fun with audio transformers

[samhay]

So as to make sure the 'Powers that Be' archive this thread, I figure it can have the running title of "The Axis of Evil Nuclear Terrorist Booster".

Anyway. I re-found a couple of audio transformers recently that I had picked up a while ago for a ring modulator (which I discovered I didn't care for). Feeling like a change, I put the transformers back on the breadboard and started playing with booster ideas. John Hollis' 'Titan Boost' uses on of them, so having played with that I first made it smaller, then moved onto power amplifier topologies.
I used Eagle TL44 interstage transformers. They have a 20k primary and 2x 1k secondaries. They are easy to find in the UK (even Maplin has them) and have a Mouser equivalent (42TM006 I think). These are probably far from ideal, but they are a start.
I figured I would share the schem's in the hope of improving them and also to stimulate some discussion as to whether we should be using transformers more often in our pedals.

The FET design gives a nice clean little booster with a little bit of gain. I left DC on the transformer in the hope of getting some saturation - if I do, I can't see or hear it.

The power amp is a text book transformer coupled push-pull amp. By using an output transformer with a 20k output impedance (if I understand the data sheet), it will not drive a speaker, but gives a monstrous voltage boost - good if you want to hit your pre-amp tubes with a much bigger signal than you can otherwise get with a 9V supply. It draws ~5 mA at rest and ~30 mA when driven with a high gain pedal in front, so I guess this is behaving like a class AB amp. There is a little bit of cross-over distortion, which is not unpleasant, but the input impedance is pretty low (20k) and it sounds a lot brighter with a buffer in front of it.
The next step is to get the MOSFET phase splitter to replace the input transformer in this design. So far, this works, but the cross-over distortion is a lot worse, so I guess the biasing needs improvement.

Edit - I have played around with a resistor from the BJT emitters to ground in an attempt to keep the transistor gain in check. I think I need a little help to understand how to choose a sensible value for this.

[tca]

Looks interesting! You should try and make a *real* amp in this way.

Cheers.

[samhay]

Looks interesting! You should try and make a *real* amp in this way.

Cheers.

I would imagine the 2N5088s would probably not generate a lot of clean output, but I do have a LT700 output transformer that I could use for such a purpose (well, it is designed for a transistor radio I think...). Problem is that I don't have a cabinet and am loath to plug this into my hi-fi speakers.

I guess I should mention that the MOSFET design is really a clean boost pedal that would go somewhere near the start of your pedal chain. On the other hand, the pseudo-power amp would go immediately before the amp (although the jimi photon's of this world would probably enjoy pounding a fuzz with it too).

[tca]

^ I do have a cab! And a similar transformer... just not the time to build one

[PRR]

> 2x 1k secondaries

It is one 1KCT secondary.

If "matched" (which you are not doing), the push-pull amp should act like a 500 ohm load on the power supplu. Or 600 since you have another 100r. So max supply current should be 9V/600= 15mA. You see twice that, even though you are under-loading (100K load on "20K" winding).

With guitar you have two octaves of bass below the xformer's 300Hz rating, where the impedance is much less than rated.

If it sounds OK, that's OK.

These amps always have crossover glitch unless you do a lot more stuff. (Possibly starting with a much bigger xformer.)

For loudspeaker use, the 001, 008, 026, 118, 013 parts are worth a try.

[samhay]

> 2x 1k secondaries

It is one 1KCT secondary.

If "matched" (which you are not doing), the push-pull amp should act like a 500 ohm load on the power supplu. Or 600 since you have another 100r. So max supply current should be 9V/600= 15mA. You see twice that, even though you are under-loading (100K load on "20K" winding).

With guitar you have two octaves of bass below the xformer's 300Hz rating, where the impedance is much less than rated.

If it sounds OK, that's OK.

These amps always have crossover glitch unless you do a lot more stuff. (Possibly starting with a much bigger xformer.)

For loudspeaker use, the 001, 008, 026, 118, 013 parts are worth a try.

Thanks PRR, I was hoping you might swing by. Sorry - but it seems I have a few more questions.

I am not sure about nomenclature, but I guess the TL44's secondary is 2k (CT). The best info I have found is in this ebay listing:
http://www.ebay.co.uk/itm/Eagle-LT700-LT44-Radio-Output-Inter-stage-Tapped-Audio-Matching-Transformer-/120788551407

This differs from the Mouser 42TM006, which is 1k (CT). I took this part number from RG's PCB layout of the Titan Boost.
In my case, I guess the the push-pull amp should be behaving as a 1k load? so max current = 9V/(1.1k) ~ 8 mA, which is pretty close to what I see at rest. I guess I need to measure a few things and to think more carefully about 'matching' the transistors - will smallish resistors from the emittors to ground do the trick, or is this just going balance them better?

I have played with the 'output' loading, and found it sounded better being at 100k vs 10k (the audio pots I had at hand). I will try ~20k and see how that sounds.

I haven't found any specs on the TL44's frequency rating - is the 300 Hz - 3.4 kHz for the 42TM006 likely to be similar, seeing as the transformers are both of similar size and construction?
I hadn't really considered the bass roll off. As the impedance is low, I figured it would be the treble that I lost. I guess it is more complicated than that. However, I have quickly run some white noise through it and looked at the frequency response. It looks reasonably flat, which relects how it sounds (which is OK, but could probably be better). The crossover noise is not too bad. With moderate input voltage, it is pretty clean and with bigger signals it behaves reasobaly gracefully.

I had a play with a resistor from the emitters to ground. With a C10k pot, you can get a fairly useful gain/volume control. Any reason why this is a bad idea?

[samhay]

^ I do have a cab! And a similar transformer... just not the time to build one

That sounds familiar. As of late, I have plenty of time to run sims, a little time to play with the breadboard, and virually no time to get the soldering iron out. Not a bad way to actually stop and think about what you are doint though.

[earthtonesaudio]

I left DC on the transformer in the hope of getting some saturation - if I do, I can't see or hear it.

...The cap (C2) prevents DC from flowing through the transformer.

Another unrelated comment: you could use a transformer as a choke to filter the power supply in the BJT version, and do away with the filter cap, resulting in a circuit with exactly zero capacitors!  Note I am not against capacitors, just very much in favor of the novelty of this idea.

[samhay]

I left DC on the transformer in the hope of getting some saturation - if I do, I can't see or hear it.

...The cap (C2) prevents DC from flowing through the transformer.

Another unrelated comment: you could use a transformer as a choke to filter the power supply in the BJT version, and do away with the filter cap, resulting in a circuit with exactly zero capacitors!  Note I am not against capacitors, just very much in favor of the novelty of this idea.

Good point about C2 - that's largely a legacy issue. I did have a large resistor from the secondary CT to ground to try and trickle DC through it, but it wasn't very successful, so I left it out of this schematic. A cap on one or both inputs to the transformer is necessary, however, as they are separated by a few volts DC and one cap vs. two caps wins on the least parts-count front.

The push-pull amp doesn't have any caps in the audio path. Somewhere out there, there is a guitar playing audiophile that might find that appealing.

[samhay]

Spent a little more time with the transformer-coupled push-pull booster on the breadboard.
I replaced the output pot with a 22k resistor. Not sure I can hear much of a difference.

Driven with a buffered signal, there is no significant bass roll-off, the frequency response is reasonably flat out to ~15 kHz, where it drops off.
If I add 100k series resistance to the input, the treble response doesn't change, but the bass rolls-off from ~300 Hz.
Despite this, when driven straight from the guitar, the bass doesn't sound especially thin, so I guess the frequency response is sufficient.

Current draw with the 'sag' LED (which draws ~0.7 mA) is 6.5 mA with no input and 23 mA when driven as hard as I can from a 9V supply. It is quite gratifying that this booster will take the signal from an a clipping op-amp and make it significantly louder.

A C10k pot from the emitter junction to ground seems like a pretty good volume control. I doesn't completely cut the volume at min, but gives a fairly linear response and reigns in the current draw nicely.

[samhay]

Probably should have done this sooner, but found some previous threads involving much talk of transformers. Here they are for reference:
http://www.diystompboxes.com/smfforum/index.php?topic=100343.0
http://www.diystompboxes.com/smfforum/index.php?topic=93514.msg808971#msg808971
http://www.diystompboxes.com/smfforum/index.php?topic=99611.0

[tca]

^ I've also collected a few circuits of transformer coupled amplifiers... I'll dig up some of them if you are interested. They are from Mullard and a old article from Popular Mechanics.

Cheers

P.S. (edit)

Found them:

- http://www.scribd.com/doc/51222751/punch-for-portables-amplifier
- http://www.scribd.com/doc/50982632/0-5-4-push-pull-amplifiers
- http://www.scribd.com/doc/50803306/Mullard-Circuits-examples

[samhay]

Thanks.
I came across your TRANSFET in one of the threads I linked. If we don't want to drive a speaker, we could do something similar with a MOSFET using the transformers I have been playing with:



Edit - tried this on the breadboard. Sounds pretty good - a little more volume, and possible a little less harsh (need to see what the scope says) than tapping the drain like a typical MOSFET booster.

[samhay]

I liked MOSFET Transformer Booster in the above post enough to build it.
I have updated the schematic above - I didn't use the ground lift in the end, but otherwise this is what I built. I built it on vero with a board-mounted pot and I managed to squeeze into a 1590A. The layout I used was done on the fly and the one below is similar, but should be a little more user-friendly. If you use a different transformer then it may, or may not, fit.

You get a clean boost until abut 3pm when the MOSFET starts clipping. At max gain, the waveform's are nice and round and you get a pleasing overdrive out of it.
It is significantly louder than a SHO - a sim with an ideal transformer says ~25 V peak-to-peak, but this is probably grossly over optimistic. Nevertheless you do get more than 9V peak-to-peak out of it, so it would benefit from a volume control (A100k should work quite well). As an alternative, I added the 'pad' option, which brings it down to more manageable levels, but reduces its clean headroom at the same time. The clipping LEDs are a nice way to stop it overdriving my audio interface and give it a bit more conventional clipping sound if you put a booster in front of it.
The input impedance can be quite high - I used 2.2||10M - but the transformer's somewhat limited frequency response means that you will not get much >15 kHz. The low frequency roll-off from 300 Hz doesn't seem to hurt the sound to much, but it might not be a great choice for bass.
Current draw is ~4 mA.

layout - not entirely verified:


Edit - I added the 2.2k 'pad' resistor and clipping LEDs to the switch, so they are not in the layout.
I also just noticed that there is no protection diode in the layout (or in my build), so you may wish to add one if this bothers you.

[tubegeek]

Since you are looking for some gain reduction and some grunge (I realize, not necessarily both at once, but at least this idea would be easy to try) you might want to see what happens if you replace the 22K with a 50K pot wired as a variable resistor across the secondary. As you turn it towards lower resistance you'd be loading down the MOSFET more and more and you might get some nice  and unusual distortion sounds out of that.

As a secondary benefit ("secondary," get it? I got jokes!) this experiment would allow you to dial in a sweet spot for that 22K load by ear....

Or else you might hate it and put it back the way it was.

[samhay]

I played with the value of the 22k load resistor quite a lot - an A100k pot in its place works as a pretty useful load/volume control. The 'grunge' kicks in at lower signal levels as you decrease the load.
Because I wanted a 1 knob box, I compromised with the 'pad' switch, which puts a 2.2k resistor in parallel to the 22k load to both drop the volume and load down the transformer/MOSFET.

To give you an idea what it sounds like, here is a quick and dirty sound clip. Pad switch on; gain at noon, 3pm, max, then clean. Signal chain is telecaster (both pickups), booster, audio interface. No post processessing.
https://dl.dropboxusercontent.com/u/11996927/super-SHO.mp3