Need suggestion in building a stereo power amp

[Alan0354]

I took a tour in the SS side of this forum. I discovered most of the SS amps are DC coupled. Anything goes wrong, the speaker likely to go along with it. That is an absolutely no no for me. I guess it's back to tube amps.

 

[costis_n]

I would recommend to get some nice output transformer and build a conventional design at first. I believe you will like what you are hearing

 

[Alan0354]

I would recommend to get some nice output transformer and build a conventional design at first. I believe you will like what you are hearing

That is my planning. Get two very nice OT, start building the amp with the Mu follower in mind and leave enough space for that. Then build something basic. I am a true believer of less is more.

The next thing is to identify the OT and the chassis. My speaker is nice, but not the top end, I just need one to be better than my Acurus. No point in making the amp much higher quality than the speaker.........then I would have to think about a better pair and that's going to hurt the pocket!!!!

I just spent the last two day taking out one channel of the Acurus and put in more filtering at different points, beefing up the high current traces and beefing up the ground trace on the input and pre driver side. Also doubling up the output wires. Next, I am going to increase the bias current and see how much improvement I can get out of it. This will buy some time for me to build the new amp.

 

[Alan0354]

I have a question after reading a thread here that someone got a new pair of B&W speaker and his tube amp did not sound good with the speakers. That get me worry. As in the beginning of this thread, I have a pair of JM Lab, what do I have to look out in designing a tube amp for that speakers? I am narrowing down to Lundahl LL1623 OT, but before I even consider putting out $500 for a pair of OT, I want to make sure I have the right design. I have no idea how to match the design to a pair of speakers, can anyone give me some pointers what to look for?

That is the pair of speaker I use, not as if I can choose a pair of speakers to fit the amp. So the amp has to work with that pair of speakers.

Thanks

 

[costis_n]

Difficult speakers means a varying speaker Z (impedance) with frequency and low-ish sensitivity.

To combat the first, you need an amp with as low an output impedance as possible. This means triode or Ultralinear operation or triode in the output stage AND feedback around the output stage (actually both Ultralinear and triode operation are forms of local feedback). If you combine both, you ll get an output Z of under 1 Ohm.

For the second requirement, you just need enough watts. Decide the output power you really need at your average listening level (as determined by some links they were posted here) and double that for headroom.

Themost important thing you need IMHO is a good, silent low impedance power supply that has very little sag.

 

[Alan0354]

Difficult speakers means a varying speaker Z (impedance) with frequency and low-ish sensitivity.

To combat the first, you need an amp with as low an output impedance as possible. This means triode or Ultralinear operation or triode in the output stage AND feedback around the output stage (actually both Ultralinear and triode operation are forms of local feedback). If you combine both, you ll get an output Z of under 1 Ohm.

So UL tap connect to screen creates a local feedback and lower the impedance. Then using global NFB to further lower the impedance. Together I can conquer the problem.

For the second requirement, you just need enough watts. Decide the output power you really need at your average listening level (as determined by some links they were posted here) and double that for headroom.

I don't think I'll ever go over 5W per channel, so I am using a pair of EL34 in Class AB to get 50W. But at 5W output, it is really running in Class A.

The most important thing you need IMHO is a good, silent low impedance power supply that has very little sag.

Thanks
I still thinking about the power supply. I have a Fender Twin Reverb and a Marshall 100W power transformer. Both are designed for 4 6L6 power tubes. I am planning to use one of them for the PT. I know it is cutting it close, but I just figure I never going to crank above 10W total, I think it should be ok. I just need a lot of filter caps.

Thanks

 

[MrCurwen]

[Y]ou need an amp with as low an output impedance as possible. This means triode or Ultralinear operation or triode in the output stage AND feedback around the output stage

I completely agree on the impedance. A low impedance driving the OT solves a lot of headaches.

However I disagree that UL or NFB is the best sounding way to get that. My ears find NFB to most times sound a bit SS.

Themost important thing you need IMHO is a good, silent low impedance power supply that has very little sag.

Second that. Every good amp starts with a low impedance PSU. Even a simple two MOSFET regulator makes a big difference.

 

[Alan0354]

Second that. Every good amp starts with a low impedance PSU. Even a simple two MOSFET regulator makes a big difference.

That's what I am working on. Using a few zener to provide a 400V reference to drive a MOSFET source follower to give a stable 400V.

I rather spend big money on the OT, and try to use a PT for 100W guitar amp that I have to save money. I have plenty of high power high voltage MOSFET from my guitar amp project, so that does not cost money. Just need to buy heatsink.

 

[costis_n]

Good points here. And, I repeat. you should sit down and make a test amp, with possibly guitar transformers. (must be stereo) to get an idea if you are OK with how a classic design with feedback all sounds in your own system. These amps are coveted by many . You can maybe then remove the feedback, and switch it to triode to compare.

How about a Radford STA-25 that uses EL34's and is loved by a lot of people in here?

 

[Alan0354]

Some picture demonstrations might help my explanation.

These current measurements are from the attached schematic. The two center traces (other one's a bit dim) are the currents thru the tubes, measured from the small resistors right above the tube plates. As you can see, they're pretty much constant; about 50µA swing with a voltage swing of about 70V (individual tube).

So the tubes operate at constant current, variable voltage.

The big swing traces are measured from the MOSFET sources. As you can see, even though the tubes have horizontal load lines, there is some 15mA current swinging thru the OT primary.

MOSFETs swing the current. Tubes handle voltage. Everybody does what they do best! OT is not drawn at all on the tubes' load lines; they are horizontal for maximum linearity and best transparency.

Power is limited by amount of available voltage headroom, and the amount of current a balanced pair passes (both tubes combined).

Hi MrCurwen
I know it's been a while, I spent my time studying SS power amp first using Cordell's book, took me a few weeks to finish that. I just started studying tube, I just read the mu-follower. I have a few questions:

1) If I run say 100mA per tube, you are going to have 82V drop across the 820 ohm resistor. So you eat up 82 head room. Seems like using Beta follower is better.
2) Do you have problem with the input capacitance of of the MOSFET? Cin is about a few thousand pF. So you really don't have a horizontal load line at the more important high frequency.
3) The impedance loading the tube is not infinite even at low frequency ( without the effect of the input capacitance of the MOSFET). This is because Av of the MOSFET is less than 1.

 

[Alan0354]

Another thing about using mu-follower and Beta follower. If you drive the OT with the source of the MOSFET, the impedance is low, this will lower the output impedance at the output of the OT. This will create gm doubling effect as described in page 101 to 103 of Bob Cordell's book. This will create a kink in output impedance and cause crossover distortion. So this will benefit only for Class A that never have to deal with crossover distortion. In normal tube output, you don't have to worry about crossover distortion and you can run Class AB.

 

[MrCurwen]

I have a few questions:

1) If I run say 100mA per tube, you are going to have 82V drop across the 820 ohm resistor. So you eat up 82 head room. Seems like using Beta follower is better.
2) Do you have problem with the input capacitance of of the MOSFET? Cin is about a few thousand pF. So you really don't have a horizontal load line at the more important high frequency.
3) The impedance loading the tube is not infinite even at low frequency ( without the effect of the input capacitance of the MOSFET). This is because Av of the MOSFET is less than 1.

Hello again!

1) That resistor is sized to drop about 10V. So at 100mA you make it 100 ohms (2W).

2) No problem. Distortion stays very low up to 20kHz. The follower topology is not so concerned with input capacitance, also high voltage headroom helps apparently. I am not an expert on MOSFET selection (I mainly use IRF820 / IRF830). In practise (ears & measurements) this is not a problem.

3) Nothing is infinite. =) Like I've stated many times in this thread, it's "some hundreds of kilo-ohms". That's 'infinite' for the tube. I always write "nearly horizontal" and not 'horizontal', to try to be exact.

 

[MrCurwen]

Another thing about using mu-follower and Beta follower. If you drive the OT with the source of the MOSFET, the impedance is low, this will lower the output impedance at the output of the OT. This will create gm doubling effect as described in page 101 to 103 of Bob Cordell's book. This will create a kink in output impedance and cause crossover distortion. So this will benefit only for Class A that never have to deal with crossover distortion. In normal tube output, you don't have to worry about crossover distortion and you can run Class AB.

As stated before, my design is stricly Class A. If that is a problem, i.e. if design priority is other than sound quality, then other designs should be looked into. For best sound, Class A no feedback etc. is best, in my experience.

It goes back to design priorities; power / heat / fidelity / cost / aesthetics ("no sand", "traditional tube design", "no HYBRID" etc.). A builder needs to know what they prioritise over what else.

As per your question, I would not use MOSFET followers to drive an OT in AB or B.