• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Why use a pentode?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
37Kdc3r.jpg


The question was approximately why pentodes ?

Please forgive anything unworkable in this diagram . I sketched it for fun to indicate the strengths of pentodes and pentode equivalents .

The idea is to use something like the 211 triode that is a pentode . Many will use an inter-stage transformer . These are seen as preferable to other ways of doing it if enough money is spent . An opportunity exists to make a considerable virtue out of that . UL or distributed load ( curve B ) .

The exact UL point for any pentode valve is not always easy to find out .

The 803 I choose for fun . No idea if it is a good choice ? It might need very special driving to work .

If anyone has had a bad experiences of UL try it without loop feedback .

The amp sketched I suspect will offend most lovers of valves ? It shouldn't . I suspect there is a very great danger that it might work !
 
MerlinB said:
But RCA made any number of small signal pentodes, so didn't they have to pay the rights after all? Why didn't they come up with teeny little small-signal beam tetrodes?
If you look inside some small-signal 'pentodes' they are actually small beam tetrodes. I remember first noticing this in the 6BZ6.

Maybe Philips would only license the pentode to others for small-signal use? I don't know.
 
8iMmU95.jpg


I have seen lots of words written about the RH series of amps , but no graphs . Strange people should fight out a long battle and not challenge the implied performance figures . I think the originator suggested 1.84 V sensitivity and 12 watt plus for the RH88 . Here are my tests . I abandoned it as a result . The circuit as posted previously and a 2K4 load at 450 V . If I misinterpreted the intentions I apologies in advance .

My amp has neither Schade feedback nor triode input . One of the other writers here pointed the way on it ( thanks ) .

The 2 watt test is the signal generator directly into the KT88 . There is much to commend in the design . I feel the ultimate distortion is too high and the sensitivity too low . The figures I show are just single harmonics , THD will be higher .

As you can see the resultant distortion is a composite of two stages . They interact and cancel . Transformer coupling allows some manipulation of that .
 
807 and LS50 true .

As I understand it EL 34 was a product of hurt pride . The other people had stolen the publicity war . EL 34 was to be almost as good for less money . It is my favorite by a long way . As far as I know all beam tetrodes are kink-less . KT is just a name . Perhaps MO had already registered the name before selling to RCA ? New stock EL34 are as little as $20 a piece even in the UK . I have some old ones marked Marshall that are very good . I wonder if they are Mullard ? I used to sell new production MO KT 66 and 88 in 1974 . I bought them from Lugtons I think ? At $12 a piece ( 88's ) I was embarrassed to say how much when people bought them . The book my friend had was new for 1972 promoting designs using KT88 .

Having read Mr Schade's paper he certainly understood the device . The curves to my eyes look like pentodes .

The TT21 is a nice device . The story I heard is some young technicians had some KT 88 modified to resemble 807 . It fitted maximum allowed transmitter power for certain licenses . They offered the production ladies some dance tickets to say thank you . They showed them to their boss on Monday morning who instantly put them into production .

I think even to this day beam tetrodes seem to suggest superiority . One has to argue the case for pentodes strongly as it always seems to be implied that it is the poor relation .

Beam Power Tetrode

Shunt feedback as a concept has been known a long time before Mr. Schade came along (actually, Dr. Schade), but he was the one thet shows it makes a pentode type device look like a triode.

Re 6L6 and 807, this is essentially the same tube packed into different envelopes and sockets for different requirements. In particular, with a high-voltage anode you are risking inter-electrode arcing if you route it to a standard socket, hence the top cap - far away from the other pins. TT21 ihas a top cap for this reason.

LS50 -> GU50 is an entirely different thing, a 'proper' pentode, but with some tricks from a BPT that make it perform even better. With BPTs aligned G1 and G2 are a necessity to produce the virtual G3. Don't confuse G3 with the beam plates. Because the grids in a BPT were wound around two support rods just like in regular tubes (later on some got frame grids), the effective geometry of the grid changes a lot near the rods and in fact the rods themselves make for a large discontinuity in the shape of the electric fields. The beam plates keep a large majority of electrons emitted off the cathode on a path to the plate where they pass through where the grids have proper relative distances.
However in order for that to happen, G1 and G2 have to be wound with the same pitch and aligned so that G2 wires 'hide behind' G1. This has a very welcome side effect of keeping G2 current low. When expressed as a fraction of total cathode current, BPTs often offer several times lower G2 currents compared to straight pentodes, which potentially makes them more efficient in 'pentode mode'.
The LS50 in particular is one of the rare straight pentodes that has aligned grids (G1, G2, G3) and amongst other reasons this was done for efficiency, since it was used in portable equipment. Apparently the GU50 was initially made on original German equipment but it was refined somewhat (different socket, grid and anode doping for higher temperature operation). To put thing simply, the GU50 is unbeatable when it comes to performance/price and it's also EXTREMELY difficult to kill. THe LS50 also had civilian versions but these were not very popular due to a difficult and expensive socket, added to a rather expansive to make tube.

Regarding the RH series amps - the original ECC81 driver is not the best choice but I suspect it was used partially because it's easily available and also because it's relative non-linearity can be used to partially cancel out the nonlinearity of the output stage. Regarding the possibility of getting lower Rp than in a straight triode G2 to anode strap - IIRC Schade hinted at this but there is a caveat. With G2 strapped to anode, the feedback gives you atriode with basically muG1G2, whereas you can easily get other combinations from shunt feedback - BUT then your driver has to do a whole lot more (and in fact even impossibly more in real world situations). G2 also sits in a rather different place compared to G1, and has it's own mu, gm, rp parameters which are not linear and may not track too well with G1 parameters. So, YMMV.
That being said, ECC81, even wit an unbypassed cathode resistor is not the best input stage WRT gain but has it's merit depending on the output tube used. It's resulting Rp is quite low so the feedback signal is attenuated and - this is the important bit - because it itself sees a low load resistance, the distortion is high - but, in this case it's needed to cancel ot the output stage distortion. So, while a pentode is a better choice WRT high Rp and gain, making it very linear might not be the best strategy. The downside of using a less than linear device or operating point is the same as for the ECC81 - tube variability between instances of the same type. Even with the RH series, plug in a different ECC81 and you get a different situation - exact matching was not one of the design requirements of this tube.
 
A caveat of harmonic cancellation is that you may end up with lower total harmonic content at the expense of listen-ability.

In my experience the rate at which each harmonic is cancelled is very bias point dependent.

For example, I have seen lower second and fourth harmonics at the expense of greater fifth harmonic and marginal lower third harmonic. Change bias by 5% and everything changes.
 
Yes, harmonic cancellation is fine if you love tube rolling and are happy to indulge in it frequently as the valves wear out and change their harmonic pattern. Bad engineering, though.

One exception is LTP, which relies on it - but there the two valves carry the same signal (approximately) and are usually in the same envelope. Another exception is a genuine balanced SRPP, but very few people seem to build these. Also push-pull, of course. All are balanced stages, carrying the same signal in each half.
 
Last edited:
Yes, harmonic cancellation is fine if you love tube rolling and are happy to indulge in it frequently as the valves wear out and change their harmonic pattern. Bad engineering, though.

One exception is LTP, which relies on it - but there the two valves carry the same signal (approximately) and are usually in the same envelope. Another exception is a genuine balanced SRPP, but very few people seem to build these. Also push-pull, of course. All are balanced stages, carrying the same signal in each half.

Agreed, that's why I specifically mention it in conjunction with the RH designs. One would need to tweak the operating points for every tube and also as they age. Add to that that ECC81/12AT7 was never intended to be tightly specified, and is nonlinear to begin with. Also good point re lower distortion but possibly a not so nice harmonic profile of the distortion. Some wrk is required already just to get two same type or dual tubes to do this right, much more difficult finding a pair of completely disparate ones that fit well, tolerances notwithstanding. One may end with good cancellation only to the first approximation, which is the second harmonic, resulting in an unfavorable distribution of the other harmonics.
 
A caveat of harmonic cancellation is that you may end up with lower total harmonic content at the expense of listen-ability.

In my experience the rate at which each harmonic is cancelled is very bias point dependent.

For example, I have seen lower second and fourth harmonics at the expense of greater fifth harmonic and marginal lower third harmonic. Change bias by 5% and everything changes.


That is why an analyzer is important , I 100 % agree . 90% of things I tried were bad at best . They had to be tried as I wanted to honestly say I had . I don't use a simulator , perhaps I waste more time on this than most ? 70% of the time things work out as one might think . 20 % has an easy answer . 10% has a slightly unexpected outcome .

Joy of joy in my design I can repeat results with many different makes of valve and some variation of type . EL 34's of different makes do sound different yet measure about the same . A slight mystery .

Some have been a bit pedantic I feel about what is a pentode . Sure there will be areas where the Beam Tetrode and Power Pentode part company . To me they all offer pentode possibilities . FET's , Triodes , Transistors don't offer these possibilities .Sometimes gaining that advantage is not seen by others as an advantage . I put up a pentode and triode curve . No one commented on the fact that neither is a great curve . Happens to be that triode is the least bad easy choice . The UL curve is a nice curve . It is in the " pentode "devices to have that . If an inter-stage transformer can be used the UL pentode is an exciting choice . It even allows phase change if that would be useful ( I could list a few uses ) .

I am told that 6L6 and 807 were a little different in reality . There were many 6L6 types ( I think 6L6GC was nearest to KT 66 ) . Some say 5881 = 6L6GC , others say nonsense . The renumbering of devices in a shrinking market probably is to blame ?

Anyone who is contemplating a " zero feedback " amplifier ( loop ) do not dismiss UL . If you learnt triode sounds better than UL , better still with no loop you might say you know something you don't . Just give UL a try with no loop .

Not recognizing pentodes would be like ignoring athletes from Kenya . There is something remarkable to know about pentodes even if everyday life is not asking for that these days .
 
Last edited:
I am told that 6L6 and 807 were a little different in reality . There were many 6L6 types ( I think 6L6GC was nearest to KT 66 ) . Some say 5881 = 6L6GC , others say nonsense . The renumbering of devices in a shrinking market probably is to blame ?

Actually, the numbers are different for NOS US and Western Europe tubes, and for Russian equivalents. Mr. Wavebourn may weigh in with better info on the Russian types.

The US-made types:

6L6 was the original metal type. 19W max plate dissipation, 360 max plate volt rating, iirc.

6L6G was the version with a large ST shaped glass bulb ("Coke bottle" shape). I think the 6L6GA had a smaller ST shaped glass bulb.

6L6GB was the GT shaped glass bulb version (straight sided bottle). The 5881 was the industrial version of these, also labeled 6L6WGB. (The "W" = "ruggedized")

6L6GC was the last version of glass bottled 6L6 type. The GC version has higher max plate voltage and max plate dissipation of 30 watts. This version looks very similar to type 7027A.

The Russian types:

6P3S is a GT bottle type (straight sided glass), equivalent to 6L6GB. About 20 watts max plate dissipation, about 400V max plate volts. That's the one usually sold as a 6L6GC, but it's not quite up to those specs. I think that's why you see so many of these 'red plate' in Fender Twin Reverb guitar amps with 450V or higher on their plates.

63PS-E is the heavy glass bottled, wafer bottom type that Sovtek was selling as 5881 (also called 6L6WGC, which is a type that was never made by the US companies). But the 6P3S-E is a quite different tube from the old Tung Sol 5881. 63PS-E seems to perform fine at over 450V on the plates, and 30 watts plate dissipation. I've used 6P3S-E to replace EL34 in a Dynaco Stereo 70. 63PS-E is not equivalent at all to EL34, but it does bias up nearly the same as an EL34 in that circuit. Put a real 6L6GC (or a 63PS with no "E") in that circuit, and that tube will bias up with a much higher quiescent current draw. Obviously, if you use 63PS-E (Sovtek 5881) as a drop in replacement for a NOS 6L6GC, it will make your 6L6 amp perform very differently.


Anyone who is contemplating a " zero feedback " amplifier ( loop ) do not dismiss UL . If you learnt triode sounds better than UL , better still with no loop you might say you know something you don't . Just give UL a try with no loop .

I have. Both in a Dyna ST35 and in a Dyna ST70, and in a new circuit I built in the ST70. The "sound" of the UL output stage is sort of intermediate between triode and pentode. The curves I've seen for KT88 in triode and in UL seem to bear this out. I couldn't live with the sound of a UL stage with no gNFB, unless there was some other kind of NFB employed to decrease distortion. It could be that the A470 OPT's in the ST70 are designed to be used with gNFB, so it's actually necessary in any amp using those OPT's. But I can tell you that the amp with triode-wired EL34's sounded cleaner than the same amp wired UL.

Maybe some other folks have tried this too.

A friend made an amp with EL34's into a pair of Hashimoto OPT's (very nice), with a triode/UL switch. Once -6dB of gNFB was applied, we could not hear the difference between triode and UL, although the UL position delivered about 10 more watts per channel.


-=|=-
 
I am told that 6L6 and 807 were a little different in reality .

I was given over 100,000 tubes about ten years ago. They were all loose tubes that had been dumped into about 30 eight cubic foot boxes and then tossed around with fork lift trucks until they were headed to scrap when my friend bought the entire warehouse. There were thousands of broken tubes, most of which I took apart to see how they were made.

The 6L6 / 807/ 6BG6 types have been around for a long time. Some variation has been in continuous production since 1939. Over the years a lot of different things were stuffed into the glass with these numbers on them.

The original 807 and 1625 contained the exact same plate structure as the 6L6GB with some additional shielding around the bottom mica insulator. The 6BG6G had exactly the same structure as the 6L6GB without the extra shielding. After looking at enough tubes, including several hundred rush produced for WWII I have seen just about every permutation of 6L6 type guts inside an "807". So there can be differences, but it is possible to find 6L6GC guts in an 807 and 7027 guts in a 6GB6GA. I have working examples of both.

Just give UL a try with no loop .

My experiences are all with a Tubelab SSE. I like the sound of a triode EL34 for some music, in fact most music that does not have extreme dynamics or where 5 watts is enough. I have tried a UL EL34 with and without cathode feedback, and it lacks something. I can't say just what is missing, but it is louder, but the SE magic is missing. A UL KT88 with zero feedback is better, but a UL KT88 with CFB just rocks the house. Feed these 15 watts to some 15 inch OB coaxial speakers with 96 db sensitivity and the bass shakes the house across the street. Want to play Pink Floyd or Depeche Mode through an SE amp? Use a low impedance tube like a KT88 in UL with CFB through a big OPT, GNFB is not needed.
 
Thanks for valve info .

UL without a loop wouldn't be modifying existing UL amplifiers . Also UL in every stage is a suggestion . I see a transformer as the purists route if so . Morph controls seem too complex . I have to stress that many loudspeakers demand a triode solution . On paper even that won't work . Works quite well in practice . Loudness control ? Perahps . A 1950's Wireless World ( 55 ? ) I read suggested a damping factor of 3 as a minimum . This was by experiment using a lightweight foil contact on the cone . The amplifier had variable damping factor . The cone of the speaker would have a stiffer suspension than now is my guess ?

Dynaco is an excellent example of the seemingly impossible made easy . The output transformers look too small . The rectifier valve is run hard . The input pentode does all the work in the amp . Only UL and loop feedback make it possible . The fact it works so well is testament to the pentode and Hafler . The inspiration for it comes from the maker of the 7199 if I am right ? The same generic idea in other amps I have seen . What the Dynaco proves is that mathematically correct solutions work exactly as predicted if the machine is a good example . Having built transistor amps all my life I am astonished any loop feedback valve amp will work , to work well is a miracle . My total respect for those who got it right .

I found by accident that UL of about 80% is nice with an EL 34 ( almost triode ) . My transformer had a spare tap ( SE ) . It was a bit like the high jump . It gave me enough to rattle the pole I had set as my target . I used KT 88 also . I have a mild preference for EL 34 . I use conventional cathode bias with a resistor and big electrolytic . I read somewhere that this bias method is to an extent auto correcting . My observation is it must be . Some of my valves are so old as to have no markings . My brother put them in ice cream tubs , otherwise I wouldn't have been able to tell . I run one valve at about 25% full power . There are no obvious differences except gain . Output valves if KT 88 varied between 39 and 45 mA . The spectra remained about the same . The EL 34 samples were closer , all about 65 mA . I would suggest the EL34 is a better understood device where cloning is concerned . The amp has no plate to plate nor loop feedback . All it has is the interplay between devices .

A question someone might answer . The friend who I made the amplifier for said that when valves die they go Super Nova . He said some designers even solder the cathode resistors in such a way so as to unsolder themselves . My cathode resistor is either 560 or 470 R ( 7 W ) . I propose to put a simple heat fuse next to it . Without going the transistor route any better solution ? I have proposed a meter with a red zone as I feel this doesn't happen overnight . I have seen this in the past and assumed it was grid failure .

Fixed bias is strange . I won't give up with it . I used a mixture of fixed and cathode bias to have an overdrive facility . It allows the valves to idle at about 70% the usual . It takes them to about 150 % if clipping . No positive or negative outcome . It did do one thing . It allowed an EL 37 to keep up with a KT 88 . As a transistor class B addict I know music is about 15% of the full sine-wave power . Valve amps are class A ( add numbers if you like ) . The rules can be bent a little . I would expect the EL 37 to have a long life in this arrangement if real music . The circuit I used is very much simpler than the TubeCad version and uses no semiconductors . Time constants are it's only problem .
 
It's About Time & Ultra-Linear Line Stages

The pentode has the special possibilities for Ultra Linear feedback . The cascode could be called a fake pentode . Sometimes a better pentode . One problem I had was getting the high gain triodes to make usable cascodes . For example an ECC82 makes a good casode . Problem for me was it wasn't the step forward from a single ECC 81 that I had hoped for . ECC81 wasn't a good cascode , although I accept I know very little about them and probably got it wrong . I did make a very good transistor + ECC81 cascode .

It is rare to see UL shown as if it is a natural curve of a pentode . Here below it is shown . In a very early text about UL the curves where shown in triode and pentode with the comment that they seem to be opposites . A rather wonderful curve of UL was shown as a perfect set of curves . The reality is a little different . As the author said the perfect UL point would vary between valve types . Beam Tetrodes are to all intents and purposes pentodes in this application . I must get some high voltage MOS FET's and connect them in UL cascode , I have a hunch it might work . I have ISOFET's , they might bolt to the chassis and survive for a while . EL 34's cost about $20 ( Ruby ) , FET's are not worth the bother . Just to know mostly .

A question . Did anyone ever try a triode with FET source follower as an output stage ? The idea being to use a standard SE transformer from source to ground . It should have mostly triode spectrum and a nice low impedance . I said to a friend it should be better than a 300 B by a long way ( as linear and better bass , less hum ) . 2A3 would be a nice driving triode . The amplifier I just built has a stereo PSU . If I want 1000 V it is not a problem , 500 CT 500 might suit what I describe .

You might think I eat sleep and drink valves ? Not so . I found some unfinished chassis of my late brother a few months ago . Where I am now and being a novice is a few months apart . My brother was rather conservative . What he built was not likely to please him . He was by IQ a genius . I wish I had bought him a spectrum analyzer . He would have rejected what he had built . I saw many attempts to make it work in the circuit layout . Sad thing is I gave most of the spare parts to my engineer friend John . He said with glee he will build a Mullard circuit . My brother and John , what can you do ?

Did anyone use a transistor cascode in UL ?

http://www.audiomatica.com/tubes/el34.htm
 
The cascode can substitute for a pentode when reduction of feedback capacitance and high mu is the aim. It is not sufficiently similar to properly work as UL; the upper g1 is only a very rough approximation to a pentode g2. Trying to run a cascode as 'UL' is almost bound to fail whichever valve you use. The cascode makes a poor output stage, anyway, as too much voltage is lost.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.