Looking at transistor/tube hybrids, there is the nagging suspicion that you are getting the worst of both worlds. The output stage might clip before the tubes do. You then get all of the transistor nastiness without the tube warmth.
What is needed is a way to ensure that the tube go into overload first. An example of this is the Tube Clipper published in Elektor 10/2001. It is a novel design that uses low voltage tubes and has a pot to adjust the level of distortion all the way up to 30%. This got me thinking. Can the principle be used in a power amp?
The Tube Clipper uses a triode connected as a cathode-follower. An opamp forms a buffered voltage source that offsets the cathode DC voltage voltage negatively with respect to the grid so
that the triode can be overdriven to a greater or lesser degree. Only one half of the signal gets clipped, so that another cathode-follower, driven by an inverted signal, is needed.
To implement this principal on a bigger scale, I based my design on the Mugen published in Elektor 10/2007. It uses a long tailed pair as input stage, with a cathode-follower as buffer. This in turn drives a solid-state current amplifier. See where this going?
What I have done is to double up the on the cathode-followers. Each one drives its own current amplifier stage. The outputs are out of phase, making for a bridged amplifier. I ditched the CFP output of the Mugen for a more conventional Blameless type. The problem that the bridged amp brings is that power levels go up substantially. This bad boy can put out over 200W into 8 ohms. I therefore added a second pair of output devices to each half of the amp to keep the magic smoke where it belongs.
The opamp was changed to a discrete design, since an IC capable of handling the voltages involved would he very expensive even if it existed. The design is based on one found on the Elliot Sound Products website, but with polarities swopped around and high voltage transistors used.
I have doodled up the whole thing in LTspice and it seems to work, but then again a simulation doesn't tell the whole story. A concern is that the phase margin is only about 20 degrees. The pot changes THD from 0,3% to 9% - a good range. There is room for even more power - just add a bigger, higher voltage power supply and pairs of output devices. Perhaps I should scale things down a bit to make it possible to build a prototype without breaking the bank.
I doubt that this has any practical use for listening to music in a home environment, but as a guitar amp it might have a future. Any thoughts?
What is needed is a way to ensure that the tube go into overload first. An example of this is the Tube Clipper published in Elektor 10/2001. It is a novel design that uses low voltage tubes and has a pot to adjust the level of distortion all the way up to 30%. This got me thinking. Can the principle be used in a power amp?
The Tube Clipper uses a triode connected as a cathode-follower. An opamp forms a buffered voltage source that offsets the cathode DC voltage voltage negatively with respect to the grid so
that the triode can be overdriven to a greater or lesser degree. Only one half of the signal gets clipped, so that another cathode-follower, driven by an inverted signal, is needed.
To implement this principal on a bigger scale, I based my design on the Mugen published in Elektor 10/2007. It uses a long tailed pair as input stage, with a cathode-follower as buffer. This in turn drives a solid-state current amplifier. See where this going?
What I have done is to double up the on the cathode-followers. Each one drives its own current amplifier stage. The outputs are out of phase, making for a bridged amplifier. I ditched the CFP output of the Mugen for a more conventional Blameless type. The problem that the bridged amp brings is that power levels go up substantially. This bad boy can put out over 200W into 8 ohms. I therefore added a second pair of output devices to each half of the amp to keep the magic smoke where it belongs.
The opamp was changed to a discrete design, since an IC capable of handling the voltages involved would he very expensive even if it existed. The design is based on one found on the Elliot Sound Products website, but with polarities swopped around and high voltage transistors used.
I have doodled up the whole thing in LTspice and it seems to work, but then again a simulation doesn't tell the whole story. A concern is that the phase margin is only about 20 degrees. The pot changes THD from 0,3% to 9% - a good range. There is room for even more power - just add a bigger, higher voltage power supply and pairs of output devices. Perhaps I should scale things down a bit to make it possible to build a prototype without breaking the bank.
I doubt that this has any practical use for listening to music in a home environment, but as a guitar amp it might have a future. Any thoughts?
Attachments
Last edited:
I'd be very worried about that phase margin, with real world loads it could oscillate.
If you are thinking this might be applicable to the guitar amplifier scene I could move it to Instruments & Amps which is where such things should be discussed. Might also get more notice there.
If you are thinking this might be applicable to the guitar amplifier scene I could move it to Instruments & Amps which is where such things should be discussed. Might also get more notice there.
Thanks, kevinkr!
A bit of reading up and I was able to improve the phase margin to 34 degrees with the addition of a resistor in series with the compensation cap.
The original Mugen design actually had no compensation of any kind in open loop mode. A simulations shows that the phase is 195 degrees at 0 dB gain, for a phase margin of -15 degrees. Even with added NFB, the gain/phase plot shows no improvement. No wonder I couldn't find any mention of anyone building it on this forum. 🙄
Here is the new and improved version of my yet unnamed baby.
A bit of reading up and I was able to improve the phase margin to 34 degrees with the addition of a resistor in series with the compensation cap.
The original Mugen design actually had no compensation of any kind in open loop mode. A simulations shows that the phase is 195 degrees at 0 dB gain, for a phase margin of -15 degrees. Even with added NFB, the gain/phase plot shows no improvement. No wonder I couldn't find any mention of anyone building it on this forum. 🙄
Here is the new and improved version of my yet unnamed baby.
Attachments
As a reference, here is what happens when a Fender 57E Bandmaster clips. Note that the input signal has been scaled. The FFT shows that odd order harmonics dominate - the magic valve sound?
I have also added the LTspice file for those who want to see for themselves. Can't remember where I got it, might even have been on this forum.
I have also added the LTspice file for those who want to see for themselves. Can't remember where I got it, might even have been on this forum.
Attachments
My early attempts at hybrid amplifiers just saw a very clean sounding amplifier.
It wasn't until I designed a pure valve amplifier that I started to get a more valve sounding amp.
A very cheap way to get a valve sound is by using a soft limiter.
It wasn't until I designed a pure valve amplifier that I started to get a more valve sounding amp.
A very cheap way to get a valve sound is by using a soft limiter.
An externally hosted image should be here but it was not working when we last tested it.
What you are showing is the output waverform of the push-pull stage, with most of the even order harmonics cancelled out in the process - so mostly power amp distortion. For what most people think of as "tube sound", please take a look at the preamp gain stages, and you will see that the even order harmonics are higher than the odd order harmonics. Here is the same amp's FFT (with 2nd-3rd stage LNFB removed, which is more representative of the majority of guitar amplifiers' design) taken at the input of the power tube for comparison.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Thanks for clearing that up. I will have read up on the odds and evens of valve amp distortion - getting myself confused here, hehehe.
Indeed a good way to do it, but what do the purists make of it? For most ardent only an all glass amp will do, I guess. What I am after is a way of meeting the needs of those who would compromise just a little.
After the above exchange of ideas, I came up with another circuit. Not very original, it combines the front end of the Bandmaster with a MOSFET push-pull buffer amplifier. I am having trouble posting images right now, but the buffer circuit can be found here:
MOSFET push-pull buffer/amplifier
It replaces tubes U5, U6 and U7 and associated components in the schematic of the Bandmaster I posted previously. It does work in LTspice. Most impotantly, the tube sound seem to be preserved. The output spectrum shows up a whole lot of second order tube distortion.
Using MOSFETS means that the expense of the output transformer and power supply choke is avoided. In their place, you now have to supply a heat sink. Unfortunately, the BUZ901 and BUZ906 lateral MOSFETs are about the same price as the 6L6 tubes! The power transformer should be easier to source, but one can't help wondering if there is any point to the exercise. One consolation is that the MOSFET output stage will produce more power. More power can't be all bad. 😎
MOSFET push-pull buffer/amplifier
It replaces tubes U5, U6 and U7 and associated components in the schematic of the Bandmaster I posted previously. It does work in LTspice. Most impotantly, the tube sound seem to be preserved. The output spectrum shows up a whole lot of second order tube distortion.
Using MOSFETS means that the expense of the output transformer and power supply choke is avoided. In their place, you now have to supply a heat sink. Unfortunately, the BUZ901 and BUZ906 lateral MOSFETs are about the same price as the 6L6 tubes! The power transformer should be easier to source, but one can't help wondering if there is any point to the exercise. One consolation is that the MOSFET output stage will produce more power. More power can't be all bad. 😎
Last edited:
The imperfections of tubes that make a tube guitar amp sound good are many. All kinds of things have been tried to drive a solid state power amp below it's clipping point, none have achieved universal acceptance. They all fall short in some aspect. Without regard to the technology used, consider the differences between a typical tube and solid state guitar amp:
Tube:
Low open loop gain, low or no feedback, moderate to high output impedance. When overdriven, the resulting square wave does not have a 50% duty cycle. 40/60 is more typical.
Solid State:
High open loop gain, tons of feedback, very low output impedance. When overdriven, the result is usually a 50% duty cycle square wave. (Note, a 50% duty cycle sounds hollow.)
Some Solid State guitar amps use mixed current and voltage feedback to raise the output impedance. These generally sound more tube like. In this book it's called mixed mode feedback:
http://www.thatraymond.com/downloads/solidstate_guitar_amplifiers_teemu_kyttala_v1.0.pdf
Lastly, I think you would benefit from looking at the simple circuit posted in this thread: http://www.diyaudio.com/forums/solid-state/226789-tube-sound-transistor-amp-its-possible.html
Although the circuit is outdated, it sounds quite good for guitar, better than some tube amps. It does this because it sidesteps some of the Soild State amp characteristics noted above. The MOSFET power amps in the link you posted could do quite well, but the transformers may be difficult to obtain. 200 Watts is too big for all but the bigest guitar amps. It's better to build a 25 Watt amp that sounds good than a 200 Watt amp that sounds so so.
Tube:
Low open loop gain, low or no feedback, moderate to high output impedance. When overdriven, the resulting square wave does not have a 50% duty cycle. 40/60 is more typical.
Solid State:
High open loop gain, tons of feedback, very low output impedance. When overdriven, the result is usually a 50% duty cycle square wave. (Note, a 50% duty cycle sounds hollow.)
Some Solid State guitar amps use mixed current and voltage feedback to raise the output impedance. These generally sound more tube like. In this book it's called mixed mode feedback:
http://www.thatraymond.com/downloads/solidstate_guitar_amplifiers_teemu_kyttala_v1.0.pdf
Lastly, I think you would benefit from looking at the simple circuit posted in this thread: http://www.diyaudio.com/forums/solid-state/226789-tube-sound-transistor-amp-its-possible.html
Although the circuit is outdated, it sounds quite good for guitar, better than some tube amps. It does this because it sidesteps some of the Soild State amp characteristics noted above. The MOSFET power amps in the link you posted could do quite well, but the transformers may be difficult to obtain. 200 Watts is too big for all but the bigest guitar amps. It's better to build a 25 Watt amp that sounds good than a 200 Watt amp that sounds so so.
Last edited:
- Status
- Not open for further replies.