Johan, Stan White's home page has links to his articles, including 5 articles on the POWRTRON (note the spelling) - just let the mouse pointer hover over the 'documents' link on Stan's home page, and a pop-up menu will appear.
In addition to his liking for UL, Stan expounds also on the Van Scoyoc cross-coupled phase-splitter, the need for both voltage and current NFB and his view of the mentality of Electrical Engineers. Incidentally, I think the Acrosound 100 watt 6146 pp amplifier design is very similar to Stan's POWRTRON.
BTW, I fully agree in the oddities of the English language. All we can hope to do is to plough through rough thoughts, though.
ray_moth said:
Ray,
You mean like reading through Stan White's narrative? In vain, in vain I sought for a circuit (like Shakespeare, I think, was searching for something else). But thanks for helping me, Ray, with Googling, spelling and such. Me being a poor engineer who understands nothing. And if I do, I must have been on a Ph.D course .... come to think of it: With this guy's confidence in his opinion, he must be right - so I think I will send a copy of his letter to my Alma Mater, asking for that Ph.D.....
Over to more engineering matters:
The actual dimensions inside power tubes might be of interest, seeing as how Mr Grimwood came to certain conclusions regarding the same. I have a display of cut-open tubes for the very post-war folks to marvel at. The following are the distances of G2 and anode from the cathode:
6L6: C-G2 = 1,32mm C-A = 6,8mm G2 at 19,4% distance
KT66: C-G2 = 1,4 mm C-A = 6,7mm G2 at 21%
6550: C-G2 = 1,12mm C-A = 6,3mm G2 at 17,8%
EL34: C-G2 = 1,1 mm C-A = 6,4mm G2 at 17,2%
It would be appreciated that the measurements were difficult to do accurately, but I believe they are ball-park. 6L6GC was a GE tube, KT66 was Osram, 6550 was Tung-sol and EL34 was Philips. I am afraid nothing was at 43%. I also find that Russian tubes take liberties with the shape of anodes, but they test the same as the data (sort of). E.g. they use the same innards for KT66 and 6L6WXT.
Regards.
Johan,
Here is the original Stan White POWRTRON article, complete with schematic.
I must admit to being a puzzled about the purpose of R10 and R14. The rest of it is quite straightforward, although it took me a while to figure out how the Van Scoyoc splitter works. (BTW, Stan spells VS's name incorrectly).
Here is the original Stan White POWRTRON article, complete with schematic.
I must admit to being a puzzled about the purpose of R10 and R14. The rest of it is quite straightforward, although it took me a while to figure out how the Van Scoyoc splitter works. (BTW, Stan spells VS's name incorrectly).
ray_moth said:
Looks like R10 and R14 are either simple series dropping resistors, or decouplers, and the capacitors somehow weren't included in the schemo. Looks like a fairly straightforward Ultralinear design with a funkey phase splitter with an exotic name, and a whole bunch of BS to convince the atechnological public that there's something grand and glorious here when there really is not.
For all the complaining he does about the 6SN7s in the original Williamson, he uses 12AU7s, which aren't any better so far as frequency goes (the r(p)s of both types are similar) and considerably worse so far as linearity is concerned.
Besides, how many folks out there have an o'scope to align that Van Scoyoc anyway?
The whole "Powrtron" schtick is nothing more than a marketing gimmick.
An externally hosted image should be here but it was not working when we last tested it.
So there we have it 😎
If the man did not need to promote it at the cost of the landmark design of his time ......
I have Van Scoyoc's article (Electronics, June 1950) showing his cross-coupled circuit originally developed for an instrumentation amplifier. There he used 12AU7s driving 12AX7s; for once a good choice because of the driving impedance ratio and C.miller not doing much damage there. The original had a -2dB point at 100KHz, but only when followed by other cross-neutralised stages and CF.
R10 and R14 are attempts at balancing the anode outputs; no Cs omitted. The Powrtron values are a rather pious attempt at this; the original had 33K as R10 with 200K load resistors. Then, I doubt if much NFB is in play. The impedance at the junction of R3, R6 is quite low because of the double cathode follower action. But resistor addition, not through an active interface; like an inverter - that is fine .... if he intended that (can't recall SW commenting on that). And heaters floating?
Lastly R20: Good, motional feedback applied. But with a fixed value resistor? My experience is that to use this with any good effect, one had better make it variable to match the loudspeaker, or somewhere something might cause spurious oscillations. (This requirement makes it user unfriendly - instruments needed.) OK, I did not investigate this circuit, so must not follow SW's attitude regarding the Williamson.
All in all quite a good circuit, provided it works as it should. (No measurements published?) Superior to the Williamson? I doubt it. Yes, the latter needed a change in coupling time constants, and I wonder if higher value power supply caps were not available then, and the LS current feedback could be added there as well. My own preference in those days were EF37/6SJ7 input with 6SN7 or ECC33 (I think) as a ltp splitter (later EF86 and ECC81; UL output as soon as it became known). But I would not claim particular superiority - we did not then have sophisticated spectrum analysers. Thd was low enough.
I am a little off topic with regard to UL, but complete circuits go with it, and I often wondered what those contributed to the ultimate sound, glibly branded as UL being flat or whatever.
If the man did not need to promote it at the cost of the landmark design of his time ......
I have Van Scoyoc's article (Electronics, June 1950) showing his cross-coupled circuit originally developed for an instrumentation amplifier. There he used 12AU7s driving 12AX7s; for once a good choice because of the driving impedance ratio and C.miller not doing much damage there. The original had a -2dB point at 100KHz, but only when followed by other cross-neutralised stages and CF.
R10 and R14 are attempts at balancing the anode outputs; no Cs omitted. The Powrtron values are a rather pious attempt at this; the original had 33K as R10 with 200K load resistors. Then, I doubt if much NFB is in play. The impedance at the junction of R3, R6 is quite low because of the double cathode follower action. But resistor addition, not through an active interface; like an inverter - that is fine .... if he intended that (can't recall SW commenting on that). And heaters floating?
Lastly R20: Good, motional feedback applied. But with a fixed value resistor? My experience is that to use this with any good effect, one had better make it variable to match the loudspeaker, or somewhere something might cause spurious oscillations. (This requirement makes it user unfriendly - instruments needed.) OK, I did not investigate this circuit, so must not follow SW's attitude regarding the Williamson.
All in all quite a good circuit, provided it works as it should. (No measurements published?) Superior to the Williamson? I doubt it. Yes, the latter needed a change in coupling time constants, and I wonder if higher value power supply caps were not available then, and the LS current feedback could be added there as well. My own preference in those days were EF37/6SJ7 input with 6SN7 or ECC33 (I think) as a ltp splitter (later EF86 and ECC81; UL output as soon as it became known). But I would not claim particular superiority - we did not then have sophisticated spectrum analysers. Thd was low enough.
I am a little off topic with regard to UL, but complete circuits go with it, and I often wondered what those contributed to the ultimate sound, glibly branded as UL being flat or whatever.
Wrong!! Write out 100 times: "I am not off topic."
I have read that the VS X-coupled splitter is fundamentally unbalanced, like an LTP splitter with a resistor instead of a CCS in the tail. I can't recall why the VS splitter is supposed to be unbalanced but it shows up clearly when modeled using SPICE. Now,that was off-topic! 😀
To get back on topic, it seems to me that UL ought to be superior to triode-mode, for the same tube, because the power is there and the low distortion is there too - a 'perfect' compromise between pentode and triode. So, what more could one ask? UL has to be the best arrangement, right?
Yet many people who have amps that can be switched easily from UL to triode-mode report that they prefer the latter. The reasons they give are usually somewhat nebulous but I suppose that's only to be expected in such a subjective area. A lot probably depends on personal sound quality preferences and tastes in music. Then there are external factors, such as the speakers used, the accoustics and size of the listening area, etc. Most people don't get the chance to try all permutations and combinations of these things, to see if UL could sound as good as or even better than triode, given the chance.
The x-coupled inverter will be unbalanced (by about mu/1+mu) unless either an extra inverter stage is used on one input (like ARC) or a balanced input transformer feeds the two sides. When you draw out the input stage and see where the signals are referenced, it will become clear why this is so.
Re: UL to triode switching, I'm always wary when I hear reports of "sounds better" without respect to whether or not the output sounds like the input. Inserting a new pole into a loop where compensation is not adjusted or (in the case of an amp without global feedback) putting in a new lowpass filter will certainly change the sound, but not necessarily in the direction of more faithfully replicating the input. Another variable is introduced when the stages are run in AB1, where Millering changes during that transition. And let's not talk about optimum loading... Even if one mode "sounds better" than the other, there are at least two variables changing, so attributing the change to some magic property of UL versus triodes is unjustified.
Now, if someone does a REAL comparison, switching between the modes but at the same time changing the load impedance and redoing the compensation so that each mode is optimized, I'll take that seriously.
Re: UL to triode switching, I'm always wary when I hear reports of "sounds better" without respect to whether or not the output sounds like the input. Inserting a new pole into a loop where compensation is not adjusted or (in the case of an amp without global feedback) putting in a new lowpass filter will certainly change the sound, but not necessarily in the direction of more faithfully replicating the input. Another variable is introduced when the stages are run in AB1, where Millering changes during that transition. And let's not talk about optimum loading... Even if one mode "sounds better" than the other, there are at least two variables changing, so attributing the change to some magic property of UL versus triodes is unjustified.
Now, if someone does a REAL comparison, switching between the modes but at the same time changing the load impedance and redoing the compensation so that each mode is optimized, I'll take that seriously.
Listening with good headphones is always a valid reality check when evaluating the main amp & speaker system results. Not for soundstage but for tonality and distortion.
tubelab.com said:
And you know, I plan to waste more time and money building hybrid class A single ended amps!
And on and on the controversy goes ... hopefully to the enjoyment of members and high-lighting all sides, and not drawing swords (that would be unfair, since I don't even own one).
But replace controversy by hi-liting. Salas, I would say that is fair. That gets rid of some extraneous factors (okay, so headphones are also not linear - but there). Only one matter in that instance: It would help to load the output with an adjustable 100W resistor, otherwise one is always listening to hardly 1W. Never tried it myself, but, come to think of it, worth a try - also easier on the neighbours. Calibrate the pot in watts.
There is one way of checking whether the difference UL/triode is real. See, in my transistor amp I "had" to include a tone control defeat switch ... sure you understand. At more than one demo I had a few intelligent brethren switch this back and forth, pronouncing with a suitable facial expression
: "You see, one can hear the difference with a keen ear!" I kept quiet, also with a suitable facial expression 😎. You see, that switch was not wired up at all.
So just wondering if one would reverse the connections to the UL/triode switch ..... That is not just naughty. It is also painless, embarrassment-less and pre-conception-less. I would bet that in a blind test the difference would not be distinguishable, because even with the rest being equal, there is not enough difference.
So why advocate it? Because now and then what difference there is, would come out in a certain piece of music, if not in general. Even that is marginal - it is mainly a 3 - 6dB difference in maximum output. It would also be an improved handling of a varying loudspeaker load. But again, not evident very often. And, from the design side, it is easy. So simply, why not use it? But my bet would be that an audible improvement from a switchable option would not be evident every time.
But I am glossing over a fact pointed out by SY. There is enough difference in the topology that NFB compensation will be out for one of the options. And depending on the complete design ....?
Regards
But replace controversy by hi-liting. Salas, I would say that is fair. That gets rid of some extraneous factors (okay, so headphones are also not linear - but there). Only one matter in that instance: It would help to load the output with an adjustable 100W resistor, otherwise one is always listening to hardly 1W. Never tried it myself, but, come to think of it, worth a try - also easier on the neighbours. Calibrate the pot in watts.
There is one way of checking whether the difference UL/triode is real. See, in my transistor amp I "had" to include a tone control defeat switch ... sure you understand. At more than one demo I had a few intelligent brethren switch this back and forth, pronouncing with a suitable facial expression
: "You see, one can hear the difference with a keen ear!" I kept quiet, also with a suitable facial expression 😎. You see, that switch was not wired up at all. So just wondering if one would reverse the connections to the UL/triode switch ..... That is not just naughty. It is also painless, embarrassment-less and pre-conception-less. I would bet that in a blind test the difference would not be distinguishable, because even with the rest being equal, there is not enough difference.
So why advocate it? Because now and then what difference there is, would come out in a certain piece of music, if not in general. Even that is marginal - it is mainly a 3 - 6dB difference in maximum output. It would also be an improved handling of a varying loudspeaker load. But again, not evident very often. And, from the design side, it is easy. So simply, why not use it? But my bet would be that an audible improvement from a switchable option would not be evident every time.
But I am glossing over a fact pointed out by SY. There is enough difference in the topology that NFB compensation will be out for one of the options. And depending on the complete design ....?
Regards
ray_moth said:
Funky symmetrically - looking input (waste of tubes in my mind) DC coupled to a differencial phase splitter. R10 and R14 may give a single advantage: to limit maximal anode voltages on tubes. What is good in the amp, is small number of phase shifts on low end introduced: it means good stability on low end with deep global negative feedback possibility, and a positive feedback by current that helps to damp speakers, up to a negative output resistance. I use such feedback when driving subwoofers, results are remarkable. In his case the feedback is wide-band, so electrostatic amps may benefit as well.
ray_moth said:
You are basically asking why people have different opinions. Good luck on getting a satisfactory answer.
People's opinions are based to a heavy degree on their systems. When a speaker is under 90dB sens and a room is rather big the chances are a good UL will prevail upon a good triode mode amp.
salas said:
Well, as one with ~100 dB sensitive speakers, and a DIY 40W U-L amp, I wanted to say that there have been no triode amps show up that sounded better.
I suspect that most UL amps are being done for power reasons, and that in that vein, other compromises are being made. AB1 operation, 'lytics in the cap-input power supplies and use of a loop of NFB to bring it to reasonable output Z numbers.
Mine isn't done that way. The only 'lytics in mine are in the bias supply and they're decent low esr and they're hiding on the other side of a very fine grid choke. Finals are DH pentodes and they're thoroughly Class A biased. They're heavy too at nearly 10 pounds per watt.
Being able to deliver some power has made a difference, even at levels where headroom shouldn't be an issue. It is rare music that doesn't have 20:1 power peaks in it. With a low power amp, running at 30-50% of its rated power, that is just going to be forced into overload. Sometimes it is easy to see/hear, and sometimes it's not. That doesn't mean it is not effecting the music for the worse.
Pass the 813's please, I need more power...something like 3dB more!
cheers.
Douglas
An out of the norm DHP UL then. Built to an exemplary UL amp standard. But popular opinion is based on this dinky lil amp archetype with the non optimized 'flip da switch' UL mode...
AFAIK, those amps that have a UL/Triode switch were mostly, if not all, designed first and foremost for UL operation. They were optimized for UL, not triode. That probably means that, when used in triode-mode, the load impedance, the NFB and the frequency compensation networks are all sub-optimal.
For these reasons, such amps probably do not perform as well or sound as good in triode-mode as they would had they been designed and optimized specifically for that purpose. Yet people still say they prefer the triode-mode sound. Odd, isn't it?
For these reasons, such amps probably do not perform as well or sound as good in triode-mode as they would had they been designed and optimized specifically for that purpose. Yet people still say they prefer the triode-mode sound. Odd, isn't it?
ray_moth said:
Or the reason could be that UL sounds so bad even a poorly designed triode connected amp sounds better.😉
- Status
- Not open for further replies.