properly, perhaps it is me being thick (probably), perhaps because of the valve I am using the charts do not show a lot of detail.

I am running each valve at 6 mA anode current, I am running the HT at 150 Volts, that is the minimum the buck converter will turn down to.

I have attached the spec sheet for the valves this little amp uses. On page 5 there are two charts, both are Average plate characteristics the top: plate current against grid volts and the bottom: plate current and grid current against grid volts.

I have also copied and blown up the bottom graph, making it bigger so I could print this out for myself, it was this I used to decide on my cathode resistor to give the required -ve grid bias. I determined that to run at 6 mA I needed a grid Voltage of -2.4 Volts, so would be correct with a 400 Ohm resistor, because the circuit calls for all 4 valves to share the same cathode resistor, this is divided by 4 to give 100 Ohms and when in circuit I get very close to the 2.4 Volts across this resistor 24 mA through 4 valves (hope they are balanced?).

I am not sure how these valves are operating, It was suggested to me class A as they are in the circuit, but I am not sure?

Could you please, or anyone, do me a favour and plot on the chart the load lines as per your description, I am hoping it then becomes clearer to me.

Many thanks

Adrian

[retailer]

Some guitar amps do have distortion built in, one of the amps that I copied the 1965 Fender Deluxe (21 watts output) was not made with built in distortion Fender amps from that era are known for their clean head room, they can be overdriven which is where the distortion comes from and can also be used to give a clean non overdriven sound, depends on the skill player. Guitar amps have a tone stack that alters the frequency response from being linear but does not introduce distortion.

I have the calcs I used for the output transformer - the core was 25.4mm x 32mm, using a variacI determined that the turns per volt would be 6 turns per volt, I went with same anode load that Fender used so primary impedance was 6600ohms. To acheive 21 watts output with 6V6's I needed around 350-360 volts so 6 x 350 = 2100 turns, this was the number of turns I used for the primary. With no load my B+ anode voltage was 415v, at clipping that dropped back to around 395v, output measured into a dummy 8 ohm load was just over 13 volts, pushing it just past clipping I had just over 14 volts - near enough to 21 watts. I didn't mention inductance, as it may have added an extra thing to think about, I recall agonising over this for a long time before I actually wound a transformer.

There are many ways to design your transformer, this is just one of them, it may not suit everyone but it does give me the results that I want.

With regard to the total number of turns, it would be 1350 and will need to be split with a centre tap so 675 turns for each half of the primary, with my Fender transformer I split it into 2 windings of 525 turns each and a 3rd winding of 1050 turns.

[AdrianH]

On my transformer, the primary turns are close to yours, my primary was 685 turns for each primary so across plate to plate (anode to anode) I had 1370 turns. Actually I dropped this sightly as I changed efficiency from 0.9 to 0.95.

I think there is a difference in my terms with the norm, I am assuming the impedance is per valve or per side of the transformer. I just wound both primaries at the same time (343 then the secondary then the 343 for the next side.

Any way it works and sounds OK, but the amp is operating in class A With the grid at -2.4 wrt cathode if I start to over drive the input then the input starts to rectify and drive negative as the grid will not go above around 0.5-0.6 volt positive.

A 6 Volt peak to peak voltage on the grid of the finals gives a 60 Volt p-p voltage on the anodes and 2.6 Volts p-p on a 4 Ohm load.

This is not driving it into distortion and keeping it clean.

So the output is flea power at 210 milli-Watt.

I will change the cathode bias and try to get the cathode at 4.5 Volts dropping the anode current down to 1.5 mA and see what happens, this is all about learning, I may be pushing class AB but really not sure if the cathode biasing will allow it?

Adrian

[AdrianH]

The change of cathode bias did not work, looks like the unit will not work in class anything but A, which I guess is down to the very simple means of having the push pull arrangement, so at some time i will have to look at a more complicated set-up.

Adrian

[AdrianH]

This should be my last post on this and this is just a measure to see how it performs with the home made transformer.

Not really sure how others check out responses, but I am using a Com120B radio test set to both generate the tone and measure it, it has wide band settings in the audio section. I also used a 4 Ohm resistive load, rather than annoy the wife.

I set up at 1KHz to provide 5% distortion through the amp, (as setting I was used to using in PMR)
The test set will measure dBm into 4 Ohms, then with a constant set output went down and up to 20KHz output frequency.

20Hz 10.2 dBm*
50Hz 17.1 dBm*
100Hz 19.1 dBm*
200Hz 20.2 dBm*
250Hz 20.4 dBm
500Hz 21.0 dBm
1KHz 21.2 dBm
2KHz 21.4 dBm
4KHz 21.5 dBm
5KHz 21.5 dBm
8KHz 21.4 dBm
10KHz 21.2 dBm
15KHz 20.3 dBm
16KHz 20.1 dBm
20KHz 19.0 dBm

* I could notice some distortion on the waveform at these frequencies suspect hitting saturation, a sort of flat topping of the sine wave but only on peaks.

Looking OK to me, no not hi-fi but not that bad, I hope the roll off is down to the cap I put in across an input to knock off top end, but for my first attempt at winding a transformer, I will be trying again when i build a better amp and will try to understand more of the theory and practical aspects of it.

Adrian

[Ed_Dinning]

Hi Adrian, close enough for practical purposes; you can see the effect of the low number of turns at the LF end.

Ed

[AdrianH]

I am reading a couple or Wolpert articles on transformers, Slowly!

I am wondering if margins are used on windings when using the plastic bobbins, or do the side of the bobbins act as the margins.

Of the couple of small mains transformers I have stripped down to check they are using the full width of the bobbins. OK they are in sections primary is separate to secondary, but when doing an audio transformer in a plastic bobbin would you still have margins?

Last question is there a simple way of determining the material used in the laminations, some are a brown colour some are a clean steel finish?

Adrian

[retailer]

You don't need a margin if using a bobbin, the bobbin cheeks support the sides of the winding and also provide insulation between the winding and the core. Winding a transformer by hand with a tube instead of a bobbin is do-able but not easy, if you are starting out winding transformers just stick to bobbins.

Some may be able to determine the core material just by looking, I can't, the brown colour you refer to may be a thin layer of potting varnish, if repurposing a transformer the varnish should be removed, don't sand or scrape it off as you will most probably destroy the thin insulating layer on the laminations, use solvent, it is a smelly and pleasant job, leaving it there will affect the stacking factor and also make it difficult to get all of the laminations back in, this is important if you are rewinding a transformer that is part of a vintage amplifier or similar as you really need to rebuild the transformer so it is identical to the original, if it is a new wind on a salvaged core and you don't remove the varnish the transformer will still work but may lose some efficiency.

[AdrianH]

Thanks retailer, I think the brown colour is not a varnish or lacquer as it is the colour even in the centre of the transformer, I should have done this before, so here is a picture a bit late sorry.

Thanks for the answer on the margins. I have purchased a cheap winder from Ebay, it was less than £19.00 (UKP). It needs a bit of fettling to make it work a little better and I have done some work on it, I am probably not going to be winding many transformers, but it will certainly save a lot of effort. I am on the look out for a small variac and it will probably take a while to get one at a reasonable cost. I find it strange that I can get small transformers cheaper then buying laminations and bobbins separately, I guess down to scale.

Adrian

I should say that that of the 6 laminations shown, the three on the top row are very similar finish, being a brownish colour and the 3 bottom ones look more like plain mild steel to me although they are probably not.

Sizes are for the two top larger ones 1 7/8 x 1 1/4 inch
The middle top is the size I used on my audio transformer is 1 5/8 x 1 1/16 inch
The three smaller ones are 1 3/8 x 15/16 inch and that transformer was 1.5VA 15 Volt at 100mA.

[kalee20]

What is important, even using plastic bobbins with cheeks, is to use a mandrel which supports the bobbin as much as possible.

Otherwise, you find that the pressure of the wire forces the cheeks apart, and they splay further and further out. The insulation, which you cut wide enough for the initial layers, no longer works for the later layers and interlayer short-circuits become a real possibility.

At the most extreme, the cheeks are forced so far apart that the laminations can't be fully inserted, enough to butt together.

Guess how I know all that!

[AdrianH]

I am slowly building up a few different bits to have another go at it, hopefully soon some more wire will arrive so I should have some basics,
34 AWG 0.16mm
36 SWG 0.19mm
26 SWG 0.45mm
22 SWG 0.71mm all some dual polyester grade 2 wire
I have some polyester high temp tape used for powder coating which is 3 thou think, I used that between primary and secondary windings when I did the audio transformer.

I should ask about putting wire on every layer, but when I stripped the old bobbins down, non of them had any interlayer tape and they were running 240V ac.

It has not been an issue yet on my audio transformer I guess because of the lowish voltages I am using, but I wonder at what voltage the coatings would break down?

OK kalee20 on the cheeks being forced open, I admit I shaved a piece of wood down to fit inside the bobbin with an M10 bolt going through, spin the bolt type thing, I have recently been making a couple of new cones for the new winder. It is all a learning curve as it is for me on valves in general, but it is fun during these times.

Adrian

[Diabolical Artificer]

Brocott is a good place for copper wire. I've tested my homewound tfmrs @ 10kv and had no issues, this between windings and also tested some grade 2 insulation.

Andy.

[AdrianH]

Hi Andy;

Brocott is where I have gone to for two coils, I had the 36 swg from Maplin many years ago, the 26 from scientific wire on Ebay as it was cheaper then direct.

[Ed_Dinning]

Hi folks, Brocott is sometimes a bit expensive. I tend to use wires.co.uk as they have a wide range including Litz

Ed

[AdrianH]

I found it cheaper at Brocott for the thwo I go, wires.co.uk and scientific wires are the same out fit one for bulk I guess and one for the fancy knitted coloured wires!

As always it can pay to shop around.

[AdrianH]

I have found a couple of old transformers in my boxes that were for 6VA mains, both are the same size at EI48, I will have another go at winding some audio transformers as I want two of the same, spurred on by a few of the other posts in homemade about audio amps, I have built up two of the mini boards on to a bit of bent aluminium.

My TIG welding is not fantastic but the chassis will hold up. I picked up some Panasonic speakers from the bay so thought I need a stereo valve amp.

The larger laminations will hopefully alow me to do more turns per volt.

Adrian

[AdrianH]

As my thread has not been shut it gives me a chance to ask a few more questions and just check on things. I am about to wind another couple of transformers to finish this little project off for myself.

I have 'butchered' three EI-48 sized transformers, these from what application they were in were typically 6 to 7VA rated and I plan to wind as a 1 watt audio transformer, so sticking to the previous impedances 57 Volts primary and 2 Volts secondary, 3250 and 4 ohms respectively.

The actual amp can only do 200mW clean, so that gives me loads of head room and a chance to use the transformers again later.

I have made a couple of bobbins, as I seem to struggle finding anywhere that sells then privately.

When I measured outputs of my small amps using my scope, one with the hammond transformer and one with my first self wound tfmr, the one with the Hammond seemed to distort the output sooner. I think this was down to two things, the see-saw push-pull output, is simple but not the best design and the fact that one side of the primary had a higher DC resistance.

Looking on the scope one output pair had a larger peak to peak voltage then the other output pair, and it seemed to follow the transformer if I swapped the primaries around.

The one I wound was done with equal windings as both primaries were wound together so at the end the same DC resistance in both sides. The scope showed the peak to peak voltages on both legs to be very close, within a volt or two.

I did a few calculations for turns per volt.

For the 1 watt idea:-
1/(4.44 x 40Hz x 0.00032m^2 x 0.7)

is 25.14 turns per Volt. primary would be:-
57 x 25 = 1433 turns per side so 2866 total primary,
My window is 22mm by 6.2mm and I have 122 turns along the 22mm with my wire so height of primaries is 4.2 mm so then I would need 50 turns of secondary so if I am very careful I might just squeeze all that in with some insulation.

If I worked it out for the 200mW I could go down to 20Hz
TPV=1/(4.44 x 20Hz x 0.00032m^2 x 0.7)

is 50.2 turns per volt, but at 200mW the voltage would be 25.5 Volts rms so a total of 1280, both primaries 2560, same wire a little bit more space.

But, a couple of things I am not sure of, is the formula anywhere near correct?

Will I be just making up a lot of capacitance between each push pull pair winding the two primaries at the same time?

All this lot is basically me putting all my thoughts down to see if I am on the correct train of thought, and a way my brain seems to work. The more I read the more confused I can get and there is a lot of information out there.

Adrian

[retailer]

Larger laminations or bigger core area usually means you can go less turns per volt. You might find the attached file interesting.

[AdrianH]

Hello retailer;

Thank you for the article, it looks like a good read, I will print and read it later, I am finding reading from paper better.

I have wound two transformers to complete my small amp project and apart from some tyding up it is completed enough for me to move on to something else.

I will keep my eyes open for some larger transformers to re-use or find a local supplier of laminations and bobbins for winding sessions.

Many thanks

Adrian