I have a UL transformer so it's where I might start from. Problem with these forums is rarely do the best ideas get shared. Some will know a good idea, they won't want to give it away.That's why a road map is good. Do your own journey and hope someone has pointed out what you might see along the way. Have you noticed that the people with most distain actually say the least of real use.
What has come to light here is various very local feedback types can be used. I have learnt we could use more than one as long as we take each step very carefully. The mistake I made was not to try reinvented Schade feedback care of Mr Kitic and UL together. He got so close to doing the same in his quest. One observer said had we noticed the pentode like behaviour of the Kitic amps. I had but it was a nice version. If we used UL as our starting point it could be a real gift of an idea as then we maintain UL traits. Personally I like a SE triode driver as it then looks like a SE amp. I don't really care if in some Hogwarts way I am wrong. If it walks like a duck and quacks like a duck it's a duck, flying comes later. I was told one of my deisgns could not reliably be made if more than one built. That was wrong and could even use very low emission valves. It was always a duck whatever I did. BTW. Loadlines are mostly Hogwarts. As most ideas have been researched to death just find out what most people agree on. The maker of the device very often has the best uses, how strange. You can always do the load lines later.
The reason I posted the op amp idea was to say it would be the only time for most where just the output stage was being spotlighted. As long as the op amp is not in the feedback loop it should sound like a cable with gain. Subtly it might not. I had this problem when analysing the Kitic design. I had to mathamatically work out what my KT 88 was doing. Not very nice was the answer. Putting the ECC82 in gave me a more conventional result. I notice all the armchair designers failed to say that. You have to be Columbo when designing things. Look for the daftest clues. It's not about being right or wrong. It's about trying hard enough. 450 VDC is dangerous is the expert bit to know.
The two UL % I favour are about 50 % ( 43 ) and 80 % triode. I suspect the latter the best trade off for how triodes drives the transformer. As 43 % is typical that seems the one to talk about. I was very wrong about UL so perhaps say too much about it. I have been thinking if the secondary ( speaker out )could make a Quad style UL cathode feedback. All my ideas so far don't look right. It is said Quad used 10%, no idea what that means exactly. ( 400/2 x.7 )/ 10 ? If so we would be in the right region. Trilogy's designer Nick Poulson told me that Quad UL has double the gain for a certain distortion level.
What has come to light here is various very local feedback types can be used. I have learnt we could use more than one as long as we take each step very carefully. The mistake I made was not to try reinvented Schade feedback care of Mr Kitic and UL together. He got so close to doing the same in his quest. One observer said had we noticed the pentode like behaviour of the Kitic amps. I had but it was a nice version. If we used UL as our starting point it could be a real gift of an idea as then we maintain UL traits. Personally I like a SE triode driver as it then looks like a SE amp. I don't really care if in some Hogwarts way I am wrong. If it walks like a duck and quacks like a duck it's a duck, flying comes later. I was told one of my deisgns could not reliably be made if more than one built. That was wrong and could even use very low emission valves. It was always a duck whatever I did. BTW. Loadlines are mostly Hogwarts. As most ideas have been researched to death just find out what most people agree on. The maker of the device very often has the best uses, how strange. You can always do the load lines later.
The reason I posted the op amp idea was to say it would be the only time for most where just the output stage was being spotlighted. As long as the op amp is not in the feedback loop it should sound like a cable with gain. Subtly it might not. I had this problem when analysing the Kitic design. I had to mathamatically work out what my KT 88 was doing. Not very nice was the answer. Putting the ECC82 in gave me a more conventional result. I notice all the armchair designers failed to say that. You have to be Columbo when designing things. Look for the daftest clues. It's not about being right or wrong. It's about trying hard enough. 450 VDC is dangerous is the expert bit to know.
The two UL % I favour are about 50 % ( 43 ) and 80 % triode. I suspect the latter the best trade off for how triodes drives the transformer. As 43 % is typical that seems the one to talk about. I was very wrong about UL so perhaps say too much about it. I have been thinking if the secondary ( speaker out )could make a Quad style UL cathode feedback. All my ideas so far don't look right. It is said Quad used 10%, no idea what that means exactly. ( 400/2 x.7 )/ 10 ? If so we would be in the right region. Trilogy's designer Nick Poulson told me that Quad UL has double the gain for a certain distortion level.
Last edited:
BTW. Hafler and Walker ( Quad ) went to court over UL. Strange as they look very different in how it's done. Blumlein already had described it in about 1937 in a patent. The judgement was that they had the same idea without knowing the others work. I think that's just Folklaw as any fool can see the Blumlein paper destroyed the case. I suspect it was the use of Ultra Linear wording that was disputed. It was the same when stereo was being patented. Mr Blumlein had already got there in the 1930's. Blumlein was a war hero of the first order. Taking UL wording as a hint we should be happy with either type.
That's not Hogwarts. How very kind of you to share that. I know we go away from the question. Still think that is worthwhile. Hafler UL + Kitic seems workable. The Kitic just for lower Rp that could be better than triode compromise. The EL84 being spectacular in UL. Perhaps it was backwards engineered to be that. I have to say I prefer EL34 over KT88 and I assume 6550A. As most amps will work for either with a small adjument, it isn't expensive to find out. Although I have never tried them Ruby EL 34 in matched 4's are very cheap.
This was the last time I tried to use these bits. The version I forgot. See how it works just fine. 8 watts is not much. Nice result I think. The 1 watt very nice. I had used a LM317. That's not what I want to do. 3 x 1N4148 and BD139 measures the same ( 2 diodes has more second harmonic ) . Used this way transistors are nice devices. Again I wanted to see where it could go Note the mismatch. Both valves from the old Chipping Norton Studios. The HT was very quiet.
This was to mimic a 211 triode in driver required. It was only later on I realsied it was a nice design.
Note this has no need of UL or Schade. It probably has better bass compared with SE designs even if a bit unmatched. A matched version looks indentical except slightly lower distortion.
I have a set of vintage PP-OPT's with a tertiary winding; it is specified by the manufacturer to be 10%.
I have not yet used them, so I can't comment on the best way to utilize the tertiary winding.
Win W5JAG
Hi
IMHO pair of power pentodes like 6550 ,EL34, ...,ususally don`t need any interstage drive transf. ,I`m convinced that IS- drive transf. is good to drive but direct heated power triodes as 2A3 , 300B , 211 , ....
here is one interesting article about HQ-PP tube power amp using power pentodes and about using OPT with separate tertiary windings , Analog Audio Amplifier Design
Regards
Last edited:
Hey All,
I have these interstage transformers that I designed originally to use with 6N6P and 6L6. I have since heard from people I trust that 6L6 can be nasty sounding without lots of feedback. And with an interstage transformer global feedback is a no go. I started this project with the simple idea that I could build an amplifier without coupling caps. The signal path will be nothing but wire and tubes.
By the way, I'm talking about a different amplifier now so Schade is not the issue in this post but feedback is.
I will have to limit the feedback to the output stage for stability and also for time smearing. And in reading Nigel's post I started thinking about UL. And the RCA handbook schematic with the Schade to the plate andthe cathode really got me thinking. What I would like is an amplifier without feedback. Feedback might linearize and reduce distortion but an amp sounds more 'live' without it.
So this is my current track. Since the interstage transformers were designed to work with 6600k pp I need to stay within that arena. I looked at all the readily available pentodes and their distortion specs. In the 450 volt B+ area the best two seem to be 7591 and KT77. 7591 is more of a true pentode and for that reason may not lend itself to ultralinear? KT77 was designed for UL use so it wins out. I think I can get the p-p to about 5500 ohms which should be close enough. In checking these power tubes, (KT66,KT77,7591,6550,KT88,6L6,5881 and EL34) I noticed that the beam tetrodes seemed to work best with UL and at around 450 B+ KT77 is listed at 1%THD. So KT77 is the tube.
w5jag, please, what transformers were you talking about? I would love to take a look at them!!! I'm ordering a set of transformers with 43% UL taps and was thinking about a tap for feedback to the KT77 cathode? (BTW I was guessing at 20%) When you say 'tertiary winding' do you mean a separate winding for the cathode feedback? Or a tap on the primary? Or is it on the secondary side of the output transformers?
Thanks, Kevin
I have these interstage transformers that I designed originally to use with 6N6P and 6L6. I have since heard from people I trust that 6L6 can be nasty sounding without lots of feedback. And with an interstage transformer global feedback is a no go. I started this project with the simple idea that I could build an amplifier without coupling caps. The signal path will be nothing but wire and tubes.
By the way, I'm talking about a different amplifier now so Schade is not the issue in this post but feedback is.
I will have to limit the feedback to the output stage for stability and also for time smearing. And in reading Nigel's post I started thinking about UL. And the RCA handbook schematic with the Schade to the plate andthe cathode really got me thinking. What I would like is an amplifier without feedback. Feedback might linearize and reduce distortion but an amp sounds more 'live' without it.
So this is my current track. Since the interstage transformers were designed to work with 6600k pp I need to stay within that arena. I looked at all the readily available pentodes and their distortion specs. In the 450 volt B+ area the best two seem to be 7591 and KT77. 7591 is more of a true pentode and for that reason may not lend itself to ultralinear? KT77 was designed for UL use so it wins out. I think I can get the p-p to about 5500 ohms which should be close enough. In checking these power tubes, (KT66,KT77,7591,6550,KT88,6L6,5881 and EL34) I noticed that the beam tetrodes seemed to work best with UL and at around 450 B+ KT77 is listed at 1%THD. So KT77 is the tube.
w5jag, please, what transformers were you talking about? I would love to take a look at them!!! I'm ordering a set of transformers with 43% UL taps and was thinking about a tap for feedback to the KT77 cathode? (BTW I was guessing at 20%) When you say 'tertiary winding' do you mean a separate winding for the cathode feedback? Or a tap on the primary? Or is it on the secondary side of the output transformers?
Thanks, Kevin
banat, I looked at DePalma's schematic and I assume his tertiary winding is more for the plate voltage being twice the grid voltage than anything else, though i might be wrong here. And I suspect that as feedback via UL it may be compromised although I'm not sure why I think that?
mr2racer
B.DePalma output power stage (OPS) work biased very close to class B of operation with screen grids supplied via separate UL windings and with half B+ DC voltage as from anodes B+ DC supply , 800V/400V ,
this is also one mode of operation original suggested by Philips and Telefunken for EL34 tube , but must say that modern EL34 tubes variants usually can not work with so high voltages , 650/325 is some maximum ,
any way in that mode of operation long live of power tubes can be expected altogether with relative low generated heat and distortion from OPS .
B.DePalma output power stage (OPS) work biased very close to class B of operation with screen grids supplied via separate UL windings and with half B+ DC voltage as from anodes B+ DC supply , 800V/400V ,
this is also one mode of operation original suggested by Philips and Telefunken for EL34 tube , but must say that modern EL34 tubes variants usually can not work with so high voltages , 650/325 is some maximum ,
any way in that mode of operation long live of power tubes can be expected altogether with relative low generated heat and distortion from OPS .
I have never built this, there could be hidden problems. I have tried to keep it inside the Kitic concept and the question asked. My problem with it and my previous post is they both need a near zero distortion amplifier stage to keep to below 1% THD. Using OP2604 or similar it should be quite nice. I dare say a 1K triode strap could be a lower value. Kitic used 1K in his RH88 when pentode ( has been 3 K ). There is no ECC81 cathode capacitor as per his idea. Unlike him I would not use a LM317 to the KT88 cathodes. It is an LM741 op amp with power transistor and 1.25 V band gap. Granted it measures well. I drew this as it was hinted at without detail. Using Kitic + UL might be interesting. The speaker can be fed back to the 330R if wanting a more local feedback loop than most. You migh get lucky and feed it directly like I did speaker to 330R ( in correct phase ).
If using my BD139 idea ( to chassis as it will get hot ) a LED is an ideal voltage reference. I don't like them to cathodes as it seems to fight against a better idea of using a resistor. If I can hear the difference I am not sure.
Modern meters show mV, 1R is enough for balance measurment at the KT88 cathodes. If the distortion cancellation of PP is a good idea when this design it is hard to say. My previous post looks SE ( like a Naim NAP 250 in fact albeit 30 dB higher, contrived in the Naim using 1K and 12 K LTP resistors ) . This one might be a bit ugly. An ECC82 might give us a gain of 10 and give what we need in distortion balance.
The all triode circuit I show has some merit. If it had a conventional cathode resistor ( 270/300 R ) and cap I would hope for 12 watts. I can't think of a better triode than EL84 as shown, it will swing the voltage required ( super ECC82 ). I tried it in pentode and had the better gain. Alas it looked ugly on the analyser. No real need for the MJE 350 transistor. One could run shunt feedback to a EL84 as pentode and maybe get something better. The Valve Wizard says well enough when triodes how it works ( below ), the pentode likes feedback the same way to g1. The point to make is the lower Rp of the output triode is not a bad thing to have and the specrum a bonus. The complex reasons people like triode seem to overlook this. An uncontrolled speaker cone is not the best sound. UL if it works correctly seems to get a better distortion Rp compromise although maybe not as controlled as triode. Looking at it again Kitic feedback with UL could work. Kitic always thought that was the big deal. I think he argued a certain valve had a Rp of 1K2 in fake triode ( g2 to anode ) and 900R Kitic. The spectrum looks pentode like. As long as the 5 th harmonic isn't way up I don't think that is a problem. It is easy to mistake the lack of control of a pentode for conventional distortion theory. If a pentode is used to drive the middle parts of the amplifer often the sound is not too bad.
When pentodes came to the market people were able to see how a triode worked. Inside a triode is a pentode. The linearity of a triode is making the best of it's defects. An ECC81 strarts to go towards where a pentode would take us. UL is a point where the triode curve and pentode curve fight it out. The resulatant curve is almost a nice curve. One can make a similar curve input to output of an amp using pentode in and triode out. The problem is the PP cancellation destroyes that nice idea. Up and down UL as in Hafler style UL PP is just fine. You might see Zobel circuits on UL taps ( GEC design book 1972 ) As far as I know not needed if no loop feedback used. The 1 K resitor works as an OK 1st order filter and current limiter. I have seen a forward biased 1N4007 used to the UL tap. It was argued it might help where speaker back EMF is concerned ( seems possible ). As these valves will tollerate 800V I suspect any voltage doubling effect if it exists will be safe. Use ears also!!!! wouldn't it be daft if my dislike of UL ( nasal sound ) was back EMF.
The Valve Wizard
When pentodes came to the market people were able to see how a triode worked. Inside a triode is a pentode. The linearity of a triode is making the best of it's defects. An ECC81 strarts to go towards where a pentode would take us. UL is a point where the triode curve and pentode curve fight it out. The resulatant curve is almost a nice curve. One can make a similar curve input to output of an amp using pentode in and triode out. The problem is the PP cancellation destroyes that nice idea. Up and down UL as in Hafler style UL PP is just fine. You might see Zobel circuits on UL taps ( GEC design book 1972 ) As far as I know not needed if no loop feedback used. The 1 K resitor works as an OK 1st order filter and current limiter. I have seen a forward biased 1N4007 used to the UL tap. It was argued it might help where speaker back EMF is concerned ( seems possible ). As these valves will tollerate 800V I suspect any voltage doubling effect if it exists will be safe. Use ears also!!!! wouldn't it be daft if my dislike of UL ( nasal sound ) was back EMF.
The Valve Wizard
m2racer and Nige
It can be interesting , here is the schematic of one simple and small PP tube amp front end ,
those amps I have used long time ago for 24/7 listenining , sound was OK with nice midrange and highs , bass spectrum was OK to ,soundstage presentation was amazing ,
it is one my variation of Mullard 5-20 circuit , except that these amps works GNFB-free ,
input stage is CF or active impedance transformer driving directly long tail pair or diff.amp phase spliter ,
amp OLG value is not so high but just enough so it can be easy directly driven from any standard digital signal source as PC card , CD player, ....
,the best way to found Rx-values is by experiment , both by measurements and listening test .
It can be interesting , here is the schematic of one simple and small PP tube amp front end ,
those amps I have used long time ago for 24/7 listenining , sound was OK with nice midrange and highs , bass spectrum was OK to ,soundstage presentation was amazing ,
it is one my variation of Mullard 5-20 circuit , except that these amps works GNFB-free ,
input stage is CF or active impedance transformer driving directly long tail pair or diff.amp phase spliter ,
amp OLG value is not so high but just enough so it can be easy directly driven from any standard digital signal source as PC card , CD player, ....
,the best way to found Rx-values is by experiment , both by measurements and listening test .
Attachments
Last edited:
That's very interesting. Like the -110V tail of the splitter.
Here is a concept ( yours better as to how side two is driven perhaps ). The idea is to have pure Kitic at the power amp end. The splitter will need some playing with to get it to balance. The idea is to combine the curves as in UL. Have easy pentode to drive. Lots of gain and play with g2 to get what we need. It might just work. $10 for a 6U8A. The ECC81 usually is 680R I think for the cathode resistors. Note this will have Schade/Kitic feedback.
Here is a concept ( yours better as to how side two is driven perhaps ). The idea is to have pure Kitic at the power amp end. The splitter will need some playing with to get it to balance. The idea is to combine the curves as in UL. Have easy pentode to drive. Lots of gain and play with g2 to get what we need. It might just work. $10 for a 6U8A. The ECC81 usually is 680R I think for the cathode resistors. Note this will have Schade/Kitic feedback.
Nige
I think that`s schematic is basically OK , except that input 6U8A tube have to be reconected in next way ,...
pentode part to be connected as standard pentode voltage gain stage further from anode DC coupled to grid of triode part but now reconfigured as standard split load phase inverter .
I think that`s schematic is basically OK , except that input 6U8A tube have to be reconected in next way ,...
pentode part to be connected as standard pentode voltage gain stage further from anode DC coupled to grid of triode part but now reconfigured as standard split load phase inverter .
Last edited:
I took the idea from Radford except I think he drives the triode side. I knew Richard Hey of Radford and Nytech. He died recently very yonge. He taught me to look at current waveforms.
Nige
I look again in your schematic from above and I cant see where phase inversion is done on input 6U8A tube , maybe if that common cathodes resistors is replaced with some CCS ?,
but than the question is how simetric will be output signal from anode of pentode part against signal from anode of triode part ,
any way that middle ECC81 driver tube gain is heavy reduced by connecting those two 100K anode resistors to plates of output power tubes ( Shade feedback) so than some significant voltage gain have to come from input 6U8A tube to get overall enough OLG .
I look again in your schematic from above and I cant see where phase inversion is done on input 6U8A tube , maybe if that common cathodes resistors is replaced with some CCS ?,
but than the question is how simetric will be output signal from anode of pentode part against signal from anode of triode part ,
any way that middle ECC81 driver tube gain is heavy reduced by connecting those two 100K anode resistors to plates of output power tubes ( Shade feedback) so than some significant voltage gain have to come from input 6U8A tube to get overall enough OLG .
Last edited:
Here is the Radford. I've moved it back so as to allow Schade feedback. If 6U8A is similar to the 7199 of Dynaco it might have the gain needed. I understand it is a mini beam tetrode! Looking again he does it the way I would. R8 C6 as per your version. R10 could be a current source. One supplier says it is the same valve. Seems not. As best I know. R10 8K2 R11 33K R12 39K ( !! )
A Quote.
The change to a pentode-triode was a development by Arthur Bailey and has a dramatic effect on the open loop gain of the amplifier. In effect, although the double-triode only had a flat gain bandwidth of around 15 kHz, the pentode-triode extended this to above 150kHz. As with the previous design, this flattening of the open loop response in terms of amplitude and phase made the amplifier inherently better, and allowed global feedback to provide further improvements without the amplifier oscillating on some loudspeaker loads.
6U8A @ The Valve Museum
A Quote.
The change to a pentode-triode was a development by Arthur Bailey and has a dramatic effect on the open loop gain of the amplifier. In effect, although the double-triode only had a flat gain bandwidth of around 15 kHz, the pentode-triode extended this to above 150kHz. As with the previous design, this flattening of the open loop response in terms of amplitude and phase made the amplifier inherently better, and allowed global feedback to provide further improvements without the amplifier oscillating on some loudspeaker loads.
6U8A @ The Valve Museum
They are PA transformers intended for P-P 6L6's, and likely predate the UL patent hoopla.
The tertiary winding is a 10% separate winding on the primary side, probably intended to energize the screens at a lower voltage, but I don't see why they couldn't be used for other feedback schemes.
The secondary side has two PA line windings, and typical 1950's speaker impedances. I expect these could also be useful for feedback.
They turn up on ebay as NOS once in a while, and the hi fi people ignore them because they are marketed as PA OPT's. I'm not an audiophile, so this is fine with me.
I don't have a PP amp, so I haven't had an opportunity to play with them yet.
Win W5JAG
Here is the Bailey paper. How it is open loop is another question. 6GH8 might be another choice ( better ) and seems cheap at $8 typical NOS. If set up carefully it might have something. ECF82 listed here. 7199 I am told was a shielded version of a typical TV valve. Different pin out.
New Phase Splitter
New Phase Splitter
- Status
- Not open for further replies.