Interesting idea. I guess I would be looking at 4* 6550 vs 2 * 2A3 aside, which might tilt the sonics to the 2 * 2A3.
What about PP v PSE?
(Audio Note make several PSE, eg the Ongaku)
diy eg Andrea Ciuffoli’s www.audiodesignguide.com/my/pse3.html “two stages only from DAC to the speakers”, output resistance 1.6 ohm.
or SE UL
www.geocities.com/n9zes@sbcglobal.net/EL84PSA.html?
Jeff, we've got to stop meeting like this!
Honestly, I don't know exactly what the optimum way to run a 6550 is in triode; and I don't have any numbers on its distortion in, say, push-pull parallel, triode connected. It certainly will require a different load than 300Bs.
But I do have a lot of data on the 7027A, a tube I used for years in almost all my pp designs. With a 4000 ohm p-p load, fixed bias, AB, triode-strapped, paralleled, I get about 1.5% THD (totally third dominated) at 30 watts. Not bad at all. I'd put that up against a 300B pp design.
Honestly, I don't know exactly what the optimum way to run a 6550 is in triode; and I don't have any numbers on its distortion in, say, push-pull parallel, triode connected. It certainly will require a different load than 300Bs.
But I do have a lot of data on the 7027A, a tube I used for years in almost all my pp designs. With a 4000 ohm p-p load, fixed bias, AB, triode-strapped, paralleled, I get about 1.5% THD (totally third dominated) at 30 watts. Not bad at all. I'd put that up against a 300B pp design.
Thanks Jason (good to have up to date info – nice that it's split H2 and H3.
Per Patrick Turner www.turneraudio.com.au/htmlwebpgs02/eh6550pographs.htm re a PP 6550 Triode:
Without any global NFB, the Rout will be less than 1% at 5 ohms and 18 watts,
and Rout will be about 1.4 ohms . . higher than the SE amp.
If 12 dB of global NFB was added, the Rout would fall to about 0.4 ohms,
and THD < 0.3% at 18 watts.
I wonder if the AWV figures used NFB?
– nice that it's split H2 and H3.Per Patrick Turner www.turneraudio.com.au/htmlwebpgs02/eh6550pographs.htm re a PP 6550 Triode:
Without any global NFB, the Rout will be less than 1% at 5 ohms and 18 watts,
and Rout will be about 1.4 ohms . . higher than the SE amp.
If 12 dB of global NFB was added, the Rout would fall to about 0.4 ohms,
and THD < 0.3% at 18 watts.
I wonder if the AWV figures used NFB?
Konnichiwa,
I have figures from a number of different vintage WE Datasheets.
The 300B operated at 350V/80mA in PP with a 4K A-A Load will give 20W with a claimed 2% THD, which will be dominated by 3rd Harmonics, so one might say 2% 3rd harmonics with good certainty. (source 1950's AT&T Datasheet)
Used in SE with 350V/80mA/2K2 Load (which is a close equaivalent) you get 9.6W with 5% (-26db) 2nd and a little over 1% (-39db) 3rd HD. Making this PSE into 1K1 will give 19.2W with 1% 3rd HD and 5% THD. (Source 1939 WE Datasheet)
BTW, if the output transformers had similar losses the output impedance would be closely similar too, for a "lossless" Transformer" around 2.5 ohm for matching to an 8Ohm Speaker.
Due to the behaviour of the auditory system we can ignore the 5% 2nd HD as the audibility of the distortion will be determined by the 3rd HD in both cases and there PP shows a 6db higher level of 3rd HD (as predicatble) for around the same Output Power.
Or in other words, a significnatly higher audibility of the distortion produced by the PP Stage opertaing in otherwise close enough to identical fasion to allow apples for apples comparison in a condition where 6db difference cannot be brushed off as "experiemntal error".
The numbers broadly match my own experience and suggest that PP always requires at least some loop feedback (6db) to drop the audible distortion for a given output power equivalent to PSE for the same valves operated under identical conditions.
Summa summarum technologica then, if percieved good sound is the goal, use SE, if low measured at a given power THD is the goal (not that I can see for the life of me why anyone would want to do that) use PP.
Of course in a complete amplifier, overall, you may find the distortion from other parts of the chain significant enough to skew the results significantly for a complete amplifier, but that as they say is another story.
Sayonara
rick57 said:
I have figures from a number of different vintage WE Datasheets.
The 300B operated at 350V/80mA in PP with a 4K A-A Load will give 20W with a claimed 2% THD, which will be dominated by 3rd Harmonics, so one might say 2% 3rd harmonics with good certainty. (source 1950's AT&T Datasheet)
Used in SE with 350V/80mA/2K2 Load (which is a close equaivalent) you get 9.6W with 5% (-26db) 2nd and a little over 1% (-39db) 3rd HD. Making this PSE into 1K1 will give 19.2W with 1% 3rd HD and 5% THD. (Source 1939 WE Datasheet)
BTW, if the output transformers had similar losses the output impedance would be closely similar too, for a "lossless" Transformer" around 2.5 ohm for matching to an 8Ohm Speaker.
Due to the behaviour of the auditory system we can ignore the 5% 2nd HD as the audibility of the distortion will be determined by the 3rd HD in both cases and there PP shows a 6db higher level of 3rd HD (as predicatble) for around the same Output Power.
Or in other words, a significnatly higher audibility of the distortion produced by the PP Stage opertaing in otherwise close enough to identical fasion to allow apples for apples comparison in a condition where 6db difference cannot be brushed off as "experiemntal error".
The numbers broadly match my own experience and suggest that PP always requires at least some loop feedback (6db) to drop the audible distortion for a given output power equivalent to PSE for the same valves operated under identical conditions.
Summa summarum technologica then, if percieved good sound is the goal, use SE, if low measured at a given power THD is the goal (not that I can see for the life of me why anyone would want to do that) use PP.
Of course in a complete amplifier, overall, you may find the distortion from other parts of the chain significant enough to skew the results significantly for a complete amplifier, but that as they say is another story.
Sayonara
I think that's a bit misleading- first, I would challenge anyone to hear 1.5% third on program. Maybe you can, maybe you can't, but it's awfully near threshold. I can't. Test tones are a different story- that amount of distortion can easily be heard on a single 1kHz tone.
Second, the distortion figures tossed around are at full power. Again, that's great if you listen to test tones at a fixed level. On program, the average level will be some 20-40dB lower, where the distortion is well below threshold. At, say, one watt out, the distortion of any of the stages considered here will be inaudible.
Second, the distortion figures tossed around are at full power. Again, that's great if you listen to test tones at a fixed level. On program, the average level will be some 20-40dB lower, where the distortion is well below threshold. At, say, one watt out, the distortion of any of the stages considered here will be inaudible.
Konnichiwa,
Maybe, maybe not. It depends on the program, the problem at any case which is audible is the resulting IMD.
Hence any argument favouring low THD or low distrotion per se is specious? Is that what you are saying? Any arguments on linearity are merely academic mental masturbation?
I think the HD figure must be related to the SPL at which the signal is present, for the very least. For arguments sake, take my 300B PP/PSE example and a pair of LS3/5 @ 82db/1W/1m.
Driven as stereo pair in a 3m listening distance with the full 20W we obtain around 92db at the suggested listening position. At that SPL 1% 3rd HD is likely audible, moreso if no masking from any present 2nd HD applies.
I agree the specific example is extreme, it was meant to illustrate that any one dimensional numbering is of little use in any determination of what does and what does not make sense.
Finally, the PP retains it's 6db higher 3rd HD compared to PSE at any given Power Level and thus whenever we are near sudibility of HD for a given SPL the PP stage will trail. Further, the cancellation of 2nd HD in the PP Stage has the additional effect of producing significant levels of 4th (largely suppressed) and 5th (fully present) HD thus complicating matters of audibility further.
Sayonara
SY said:
Maybe, maybe not. It depends on the program, the problem at any case which is audible is the resulting IMD.
SY said:
Hence any argument favouring low THD or low distrotion per se is specious? Is that what you are saying? Any arguments on linearity are merely academic mental masturbation?
I think the HD figure must be related to the SPL at which the signal is present, for the very least. For arguments sake, take my 300B PP/PSE example and a pair of LS3/5 @ 82db/1W/1m.
Driven as stereo pair in a 3m listening distance with the full 20W we obtain around 92db at the suggested listening position. At that SPL 1% 3rd HD is likely audible, moreso if no masking from any present 2nd HD applies.
I agree the specific example is extreme, it was meant to illustrate that any one dimensional numbering is of little use in any determination of what does and what does not make sense.
Finally, the PP retains it's 6db higher 3rd HD compared to PSE at any given Power Level and thus whenever we are near sudibility of HD for a given SPL the PP stage will trail. Further, the cancellation of 2nd HD in the PP Stage has the additional effect of producing significant levels of 4th (largely suppressed) and 5th (fully present) HD thus complicating matters of audibility further.
Sayonara
Apparently not, since I didn't say that. I would say, though, that any argument about distortion below audible thresholds is more for engineering esthetics than actual performance.
Listening to test tones again, are we? In this example, I'd be far more interested in clipping behavior and overload recovery.
Konnichiwa,
Good, in which case you could provide for others (I do have my own references FWIW) what you consider the audibility limits, referenced to SPL and where neccesary including the masking effect of even harmonics on the odd please. Then we actually have somthing that bears discussion and can be used to define engineering goals aimed at good sound, as opposed to engineering esthetics.
No, music. You should study D.E.L. Shorter of the BBC more I guess.
I would too, actually, or rather I'd be concerend of finding some speakers worth bothering with in the farfield, but it makes a good point of illustartion.
Sayonara
SY said:
Good, in which case you could provide for others (I do have my own references FWIW) what you consider the audibility limits, referenced to SPL and where neccesary including the masking effect of even harmonics on the odd please. Then we actually have somthing that bears discussion and can be used to define engineering goals aimed at good sound, as opposed to engineering esthetics.
SY said:
No, music. You should study D.E.L. Shorter of the BBC more I guess.
SY said:
I would too, actually, or rather I'd be concerend of finding some speakers worth bothering with in the farfield, but it makes a good point of illustartion.
Sayonara
EC8010 said:
The problem is that I'd need 8 of them to equal 4 7027...
The distribution of harmonics is interesting. Why do you think second is so (relatively) high?
If you want to pin that down, you'd need to be very specific on test conditions. And it's useless to pin down whether it's 4% for 64% of listeners at 86dB average with a Island pressing of Stand Up or 8% for 37% of listeners at 90dB average with a Mercury pressing of Beethoven's Second when any decent quality amp is running less than 0.1% at average levels.
If you want to talk about audibility, that's a different subject, and as hinted at before, is FAR more related to clipping and overload than anything else once one has the distortion below parts per hundred. It would be better to start a new thread than to continue to take this one off topic.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.