My modelling of this approach using a triode driver showed a similar behaviour with 2nd and 3rd harmonics - but there was significant tail of higher order harmonics which carried out way beyond the 10th at about equal levels.
Using a pentode driver in a similar position produced almost identical 2nd and 3rd harmonics (depending on the choice of pentode - 6AU6 increased the 2nd) but there was no corresponding tail of higher order harmonics.
This was in an SE though so a PP might behave differently. It seems to me that pentodes can give a more benign harmonic signature.
Shoog
Hi you might try a phase splitter input transformer like a Lundahl with the 2 out of phase signal driving 2 RH type amps into a push pull output transformer. The amp would be class A operation. I am considering building an amp like this using a pair of EICO HF87 output transformers. But have not gotten around to ordering the input transformers.
Why do you find it strange? Perhaps I've got some design issues, or am I missing something?
Russian Audio tubes can be some of the best available, especially the Winged C Sventlana brand. It is also my experience that quality control is usually very good. Here's the clincher though - ordered directly from Eastern Europe they usually cost a quarter of any western european or American made brands. Since valve production carried on for much longer in Russia there are still vast quantities of most valves unlike the dwindling supplies of western NOS valves.it seems you prefer Russian audio tubes .
I ask you why you are scratching your head in light of this information.
PS - Waveborne comes from Russia so it would be surprising if he didn't use Russian valves.
Shoog
Last edited:
I prefer great tubes that are not greatly overpriced.
I could use FL152 or LS-50 instead of GU-50, but they cost much more, and are rare.
new production are chipper.
I cannot do anything else than to agree with what Shoog said over here.
Most of the soviet era tubes are high quality stuff, with great price/performance ratio. There are still plenty available, at accessible prices.
From my personal experience, I have been very much surprised by 6N1P-VI, 6N2P-EV, 6N6P-I, 6N23P, 6P36S, 6P42S, 6P44S.
I built a anode shunt and joined cathode amp. It needed careful measurement to get it right. As soon as loop feedback was added I felt it unworkable. UL without loop can be very good and seems not to ring. I have used 82% triode and was pleased it so like triode with useful extra gain. Loop feedback is better if Pye Mozart style. If analysed it looks as if no feedback is being used. I used 16 dB. The output winding 16 R output was used rather than a new winding. I suspect it would allow coupled cathodes, I didn't try. The RCA shown here is a repeat of 1938 designs hybrided with the use of 7199 moded TV tube. The Dynaco could be moded this way. Most do not realise the Dynaco is the RCA idea they used to sell 7199. Hafler just got the ideal price verses quality correct with his own UL used. Perhaps he gave RCA the nod.
Maybe he meant "cheaper"
But i'm not aware of current production of GU50...
I very much agree with you.
Also the UL can be made differently from usual. Instead of connecting the screen to the transformer this is directly supplied like in normal pentode operation. So more freedom for operative conditions and choice in devices. Then the optimal plate load for normal (conventional) operation will be split between plate and cathode in unequal parts. The screen becomes the reference (cathode-screen and plate-screen voltage swings). Crowhurst called this "modified ultralinear". It is possibly the best output stage arrangement to drive difficult loads together with the UL cathode follower where the speakers impedance drops notably in module and is also reactive. Standard UL and triode come second in this department. However the UL follower is in most cases not practical because of the large driving voltage. Instead the modified UL power stage is actually less demanding than the McIntosh unity coupling (i.e. voltage gain >2 ) provided that the plate has more turns. There is no fixed %. The only price one has to pay is, like the unity coupling, the transformer quality. This has to have best possible coupling between the plate and cathode windings.
One thing I did was make the output stage SE in PP out, that is g1 of EL34 No 2 is grounded and the pair are a long tail pair. This requied 56 V rms drive. Just what I wanted as it was to be a 211 driver for another day. That allowed PYE Mozart style feedback into the driver pentode. In the end that one ended up triode in to get a nice harmonic balance ( I never did the feedback ). I had UL option which if zero feedback can work. My speakers have no real need for damping so happy to go this way. I got 12 W triode so was happy. I would guess 20 watts UL and 40 pentode. 450V HT. The transformer a Danbury sold by Maplin. The pentode looked workable. With some Kitic feedback I think it could work. The output looked SE in style. My better choice. If not a transistor amp is fine. 1% THD at 12 W no feedback I seem to remember. I would love to post it if the file can be found. After a computer rebuild half seem missing although do turn up in weird places. I tried a 20 % mismatch of valves and had an even better results at 10 watts. The transformer didn't mind. I suspect it's BH curve actually better ( before I get told off this is my specialised field ). If you think about it that is possible. DC Biasing as in cheap tape recorders. M6 metal will cope with that. Sure you can hear the music in the core as a result. One could try 400 kHz also. That should be better and would favour balanced valves, over to you guys on that one as I have said all I can without getting told off by my boss.
I ran 2 x 24 R resistors in the cathodes as in Dynaco to get 1.5 V to measure valve health. Then fixed bias to g1 of -35V. This was on a doubler to give - 70V standby and ramped switch on to save the cathodes. This came from a Siemens amp that was for cinema's belonging to PU arm maker Frank Schroeder ( his collection, Berlin is Siemens as is Frank ). 820 V HT all pentode EF 45 ( think ) in LTP like Quad and 2 x EL34 ( Siemens 34's don't go blue at 820V, it would be 800 in Berlin at 227VAC, I have 246VAC ). 9 dB feedback. About 1Vrms in. 13 % THD 100 watts. 1.3% 10 watts ( cinema loud levels) 0.13% 1 watt ( average cinema level ). 100V line with 8R tap at 10 watts ( not in cct, was deduced ). Circa 1959. Sound was as we remember if like me 60 as typical cinema. A bit constricted and yet open. Great fun. It measured well enough. The 100 watts for big sounds. That was an amp designed by an expert or team. Choke PSU to EF45's and raw DC to EL34's as pentodes. That uses the pentode as a combined regulator amplifier. A power op-amp in valves as did Quad. Mr HC Lin was a year behind on the 741 type devices. They were declassified about 1960. Blumlein knew them in 1935. To be honest fixed bias isn't the best way. The cathode resistor if you look up constant current sources in JFET's ( better written up than other indentical opperation depleation devices) shows the cathode resistor to be doing more than you might think. DTN Williamson prefered a single resistor as some extra cancellation takes place to his way of thinking. Not bad for a lad of 19 ( 1947 ) when people still don't believe in TID. If so add 2 x 24R also. That might be 560R + 2 x 24 R. You can add 1000 uF 50V( 63 ) Panasonic FC to bootstrap the 560R. 470R for KT88 ( but why would you ).
I always do feedback by guessing. Not much at first and an oscilloscope to be sure. Even a wirewound resistor is enough to show problems. Back off 6 dB when you find them and 6 db more by ear is usual. That is 12 dB less than the oscillation point, if that is at - 10 dB have a redesign. The reason zero feedback amps sound better is applying feedback with a transformer in the way needs PHD maths or hours of testing. The Mozart trick seems to side step that a bit. Obvious why it should if drawing the reactive parts of an amp as filters. It doesn't take much to get the amp to ring. If UL sounds nasal you have ringing. Most often you do. Last book of GEC designs 1972 show they understood that. Transistor amps always sound best with high feedback levels if an OK design. Valves don't as the transformer is the killer problem. The Schade-Kitic feedback is a very very good idea. Damping factors > 3 already useful ( Wireless World 1955 paper cone speakers like 12Lta use, other Magnepan SMGa ). Ordered distortion < 1% is inaudible ( sorry transistor guys it is ), read Hiraga. If you have panel speakers they might be 0.1%@ 90 dB 1 metre. With an OK CD ( If only, Quad CD67 not bad with 20 bit Crystal DAC ) one can have perfection and not use feedback. Don't try with Quad 2/22, they need feedback, Dynaco also I suspect although less so. The sound without belongs to Joe Meek.
Officially I don't write on DIY audio. Officially I don't know about valves. I fixed some so know a bit. I still get stuff tested by the old PYE labs in Cambridge. Philips were bonkers enough to close it down when RFi was starting to be a problem. They have an RF quiet room so as to measure down to silly levels. That was built before AP Analyser existed.It can reach the limits of that. PYE was a very special company. All cold war stuff.
Zero feedback = no loop feedback I thought I should add. UL and Kitic is good feedback and PYE Mozart feedback" looks " good. UL + Loop is hard work and even when it works often you won't use it. In my case as it gets rid of a gain stage not to ( better transparency ). PSU design is critical. I use an FET multiplier. It adds a bit of second harmonic, that isn't so bad. Add a series input drain resistor to get it reliable 47R is about right at 63 mA and don't use posh ones. drop enough voltage on the gate divider to loose the hum ( I think I used 10K and 3M, That was 3 x 47 uF 200V with 1M each as a sausage string to the gate ). I had a 10R + 100 nF Zobel to the output ( very nice tweak to be much like a choke ). Don't be tempted to go down to near zero gate current for less hum. Suddenly your tweeters will smoke when you least exspect it. Add a zener before the gate stopper ( 220R discuss where if you like as it could be either side in theory, as a coward I put it before ). Zener is best ( 12 V 1.3 W will do) as it can go through the origine to stop depletion working when it should be enhancement. Most FET's have a back EMF diode. Use plastic cased ones as it is easier. 100OV N types are easy to get.
I ran 2 x 24 R resistors in the cathodes as in Dynaco to get 1.5 V to measure valve health. Then fixed bias to g1 of -35V. This was on a doubler to give - 70V standby and ramped switch on to save the cathodes. This came from a Siemens amp that was for cinema's belonging to PU arm maker Frank Schroeder ( his collection, Berlin is Siemens as is Frank ). 820 V HT all pentode EF 45 ( think ) in LTP like Quad and 2 x EL34 ( Siemens 34's don't go blue at 820V, it would be 800 in Berlin at 227VAC, I have 246VAC ). 9 dB feedback. About 1Vrms in. 13 % THD 100 watts. 1.3% 10 watts ( cinema loud levels) 0.13% 1 watt ( average cinema level ). 100V line with 8R tap at 10 watts ( not in cct, was deduced ). Circa 1959. Sound was as we remember if like me 60 as typical cinema. A bit constricted and yet open. Great fun. It measured well enough. The 100 watts for big sounds. That was an amp designed by an expert or team. Choke PSU to EF45's and raw DC to EL34's as pentodes. That uses the pentode as a combined regulator amplifier. A power op-amp in valves as did Quad. Mr HC Lin was a year behind on the 741 type devices. They were declassified about 1960. Blumlein knew them in 1935. To be honest fixed bias isn't the best way. The cathode resistor if you look up constant current sources in JFET's ( better written up than other indentical opperation depleation devices) shows the cathode resistor to be doing more than you might think. DTN Williamson prefered a single resistor as some extra cancellation takes place to his way of thinking. Not bad for a lad of 19 ( 1947 ) when people still don't believe in TID. If so add 2 x 24R also. That might be 560R + 2 x 24 R. You can add 1000 uF 50V( 63 ) Panasonic FC to bootstrap the 560R. 470R for KT88 ( but why would you ).
I always do feedback by guessing. Not much at first and an oscilloscope to be sure. Even a wirewound resistor is enough to show problems. Back off 6 dB when you find them and 6 db more by ear is usual. That is 12 dB less than the oscillation point, if that is at - 10 dB have a redesign. The reason zero feedback amps sound better is applying feedback with a transformer in the way needs PHD maths or hours of testing. The Mozart trick seems to side step that a bit. Obvious why it should if drawing the reactive parts of an amp as filters. It doesn't take much to get the amp to ring. If UL sounds nasal you have ringing. Most often you do. Last book of GEC designs 1972 show they understood that. Transistor amps always sound best with high feedback levels if an OK design. Valves don't as the transformer is the killer problem. The Schade-Kitic feedback is a very very good idea. Damping factors > 3 already useful ( Wireless World 1955 paper cone speakers like 12Lta use, other Magnepan SMGa ). Ordered distortion < 1% is inaudible ( sorry transistor guys it is ), read Hiraga. If you have panel speakers they might be 0.1%@ 90 dB 1 metre. With an OK CD ( If only, Quad CD67 not bad with 20 bit Crystal DAC ) one can have perfection and not use feedback. Don't try with Quad 2/22, they need feedback, Dynaco also I suspect although less so. The sound without belongs to Joe Meek.
Officially I don't write on DIY audio. Officially I don't know about valves. I fixed some so know a bit. I still get stuff tested by the old PYE labs in Cambridge. Philips were bonkers enough to close it down when RFi was starting to be a problem. They have an RF quiet room so as to measure down to silly levels. That was built before AP Analyser existed.It can reach the limits of that. PYE was a very special company. All cold war stuff.
Zero feedback = no loop feedback I thought I should add. UL and Kitic is good feedback and PYE Mozart feedback" looks " good. UL + Loop is hard work and even when it works often you won't use it. In my case as it gets rid of a gain stage not to ( better transparency ). PSU design is critical. I use an FET multiplier. It adds a bit of second harmonic, that isn't so bad. Add a series input drain resistor to get it reliable 47R is about right at 63 mA and don't use posh ones. drop enough voltage on the gate divider to loose the hum ( I think I used 10K and 3M, That was 3 x 47 uF 200V with 1M each as a sausage string to the gate ). I had a 10R + 100 nF Zobel to the output ( very nice tweak to be much like a choke ). Don't be tempted to go down to near zero gate current for less hum. Suddenly your tweeters will smoke when you least exspect it. Add a zener before the gate stopper ( 220R discuss where if you like as it could be either side in theory, as a coward I put it before ). Zener is best ( 12 V 1.3 W will do) as it can go through the origine to stop depletion working when it should be enhancement. Most FET's have a back EMF diode. Use plastic cased ones as it is easier. 100OV N types are easy to get.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.