Thanks SY. That seems to be a good direction to go into. By the way do you have the book ? How is it ?
So assuming all else is same(tubes, design, etc..) there really isnt much difference between the way one sounds than the other ?
I really wanted to get some impressions from people who have had a chance to listen to these types of output stages.
So assuming all else is same(tubes, design, etc..) there really isnt much difference between the way one sounds than the other ?
I really wanted to get some impressions from people who have had a chance to listen to these types of output stages.
It's a GREAT book, one that any tube aficionado should own.
As for sonic differences, there are good and bad amps out there with all of those topologies (and a few more you haven't mentioned). The universal best choice is... well, it's more a matter of faith than of actual data. Maybe that's why there's so many different kinds of amps.
UL vs Triode PP
I can give my own personal views ONLY based upon -
1) a EL84 Push Pull Amp I built and have run in Ultralinear and "strapped as Triodes". Hi quality 1970's EI Output transformers.
2) A prototype monoblock using 4 x KT88 driving Plitron PAT4006 Toroidal Output Transformers.
You are trying to balance a number of issues:
For low distortion the impedance of the Output Transformers Primary Inductance needs to be much higher than the output impedance of the output tubes. At low frequencies this is often not the case and distortion (2nd Harmonic in particular) rises. When this is excessive it gives the amp a "covered" sound. To get over this you need to do one of 2 things.
a) Get an output transformer with higher primary inductance
OR
b) Lower the output impedance of the output tubes.
Most people opt for the second solution as it has an additional benefit that the reduced output impedance of the tubes reflect to the secondary of the output transformer as lower amplifier output impedance giving you a better damping factor.
The tube output impedance is about 2.5 times less in strapped as triodes mode than in ultalinear mode.
With the little EL84 Amp (10W per channel) I run in Ultralinear Mode and put up with a slightly covered (but glorious) sound. I tried it in strapped as triodes mode but it didn't have enough power.
With the big 4 x KT88 Amp power wasn't an issue and the output transformer has a massive 590H primary inductance so I wasn't chasing impedance issues and could compare the sounds more easily.
The Ultralinear is lightening fast and has very a "live" feel.
The strapped as triodes mode is subjectively "slow" which is a complaint that some make about Single Ended Triode Amps too.
I prefer the Ultralinear Mode BUT the damping factor is NOT as good as I would wish and some speaker colouration is evident at higher frequencies (gets stronger as the amp delivers more power imto the higher impedance load at higher frequencie)
All of the above should have one other caveat. I run all my amps with zero global feedback. This means that there is no extra reduction in output impedance of the amp and so damping factor is usually an issue. I have found that any global feedback destroys time coherency and therefore sound stage (stereo imaging) - actually up to about 5 dB is sometimes an acceptable compromise.
I am currently running the 4 x KT88 in strapped as triodes mode (even though I like the pace less than ultralinear) as a compromise to get better damping. In Ultralinear Mode it just a little too raw. In the mean time I have ordered a pair of VDV2100-CFB/H Output Transformers to try inplace of the PAT4006. The VDV2100-CFB/H have 30% Ultralinear Taps and 10% Cathode Feedback whereas the PAT4006 I have now have only 40% ultralinear.
My target - ultralinear plus cathode feedback
A bit of a rave - BUT I hope there is something useful in there.
Cheers
I can give my own personal views ONLY based upon -
1) a EL84 Push Pull Amp I built and have run in Ultralinear and "strapped as Triodes". Hi quality 1970's EI Output transformers.
2) A prototype monoblock using 4 x KT88 driving Plitron PAT4006 Toroidal Output Transformers.
You are trying to balance a number of issues:
For low distortion the impedance of the Output Transformers Primary Inductance needs to be much higher than the output impedance of the output tubes. At low frequencies this is often not the case and distortion (2nd Harmonic in particular) rises. When this is excessive it gives the amp a "covered" sound. To get over this you need to do one of 2 things.
a) Get an output transformer with higher primary inductance
OR
b) Lower the output impedance of the output tubes.
Most people opt for the second solution as it has an additional benefit that the reduced output impedance of the tubes reflect to the secondary of the output transformer as lower amplifier output impedance giving you a better damping factor.
The tube output impedance is about 2.5 times less in strapped as triodes mode than in ultalinear mode.
With the little EL84 Amp (10W per channel) I run in Ultralinear Mode and put up with a slightly covered (but glorious) sound. I tried it in strapped as triodes mode but it didn't have enough power.
With the big 4 x KT88 Amp power wasn't an issue and the output transformer has a massive 590H primary inductance so I wasn't chasing impedance issues and could compare the sounds more easily.
The Ultralinear is lightening fast and has very a "live" feel.
The strapped as triodes mode is subjectively "slow" which is a complaint that some make about Single Ended Triode Amps too.
I prefer the Ultralinear Mode BUT the damping factor is NOT as good as I would wish and some speaker colouration is evident at higher frequencies (gets stronger as the amp delivers more power imto the higher impedance load at higher frequencie)
All of the above should have one other caveat. I run all my amps with zero global feedback. This means that there is no extra reduction in output impedance of the amp and so damping factor is usually an issue. I have found that any global feedback destroys time coherency and therefore sound stage (stereo imaging) - actually up to about 5 dB is sometimes an acceptable compromise.
I am currently running the 4 x KT88 in strapped as triodes mode (even though I like the pace less than ultralinear) as a compromise to get better damping. In Ultralinear Mode it just a little too raw. In the mean time I have ordered a pair of VDV2100-CFB/H Output Transformers to try inplace of the PAT4006. The VDV2100-CFB/H have 30% Ultralinear Taps and 10% Cathode Feedback whereas the PAT4006 I have now have only 40% ultralinear.
My target - ultralinear plus cathode feedback
A bit of a rave - BUT I hope there is something useful in there.
Cheers
@percy
www.audioxpress.com/resource/audioclass/
It's not the book, but it's a good read by Norman Crowhurst.
Nine PDF files, Audio Classroom: Desiging Your Own Amplifier
Wayne
www.audioxpress.com/resource/audioclass/
It's not the book, but it's a good read by Norman Crowhurst.
Nine PDF files, Audio Classroom: Desiging Your Own Amplifier
Wayne
Mr. Stan White, who designed and marketed the POWRTRON amplifier in the early 1950s, claims that a triode is a voltage ampilifer, a pentode is a current amplifier and an ultralinear connected pentode is a power amplifier. He says that only a power amplifier is suitable to drive a dynamic load such as loudspeaker.
He also had a few scathing things to say about D.T.N. Williamson's amp design. If you're interested, you can read more of Stan's opinions at this link.
He also had a few scathing things to say about D.T.N. Williamson's amp design. If you're interested, you can read more of Stan's opinions at this link.
This article by Stan White is full of misinformation and show a lack of technical knowledge.
Some things...
In the article it says that Williamson suggested a 2 chassis design as he thought that this would be reduce hum, this is not true. In the original article in Wireless World Williamson described a single chassis with amplifier and power supply, in the 2nd follow-up article he also described an alternative design using 2 chassis but this is not described as superiour and is not specifically recommended.
In the article an amplifier with 807's and "an American version of the Williamson" are described. These amplifiers are not Williamson amplifiers, they could be based on the Williamson circuit but was never designed by Williamson. 807's even though they are very good tubes are far away in performance from the KT66's used in the original.
The description of the split load phase inverter used in the Williamson repeats the same common mistake that is abundant in descriptions of this circuit, that is claiming that output impedances from anode and cathode are different.
This is not correct and has been described many times, for instance by Morgan Jones in his book, a more thorough analysis can be found here http://www.aikenamps.com/cathodyne.pdf
The Williamson amplifier of course have its weak points that can be improved but I dont think these kind of articles with many errors can help anyone.
Regarding Stan Whites claim that an amplifier should have what he call "power feedback" the fact is that most modern speakers are designed to be driven by a voltage source with low impedance, it is of course possible that it would be better to design speakers and amplifiers in other ways but with comercially available speakers I believe that a low impedance amplifier, (with voltage feedback) is the best solution.
It is noticeable that even though it is 50 years since the Powertron came on the market no one else have presented a similar amplifier, if it would have been soo good as Stan White say it is I think many companies would have taken the chance to copy the concept....
Regards Hans
Some things...
In the article it says that Williamson suggested a 2 chassis design as he thought that this would be reduce hum, this is not true. In the original article in Wireless World Williamson described a single chassis with amplifier and power supply, in the 2nd follow-up article he also described an alternative design using 2 chassis but this is not described as superiour and is not specifically recommended.
In the article an amplifier with 807's and "an American version of the Williamson" are described. These amplifiers are not Williamson amplifiers, they could be based on the Williamson circuit but was never designed by Williamson. 807's even though they are very good tubes are far away in performance from the KT66's used in the original.
The description of the split load phase inverter used in the Williamson repeats the same common mistake that is abundant in descriptions of this circuit, that is claiming that output impedances from anode and cathode are different.
This is not correct and has been described many times, for instance by Morgan Jones in his book, a more thorough analysis can be found here http://www.aikenamps.com/cathodyne.pdf
The Williamson amplifier of course have its weak points that can be improved but I dont think these kind of articles with many errors can help anyone.
Regarding Stan Whites claim that an amplifier should have what he call "power feedback" the fact is that most modern speakers are designed to be driven by a voltage source with low impedance, it is of course possible that it would be better to design speakers and amplifiers in other ways but with comercially available speakers I believe that a low impedance amplifier, (with voltage feedback) is the best solution.
It is noticeable that even though it is 50 years since the Powertron came on the market no one else have presented a similar amplifier, if it would have been soo good as Stan White say it is I think many companies would have taken the chance to copy the concept....
Regards Hans
Split Load Phase Inverter
In any discussion of the split load phase inverter sooner or latter the article of Mr. Preisman is sited.
His prime arguement is that so long as the load impedances on the anode and cathode are equal the outputs will be equal in phase and magnitude REGARDLESS of the differing source impedances.
I'm about to sit down and work through this paper - for the second time, I bailed out half way thru' last time.
Has anyone done this - i.e. is there INDEPENDENT confirmation of Mr Preisman's article anywhere.
PLEASE - The last time I saw this discussed on a forum similar to this one, the debate degenerated into heated personal abuse and no conclusions were reached. Lets not repeat that here.
I confirm that Morgan Jones does'nt agree with Preisman and advocates the use of a "build out" resistor in the cathode output to equalise the source impedances. I built his little EL84 Amp which uses the split load phase inverter. In the end I connected a dual trace oscilloscope to the anode and cathode (to keep the loads equal) and then adjusted the "build out" resistor in the cathode output empirically to give exact balance at 120kHz. The resistor value I ended up with was about 1/5th the value that my calculations of the source impedances suggested would be required.
I guess that puts me in "CONFUSED" camp.
In any discussion of the split load phase inverter sooner or latter the article of Mr. Preisman is sited.
His prime arguement is that so long as the load impedances on the anode and cathode are equal the outputs will be equal in phase and magnitude REGARDLESS of the differing source impedances.
I'm about to sit down and work through this paper - for the second time, I bailed out half way thru' last time.
Has anyone done this - i.e. is there INDEPENDENT confirmation of Mr Preisman's article anywhere.
PLEASE - The last time I saw this discussed on a forum similar to this one, the debate degenerated into heated personal abuse and no conclusions were reached. Lets not repeat that here.
I confirm that Morgan Jones does'nt agree with Preisman and advocates the use of a "build out" resistor in the cathode output to equalise the source impedances. I built his little EL84 Amp which uses the split load phase inverter. In the end I connected a dual trace oscilloscope to the anode and cathode (to keep the loads equal) and then adjusted the "build out" resistor in the cathode output empirically to give exact balance at 120kHz. The resistor value I ended up with was about 1/5th the value that my calculations of the source impedances suggested would be required.
I guess that puts me in "CONFUSED" camp.
I think that Morgan Jones changed the text in later editions of his book, in the late edition I have he explains that the output impedances are equal except for very high frequencies where there is an unbalance due to different capacitances, but this doesn't come into effect until very high frequencies like several MHz with a 6SN7, (this unbalance can also be completely corrected by adding a small capacitor as described in "Vacuum tube amplifiers" by Valley and Wallman).
I am not either going to be involved into a debate about Preismans text, but he explains why the output impedances are equal as output impedances are usually defined and measured.
If you have Spice it is very easy to confirm that Preisman, (and Morgan Jones's later text) is correct both for the equality of the impedances and the unexpectedly low value of it.
I have done Spice simulations and also built split-load inverters with various tubes where I have confirmed both the value of output impedance and that they are indeed equal.
However I can't explain why you come to a different result with your measurements.
Regards Hans
I'm not surprised at your reaction to Stan's article, nor do I disagree with you; I only referred to it because I was entertained by what he had to say and I thought others might be too. It did make me interested, for a while, in the Van Scoyoc (mis-spelled by Stan) cross-coupled phase splitter design, until I realized that nobody else seemed to think much of it.
There seems to have been a bit of sniping going on in those days, including Stan White's attacks on electrical engineers in general and on D.T.N. Williamson in particular, plus Williamson's own article decrying the "so-called ultralinear" concept of Hafler & Keroes. Professional jealousy, perhaps?
It's pretty easy to measure for yourself that the split-load really does balance and that White is full of it. BUT... you absolutely MUST ensure that the plate and cathode loads are equal and that means buffering BOTH identically and probing at the output of the buffer- sticking most scope probes right on the cathode or plate will unbalance things sufficiently to throw you off. That implies that the balance will only hold if the load is a class A stage; either a buffer, class A diff amp, or class A output stage.
Conceptually, you're forced to have cathode and plate balance because of Kirchoff and Ohm's laws.
Conceptually, you're forced to have cathode and plate balance because of Kirchoff and Ohm's laws.
I just read Stan White's document and here are the errors I found just from the top of my head.
1. There was a pure Williamson amplifier offered on the American market - the Radio Craftsmen 500.
2. Williamson did indeed work for a tube company when he designed the amplifier but he was actually working for the Ferranti transformer company by the time the article appeared in Wireless World.
3. Marconi Osram did at the time produce a triode similar (and better) than the 2A3 - the PX4. Williamson in the article states that the KT66 was chosen because of low cost and availability and its triode-connected characteristics were similar to the PX4.
I am sure there are other errors in the article. The basic improvements one can make in the Williamson are simply changing the operating points on the 6SN7's to a more linear region of the tube's characteristics.
John
1. There was a pure Williamson amplifier offered on the American market - the Radio Craftsmen 500.
2. Williamson did indeed work for a tube company when he designed the amplifier but he was actually working for the Ferranti transformer company by the time the article appeared in Wireless World.
3. Marconi Osram did at the time produce a triode similar (and better) than the 2A3 - the PX4. Williamson in the article states that the KT66 was chosen because of low cost and availability and its triode-connected characteristics were similar to the PX4.
I am sure there are other errors in the article. The basic improvements one can make in the Williamson are simply changing the operating points on the 6SN7's to a more linear region of the tube's characteristics.
John
Hans
<
It is noticeable that even though it is 50 years since the Powertron came on the market no one else have presented a similar amplifier, if it would have been soo good as Stan White say it is I think many companies would have taken the chance to copy the concept....
>
Hello Hans,
You are wrong. The best tube amplifiers made by AUDIO RESEARCH the Powrtron power-feedback topology.
see Jean Hiraga: "Initiation aux amplis à tubes"
Cheers,
PP
<
It is noticeable that even though it is 50 years since the Powertron came on the market no one else have presented a similar amplifier, if it would have been soo good as Stan White say it is I think many companies would have taken the chance to copy the concept....
>
Hello Hans,
You are wrong. The best tube amplifiers made by AUDIO RESEARCH the Powrtron power-feedback topology.
see Jean Hiraga: "Initiation aux amplis à tubes"
Cheers,
PP
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.