Not a stompbox: idea for a power amplifier

[fryingpan]

I was thinking of making a low-power amp (for a solid state amplifier at least). Say, around 25W maximum or so. Something to be used only for recording purposes. I was thinking of making one for bass guitar, but obviously a power amp that is adequate for bass guitar should be equally adequate for guitar. Now, since I don't really care about linearity that much (we're talking about guitar and bass after all), and since there are some amp topologies that require two inverted signals into two power transistors... Why not achieve this with a high-voltage opamp (or just a transistor as a voltage amplifier, even though it may be harder to supply voltage to) with a line output transformer whose two inverted outputs are fed into the output transistors? Considering that many relatively cheap transformers (~€50) should be able to withstand voltages of 25Vp-p without excessive distortion especially if low frequency content isn't very high (and even if we look at bass guitar, most of the energy is focused between 80 and 200Hz) for a low power amp it could be worth pursuing, since you could also employ transformer saturation as a form of distortion (and also band limiting). Thoughts?

[PRR]

You are years too late. This was common 60 years back. In the US, Delco pushed the idea. Used in early Rhodes and I think in Beatles VOX?

http://www.introni.it/pdf/RCA%20Transistor%20Manual%201964.pdf

[Steben]

Very common. Deacy amp is like that.

[teemuk]

...and when that transformer is inside the closed feedback loop it's going to effect phase (like in tube amps) and limit how much feedback one can reliably apply. This in turn is going to effect your design of the input stage in order to keep the amp stabile. There's a sound reasoning behind of why "modern" designs got rid of interstage transformers as soon as it was possible.

Saturation is a myth that just keeps on prevailing. Ironucally, once upon a time there were tons of solid-state radios, amps, PA's and paging systems with audio transformers here and there in the signal path and yet no hyping whatsoever of how great they sounded due to all that supposed transformer saturation. Even a tube OT is difficult to saturate with typical -guitar- signals, and when it does saturate the effect isn't very nice.

There are certain merits in the differential / push-pull  arrangement that is often found from transformer coupled circuits - particularly in how such circuits distort - but as shown by i.e. Vox (Valvereactor, VBM1 "Deacy" clone, etc) or Hughes&Kettner (Quantum) a real transformer can be replaced by a simple differential amp).

[amptramp]

My old Univox G-65 has an output similar to the second example Paul showed:



It shows feedback around a driver transformer but you have to be very careful that the magnitude of the feedback at a 180 degree phase shift is less than unity or the circuit will oscillate at low frequencies.  There is a 60 pF feedback capacitor in parallel with the 50 K feedback resistor just to keep the high-frequency phase shift from causing oscillations.  One rule of thumb for designing power audio stages is if the frequency response shows any sign of peaks at the lower or upper end of the response, the design is marginally stable and should be corrected.

[fryingpan]

...and when that transformer is inside the closed feedback loop it's going to effect phase (like in tube amps) and limit how much feedback one can reliably apply. This in turn is going to effect your design of the input stage in order to keep the amp stabile. There's a sound reasoning behind of why "modern" designs got rid of interstage transformers as soon as it was possible.

Saturation is a myth that just keeps on prevailing. Ironucally, once upon a time there were tons of solid-state radios, amps, PA's and paging systems with audio transformers here and there in the signal path and yet no hyping whatsoever of how great they sounded due to all that supposed transformer saturation. Even a tube OT is difficult to saturate with typical -guitar- signals, and when it does saturate the effect isn't very nice.

There are certain merits in the differential / push-pull  arrangement that is often found from transformer coupled circuits - particularly in how such circuits distort - but as shown by i.e. Vox (Valvereactor, VBM1 "Deacy" clone, etc) or Hughes&Kettner (Quantum) a real transformer can be replaced by a simple differential amp).

Yes, I realise that the "merits" of transformer distortion are overstated, but the advantages I see with the use of interstage transformers are also safety and isolation. I don't know how significant they are and if they are worth pursuing, I was thinking of using an interstage transformer with a high-voltage yet relatively low-current signal so that the costs could also be kept down, possibly? In the end it was all a (newbie-class) experiment. The fact that impedance has to be kept relatively high is not necessarily a problem with a low-power design too, and having less feedback (therefore less error correction) should result in higher 2nd harmonic distortion especially (even though we are talking about a push-pull configuration, so that's debatable), and when it comes to guitar or bass that shouldn't be undesirable, perhaps?

As for transformer distortion, apparently though Marshall amps are (in)famous for having very undersized transformers and yet that gives those amps their "compressed" sound, I don't know how much truth there is in it.

[Steben]

...and when that transformer is inside the closed feedback loop it's going to effect phase (like in tube amps) and limit how much feedback one can reliably apply. This in turn is going to effect your design of the input stage in order to keep the amp stabile. There's a sound reasoning behind of why "modern" designs got rid of interstage transformers as soon as it was possible.

Saturation is a myth that just keeps on prevailing. Ironucally, once upon a time there were tons of solid-state radios, amps, PA's and paging systems with audio transformers here and there in the signal path and yet no hyping whatsoever of how great they sounded due to all that supposed transformer saturation. Even a tube OT is difficult to saturate with typical -guitar- signals, and when it does saturate the effect isn't very nice.

There are certain merits in the differential / push-pull  arrangement that is often found from transformer coupled circuits - particularly in how such circuits distort - but as shown by i.e. Vox (Valvereactor, VBM1 "Deacy" clone, etc) or Hughes&Kettner (Quantum) a real transformer can be replaced by a simple differential amp).

Yes, I realise that the "merits" of transformer distortion are overstated, but the advantages I see with the use of interstage transformers are also safety and isolation. I don't know how significant they are and if they are worth pursuing, I was thinking of using an interstage transformer with a high-voltage yet relatively low-current signal so that the costs could also be kept down, possibly? In the end it was all a (newbie-class) experiment. The fact that impedance has to be kept relatively high is not necessarily a problem with a low-power design too, and having less feedback (therefore less error correction) should result in higher 2nd harmonic distortion especially (even though we are talking about a push-pull configuration, so that's debatable), and when it comes to guitar or bass that shouldn't be undesirable, perhaps?

As for transformer distortion, apparently though Marshall amps are (in)famous for having very undersized transformers and yet that gives those amps their "compressed" sound, I don't know how much truth there is in it.

It is very easy to have 2nd harmonic added to a push pull: just misbias the halves. 

Feedback makes for more linear with harsher onset into clipping. That is true for tubes as well as for transistor amps.
Dynamic cross over and power sag can happen in solid state push pull as well. It is all about design.

[teemuk]

"Isolation" isn't neccessarily a merit in solid-state power amp because you need to keep DC offset of the output stabile. This is usually achieved with negative AC+DC feedback that makes the input stage a "servo" that corrects possible offset shifting of the output. When you isolate the DC conditions this no longer works.
..and then you probably need to add a specified DC servo for the task.

And regarding compression of some Marshall amps... I have a hunch it's more result of overall power supply design (i.e. lower filtering capacitance of JMP vs. JCM IIRC) than that of saturation of undersized OT's. Weren't the Drake OT's supposed to be sort of revered in comparison to noname OT's of cheaper amps anyway?

Yes, as shown you can definitely make an amp with an interstage transformer driving output transistors of same polarity, but why not just exploit complementary devices now that we have them? The topologies aren't essentially that much different.

[Steben]

A classic tube push-pull amp does include the better part of tubes: gradually compressive clipping. No transistor can achieve that exact same part of the tube DNA.
The output transformer is necessary when using 2 exact same output tubes. In solid state you don't need the architecture since you don't need to use the output stage clipping with same parts.