Is there an advantage to more tubes in an amp?

Good stuff! Thanks for all of the replies.

The links are great...I'm working my way through them.

Clueless: My experience has been that transistors made by the same manufacturer from the same batch are relatively close in operation. In fact, so long as the manufacturer maintains consistent standards, i've found them to be pretty good from batch to batch. As a case in point, i've been able to pull a dead transistor(s) out of a circuit and install a new one of the same brand with nary a problem. Bare in mind that this is taking place at RF frequencies with multiple pairs ( up to 96 output devices ) running in parallel. Should something be funny, i.e. dis-similar conductance characteristics, the results would show up VERY quickly in terms of distortion and circuit instability.

Obviously, others share similar experiences / thoughts as to what you stated about mismatches, especially between npn & pnp's. Some manufacturers have gone WAY out of their way to avoid such situations. I know that Plinius uses outputs of only one type ( can't remember if it's PNP or NPN ) in their output stages.

I can't say the same thing about tubes based on my own past experiences and those of others. One of the EE's that i used to work with at another job used to work for RCA in one of their tube production facilities. He told me that tubes were all over the place in terms of specs and that they varied the "acceptable performance standards" based upon how each batch of tubes turned out. As such, the "best" of one batch might be equal to the "mediocre" tubes from the last batch. As such, not only did the "best" tubes from each batch vary, so did those that were at the bottom of the barrel.

I found much of what i know about tubes the hard way after purchasing a company that manufactured tube based RF amplifiers. The funniest thing that i've noticed is that specific tubes that are highly desireable in RF circuits in terms of high output levels and ruggedness are supposedly less desirable in audio circuits. The green label Sylvania's come to mind here. Another thing was that tubes that we knew to be of very high conductance were rarely used in audio circuits. While some amps did use 8417's, i've never seen an 8908 or 2057 in an audio amp. I have to imagine that there are reasons why, but my thoughts are that a tube that had high output levels up that high and was quite rugged would be coasting down low in the audio band. The interesting thing is that i'm beginning to see more tube amps using tubes designed for transmitting purposes pop up lately, so maybe my thinking was that far out of line.

Marakanetz: There are SS devices that can put out GOBS of power by themselves. The only problem is that they require both very high voltage and very high current and require tons of heat sinking. This means that the manufacturer has to have very sturdy power supplies capable of very high steady state current levels with enough heatsinking to handle such tasks. As such, they are not suitable for high profit margin operations or where one would need mass production / mass distribution. The weight alone would kill them in terms of shipping. As to whether or not such devices are suitable for use in the AF ( Audio Frequency ) range is also another matter.

As a side note, what are commonly called "ceramic" or "steel" broadcast tubes are basically a hybrid between a tube and a transistor. There are devices like this that can output 15 KW on a steady state basis that you can hold with one hand. The amount of heat generated by such a device is absolutely incredible though and i would not want to be in the same room that it was being used in : ) Sean
>

Hey thanks for the kind words folks. I learn more here than anything else and a lot from the folks in this thread.

above by Mark >> "The only thing I don't dig is how 300b or any pre-historic-triode SE amp can play MUSIC? To me it's the same as to build 1w radio that also can play music quite loud especially if placed soewhere in the corner of empty room..."

What do you mean you "don't dig" Mark? I'm not sure I get your point.? Do you think low powered tube amps are nonsense? I don't think they are for everybody but they do a lot of things quite well.

>> said above by Sean .."On top of this, i've found that tubes (even from the same manufacturer ) have a far wider level of operating tolerances than SS devices"

I thought that fets and mosfets were far less consistent than tubes? A 30% tolerance in transistors is fantastically good isn't it?

I'm not much re SS but I'm told the specs show that one transistor will have twice the capacitance of the other - thus half the speed...ie the NPN and PNP "complements" really aren't a true complement. MOSFETs being the worst in this regard. These capacitance nonlinearities get worse when you have more than two transistors in the output stage. You have large, and dissimilar, capacitances that are all over the place, and drifting slowly with temperature. It takes a lot of current to charge them. I’m not sure that SS is an answer to poor tolerances or stray capacitance issues. They are different problems but they are still there.

Anyway I’m really NOT that well informed about the SS stuff as I don’t really mess with it much. Really just asking. your comment re tolerance surprised me. Tubes have some weaknesses I wasn't sure that was one. I'll have to check it out.

There are lots of handy tricks to bring aging a tubes back into line such as simple cathode biasing which makes a stage almost self adjusting (if the current goes up in one it increases the bias which, in turn, decreases the current.)

Anyway
Cheers
I remain,

Sean,
Although tubes have substantially larger operating tolerances the number of tubes for the same power is substantially smaller than the number of transistors that will add-up in result the same or even larger operating tolerances.
In all cases the number of output devices should be optimized and that's how we can approach synergy, spead and clarity.

The only thing I don't dig is how 300b or any pre-historic-triode SE amp can play MUSIC?
To me it's the same as to build 1w radio that also can play music quite loud especially if placed soewhere in the corner of empty room...

Clueless did a very nice job of presenting a great yet brief over-view of the subject AND took the time to provide a vast amount of resources for those that are interested in learning. Kudo's to him for helping to educate those that are willing to learn and sharing his love for tube based audio devices. His post really was "glowing" in content : )

Having said that, I would like to highlight one specific point that Gs brings up. At the same time, this one sentence is phenomenally simple yet highly elegant in terms of the idea expressed:

"But as a rule of thumb, the more output devices means the more careful (read: expensive)the circuit design, since there are more devices it's easier to corrupt the current flow."

If the circuit is not fully optimized, the chances of circuit instability, distortion, saturation of power supplies / output transformers, etc.. becomes more significant as output and number of output devices increase. On top of this, i've found that tubes ( even from the same manufacturer ) have a far wider level of operating tolerances than SS devices. As such, the more tubes that you run in a circuit, the greater the chances for circuit instability / reduced performance levels. Most of this has to due with differences in the internal conductance characteristics of the tubes themselves and stray capacitance within the circuit they are used in. More parts equals increased complexity of both parts and circuitry.

As Clueless hints at, simpler "may" be better but it is all about the individual quality of the design / circuit and what one is looking for / willing to accept as trade-offs in a design and a system.

As a side note, i've found some tube based RF amps that worked better after plucking a few tubes out of the output section. After re-tuning the circuit, power output was up, distortion was reduced, the load on the power supply was lowered, etc.... All of this has to do with the above-mentioned reasons of tubes being mismatched and "fighting each other", stray capacitance, transformer / power supply saturation, etc... I'm quite certain that the same thing occurs in audio based designs that try to achieve "gonzo" levels of output. The more output devices you have, the more closely they have to be matched.

Unless you really are an avid tube enthusiast, i would suggest "keeping things simple" and the tube count down to no more than two to four output devices per channel. Otherwise, the cost and complexity will DRASTICALLY increase above that point. Sean
>

Great post Clueless

A knowlagable is clueless indeed.

Good thing mentioned here about distortion that is the main factor in sonic signature of tube. And it's so close to the real-time live accoustics.

Clueless--you really have to do something about that moniker. You are far too knowledgable to remain here. Can't we get some real input here, like: "Well it really depends on what cables you are using, you should really try the xyz cables to get the most out of any tube." Seriously though, I hope A-gon picks up on your response and uses it in the FAQ section.

Tubes are also called "valves" for the reason they act like water valves. Think of the power supply as a water reservoir and the speakers water turbines. The more water that hits the turbine, the more the sound power. If you have only one valve, then the water flow to the turbine is limited to the valve let-thru. If you use two valves, you double the water flow, and the power. Three valves, four...the more water and the more power. But the ultimate limit to the water flow is the reservoir (power supply transformer and caps).

The number of output devices is picked to correspond to the power (current) desired and the power supply has to correspond to the current draw of these devices so the power is maintained. The quality of the sound depends on many many other factors. But as a rule of thumb, the more output devices means the more careful (read: expensive)the circuit design, since there are more devices it's easier to corrupt the current flow.

said above >> I'm not really familiar with tube amp design, so I really don't have an understanding of how the number of tubes affect the sound of an amp.

There are lots of things that add to the sound and I’m not sure the “number of tubes” as such really is that important although, as you note below, some folks have a minimalist philosophy that “less is more” when it comes to the signal path. It can be confusing though. Generally tubes have different functions in an amp. They are used in different amp “stages” to amplify voltage, or power (current), split phase, provide buffers and/or match impedances, etc… I listed a few online articles that are general introductions at the bottom of this post if you are curious about tubes.

said above>> Do more tubes allow you to minimize the sonic effects of a single tube within the group?

I guess it depends. It is just as possible that one tube, if off a little or used with the wrong operating points, or not right for the job (that is, it does not give the tube in the next stage of the circuit what it needs) could change the sound of all the others too. The Q is a little stilted because generally, if you do not like the sound associated with a tube, you are going to take it out of the circuit and use one that sounds good in the mix. Everything is interrelated and it is hard to hide/bury something. Especially is a SET because there just is not much there to begin with.

>>Do more output tubes give you more power?

Generally, yes. If you use a push-pull (PP) circuit (at least 2 op tubes) you will usually have more power than a single ended one output tube amp or a single ended amp with more than one output tube. If you use more than one output tube in a SET they are connected in parallel. And if you are using a P-P amp you are getting more power not just because you have more tubes but because you are running them in a different, more efficient, manner (at least in terms of pure power). Of course, a given type of tube can only provide so much power. A 300b about 8w, a 2A3 about half of a 300b and a 45 about half of a 2A3 etc….If you want more power one usually goes to a different type of tube. If you like the sound of a low power tube you “compromise” for less power. Generally all triode amps make this compromise of gaining a certain sound for power. (yes, I know some do not think it a compromise at all) Power is closely associated with the tube’s capacity to dissipate power (as heat) from its plate.

>>Are amps with fewer tubes "more pure" and "cleaner"?

That would be Single Ended Triodes. Some think they are and that they sound better too. I like them very much but basically it isn’t about pentode vs. triode, or Push Pull vs. Single Ended, or even tube vs. SS. It's about making the right design compromises and choosing the right parts to build a well thought-out circuit. There are many roads to sonic purity grasshopper. Don’t harden your brain’s arteries before you have listened for yourself.

>>How do Current Source Tubes and Input Tubes affect the sound of an amp?

They are both important. Amps are generally named after the power output(op) tubes which I think you mean by the “current source tube.” I think most folks would say the power tube is the “most important.” They generally name tube amps after the power/output stage - 300bs and 2A3s (type of output tube), Push-pull, single ended (describes how the op tube’s plate is attached to the OP transformer), class A or B (how the op tube’s grid is biased), ultralinear, ultrapath (how an op tube’s screen grid is attached to the power supply or the op transformer’s primary.) etc…all describe the output tube in one way or another and so it generally is thought of as the “defining stage.” But I’m not entirely sure that is always true. You can make an intelligent argument that the transformers are most important or the circuit designed around the tube or the driver stage. It is very hard to isolate these things and tell what you are listening to in a tube amp sometimes. Some think that the coupling caps (used to attach stages to each other and block dc current) add the strongest sonic signature. In any event, all tubes “affect the sound of an amp” because they are not perfect. They do not amplify in an even/linear fashion and each tube has it’s own unique signature in this way. That’s right, a tube’s sound is mostly related to its imperfections. If a tube has a perfectly linear response, that is, its output is an exact copy of it’s input only bigger, it is basically sonically invisible. It’ the unique distortion pattern that to a great extent defines the sound.

A few things online to read.

1) ✔ A Taste of Tubes (50 pages), Sonic Frontiers: http://www.anthemav.com/OldSitev1/frames/tubesfr.html
2) ✮ The National (British) Valve Museum. http://www.r-type.org/static/.contents.htm
3) The Cool Sound of Tubes, E. Barbour: http://www.spectrum.ieee.org/select/0898/tube.html
4) How A Vacuum Tube Works, E. Barbour: http://www.svetlana.com/docs/tubeworks.html.
5) Vacuum Tube Valley FAXs, E. Barbour: http://www.vacuumtube.com/FAQ1.htm.
6) Vacuum Tube FAX, H. Pasternack: http://home.earthlink.net/~busenitz/vac.html.
7) A History of Tube Companies: http://vintagetubeservices.com/page8.html
8) Vacuum Tube Primer, R. Hamm: http://www.audioasylum.com/audio/faq/tubeprimer.html

Cheers
I remain