Audio Wisdom: Debunking common myths

[planet10]

A push-pull amp is quite capable of doing the same thing.

Indeed. I have a little (Scott LK48 based) 4W PP class A EL84 strapped in triode that sounds wonderful. We are eagar to see how it compares to the same (or similar) amp with the output stage biased as per the Red Light District -- the extra power is a strong siren call.

dave

 

[phn]

I don't know about the V 69. But I would think it has relatively high distortion. I would think that the Telefunken engineers were as pragmatic as the Bell Labs engineers were when they designed the WE 91A.

 

[lndm]

Up until the output stage in a push pull amplifier, the amp is generally single ended class A, or exposed to the same characteristics as SE class A before balancing. If there is that much 2HD, wouldn't balancing it convert it to 3HD?

 

[phn]

This is biased and should be taken as that. So a second, more enlightened, opinion would be appreciated.

http://www.nutshellhifi.com/triode4.html

 

[Magura]

This is biased and should be taken as that. So a second, more enlightened, opinion would be appreciated.

You will find a somewhat well documented POV at

www.passlabs.com

and

www.passdiy.com

in the articles written by Nelson Pass.

Magura

 

[rdf]

Third harmonic, too. And 5th as well. In SE amps, they tend to be smaller relative to 2nd, but much greater on an absolute basis, mostly because the 2nd is so enormous in most SE designs.

Hi Sy. Why would 3rd be higher in SE, all else equal? Odd harmonics add in push-pull. Did you mean 'as typically implemented'? (The numbers for your EL84 suggest global negative feedback.)

 

[SY]

rdf: That's a reasonable qualifier. But apples to apples- the numbers have to be compared at the same power level, e.g., 1 watt. True, push-pull doesn't cause cancellation of odd order, but the output tubes don't need to swing as much for the same power level.

 

[SY]

Up until the output stage in a push pull amplifier, the amp is generally single ended class A, or exposed to the same characteristics as SE class A before balancing.

Not in most competent tube designs, especially the more modern ones. For example, the Williamson only runs the first stage single-ended. Same with the Marantz 9 (ignoring the phase reversal stage). The Crystal Palace has no single-ended stages.

 

[rdf]

Good point, never considered that aspect.

 

[lndm]

apples to apples- the numbers have to be compared at the same power level,

I disagree. I think that the devices should be in all cases at their optimum operating points and the fact that a single-ended amp needs bigger devices and transformers is a separate caveat. The problem as I see it is: are we comparing amplifiers or topologies?

 

[lndm]

The Crystal Palace has no single-ended stages.

Correct me if I'm wrong, but isn't a balanced stage just two single-ended stages operating simultaneously back-to-back, ala push-pull?

 

[SY]

That depends on your definition. In the case of differential amps, the sides are tied together absolutely by the cathode CCS, as opposed to two grounded cathode stages handling each half of the signal. The former show even harmonic cancellation in differential mode.

 

[lndm]

That depends on your definition.

...that the linearity of the devices as if biased SE remains, but is combined with its opposite (but not necessarily equal) counterpart.

 

[SY]

In that case, you've made even a better argument for push-pull: your choice of an operating point can trade off increased even order (mainly second) distortion for reduction in odd order (mainly third and fifth). And, of course, for any given output, the tubes are swinging less voltage.

 

[lndm]

How are two opposed transfer curves going to produce an assymetrical result? Are you talking of intentionally imbalancing your circuit?

 

[Bare]

Geeeez.. More Drivel from the Quasi Technical Sycophants..
Gotta love Audio Weenies.

 

[SY]

I was not clear. Suppose you have operating point A where the second harmonic is -28dB, third is -55dB, fifth is -65dB. Another operating point B shows second at -15dB, third of -65dB, fifth of -78dB.

OK, you've got an SE amp. You sure like the third and fifth improvements from point B, but geez, that second... that's too much for even the most die-hard SE fan. Ah well, you rationalize, point A isn't so terrible. I on the other hand smile, put the tubes in push-pull, use point B, cancel the second, and drop the per-tube swing to get a doubly-whammy improvement.

 

[rdf]

In that case, you've made even a better argument for push-pull....

Much hinges on how the ear responds to an even monotonic harmonic distribution vs. a decreasing sequence of odds-only at various operating levels. In other words, does one harmonic to some extent mask the next higher? (Honestly don't know.)

BTW SY, the SE 3rd and 5th thing is far from universal. On my bench a single 6CN7 - LED biased, plate load >1 meg - swings > 100 v p-p into a cathode follower at about 0.6% 2nd with all higher harmonics 90+ dB below the fundamental. And yes, I'm curious what it'll do push-pull.

 

[lndm]

Much hinges on how the ear responds to an even monotonic harmonic distribution vs. a decreasing sequence of odds-only at various operating levels. In other words, does one harmonic to some extent mask the next higher? (Honestly don't know.)

Neither can I say that I can prove it but I know what I like. I find odd and higher order harmonics to sound kind of exposed, or relatively excessive when they are out of proportion.

I also find an amp to have a character that is evident despite its distortion being very small.

SY, when you speak of the even harmonics getting removed, I think of the odd harmonics becoming exposed. If distortion is inevitable shouldn't it be even handed? (no pun intended)

On my bench a single 6CN7 - LED biased, plate load >1 meg - swings > 100 v p-p into a cathode follower at about 0.6% 2nd with all higher harmonics 90+ dB below the fundamental. And yes, I'm curious what it'll do push-pull.

or as a mu-follower?

 

[SY]

SY, when you speak of the even harmonics getting removed, I think of the odd harmonics becoming exposed.

But those harmonics are there anyway, and in the SE case, there is more limitation on their minimization. You can speculate about masking effects, but the emphasis is on "speculate." And let's not talk about the related issues of power supply and common-mode rejection. And the greater design limitations and tradeoffs necessary for an output transformer to work properly when it has a river of DC running through it.