|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
144.131.127.147
I have sometimes read that people who have tried ultralinear PP "didn't like it" and preferred triode PP operation, despite the loss of 50% of the power. I've even read cases where pentode mode sounded better than UL. However, I've never read any explanation for these findings.If we are to believe the copious material has been published in support of UL since it became popular, it should be the "best of both worlds", in terms of distortion, power, output load flexibility and damping. Does anyone know any reasons for avoiding UL (assuming the OPT used is good enough for the job)?
Follow Ups:
Any resonance in the transformer from leakage inductance and capacitance is being fedback via the UL taps. UL mode is much more susceptible parasitic oscillation than triode or pentode. (On a parallel push pull EL34 150R screen resistors worked well in triode and pentode mode but I had to change them to 1K for Ultralinear mode to supress parasitics).I've built around 8 or 10 Ultralinear Amps and have always found that the value of the series resistor between the screen and the UL tap makes a big difference to sound
BUT
the addition of zobel networks between anode and screen taps (in addition to the series resistor) makes a huge difference. These zobels are very difficult to dimension properly but even sub-optimal zobels make an incredible difference to detail and clarity.On the one occassion that I tried to do the zobels "scientifically" I sat down with this paper:
http://www.siteswithstyle.com/VoltSecond/Damping_ringing_XFMRS/Damping_ringing_in_xfmrs.html
I ran just the output tubes - no phase splitter or driver and drove one side only while the other was at DC idle. I monitored the anode on that side to draw up a Bode Plot (Phase and Amplitude). Then did the other side. It was quite apparant that one side had a lower resonant frequency than the other and so I calculated a zobel to supress that resonance. (I use what Voltsecond called Rd_opt_Zi values). having supressed the lowest resonance (confirmed by redoing the Bode Plot) I then did the other side to supress the next lowest resonance. Unsurprisingly I ebded up with different zobels on the 2 sides and that makes me suspicious that there maybe something wrong with my method. By the way using Voltseconds method I tried zobels anode to anode, each anode to primary centre tap and across the secondary and confirmed that anode to screen was the right place to put the zobels.A bit of a rave but hopefully there is something there for people to have a think about and experiment with.
Manipulation of screen voltages is a bad design choice, IMHO. The pentode was designed to be operated with a fixed screen voltage to provide more precise control over plate currents and to increase efficiency. When you start to jack around with the screen voltages it becomes very difficult to track what tube behaviors you are changing. Cathode Feedback, or the distributed load connection provides a superior design solution to pentode operation. It is far more effective at reducing plate resistance than UL connection. The drive requirements are easy to calculate, and the dampening factors are usually better than even triode connection. In addition, plate efficiency remains quite high, usually around 40%. CFB is also better at reducing THD than UL connection. From a design point, it is also easier to pick a load line with CFB than UL because you can still use the pentode plate characteristics to plot your load line. With UL connection, you need the UL plate curves, which can be hard to find for tubes other than the most popular ones. Transformers for CFB are quite easy to come by due to the many competent transformer makers we have today. All in all, I really like CFB pentode mode. It is very flexible and you can tailor your operating point to emphasize the harmonic profile that best suites your system.Rgs, JLH
Would you believe that the cathode FB design at least as implemented by PJ walker in the QUAD II (and as discussed in the polite rant by Mr.’s Walker and Williamson in the link you provided) is usually considered a form of UL. In the QII, g2 is ostensibly a fixed voltage. However, because the output tube cathodes go to ground thru a centertapped winding, there is signal voltage at each cathode. With g2 fixed but cathode voltage changing, it’s effectively the same as if the cathode voltage was fixed but g2 was changing in same magnitude but in opposite direction of cathode. IOW, a corrective feedback signal appears at g2 just like “conventional” UL using taps off the plate windings. There will also be what amounts to corrective feedback at g1 for same reason.One advantage of cathode type UL is that you get to pick your g2 static voltage. You can do this with UL too but it requires an OPT with a separate g2 winding...not all that common. Come to think of it the QII OPT config isn’t all that common either.
One conclusion is that conceptually, UL may not be any more flawed than other feedback schemes. However, since proper implementation seems to require a very good OPT, practical implementations may indeed be flawed to some degree when cost is a major consideration.
Hi Steve,I can see what you are saying, however I was thinking about single ended circuits, not push-pull. I have been able to get some very nice results with single ended high Gm pentodes with CFB. Not to let the cat completely out of the bag yet, but using 15% CFB effectively turns my output tube into the following specs: Gm = 23mA/V, Rp = 285 ohms, and Mu = 6.56. Needless to say, it plays music that makes going back to low Gm triodes impossible.
The concept of cathode feedback being UL in disguise is equally applicable to SE penthode topology. Properly applied, corrective feedback is a helpful tool as you've found. I'll be interested to hear what your output tube is when you do decide to let the cat out of the bag.
Ray,
I use UL routinely BUT coupled with some balanced shunt feedback from the output tube anodes (the "Baby Huey" scheme). With this arrangement I can generally get away with low or no global feedback although I've used up to 6dB of global feedback with minimal image degradation.
This is mainly with lower power amps using 6GW8, EL84 and 6V6
I like its sense of speed compared to triode mode.
I like its reduced global feedback requirements compaared to pentode mode.
Cheers,
Ian
I have read these arguments before and I have owned many triode, UL and pentode amps with regulated screen supplies. Lets stand back from the theory a moment and look at the results. There are many GOOD UL amp: Marantz 8B and 9, Eico HF-60, Citation II, Conrad Johnson Premier 12s, Music Reference RM -9, just to mane a few. I have owned all these amps, and many others; you can make a good sounding UL amp--it is all in the implementation. I have also heard lousy SETs, but I blame the implementation rather than the concept.
I also find it interesting that McIntosh "criss-crossed" the screens to opposite anodes. Seems nearly a polar opposite approach compared to UL.
...further demonstrates the ingenuity of the unity coupled output stage.Since the cathode and plate windings are 1:1, the cathode and plate voltages are always moving in equal but opposite directions. By connecting g2 of the one side of the output stage to the plate of the other side (which overall is 180 deg out of phase but plate 2 and cathode 1 are in phase) the voltage differential betw the g2 and cathode of the one side is essentially constant under all conditions of normal operation. This results in the equivalent of true, fixed g2 voltage pentode operation even though cathode voltage is swinging wildly.
UL alone doesn't lower OP impedance enough. To get it down enough designers use a total NFB loopSo, the reliance on just UL, and total loop NFB doesn't sound as good as it could.
Using the other available NFB paths makes ul acceptable, and even beneficial.
BIg Fun and thanks this topic has been the most interestng in weeks.
-3db
...compared to either correct pentode operation (fixed, regulated screeen voltage) or triode (screen tied to anode) operation.My *opinion* as to why this is so comes from examination of the screen current drawn in UL mode. I placed a pair of scope probes in differential mode across the circa 1k screen tap to screen R and saw a very distorted waveform, hinting at a very distorted screen current.
To me, this distorted current must "mix" with the obviously less distorted anode current in the transformer itself - and either cancel or add (or both) at a microdynamic level signals - vanishing vital signal information.
This is my take on it.
even in local form, though nowhere near as bad as global. I'm with you on UL in general, everything seems to be there but it just doesn't sound quite right. Every amp I've triode strapped sounded better to a greater or lesser degree, more relaxed and definitely more three dimensional ... more like real music.Even my old ARCs which used very expensive OP Traffos sporting both UL taps and CF windings sounded way better in Triode mode and I think NFB is the culprit. The implementation has to be near perfect in every way, especially linearity and phase for it to work properly without stuffing things up ... and that's almost impossible in the real (flawed) world.
And might I add that it was most probable that was due to the very high impedance of the screen supply. Once a person has heard a pentode with a well made and low impedance G2 supply there is no going back IMHO.I know you are very familiar with the asymmetrical plate impedance of tubes in respect to their plate current. While a normal tube's ( 300B in my example ) plate impedance might swing from 250 ohms near Vg1 = 0, up to 2000 ohms at twice the quiescent Vg1 point. The pentode's G2 impedance under goes an even more radical impedance swing which can be from a few thousand ohms to several Meg ohms. IMO, this is what causes a lot of the screen current non-linearity. However, a low impedance G2 supply greatly reduces the non-linearity and regains the lost transparency and detail.
Boo as Ivan303 does.
Jeff,I remember that you are another who's expressed a dislike for UL, or "ultra(non)linear" as you call it. What is it you don't like about it?
Jeff,You're another who's expressed a dislike for ultra(non)linear, as I remember you call it. What is wrong with it, as far as your listening goes?
AW,Have you ever tried the separate screen grid winding "route"? TANSTAAFL applies and the "iron" is expensive. However, a separate g2 winding allows g2 B+ to be regulated at a fraction of anode B+, which deals with a host of ills.
ALL NFB, local (like UL) and loop, works best when the circuitry is reasonably linear to begin with.
Eli D.
Hi."TANSTAAFL" is short of There Aain't No Such Thing As A Free Lunch".
Mind you, take care to use it as it is the official name of wire supplies company in USA, specialized in HV (25KV) & HT (538C) wires.
Agreed to your comment on UL needs high qualtiy O/P iron which are "expensive" to buy.
Here is what Norman H. Crowhurst said his his publshed paper in Nov 1959:
"In the case of Ultra-Linear the choice of the tube operation is vertually one btween pentode & triode. The tappings on the transformer primary "split the difference" between connecting the screen to B+ or directly to plates. The first is pentode, the second is triode. Connecting them to a tapping results in Ultra-Linear. The achieves practically the efficiency of a pentrode while maintain the linearity or lower-order distortion of a triode.
This would seem to be ideal. The difficulty is that, to work perfectly, the transformer must maintain the correct tapping, both in voltage & phase, of ALL audio frequencies. This is not too difficult for the low frequency end but, at the high frequency end, stray leakage inductances between different parts of the winding, along with the winding intercapacitances, can really play HAVOC
with an U-L circuit resulting in some quite weird waveforms at some specific frequencies."The solution as Norman suggested is to get a "correctly designed" U-L transformer to "avoid any spurious deviation from correct tapping up to a frequency beyond the audio range and ALSO beyond the cutoff of the transformer".
This not impossible, but only relatively few irons so built under the name of U-L achieve this objective.
Frankly, I modified my ST-70 with switchable UL/triode strapped, I always listen with the power tubes triode strapped since day one as I am not that impressed by its original U-L mode.
c-J
Thanks, c-J, that is just the sort of comment I've read from a number of people who have tried UL and didn't like it. I believe the ST-70 was designed by one of the founders of UL, if not actually its inventor, so one would think that his UL product should be as good as it gets. (I know the ST-70 front end was mediocre but, hey, we can't all be good at everything!)Yet if, as you say, the ST-70 sounds better in triode mode, then why did he use UL? Could it have been a matter of pride in his own topology, or lack of appreciation of good sound, or commercial pressure to provide more power than triode mode could give?
Hi.& was later popularized by David Haflter & Herbert Keroes in early 1950s.
I thought he designed guns for Wyatt Earp? - just kidding ;)Yes, I know what you're saying, that was why I tried to choose my words carefully. However, the fact remains that nothing much came of the idea until, as you say, H & K came along, found out more about it, calle it "Ultralinear" and popularized it.
The question still remains in my mind: Why did Hafler, of all people, design and build a UL amp that sounds significantly better in triode mode? (I'm not speakinmg from personal experience; I've never even seen a Dynaco amp, but you're not the only person who has reported a similar finding.)
Hi.I don't know why this shared O/P load topology was so called "Ultra-linear". Technically, it is sorta kinda a blend-up of a pentode efficieny & a triode low-distortion transfer.
If you compare a U-L transfer chart, you will see the transfer traces look like a very distored triode transfer curve, kinked in a way towards a pentode curve. The curve forms of course change per the tapping percentage on the primary of the O/P iron.
Although commonly it was 43%, but Mullard used 20% & LEAK used 50%.
Both Peter Walker of QUAD, & Williamson objected using the term
"ultra-linear" & preferred the more appropriate terminology of "distrributed load", to which I concur.Put asisde the O/P power efficiency being much better than a trioide, which IMO, is the objective of U-L design, its transfer curve clearly told us it can't outperform a triode's linearity.
So why called it "ultra linear"? I scratched my head on this terminology. IMO, it could be a marketing tactics, who knows?
c-J
Hi ,
Mullard used 20% taps for higher output power or 43% for minimum distortion . Leak designs used 43% screen taps , they also used series resistors between the screen and screen taps in an attempt to reduce distortion . Williamson never advocated ultra linear in the original 1947 article or the later 1949 addendum published in Wireless World . Later designs which were based on the Williamson input stage and driver did . Quad 2 topology is somewhat different to UL , the load is split 90% to the anode and 10% to the cathode , no screen taps at all , this supposedly provides a low output impedence with pentodes with far greater power efficiency than triodescheers
c-J, where did you find it? Please share it!
...have never been keen enough on any pentode/UL concept to try and get any.
If Allen used U-L in his CCS-ed cathode amps, the g2 current could indeed cause some issues.
If UL is SO bad why were the most popular tube amps, dynacos all UL among many other amps?
Then again a triode PP amp has lower power than an UL or pentode but is more linear(more accurate)
To realize full advantages of pentode connection, special stable power supply is needed for G2. UL comes at no extra cost, just taps in the transformer's primary.As it is common in industry, cost-saving solution was advertized as something that improves quality. Hence this meaningless term "ultralinear".
I think there is a flaw in your logic.A screen supply for a pentode can be as cheap as a couple pieces of wire and an r-c (with a small capacitor) section. If the maker was looking for the cheapest way out, then they certainly wouldn't invest in a sophisticated (read:expensive) screen supply. They would use a cheapie supply.
The taps off the OPT are not free - they don't cost a lot, but they do cost something.
So the cost difference between the two approaches can be negligible if the maker so desires.
And if U/L is just cheaper and offers no advantages, why do you think Stu Hegeman used it in the Cit II?
The real question is whether UL should be recommended for DIY, and from most of the posts here it should not. Sure, with a lot of professional engineering, talent, time, and equipment, almost everything can be made to sound good. Even transistors. But average DIY Joe is not Stu Hegeman, he needs something simple that works without much tweaking and he does not need circuit antics a la Citation II or Macintosh. Would you recommend someone to make a DIY Cit II?As to feeding g2 from B+ through series resistor, this is the worst way to do pentode connection. It only works with Ug2 close to Ua, and has poor regulation. Which of these two evils is worse, UL or series resistor g2, is not worthy of discussion.
"The real question is whether UL should be recommended for DIY,"The question I responded to was related to cost, and if that (cost) was the reason U/L output stages were used. So this is a different question.
Why not use it for DIY? With all the other gyrations DIY builders go through, why shouldn't optimizing a U/L stage be an acceptable endeavor? There are any number of complex and challenging issues that DIYers dive into routinely (power supply anyone?), why should this be any different?
"Would you recommend someone to make a DIY Cit II?"
Sure!! In a heartbeat. Why not? The key is that to do it will require OPTs of the highest quality - which aren't cheap. Of course, neither are the Teflon caps and silver wire that DIYers often use.
"As to feeding g2 from B+ through series resistor, this is the worst way to do pentode connection."
I didn't say it was any good, I said it was cheap. And the purpose of mentioning it was to demonstrate how the cost difference between U/L and pure pentode could be reduced to insignificant levels. I will also say that some damn fine sounding gear feeds the screens just that way. Could you improve it? Of course! But that doesn't mean it can't sound very good despite it being cheap.
I'm not a crusader/"true believer" in U/L or any given topology. I used the Cit II to illustrate a point. That point is that you can make a great sounding amp with a U/L stage - as you can with pentode, triode, or any other stage as long as it's well executed.
You certainly can have a personal preference or favorite, but just because you or I like something the best doesn't mean it is the best. Is chocolate better than vanilla because I like chocolate better??
But when building the a DIY Deuce, do it monoblock. Consider also lowering B+ to allow lower plate dissipation valves, or to go closer to Class A.866's for the rectifiers would be a nice touch while you're at it.
> And if U/L is just cheaper and offers no advantages, why do you think Stu Hegeman used it in the Cit II?Not all new topologies are implemented for sonic superiority. The popularity of UL had as much to do with marketing as anything else.
So you are saying what here? It can't work well? The Cit II is no good?Look at the marketing material of the day (you can see a 1961 brochure on my site if you want). There was no mention of UL or "distributed load" except in the spec sheet where it describes the output section in just a few words.
I'm confident the choice was made for performance reasons. There is virtually nothing about a Cit II that followed the trends of the day, there's no reason to believe the UL output was an exception to that.
> I'm confident the choice was made for performance reasons.Which of HK's product specifications improved as a result of applying the UL technique? If the answer is "none," and if you're correct that this wasn't market-driven, then perhaps it was just an excercise in innovation. Engineers do that sometimes.
"Which of HK's product specifications improved as a result of applying the UL technique?"Improved as compared to what? If there was EVER a "clean sheet of paper" design the II was one of them. And H-K wasn't U/L across the board. The Cit V was a true pentode amp (and it's marvelous!).
"If the answer is "none," and if you're correct that this wasn't market-driven, then perhaps it was just an excercise in innovation. Engineers do that sometimes."
Sure, that happens in a lot of fields.
But in addition to the published material, I've had the honor of speaking with people who were personally well acquainted with Stu before he passed away. They tell me to a man that he was absolutely driven to make the finest amplifier in the world. He could have chosen triode, pentode, U/L, whatever. He had free reign. He chose U/L. And really, by 1959-1960 U/L was hardly an innovation, it had been around for 5-6 years.
There was only one design and one construction compromise in the Cit II that Stu made anyone aware of - and it wasn't the ouput stage.
1. Construction - It had to be buildable by an average Joe - the huge majority of those amps were kits.
2. Design - Stu was not able to resolve a 1 Hz oscillation in the Cit II the way he wanted. He ended up using a different approach to solving the problem than he wanted because he ran out of time to work on it. The amp was scheduled to begin production, and he simply had to use a less elegant solution than he preferred.
With the benefit of way more time and much better parts than he had, I've been able to eliminate the 1Hz oscillation without applying Stu's fix. I must tell you, his fix may not have been as elegant as he wanted, but it is totally inaudible - and fixes the problem.
In light of all this, do you still believe he used U/L just as a marketing gimmick, or just to prove he could??
> In light of all this, do you still believe he used U/L just as a marketing gimmick, or just to prove he could??Please don't misquote me for your own purpose. You said earlier that you are confident the choice was made for performance reasons. To that, I asked which of HK's product specifications improved as a result of applying the UL technique? I was of course referring to HK's previous/parallel, non-UL power amplifier products and circuitry. If you have no definitive answer to this question, then I think it might be difficult to demonstrate that the basis for designing a UL output stage was performance-motivated.
One last issue here. You wrote in two earlier posts:"The popularity of UL had as much to do with marketing as anything else."
And later:
"If the answer is "none," and if you're correct that this wasn't market-driven, then perhaps it was just an excercise in innovation."
In my reply I wrote that "In light of all this, do you still believe he used U/L just as a marketing gimmick, or just to prove he could??"
That was not a quote, I was paraphrasing what you said. In light of your actual quotes above, do you think I paraphrased inaccurately or inappropriately? I apologize if I did, but I really don't think that's the case. Other viewers of this post can decide for themselves I guess.
You weren't misquoted - your actual quotes are in my post so there is no possible way to misquote you. No way."You said earlier that you are confident the choice was made for performance reasons."
Yes, and that was based on a number of different things as I said.
"To that, I asked which of HK's product specifications improved as a result of applying the UL technique? I was of course referring to HK's previous/parallel, non-UL power amplifier products and circuitry."
There was VERY little in the Cit II that was common to any earlier H-K stuff. No other H-K product up to that time had used that tube set, those OPTs, the multiple loop NFB, the low impedance power supply, and so on. What is it about "clean sheet of paper" that you don't understand? Do you really want to compare the Cit II to an A-300?? It's ludicrous.
If the change to U/L was the only change (or the only significant change) then your narrow-focused question could be answered. But there really is no H-K product to draw a direct comparison to!!
It doesn't mean anything, but since you seem intent on having some sort of answer (relevant or not), here is a quick list.
Compared to prior H-K designs, the Cit II...
1. Had octaves wider bandwidth
1a. Had an OPT resonant frequency in the 450K area! This made a lot of other things possible in the design.
2. Had much higher RMS and peak power
3. Had lower THD at a given power output
4. Had lower hum and noise
5. Had a higher damping factor
6. Used much higher total NFB. Used lower loop NFB.
Properly implemented UL will show better open loop performance than pentode as regards bass distortion and load tolerance, with a relatively small penalty in efficiency. Compared with triode, it will have less Millering (hence better OL bandwidth and a better spreading of the poles making feedback more stable) and markedly better efficiency. Some of these things will show up on the spec sheet (power), some won't (LF distortion, load tolerance) but will be audible. Your Procrustean restriction to the spec sheet alone is curious and telling- do you really think that all design virtues are contained therein?The two key words are "properly implemented." Slap a cheap piece of shit transformer in and you'll have grief; the best topology for shitty transformers is probably Circlotron.
disclaimer: My own design choices do not include ultralinear. But in the past I've built UL amps with excellent performance, so it's not beyond the capabilities of a DIYer.
> Properly implemented UL will show better open loop performance than pentode as regards bass distortion and load tolerance
Ahh, semantics. Very good.
they hapeened to have a U/L tap? ;-)
__________________________________________________
Boo!
Hee! Don't think so...
UL makes use of NFB. We knew that. Ok, conventional FB theory takes a sample of the output and feeds it back to the input. It's out of phase, of course, and the out-of-phasedness subtracts out whatever was right in the fedback signal leaving an error signal to perform corrections. (If we could get just the error signal, then there would be no reduction in gain with NFB.)Here's the thought: the input used in UL is not the same input used for the music signal. More to the point, the screen as an input does not present the same nonlinearities (distortion) as the control grid. The result is that the error signal is not the right one for the input being used. According to FB theory, we ought to be able to reduce distortion to an arbitrarily low level provided we have enough open loop gain (and ignoring the realities of phase shift, etc...) But it seems to me that that is not the case with UL given the fact that the FB is applied to an input with different distortion characteristics than that used to apply the music.
To look at it another way, applying a distorted signal to the input of an amplifier can't be good. We can (in theory) get away with it with ordinary FB loops because the distortion is (in theory) exactly opposite that produced by the amplifier itself. We don't have that feature with UL.
Dave,I don't know if your thinking holds up. An error correction signal is applied to g2. IF it is appropriate, things will work well.
In any case, loop NFB (hopefully only a few dB.) is needed to further lower O/P impedance and correct for "iron" flaws.
See my post down the page replying to sser2. The need to keep instantaneous plate potential above g2 potential never goes away.
Eli D.
"The need to keep instantaneous plate potential above g2 potential never goes away."That is an impossible need to meet, in either UL or pentode mode. At full signal, the plate voltage will swing from close to 0v up to twice B+. How can the screen voltage always be lower than that?
Obviously, g2 potential must remain positive. However, when the "idle" state potential difference between g2 and plate is significant, multi-grid power O/P tubes are most linear. Also, quite a bit of the distortion generated is of the HIGHLY objectionable IM variety.
Eli D.
Hi.Here what I have done to my ST-70 triode strapped EL-34.
Instead of the common practice of 100R bridging the plate & g2 I used before, I have them 32V apart by replacing it with a string of 5x6.2V zener, a SS diode & a 10R resistors.
"Most linear" or "highly objectional IM variety" generated or not, I am happy I have done so as it sounds better to me now.
I don't think its primary effect is applying NFB. It is more like modifying the pentode characteristics a bit. It makes the pentode a lot less senditive to the applied load and g2 voltage.If one takes the U-L as applying NFB to a pentode, the arguement that a triode is a pentode with NFB becomes a whole lot more believable. The triode being the case where the g2 is attached to the plate or 100% U-L taps.
The distaste for U-L can probably be traced to the amps that first used it. Their designs left a bit to be desired from the POV of an inspired builder.
"If one takes the U-L as applying NFB to a pentode, the arguement that a triode is a pentode with NFB becomes a whole lot more believable."Good point!
What's going on with the usual NFB idea is the output is fed back to *THE* input. In the pentode, the output is fed back to *AN* input. In the triode/pentode case the pentode( taken as a beam tube with beam plates attached to the cathode for simplification ) degenerates from a 4 element device to a 3 element device when the g2 reaches the anode.
Not sure I fully understand your point of "the" vs "an". In a typical global feedback scheme, the signal is returned to the input stage but not the actual input "port" e.g. The input signal is applied to g1 of input amplifier but FB signal is applied to the cathode of same amplifier. Really not conceptually unlike UL is it? The only common topology that comes immediately to my mind where "THE" output is fed back to "THE" input would be a "plate follower".
The cathode or grid input has the same effect. The same resistor that is being used in the cathode circuit to develop the voltage could be moved to the grid circuit. I was considerng it to be the same...perhaps a bit of a simplification on my part.
scratch some of that. writing w/o morning stimulant dose.but still, moving the cathode is just the same as moving the grid. In the circuit it is a whole lot easier to move the cathode( to effect grid-cathode ) than to try changing the grid-ground relationship.
Changing the voltage at the cathode not only changes Vgk, but also Vak. It might not mean much, especially in the case of a tube with high plate resistance, but something with very low ra and mu? I dunno, the cathode (even ignoring input impedance) is not *exactly* the same input as the grid and won't have *exactly* the same distortion characteristics. The difference might be genuinely negligible compared with all the other realities involved, but...
You have it right. I do think it's a small thing. As implemented with a pentode, Va-k change is trivial. When it's done with a small resistor underneath( between ground and the cathode bias and bypass ), I think I can ignore the cathode's input Z. For a triode at the front of the amp, the FB signal might amount to a few % of Va-k. I agree with your assesment that this is also likely a trivial effect. It is a good thing to realize when you're 'neglecting air resistance', yes?
"The distaste for U-L can probably be traced to the amps that first used it. Their designs left a bit to be desired from the POV of an inspired builder."That may be what I'm trying to find out. What kind of defects were in those deigns?
Let's see, marginal power supply, high impedance driver's, full loop NFB, AB1 bias for maximum power, OPT load for maximum power with distortion corrected as mentioned earlier. As a general example, look at the Dynaco St.70 modification list. To be more general, consumerist style electronics of the same sort available these days.Build an amp of any style with the care and attention that Josh at E-Luv applies and I think you'll be able to get away with a whole bunch of Forbidden methods.
Or realize that the wrong methods have come to defined 'conventional wisdom'?Just which EL amps have you heard?
actually I've only smelled one. Nice smoke though.I'm not sure how I'd define conventional wisdom. I was probably thinking of something more like conventional dogma when I wrote it.
.
__________________________________________________
Boo!
1. It abuses screens by exceeding their voltage and power ratings.2. It uses pentodes inappropriately. For proper operation of pentode, screen voltage should remain constant. In UL, both plate and screen voltages vary, causing non-linearity.
3. UL achieves lower Zout and distortion via negative feedback from plate to screen; the NFB loop involves OPT. Exactly same effects are achieved by applying global NFB from OPT secondary in most true pentode amplifiers.
If there is any advantage of UL over pentode, I would like to hear arguments. So far I see only disadvantages.
is the primary advantage. Of course, UL is not as linear as triodes are and you don't get as much power as straight pentode, But given that its not perfect it does sound better. With the increased linearity, there is less need for as much global feedback, in fact you might be able to get away without it- something you cannot consider with a pentode amp.Properly designed there is not threat to the life of the tube.
I wonder if that's its main flaw - being neither one thing nor the other? The people who complain about UL may be those who are keen on either triode or pentode; either way, they might find UL not satisfying enough.
My only question is why would U/L flaws be any different than the flaws of triode or pentode operation? All modes have flaws, so it's a matter of selecting the mode that works best for you.Ray, I think your second sentence is spot-on. I think the basis of most of the arguments here are "my favorite output stage is better than your favorite output stage". There's nothing wrong with that - albeit I think an open mind is a good thing to have.
For the record, my "McShane" amp uses a triode wired output stage. One of my very favorite amps out there is the little Citation V - it's a pentode design. And the killer Citation II is U/L, and I love it too. They all sound great, regardless of the configuration of the outputs.
The Citation II is way more amp than the Citation V (IMO); arguably the best of the vintage amplifiers, although the Citation V output transformer is no slouch!IMO (again) the UL approach, if you have to use pentodes, is a pretty good way to go. I like to cathode cross-couple them- then the distortion gets low enough you can run the driver with zero feedback.
However if you have triodes then the game is over- nothing beats them.
Hello Ralph!It's an honor to "meet" you!
I just wanted to mention that the little Cit V realy responds well to some TLC. Once the power supply (especially the screen supply) gets upgraded it's an amazing amp. I have had a couple here I redid for people I really didn't want to let go of. They were just astoundingly good.
Of course, since they use a pentode ouput stage, the improved screen supply is a big help.
UL output power is significantly less than the same tube in pentode mode utilizing "loop" NFB.
"1. It abuses screens by exceeding their voltage and power ratings."With EL34s in UL, Mullard recommended B+ that would not exceed the screen voltages, plus a 1k dropper/stopper resistor between eaxch screen and its UL tap.
"2. It uses pentodes inappropriately. For proper operation of pentode, screen voltage should remain constant. In UL, both plate and screen voltages vary, causing non-linearity."
I understood that the point of UL was that the screen waveform would follow the plate waveform, but with a lesser amplitude goverbed by the tap %.
"3. UL achieves lower Zout and distortion via negative feedback from plate to screen; the NFB loop involves OPT. Exactly same effects are achieved by applying global NFB from OPT secondary in most true pentode amplifiers."
UL designs that I've seen also use global NFB, although the amount needed is supposedly less than with pentode mode, because UL is claimed to have lower distortion and lower OP impedance by virtue of its local NFB.
All forms of NFB work best when things are reasonably linear to begin with. It's a matter of smaller error correction signals.To get MAX. performance out of UL topology, an O/P trafo with a separate screen grid winding is necessary, along with a separate, regulated, B+' supply. The objections you gave are overcome. Surprise (NOT), the expense factor surfaces, again.
UL topology is a form of local NFB. Frequently, local NFB is less problematic than loop NFB.
Eli D.
Ray, to me it sounds sloppy. Now, I haven't fooled with it since '90 or so, but at the time, I actually preferred pentode, which sounded a bit transistory, but cleaner.Maybe if I spearminted around with it now, I might find something.
So might you!
Aloha,
Hi Poinz,Maybe the sound of UL is disliked by people who like triode and/or pentode - maybe it's too much of a compromise between the two, so that it lacks the finer points of either?
Pentode has fine detail, but lacks a little juice. Triode is the daddy; it has as much or more rez, and all the natural juices of the source (or mostly), but of course only one third the power. UL, it has the power of pentode, but not the detail, and maybe a little juicier, but there's just too much data missing.Aloha,
You might want to check the following thread for a long discussion about possible hitherto unknown costs of UL operation.
This post is made possible by the generous support of people like you and our sponsors: