Why do some people dislike ultralinear?

[Bandersnatch]

hey-Hey!!!,
The switch is built for religious fanatic customers. Look, I can change it, and it sounds better. Besides, everybody knows that triodes are the Way to True Nirvana. A pentode is for those damned with poor hearing and power hungry speakers.

Well, since Nobody builds amps like me, I will happily remain nobody.
cheers,
Douglas

 

[Salas]

Nobody questions your findings and practice. We investigate the root of a possible urban myth. To echo SY with a slight twist, sociopathology and bad tweakers.😉

 

[thevoice]

More along the lines of social engineering in audio:

Before I somehow ended up on the other side of the world commissioning lighting systems, I used to mix the odd band, mostly small stuff but some people getting local radio play and so on. A lot of over inflated egos were in play, particularly amongst the evangelical christian bands, for some reason.

One particularly difficult lead man to deal with simply didn't feel complete without asking for more or less foldback, less bass or somesuch thing during each and every gig. To begin with I'd give him what he asked for. Eventually I discovered that if he couldn't see me make the change, he'd ask for it again. So I started reaching for random knobs and pretend to turn them, making sure he could see me. Without fail he would be satisfied - as long as everyone in the room knew it was him calling the shots. I worked with him and his band for 2 years, he never realised the joke was on him.

Sometimes people hear what they want to, he wasn't the last bloke I played that trick on.




Of course, some people really *can* hear the difference...

 

[Kent Smith]

I see alot of truth here....

I think people always say that triodes sound better for fear of being laughed at. Well, I have triodes, but I prefer listening to my 6V6 PP UL amp. But if the UL amp wasn't tweaked right, I would certainly prefer triodes!

I don't think the most important thing is the tube, or if UL/pentode/triode, or gold plated transformer cases..... many things go into the design, either it's good (and it sounds good) or it isn't.

Sorry, don't mean to offend anyone, that's only my opinion.

And I don't hate triodes.... but I don't worship at the SET Alter either.

 

[richwalters]

Besides, everybody knows that triodes are the Way to True Nirvana. A pentode is for those damned with poor hearing and power hungry speakers.

Well, since Nobody builds amps like me, I will happily remain nobody.
cheers,
Douglas

Avoid getting anywhere to the truth (!) I do suffer from tinitus but that wasn't due to loud music.......I've fooled many UL listeners that they were listening to triode p-p.......Lets be fair, triode operation does clean up those lazy harmonics present in UL and can tolerate considerable mismatched pairs but to the rescue of UL, I find a well matched pentode/tet' p-p o/p stage can sound like a triode o/p stage- It's better than it seems.

For those Class A triode power fanatics; it is a dreadful waste of energy for basic listening levels.

richj

 

[rdf]

For those Class A triode power fanatics; it is a dreadful waste of energy for basic listening levels.

richj

No fanaticism here, just picking the best topology for my needs based on measurements. Like many apartment dwellers a handful of watts coupled with high efficiency speakers (94 and 97 dB/watt) are a lease-breaker combination. In their linear operating ranges watt for watt every output I've built measured better in triode mode than UL, operating parameters adjusted to suit. Distortion wasn't usually much worse in UL, though the harmonics coming up tended to be the lower, more benign but harder to correct with feedback ones. Damping factor is the killer, unless the speaker load is unusually linear I don't see how to avoid global negative feedback in UL, a completely different argument but something I prefer to avoid.

My current listen is a 7591S triode-mode PP good for about 12 watts peak at ~0.5% predominantly 3rd harmonic. At 1 watt THD is below 0.1% 3rd followed by 2nd with no other harmonic visible, at typical listening levels the distortion drops well below 0.03%. UL buys me little important for 99% of my listening but added complication.

I'll toss this out more as a question than a criticism of UL. GNFB requires extra gain, which typically means more stages and more front end noise. The amps I build target the minimum front end gain required to drive the output into hard clip with a CD player. Unless I screw something up, hiss with the my ear to a high-efficiency tweeter was inaudible or at the limits of audibility with all of them. In contrast, high GNFB designs like the various SS boxes I bring home from work periodically or my MC240 all have clearly audible hiss. Is this a neccessary consequence of the extra gain and stages required of UL and pentode outputs or implementation? It sounds like a silly question but it's such a consistent observation.

 

[Bandersnatch]

One more time...global or loop NFB is *NOT* needed to bring pentode or U-L amps a low output Z and ability to deal with wacky speaker behaviour. Multiple stages of high gain is another urban legend of similar truth level.
cheers,
Douglas

 

[Johan Potgieter]

Amen to Brandersnatch - although some effect of NFB would be to get the damping factor out of a sometimes grey area. I am also on record for stating it somewhat different: One must beware of a low output impedance obtained by NFB to address loudspeaker anomalies. Again, an old article by Prof. Otala illustrated that quite clearly. (This is most prevalent in SS circuits, but also relevant to tubes.)

I must get distortion analysing ability up again! (Lost my contact at my last work. ) But as I can recall I never got much "malignant" distortion from UL - not very different to triode in content.

Rdf, would you care to give representative firgures, please? (Meaning breakdown, not thd.) For popular purposes, I always equated UL characteristics as those obtainable from slightly less powerful (higher rp) triodes. E.g. KT88 UL would look about the same as KT66 triodes, apart from max. available power output. But that was some time ago.

Regards.

 

[Salas]

Some (even very little GNFB) I find necessary to keep standard gain & surely same between L&R.

 

[rdf]

"Multiple stages of high gain is another urban legend of similar truth level."

"One more time...global or loop NFB is *NOT* needed to bring pentode or U-L amps a low output Z"

I'm not sure what you mean. By low gain stage count I mean a single gain stage (gain as described earlier) + output, perhaps with a unity gain cathode-follower driver. I've found the latter sufficient to drive even insensitive 8xx transmitter tubes hard into A2 at reasonable distortion. The most popular UL and pentode amp topologies all employ much more gain than 'just enough to clip', and more gain stages.

The relationship between gain, output impedance and feedback (loop or internal) is pretty set. Feedback always requires more gain, high output impedance always requires more of both for equivalent sensitivity. If you're objecting to 'global', point taken. E-linear? Arguably a form of interstage loop. I've tried it on a few occasions and didn't like the results. At least in my implementations it invariably reduced 2nd by bringing up higher harmonics fast, measured and simulated. Cathode feedback? Results limited and very dependent on output tube sensitivity. Tried it, liked it in some implementations (EL84), but generally prefer fixed bias and don't like the consequent DC voltage on the speaker terminals.

Neither has been required with the triode-mode amps I've built for my needs. Triode mode accomplished my objectives better on purely technical grounds, the intended point. Like so many of these discussions it was veering into psychological explanations of triode preference. If in the future I really need high power at reasonable cost (no transmitter tubes) I would certainly look to UL over pentode though.

Hi Johan, unfortunately no. I'm pretty bad at keeping documentation on things I discard but don't really see how it could be otherwise. At any given output level/tube/OPT far enough away from A2, triode-mode increases the feedback around the output and could only result in lower distortion and Zout than UL. UL does preserve much of triode's benefit but never bettered it in my tests. Although it's obviously on a case-by-case basis, for the tubes I've tried - KT100, 6BQ5, 828 and a screen rating abused GU50 off the top of my head - cathode feedback alone never compensated for the loss in damping factor compared to triode mode.
The only two pentode amps I've measured on the other hand both look like the spectra below, in this case the high GNFB Mullard 3-3 SE.

http://www.diyaudio.com/forums/attachment.php?s=&postid=1107479&stamp=1168846982

The other was a near Class-B high GNFB jukebox amp, a 7868 P-P affair that donated its iron for the 7591s amp discussed earlier. Might be high time to run the MC240 on the bench for a look-see.

 

[Bandersnatch]

hey-Hey!!!,
Please describe your E-Linear implementations. Like what sort of tube, and what the various parts were referenced to( voltage-wise ). I have not observed the things you have in current implementations, but found several ways to rig an E-Linear stage that did behave as you describe....so I wonder what you did.
cheers,
Douglas

 

[Johan Potgieter]

At any given output level/tube/OPT far enough away from A2, triode-mode increases the feedback around the output and could only result in lower distortion and Zout than UL. UL does preserve much of triode's benefit but never bettered it in my tests. Although it's obviously on a case-by-case basis, for the tubes I've tried - KT100, 6BQ5, 828 and a screen rating abused GU50 off the top of my head - cathode feedback alone never compensated for the loss in damping factor compared to triode mode.

Thanks, Rdf.
I may be dense here - like forgetting to read the last part of your previous post because I had to go elsewhere for the moment!

So, before coming to what I do not altogether follow, about noise: Noise, as a random uncorrelated signal, is not cancelled by NFB like a periodic signal. There is something like a sq. root relationship somewhere; quite long ago for me. But it means that totally random noise is cancelled by a smaller factor than periodic signal. Thus the S/N will decrease with reasonable gNFB.

By "triode mode increases the feedback..." I do not fully understand. Do you mean in a damping factor way? Also, I agree that UL cannot better triode performance. As simply put, it seems to provide most of triode and pentode benefits. I looked again, for KT88 (only good graphs I got) the rp is 4K4 for UL (40% taps) and 2k4 for triode. OK, almost twice. Distortion is about 15% more.

[But that is also a distorted picture, to be truthful. Distortion is shown as % of max. output. If triode output is only half that of pentode, than the actual distortion amplitude is also half if the same %figure is stated. This is a confusing way of stating it. In combined graphs for pentode and triode it is never clear to what %D refers. One does not hear distortion as % of some other variable. One hears it as a certain level above hearing threshold, although it can be masked.]

Then exactly what do you mean by cathode feedback? An unbypassed cathode resistor or cathode coupled to the loudspeaker winding for some local NFB? Sorry if I sound dense.

Lastly you made my day. I thought I was the only one not being very organised about keeping results! 🙂

Regards.

 

[Tweeker]

An unbypassed cathode resistor to the loudspeaker winding, unless you have a separate cathode winding. Lundahl 1620CFB an example of the later. The resistor doesnt have carry all the current, there could also by a bypassed resistor to ground for less feedback etc.

 

[Johan Potgieter]

Thanks, Tweeker.

But 🙁 - an unbypassed cathode resistor ... in an output stage? That is current feedback, with all the disadvantages one does not want in an output stage. Fine, the loudspeaker winding will give voltage feedback, but no more than a few dB (about 5dB for an EL84, 8 ohm speaker), which can easily be overridden (word sounds wrong here) by the higher current feedback. Why an unbypassed resistor?

 

[Bandersnatch]

hey-Hey!!!,
While cathode FB could be applied to any means that applies the signal to the cathode...I treat it as a very specific circuit. That is a tertiary CT coil in the output TX to which the final's cathodes are attached to. This coil is out of phase with the plate winding so as to deliver negative FB...though it could be arranced to deliver positive FB...🙂
cheers,
Douglas

don't forget that the load applied to the finals changes with the addition of these extra turns.

 

[Tweeker]

I suppose it could be bypassed.

 

[ray_moth]

Tweaker,

An unbypassed cathode resistor to the loudspeaker winding

Better to take the cathode resistor to ground and the cathode bypass cap to the speaker winding. That keeps DC, but not the signal, out of the secondary.

Bandersnatch,

global or loop NFB is *NOT* needed to bring pentode or U-L amps a low output Z and ability to deal with wacky speaker behaviour.

Pentode and UL amps are quite dissimilar in output impedance characteristics. UL OP impedance is not much higher than triode and is normally OK with only 6-10 dB of added NFB (in addition to its own NFB, that is). Pentode OP impedance is an order of magnitude higher that UL and needs much more NFB, which almost certainly needs to be global because of the large amount needed (usually at least 26 dB).

 

[Tubes4e4]

Hi Johan,

I looked again, for KT88 (only good graphs I got) the rp is 4K4 for UL (40% taps) and 2k4 for triode. OK, almost twice. Distortion is about 15% more.

Some further UL-plots are available at Audiomatica Sofia site

Plus a few smaller tubes, sometimes with several UL ratios to compare, on my site.

Regards,

Tom

 

[Bandersnatch]

Pentode OP impedance is an order of magnitude higher that UL and needs much more NFB, which almost certainly needs to be global because of the large amount needed (usually at least 26 dB).

Well...I don't see why you'd jump to the conclusion that just because you need more it has to be global. RCA made a *VERY* nice set of 6V6 amps, and they used no global FB. They did run a lot of it back a stage to the driver cathodes from the final's anodes....🙂 So far that PPP 6V6 amp was the best sounding 'off the rack' amp I have ever heard.

The global loop is *NOT* required in any case. IMO, neither is a SS-like low output Z.
cheers,
Douglas

 

[ray_moth]

The global loop is *NOT* required in any case. IMO, neither is a SS-like low output Z.

I never said a global loop was required, only "almost certainly" 😀

"Low output Z" is a relative thing and I doubt if any damping factor higher than 10 is needed. Lower DF can be OK too, but that depends on your speakers. In any case, even to get a DF of 5 or so needs a LOT of NFB when you're starting with pentodes.