[bigjppop]

The following is a cut and paste job from a thread I started over at diyaudio.com.  I'd love to get the input of some of the members that here as well.


First of all, I know dummies probably shouldn't be looking to build big push-pull amps but lets move beyond that.

I'm trying to wrap my arms/brain around the pros/cons involved in getting big power from PP amps. I've started with some SET amps that I love (Tubelab's SSE is sounding good right now) and am getting ready to start construction on Poinz' Musical Machine. I've been following:

http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp.html

http://www.diyaudio.com/forums/tubes-valves/156699-mullard-5-20-kt88-pp-blocks.html

http://www.diyaudio.com/forums/tubes-valves/161702-opus-5-0-modern-mullard.html

All three of these threads have given me some great insight but I'm still a beginner. Down the road, maybe next year, I'd like my next amp project to be a pair of PP monoblocks that can do 60-75 WPC (I've got some speakers that I really like that are only 84-85 dB senstive). I'd like to utilize the Russian 6P3S-E that I've recently become enamored with. I'm sure I'll need some specific help down the road but here I'd just like to ask some basic questions.

First of all, I've seen several commercial designs running 4 output tubes per channel (or more) and I'm wondering about the benefits/drawbacks of running more output tubes in parallel. One of the reasons I like these tubes so much is they are very reasonably priced and it wouldn't be difficult to do 4 per channel, at least not nearly as expensive as running 4 of the expensive reissue KT-88's per side. 2 per side... OK, but 4 per side starts to get pricey.

Second question, money aside, is it easier to make a better sounding amp running 2 bigger output tubes per side vs. 4 smaller output tubes per side?

I'm sure I'll come up with more questions down the road but those two should get me started for now.

Thanks in advance!

[Steve]

The following is a cut and paste job from a thread I started over at diyaudio.com.  I'd love to get the input of some of the members that here as well.


First of all, I know dummies probably shouldn't be looking to build big push-pull amps but lets move beyond that.

I'm trying to wrap my arms/brain around the pros/cons involved in getting big power from PP amps. I've started with some SET amps that I love (Tubelab's SSE is sounding good right now) and am getting ready to start construction on Poinz' Musical Machine. I've been following:

http://www.diyaudio.com/forums/tubes-valves/133034-6l6gc-ab2-amp.html

http://www.diyaudio.com/forums/tubes-valves/156699-mullard-5-20-kt88-pp-blocks.html

http://www.diyaudio.com/forums/tubes-valves/161702-opus-5-0-modern-mullard.html

All three of these threads have given me some great insight but I'm still a beginner. Down the road, maybe next year, I'd like my next amp project to be a pair of PP monoblocks that can do 60-75 WPC (I've got some speakers that I really like that are only 84-85 dB senstive). I'd like to utilize the Russian 6P3S-E that I've recently become enamored with. I'm sure I'll need some specific help down the road but here I'd just like to ask some basic questions.

First of all, I've seen several commercial designs running 4 output tubes per channel (or more) and I'm wondering about the benefits/drawbacks of running more output tubes in parallel. One of the reasons I like these tubes so much is they are very reasonably priced and it wouldn't be difficult to do 4 per channel, at least not nearly as expensive as running 4 of the expensive reissue KT-88's per side. 2 per side... OK, but 4 per side starts to get pricey.

Second question, money aside, is it easier to make a better sounding amp running 2 bigger output tubes per side vs. 4 smaller output tubes per side?

I'm sure I'll come up with more questions down the road but those two should get me started for now.

Thanks in advance!

Hi BigJ.

SETs are much much easier to design than PP. PP is a whole different critter and unless one is very very skilled the results will not be optimum to say the least. I tend to examine circuits per price is no object scenario. However, price is a real consideration which I am sure you have contemplated.

The circuits presented are ok, but not spectacular. The second schematic, the 12AU7 output coupling to the output tubes, 0.1uf/102k,  immediately sends a huge red flag concerning the accuracy of the bass response under no feedback conditions. The coupling circuit is approx 5db rolled off at 20hz.

So how is bass being compensated? Feedback will to some extent if one is into feedback. The coupling circuit maybe such as to prevent "motorboating", a severe form of oscillation, because the negative feedback becomes positive feedback, and of course phase shifting, as the signal lowers in frequency.

By the way, I am finding the EL-156 tube beats up on the output tubes presented, although not cheap. Might check into this tube.

Sorry to sound pessimestic, but it still may sound good depending upon how you wish to impliment such a design with the rest of your system.

Hope this helps.

[bigjppop]

I appreciate the input.  I'm VERY far away from being able to design my own circuit; just trying to get a handle on some of the pros/cons of the various design philosophies. 

[Steve]

I appreciate the input.  I'm VERY far away from being able to design my own circuit; just trying to get a handle on some of the pros/cons of the various design philosophies.

Your welcome BigJ. Hope this helped.

Cheers.

[Niteshade]

Push-pull amplifiers are no more difficult to design than SET's- in fact, they might be considered easier since they are less prone to noise problems.

I suggest using a combined phase splitter/voltage amplifier circuit to reduce wiring issues.

[Steve]

Push-pull amplifiers are no more difficult to design than SET's- in fact, they might be considered easier since they are less prone to noise problems.

I suggest using a combined phase splitter/voltage amplifier circuit to reduce wiring issues.

Yes, one can copy or design from a number of circuits, including using the common splitter transformer, they sound ok. Unfortunately very few use better techniques because either it is much more difficult for their expertise, they need a huge splitter drive voltage to the output tubes, or the output impedance of the splitter circuit is too high. The very best splitters have to be designed very very carefully for world class sonic quality.

They generally put out more power than SETs and can have more accurate bass and treble although not necessarily.

Cheers.

[JoshK]

I'll add my $.02, fwiw, ymmv, and all that.

First, it probably was said before, but obtain a copy of Morgan Jones's Valve Amplifiers 3rd ed and look at the crystal palace design.  This driver/splitter is up to the task of driving mean output tubes and the first thread you link has a driver/splitter that is a variation on the same theme, with just a bit more sand (not bad in this case, so don't worry).

Don't underestimate the importance of the driver/splitter because it essentially defines the performance that the amp can acheive assuming reasonable output design and iron.

60 watts is pretty doable with just 2 output tubes in class A/B2.  Take a look at that first thread again for proof.  George (tubelab.com) obtains 75wpc with 2 triode-wired KT88s using that driver board.  Myself, I'd wait until George puts out that driver board and use that as the basis of your amp unless you clearly understand its workings and can mock it in p2p.

I am waiting to get some of them myself.

[Steve]

I'll add my $.02, fwiw, ymmv, and all that.

First, it probably was said before, but obtain a copy of Morgan Jones's Valve Amplifiers 3rd ed and look at the crystal palace design.  This driver/splitter is up to the task of driving mean output tubes and the first thread you link has a driver/splitter that is a variation on the same theme, with just a bit more sand (not bad in this case, so don't worry).

Don't underestimate the importance of the driver/splitter because it essentially defines the performance that the amp can acheive assuming reasonable output design and iron.

60 watts is pretty doable with just 2 output tubes in class A/B2.  Take a look at that first thread again for proof.  George (tubelab.com) obtains 75wpc with 2 triode-wired KT88s using that driver board.  Myself, I'd wait until George puts out that driver board and use that as the basis of your amp unless you clearly understand its workings and can mock it in p2p.

I am waiting to get some of them myself.

Yes the phase splitter is all important if one is to get SET midrange magic. Driving AB2 requires a driver with guts.

Which schematic are you looking at? This one?

http://www.diytube.com/unidriver/poseidon.pdf

Cheers.

[JoshK]

Yes the phase splitter is all important if one is to get SET midrange magic. Driving AB2 requires a driver with guts.

Which schematic are you looking at? This one?

http://www.diytube.com/unidriver/poseidon.pdf

Cheers.

No, definitely not that one.  That one doesn't look like it could do it (just looked at the schematic on the first few pages).

The one I was referring to is buried in the 6L6GC AB2 thread on diyaudio but it described here.
http://www.tubelab.com/Universal%20Driver.htm

It is similar to the Crystal Palace driver exept the CFs in MJ's version are replaced with MosFET SFs.  Its not a copy but it rhymes.

[JoshK]

Gaining the midrange magic of SETs in PP amps is no small feat, as I understand it.  Its contraversial even what this is caused from, some say euphonic characteristic of lots of H2 distortion.  I tend to think it is due to the harmonic profile with lack of upper harmonics and a nice waterfall profile to the lower orders with H2 > H3.

In PP, the H2 gets greatly cancelled in the OPT, so H3 dominates.  As soon as you introduce FB, you trade lots of lower order harmonics for a bit of upper harmonics.  I think this takes some of the magic away.

Probably the most interesting technique I've seen for balancing the harmonics in PP amps is Western Electric's harmonic equalizer/balancer circuit.  Lynn Olson most noteably brings this topic back to life on his site.  Stephie [Steve] Bench though went a few steps further in his 801A pp amp that is detailed on Dave Slagle's Intactaudio forum.

[Steve]

No, definitely not that one.  That one doesn't look like it could do it (just looked at the schematic on the first few pages).

The one I was referring to is buried in the 6L6GC AB2 thread on diyaudio but it described here.
http://www.tubelab.com/Universal%20Driver.htm

It is similar to the Crystal Palace driver exept the CFs in MJ's version are replaced with MosFET SFs.  Its not a copy but it rhymes.

The tubes are driven more likely lean AB2  or B idle for maximum efficiency and minimum plate dissapation, for the output power claimed, especially for the 6BQ6..

Also crossover distortion is probably present, especially for the 6BQ6. Driving the grid 1 will save stages and sound better, plus help with other design problems.

Better have the speakers fused. 

Imo, the phase splitter is certainly questionable especially when SS devices are used.

Cheers.

[Steve]

Gaining the midrange magic of SETs in PP amps is no small feat, as I understand it.  Its contraversial even what this is caused from, some say euphonic characteristic of lots of H2 distortion.  I tend to think it is due to the harmonic profile with lack of upper harmonics and a nice waterfall profile to the lower orders with H2 > H3.

In PP, the H2 gets greatly cancelled in the OPT, so H3 dominates.  As soon as you introduce FB, you trade lots of lower order harmonics for a bit of upper harmonics.  I think this takes some of the magic away.

Probably the most interesting technique I've seen for balancing the harmonics in PP amps is Western Electric's harmonic equalizer/balancer circuit.  Lynn Olson most noteably brings this topic back to life on his site.  Stephie [Steve] Bench though went a few steps further in his 801A pp amp that is detailed on Dave Slagle's Intactaudio forum.

I was fortunate enough to perform my own testing several years ago and I was able to obtain extremely close midrange magic of my PP amps by simply reducing the highs and lows and working with the phase splitter. As you stated, the toughest part was the phase splitter. Adding too much bass would mask/mess up midrange magic compared to SETs.

You are quite correct when you stated the phase splitter design is critical. Very critical.

Cheers.

[JoshK]

The tubes are driven more likely lean AB2  or B idle for maximum efficiency and minimum plate dissapation, for the output power claimed, especially for the 6BQ6..

Also crossover distortion is probably present, especially for the 6BQ6. Driving the grid 1 will save stages and sound better, plus help with other design problems.

Better have the speakers fused. 

Imo, the phase splitter is certainly questionable especially when SS devices are used.

Cheers.

The 6BQ6 is an extreme example, to test the limits of what *could* be done, but obviously not the thrust of that design.  I am not arguing whether his design choices would be yours, but the driver I would argue will be tough to compete against (objectively given some reasonable arguing of criteria). 

You object to LTP with a CCS in the tail? That is the splitter.  The second stage is another differential stage with again a CCS in the tail.  (note the text explains what the schema pictured doesn't show).  That is then followed by Mosfet SFs.  There have been a careful selection of the SFs to ensure the most benign reverse transfer capacitance vs. freq, which is most likely the source for most people's complaints with using sand in a tube design.

[Steve]

The 6BQ6 is an extreme example, to test the limits of what *could* be done, but obviously not the thrust of that design.  I am not arguing whether his design choices would be yours, but the driver I would argue will be tough to compete against (objectively given some reasonable arguing of criteria). 

Thanks for your feedback Josh. I think this is an opportunity for viewers to see the "inside" of a design, at least partially (I don't give away secrets though.  ) and see the problems designers see. Yes, it is an extreme example using a 6BQ6 and 6L6GC, but to what end?
 
The CCS chip has several problems. First it is solid state which and of itself influences the sound as it is directly in the signal path. I don't think someone who uses a rectifier tube would place a solid state device, operating, in the direct signal path? I have seen this on commercially produced products as well. Maybe not a huge point, but something to think about.

More importantly, no capacitance value is given, but with high dissapation SS FETs (This CCS has a 40 watt dissapation rating), the capacitance of the high impedance "output stage" is high, .001uf or higher between junctions is common. This high capacitance also varies with signal voltage applied to the output of the CCS, which also influences the sound.

You object to LTP with a CCS in the tail? That is the splitter.  The second stage is another differential stage with again a CCS in the tail.  (note the text explains what the schema pictured doesn't show).

I am very well aware as I use LTP and CCS all the time. (Been designing for decades.) This design is ok for those on a budget, which is important, but is far from the best/optimum. I just hope viewers do not believe this design is the next panacea.

A question I think needs answering is, if on a budget, why use a design that is more complex than it needs to be? Is 4 stages (less than optimum design) better than fewer but better designed stages? With all the problems I see, I would have to say no.

That is then followed by Mosfet SFs.  There have been a careful selection of the SFs to ensure the most benign reverse transfer capacitance vs. freq, which is most likely the source for most people's complaints with using sand in a tube design.

I am sure care was taken with respect to selecting the FETs. However, there is a huge problem with frequency dependent feedback from every stage to every other stage through the power supply itself; absolutely rampant in this design with accompanying phase shifts well into the midrange. The frequency response is suspect as well, and the distortion produced is higher order as well. 

Imo, I would be hesitant to spend such a sum on such a complicated design that can be bettered by simpler designs already out there.

Cheers.