For either PP triode or PP UL, Tango and Hashimoto have sounded best to me. I have also messed with vintage OPT's from UTC, Dyna, Scott and Fisher. The Tango and Hashimoto OPT's I've heard or used have far surpassed the vintage OPT's. I have not heard Electraprint or Sowter.
A friend built a simple PP EL84 amp with Edcor iron and said the OPT's sounded extremely good considering the price. I'd like to try a pair of those.
I hope that's the kind of opinion you were looking for.
A good friend of mine who has quite a nice lab full of high end test equipment allegedly did a valid comparison of a bunch of output trannies, and swore that the more expensive ones sounded better. When he stepped out of the room for a moment, his side kick buddy told me he was there and heard no difference at all between any of them...
I've been having very good luck with Hammond power and output transformers for years, for both high end hi-fi and guitar amps. I've only gotten one bad one in about 20, and I returned it for a replacement. I usually buy them through Antique Electronics in Arizona. Although I haven't tried the more expensive brands personally, I suspect that they are really just fool traps, much like fancy speaker wire. Really cool looking, with rare earth elements in the core (cobalt for ex.), maybe pure silver wire, etc., and on the bottom line not audibly better than hammond cheapos, at least not to my ears. My Hi-Fi tube amp only has a max of 12dB negative feedback (adjustable 0-12), so minimal distortion correction, and it sounds dam good to my ears.
I've been a hard core audio engineer since the 60's so I would think I would know how to judge sound, but to be fair, there are those who say they can hear the difference.
My guitar amps that I built using Hammond Hi-Fi transformers sound so much better than the other guitar amps in the band I'm in (2 other guitarists, Fender and Peavey, all tubes), it borders on being embarrassing. I'd worry more about getting the impedance match to the tubes right, and use a topology that has minimal crossover distortion when in clipping.
You can check out my "Musicbox" 30 watt per ch. Hi-Fi tube power amp at:
http://www.spiritone.com/~rob_369
I've been having very good luck with Hammond power and output transformers for years, for both high end hi-fi and guitar amps. I've only gotten one bad one in about 20, and I returned it for a replacement. I usually buy them through Antique Electronics in Arizona. Although I haven't tried the more expensive brands personally, I suspect that they are really just fool traps, much like fancy speaker wire. Really cool looking, with rare earth elements in the core (cobalt for ex.), maybe pure silver wire, etc., and on the bottom line not audibly better than hammond cheapos, at least not to my ears. My Hi-Fi tube amp only has a max of 12dB negative feedback (adjustable 0-12), so minimal distortion correction, and it sounds dam good to my ears.
I've been a hard core audio engineer since the 60's so I would think I would know how to judge sound, but to be fair, there are those who say they can hear the difference.
My guitar amps that I built using Hammond Hi-Fi transformers sound so much better than the other guitar amps in the band I'm in (2 other guitarists, Fender and Peavey, all tubes), it borders on being embarrassing. I'd worry more about getting the impedance match to the tubes right, and use a topology that has minimal crossover distortion when in clipping.
You can check out my "Musicbox" 30 watt per ch. Hi-Fi tube power amp at:
http://www.spiritone.com/~rob_369
Last edited:
Any of Edcor's units are competently designed and well built. Hammond is a bit more variable. Don't bother with the "high end" boutique makers like O-Netics, Intact Audio, Tribute Audio, Sowter, Lundahl etc, until you know exactly what you do not like about the Edcor product. Chances are very good you will never need to spend more.
Bud
Bud
the Edcor I didn't like for push-pull pentodes. the single ended for a 6em7 was good for them.
the Black enclosed hammond transformers I like them for push pull DHT and pentodes on more of the budget stuff. I've had manufacturing inconsistencies with the yellow bobbin ones.
If you want to build a higher end, Plitron or Hashimoto is the way to go for push-pull.
Otherwise, the black enclosed Hammond works good.
the Black enclosed hammond transformers I like them for push pull DHT and pentodes on more of the budget stuff. I've had manufacturing inconsistencies with the yellow bobbin ones.
If you want to build a higher end, Plitron or Hashimoto is the way to go for push-pull.
Otherwise, the black enclosed Hammond works good.
Here a test from some "cheap" transformers.
Budget Output Transformers | Tubelab
Not PP but SE and you can see non of them are really good, but useable. At the other hand, those "high-end" manufactures don't give datasheets so maybe they not better.
Budget Output Transformers | Tubelab
Not PP but SE and you can see non of them are really good, but useable. At the other hand, those "high-end" manufactures don't give datasheets so maybe they not better.
Last edited:
I've tested a couple of the 15-20W Hammonds and they were not very good, especially on the 8 Ohm tap. Dynacos and their recent clones were quite good, though beat by a number of vintage transformers - including some Scott, Fisher, Triad, UTC, Stancor, H-K and even some Bogen PA transformers.
Zeroing in
Good Morning to All from beautiful Central Texas:
Thanks so much for all the input! Time to reveal my more focused thoughts and musings, and also a bit more about my experience and expertise.
I make my living as an engineer in the MWD industry. There we have to deal with very high temperatures, power magnetics, signal and sensor magnetics (magnetometers, inductive sensors, others), low noise/high gain amplifiers, logic/microprocessor circuits, and the like, all operating at temperatures in excess of 175C. No typo, over 175C all the way to 200C. In other times I have designed many transformers and magnetics of all types, including pulse, power, ferroresonant, etc.
In spite of my qualifications, I wanted the collective wisdom and experience of this group. There are lots of you who have been there, done it and probably have the tee shirt to prove it. No way was I going to miss out on that. Thanks again for all the postings!
During all this time, my research has been bringing me to a very specific topology, and a reasonable vendor.
I believe that Edcor will give me the best bang for the buck, especially since I want a bit of a custom design. Nothing way out but still a bit of a change from the standard configurations. But first, the topology so it all makes better sense.
I just love the old treatises on vacuum tubes and amplifier theory, with all the heavy math and all. While my original thinking was toward a conventional UL output stage but with a "tertiary" winding (in quotes because that appellation depends on physical placement, functionality, and other things some of which are subjective to the designer) in order to accommodate the sweep tubes with their much lower Vg2, I have since highly modified all that.
If I were building it today, it will no longer be UL implemented with the transformer. Instead, it will have a regulated Vg2 supply with the ability, via MOSFETs, to apply a percentage of plate signal to the g2s. But, that could wind up being none, just straight DC (full pentode operation). The transformer will have a very mild CFB factor, but not a separate winding for that. Instead, the speaker secondary will be center tapped, taps pulled for my 6 Ohm speakers, and further symmetrical extensions for a very mild about 5% CFB. The drivers will be 6AU6 with both output plate to driver plate feedback (so called Schade) and output plate to driver cathode feedback. The 6AU6s will be driven by a fully differential input stage with sand state CCS. Overall feedback of about 6dB will be applied to one grid and the other grid is the signal input. The jury is still out on whether to use MOSFET source followers between the 6AU6s plates and the g1 of the output tubes. Grid stopper resistors will probably be adequate, plus the margin being built but if I wind up biasing more toward A than B then grid saturation is a real concern and the buffers would come in very handy in an overload situation.
So, error correction for the output stage, error correction for the driver/output combination and error correction for the whole thing. The total nested and overall feedback factor will be in the order of some 30dB. From my days of designing and implementing closed loop analog systems, that is about right to produce a very clean system. Additional tricks of resolving the OPT's resonance and providing a corrective network, shaping the frequency response of the chain and running Bode plots to make sure all the phase shifters are tamed will of course be applied. Differential all the way to improve CMRR, stabilize operating points against aging and other factors, and inherent even order harmonic distortion cancellation. Looks cool on paper so far, can't wait to have the time to actually build the beast!
Back to the transformer, my thought is to specify a 70W stack, pick the P-P load I want, specify the secondary per text above and other than asking for the secondary to be fully symmetrical (which probably means they'll have to either make it bifilar or in identical sections suitably connected) including DCR for my bias stabilizing scheme, they get to make it their way which according to many of you, it seems they know how to do it right. The 70W will give me margin since I only want 50W. The down side of heavy feedback is no tolerance for overloads. If anything saturates, the whole thing rapidly goes down the distortion drain. Like NOW. So, a 50W rating with about 70W capability and useage more in the 5W range. Running things in class AB1 will give me some 7W or so of straight class A1.
Thanks to all, keep the comments coming!
Rene
Good Morning to All from beautiful Central Texas:
Thanks so much for all the input! Time to reveal my more focused thoughts and musings, and also a bit more about my experience and expertise.
I make my living as an engineer in the MWD industry. There we have to deal with very high temperatures, power magnetics, signal and sensor magnetics (magnetometers, inductive sensors, others), low noise/high gain amplifiers, logic/microprocessor circuits, and the like, all operating at temperatures in excess of 175C. No typo, over 175C all the way to 200C. In other times I have designed many transformers and magnetics of all types, including pulse, power, ferroresonant, etc.
In spite of my qualifications, I wanted the collective wisdom and experience of this group. There are lots of you who have been there, done it and probably have the tee shirt to prove it. No way was I going to miss out on that. Thanks again for all the postings!
During all this time, my research has been bringing me to a very specific topology, and a reasonable vendor.
I believe that Edcor will give me the best bang for the buck, especially since I want a bit of a custom design. Nothing way out but still a bit of a change from the standard configurations. But first, the topology so it all makes better sense.
I just love the old treatises on vacuum tubes and amplifier theory, with all the heavy math and all. While my original thinking was toward a conventional UL output stage but with a "tertiary" winding (in quotes because that appellation depends on physical placement, functionality, and other things some of which are subjective to the designer) in order to accommodate the sweep tubes with their much lower Vg2, I have since highly modified all that.
If I were building it today, it will no longer be UL implemented with the transformer. Instead, it will have a regulated Vg2 supply with the ability, via MOSFETs, to apply a percentage of plate signal to the g2s. But, that could wind up being none, just straight DC (full pentode operation). The transformer will have a very mild CFB factor, but not a separate winding for that. Instead, the speaker secondary will be center tapped, taps pulled for my 6 Ohm speakers, and further symmetrical extensions for a very mild about 5% CFB. The drivers will be 6AU6 with both output plate to driver plate feedback (so called Schade) and output plate to driver cathode feedback. The 6AU6s will be driven by a fully differential input stage with sand state CCS. Overall feedback of about 6dB will be applied to one grid and the other grid is the signal input. The jury is still out on whether to use MOSFET source followers between the 6AU6s plates and the g1 of the output tubes. Grid stopper resistors will probably be adequate, plus the margin being built but if I wind up biasing more toward A than B then grid saturation is a real concern and the buffers would come in very handy in an overload situation.
So, error correction for the output stage, error correction for the driver/output combination and error correction for the whole thing. The total nested and overall feedback factor will be in the order of some 30dB. From my days of designing and implementing closed loop analog systems, that is about right to produce a very clean system. Additional tricks of resolving the OPT's resonance and providing a corrective network, shaping the frequency response of the chain and running Bode plots to make sure all the phase shifters are tamed will of course be applied. Differential all the way to improve CMRR, stabilize operating points against aging and other factors, and inherent even order harmonic distortion cancellation. Looks cool on paper so far, can't wait to have the time to actually build the beast!
Back to the transformer, my thought is to specify a 70W stack, pick the P-P load I want, specify the secondary per text above and other than asking for the secondary to be fully symmetrical (which probably means they'll have to either make it bifilar or in identical sections suitably connected) including DCR for my bias stabilizing scheme, they get to make it their way which according to many of you, it seems they know how to do it right. The 70W will give me margin since I only want 50W. The down side of heavy feedback is no tolerance for overloads. If anything saturates, the whole thing rapidly goes down the distortion drain. Like NOW. So, a 50W rating with about 70W capability and useage more in the 5W range. Running things in class AB1 will give me some 7W or so of straight class A1.
Thanks to all, keep the comments coming!
Rene
Sounds like a pretty ambitious project. I've heard good things about Edcore. If I understand what you said above, you've got local negative feedback going on here and there. That's good (less of a phase margin issue).
I built several guitar amps with no loop feedback, as well as one Hi-Fi tube amp that has 0 to about 12dB of feedback around the output stage and differential driver stage, and have a front end triode (6SN7) outside any feedback loop with a current source plate load for improved linearity and power supply rejection. This amp is excellent up to clipping, but once pushed into clipping it generates push-pull crossover distortion more than I wanted, despite me following advice on minimizing "blocking distortion" (from the Aiken website) which is apparently the cause of the crossover distortion, and messing with the bias adjustment pushing it significantly into class A operation.
I found what appears to be a better topology for dealing with this blocking distortion and its resulting crossover distortion. One way is to use interstage transformers as a phase splitter driving the output tubes, or so I've been told, I haven't actually tried that. But then you've got to have 2 more expensive transformers, and the amp chassis ends up being rather large.
A friend brought me a mid 1950's Gibson guitar amp to fix one day, and I noticed that it had next to no crossover distortion when jammed well into clipping. It turns out that it has arguably the most primitive version of a phase splitter there is. Same as in the first Fender tweed amp (The Dual Professional from 1948). The phase splitter stage is a simple differential topology. The upper half of the phase splitter has it's plate capacitively coupled to the grid of the upper output tube. After that coupling cap (and this is apparently the trick), the signal is tapped off, resistively attenuated, and fed to the grid of the lower triode of the phase splitter stage, which in turn drives the grid of the lower output tube with capacitive coupling. So the lower phase splitter triode is fed a signal that is in a sense predistorted by the blocking distortion mechanism of the upper output tube, thereby causing a cancellation of the crossover distortion when jammed well into clipping...
If I build another tube amp, I plan to use this topology of phase splitter for driving the output stage. The amp did have a nice sound when over driven with a guitar. Only some of the earliest guitar amps used this topology, so there could be a drawback I don't know of yet (higher pre-clipping distortion or noise issue?), but for a guitar amp I think it's a big plus, and I don't know if the early engineers realized that they had this feature. Not everyone cares what happens beyond clipping. I do.
I've attached the Gibson circuit.
I built several guitar amps with no loop feedback, as well as one Hi-Fi tube amp that has 0 to about 12dB of feedback around the output stage and differential driver stage, and have a front end triode (6SN7) outside any feedback loop with a current source plate load for improved linearity and power supply rejection. This amp is excellent up to clipping, but once pushed into clipping it generates push-pull crossover distortion more than I wanted, despite me following advice on minimizing "blocking distortion" (from the Aiken website) which is apparently the cause of the crossover distortion, and messing with the bias adjustment pushing it significantly into class A operation.
I found what appears to be a better topology for dealing with this blocking distortion and its resulting crossover distortion. One way is to use interstage transformers as a phase splitter driving the output tubes, or so I've been told, I haven't actually tried that. But then you've got to have 2 more expensive transformers, and the amp chassis ends up being rather large.
A friend brought me a mid 1950's Gibson guitar amp to fix one day, and I noticed that it had next to no crossover distortion when jammed well into clipping. It turns out that it has arguably the most primitive version of a phase splitter there is. Same as in the first Fender tweed amp (The Dual Professional from 1948). The phase splitter stage is a simple differential topology. The upper half of the phase splitter has it's plate capacitively coupled to the grid of the upper output tube. After that coupling cap (and this is apparently the trick), the signal is tapped off, resistively attenuated, and fed to the grid of the lower triode of the phase splitter stage, which in turn drives the grid of the lower output tube with capacitive coupling. So the lower phase splitter triode is fed a signal that is in a sense predistorted by the blocking distortion mechanism of the upper output tube, thereby causing a cancellation of the crossover distortion when jammed well into clipping...
If I build another tube amp, I plan to use this topology of phase splitter for driving the output stage. The amp did have a nice sound when over driven with a guitar. Only some of the earliest guitar amps used this topology, so there could be a drawback I don't know of yet (higher pre-clipping distortion or noise issue?), but for a guitar amp I think it's a big plus, and I don't know if the early engineers realized that they had this feature. Not everyone cares what happens beyond clipping. I do.
I've attached the Gibson circuit.
Attachments
Last edited:
Yep, the old "paraphase" PI.
It seems that the "top" unit's plate would have to see the distortion caused by the grid current of the corresponding 6V6, which it would, and feed it to the "lower" unit. That lower unit's distortion products are not fed back to the top unit, nor to the upper 6V6's grid But that lower unit is also seeing grid current caused distortion due to its corresponding 6V6. Got to think about all that, considering the out of phase displacements, etc. In my case, this is for Hi Fi so I simply don't want to get into clipping deliberately. The concern is when it happens accidentally and how well the amplifier recovers.
By the way, the Hi Fi you described above, does its behavior at clipping differ significantly between with and without feedback? 12dB is significant enough to suddenly increase effective drive once anything in the chain saturates.
Thanks, something else to consider! Now back to work :-(
It seems that the "top" unit's plate would have to see the distortion caused by the grid current of the corresponding 6V6, which it would, and feed it to the "lower" unit. That lower unit's distortion products are not fed back to the top unit, nor to the upper 6V6's grid But that lower unit is also seeing grid current caused distortion due to its corresponding 6V6. Got to think about all that, considering the out of phase displacements, etc. In my case, this is for Hi Fi so I simply don't want to get into clipping deliberately. The concern is when it happens accidentally and how well the amplifier recovers.
By the way, the Hi Fi you described above, does its behavior at clipping differ significantly between with and without feedback? 12dB is significant enough to suddenly increase effective drive once anything in the chain saturates.
Thanks, something else to consider! Now back to work :-(
I add in the 12dB of feedback mostly to lower the output impedance as seen by the speaker, so the speaker doesn't have a big un-tamed resonance somewhere around 60HZ. I don't think it changed the clipping waveform much but I didn't look at that recently enough to remember. Maybe I should take another look at that.
If the amp you are building is only going to be 50 watts, there is a significant chance you will on occasion hit clipping if you turn it up loud. Transients in music are often 10dB+ higher than the average levels.
I agree, once the transformer is more or less good enough, the rest is up to the circuit.
Bob, you are so right. It seems the accepted peak to "normal" ratio is about 10:1, or 10dB in power as you say. My current amplifier is a salvaged Magnavox which only puts out 10W or so per channel. I just don't turn it up that far. Depending on which link in the chain saturates first, and whether my amp can gracefully handle the outputs drawing grid current, I may or may not hear much commotion during brief episodes of clipping. So, the question is, how much more will I overbuild this thing? It only has to please me so that makes it either harder or easier!!
Got some other things in the fire right now, might make a great winter project. I will start rendering it all in my schematic capture and publish it all in maybe the next 2-3 months. Plenty time for y'all to continue to influence me!
Rene
Those would have been my tests. They were done in the early stages of my 6 or 7 year affliction with the SE fever. I built nothing but SE tube amps in sizes from 1 WPC to a 200 Watt SE guitar amp. I have used OPT's from cheap ($10) to mid priced (just over $100 each). On at least 3 occasions I gathered some friends and played musical OPT's in one of my amps. The transformers were swapped with clip leads on a hot amp (removing the rectifier tube to avoid temporary loss of vocabulary control that happens when playing with electricity). On one occasion we swapped a Transcendar, an Electra Print, and a One Electron into a SSE amp using KT88's then EL34's. All of the participants agreed that the Transcendar and the One Electron sounded very similar. They could not be reliably identified without looking. All agreed that the Electra Print sounded different, most said it was better, and one person didn't like it.
I have recovered enough from the SE tube amp fever such that I actually built a solid state class D amp, but this is about P-P tube amps, specifically powerful sweep tube amps. I know a bit about this. Pete Millett created a 18WPC mild mannered amp he called his "Engineers Amp", since I am an engineer, I got one of his boards and told Pete that I would not be happy unless I could get at least 50 WPC. 50 WPC came and went in about 5 minutes. I stopped squeezing the thing at 250 WPC or 525 watts in paralleled mono. OPT's......Edcor.
I published a 125 WPC version using Petes circuit board. At least 10 people built it and most used Edcor 100 watt 3300 ohm OPT's. I have a set of Plitron 400 watt toriodal OPT's but they are overkill for this project, so I bought a pair of the 100 watt Edcor's.
You seem to have already made most of the major design choices. You asked about the load impedance, and stated that you want about 50 WPC. This can be done with any load from 3300 ohms to 6600 ohms. Below 3300 ohms you will need rather large output tubes to support the peak current demands, and above 6600 ohms you will need too much B+ to get headroom over your 50 watts around woofer resonance. A higher load will gain a bit more damping factor, and a lower load will gain power output for a given B+.
I tend to avoid large amounts of negative feedback since it can dull the dynamics of some music, cause all sorts of overload issues WHEN the amp does clip, and cause instability if the OPT is inside the feedback loop.
I have breadboarded a sweep tube amp that hits 125 WPC with a pair of $1 13GB5's per channel. It uses combined G1 and G2 drive and "Schade" type feedback from the plates of the output tubes to the plates of the drivers, and the input tubes. The driver circuit is a 6SL7 LTP driving a 6SN7 LTP with mosfet followers feeding the output tube grids. The final amp will not be completed for a year or so since I am in the process of moving my lab 1200 miles.
Tubelab, I'm glad you got over your SE addiction. That's been known to render guys single at the 200 watt level. Did you use a 12 step group for that? The way I see it, and its arguable that I'm wrong, but if crossover distortion during clipping (or any other time) is extremely low, then it should sound as pleasant as a SE circuit. In my case I reached a point where I just didn't have room for a whole bunch of large tube amps. For guitar distortion I still highly prefer tubes, but for Hi-Fi playback I consider transistors and opamps to be quite nice enough. Maybe better than tubes, not sure. There's tradeoffs in both cases. Tube amps are nostalgic for me, and can make a great art object.
I've been thinking of building up a class D for a ghetto blaster to take camping, but some of the chips/circuits are still questionable - there's the occasional explosion and blown speaker, so being too busy anyway with other projects, I'm putting that off for now.
There are all kinds of great tubes that are no longer mainstream, but I prefer the more common ones that the guitar industry keeps alive (still being produced somewhere). They seem fine to me, and then it's up to the circuit to use them well. My Hi-Fi tube amp (driving my Seas Millenium tweeters in my OB tri-amp'd system) uses 2 6SN7's and two EL34's per channel. The trannies are just Hammond but it sounds fine.
I've been thinking of building up a class D for a ghetto blaster to take camping, but some of the chips/circuits are still questionable - there's the occasional explosion and blown speaker, so being too busy anyway with other projects, I'm putting that off for now.
There are all kinds of great tubes that are no longer mainstream, but I prefer the more common ones that the guitar industry keeps alive (still being produced somewhere). They seem fine to me, and then it's up to the circuit to use them well. My Hi-Fi tube amp (driving my Seas Millenium tweeters in my OB tri-amp'd system) uses 2 6SN7's and two EL34's per channel. The trannies are just Hammond but it sounds fine.
Well, since nobody else mentioned it... my favorite is the Dynaco A431 clone that Triode Electronics sells:
A431S Triode USA Output Transformer for Dynaco 4300 Ohm 60W/120W for Dynaco MADE IN USA
I've used it on PPP 807's, and an 815, both pentode mode. It's way overbuilt for 50W. LF response in both of these amps (which have feedback, the 807's quite a lot) goes very low.
You can find some measurements of the amps at www.pmillett.com.
Pete
A431S Triode USA Output Transformer for Dynaco 4300 Ohm 60W/120W for Dynaco MADE IN USA
I've used it on PPP 807's, and an 815, both pentode mode. It's way overbuilt for 50W. LF response in both of these amps (which have feedback, the 807's quite a lot) goes very low.
You can find some measurements of the amps at www.pmillett.com.
Pete
I'm really not much of an expert on transformers, but I believe that many of the cheaper ones may have a lower high end rolloff frequency, which will almost certainly be the "dominant pole" setting the phase margin situation when the circuit has negative feedback. The best way to deal with that may be to bandwidth limit the input signal so the feedback doesn't ever try to force the transformer to put out more supersonic signal than it can until the amp is in clipping. Apparently the cheaper trannies also have a weak low frequency end, and may not put out a flat response down to 30HZ or so. I prefer high-feedback transistor amps for bass since they don't have any transformer issues at the low end, and provide significant and consistent damping of the woofer and it's resonance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.