Hugh,
Thanks for your comments.
I am aware that using a differential circuit would cancel even harmonics and H3, 5 would then become dominant. As I have mentioned earlier, I have built and listened to single-ended JLH's (the latest version from Geoff) and then bridged two of those as a differential amp. Though they were not Xed as in Nelson Pass XA's but simply driven by a phase splitter as published by JLH i n his 1970 "letter to the editor", I expected that H2 would still be cancelled to a large extent.
And I agree with Nelson Pass, who noted in one of his threads that he was surprised that people prefer H3 to H2. Well at least I do. : )
Shall build in due course.
Cheers,
Patrick
Thanks for your comments.
I am aware that using a differential circuit would cancel even harmonics and H3, 5 would then become dominant. As I have mentioned earlier, I have built and listened to single-ended JLH's (the latest version from Geoff) and then bridged two of those as a differential amp. Though they were not Xed as in Nelson Pass XA's but simply driven by a phase splitter as published by JLH i n his 1970 "letter to the editor", I expected that H2 would still be cancelled to a large extent.
And I agree with Nelson Pass, who noted in one of his threads that he was surprised that people prefer H3 to H2. Well at least I do. : )
Shall build in due course.
Cheers,
Patrick
Brian Donegan, i will scratch, but in reality i am waiting that kind of circuit too.
I have 43 of Graham designs, he is very creative and produces at least one circuit each day, all working circuits.
And i am always asking him to send me his "brain storms" ideas, something that is boiling into his head all the time.
Graham normally have headaches, i suppose one of the reasons is that brain is trying to grow, and intra cranian box, the Skull is limiting that.
I am waiting a single ended, double or triple darlington, with transformer output or other coupling...waiting that circuit from him, already told him.
He is researching and comparing all kind of circuits, keeping class AB with high bias...so starting class A and go on till reach some ten or 25 watts, and keeping over bias til maximum power.
I try to find the schematic you imagined...unfortunattely i could not find it.
But related that clever man, Graham, this is a matter of time, as circuits are blowing there alike a popcorn inside a can.
I have a lot of PNPs... and searching some to make them work... widow from burned NPNs.
Yesterday, Graham sent me a circuit with 0.000001567 percent distortion...that's interesting. (not finished and not published)
regards,
Carlos
I have 43 of Graham designs, he is very creative and produces at least one circuit each day, all working circuits.
And i am always asking him to send me his "brain storms" ideas, something that is boiling into his head all the time.
Graham normally have headaches, i suppose one of the reasons is that brain is trying to grow, and intra cranian box, the Skull is limiting that.
I am waiting a single ended, double or triple darlington, with transformer output or other coupling...waiting that circuit from him, already told him.
He is researching and comparing all kind of circuits, keeping class AB with high bias...so starting class A and go on till reach some ten or 25 watts, and keeping over bias til maximum power.
I try to find the schematic you imagined...unfortunattely i could not find it.
But related that clever man, Graham, this is a matter of time, as circuits are blowing there alike a popcorn inside a can.
I have a lot of PNPs... and searching some to make them work... widow from burned NPNs.
Yesterday, Graham sent me a circuit with 0.000001567 percent distortion...that's interesting. (not finished and not published)
regards,
Carlos
Hi Carlos-
I know what you mean. I have seen posts of so many of Graham's designs and tweaks it confuses me to think about them. When I posted, I was confused. I have two of the 2565s, and I had initially looked at one of the 5200 designs thinking, gee, I have these two free PNPs, I'll make this amp. But I need 4, not 2, unless I want mono. Anyway, mix that all up ans wait a few months, and I came out thinking there was a single PNP design, but then I came to my senses. I think...
I know what you mean. I have seen posts of so many of Graham's designs and tweaks it confuses me to think about them. When I posted, I was confused. I have two of the 2565s, and I had initially looked at one of the 5200 designs thinking, gee, I have these two free PNPs, I'll make this amp. But I need 4, not 2, unless I want mono. Anyway, mix that all up ans wait a few months, and I came out thinking there was a single PNP design, but then I came to my senses. I think...
Even Harmonics
Hugh,
May I correct myself.
Actually, if you were to assume that the devices on the positive and negative rails are identical, and that they were driven perfectly out of phase by 180 degrees, then the even harmonics would already be canceled. In practice, this is never perfect of course.
And I would agree that a bridged version with left and right perfectly matched (rather than top and bottom) would also be odd harmonics dominant.
Patrick
Hugh,
May I correct myself.
Actually, if you were to assume that the devices on the positive and negative rails are identical, and that they were driven perfectly out of phase by 180 degrees, then the even harmonics would already be canceled. In practice, this is never perfect of course.
And I would agree that a bridged version with left and right perfectly matched (rather than top and bottom) would also be odd harmonics dominant.
Patrick
Hi Patrick,
Yes, indeed, but the $64K question is this:
Will the symmetrical, bridge/balanced system, with its odd harmonic dominant distortion spectrum, sound better than the single ended version? And will this 'better' in fact be a matter of taste?
It will certainly sound different. In fact you can restore some of the even harmonic by making each side subtly different, by changing the operating point slightly on one side of the bridge. This enables you to shift the harmonic spectrum towards even order, and permits use of lower voltage rails for the same power, no bad thing.......
Cheers,
Hugh
Yes, indeed, but the $64K question is this:
Will the symmetrical, bridge/balanced system, with its odd harmonic dominant distortion spectrum, sound better than the single ended version? And will this 'better' in fact be a matter of taste?
It will certainly sound different. In fact you can restore some of the even harmonic by making each side subtly different, by changing the operating point slightly on one side of the bridge. This enables you to shift the harmonic spectrum towards even order, and permits use of lower voltage rails for the same power, no bad thing.......
Cheers,
Hugh
> And will this 'better' in fact be a matter of taste?
Indeed. And I do know people who prefer a not-too-well-matched single-ended JLH to bridged, or an Aleph to Aleph-X.
But then I thought I used the word "prefer" rather than better, as I knew these things are very subjective. : )
Referring back to my previous post #1186. Not having any simulation software, I tried calculating the open loop response of the output stage of the JLH with measured transistor characteristics in an Excel sheet, and observed that ideally matched transistors with a perfect phase splitter driver stage will actually cancel all even harmonics at the loudspeaker output (in the single ended configuration). The dynamic current contribution from both transistors would vary according to output voltage (or rather output current, as I assumed a pure resistive load of 4 ohms), but the loudspeaker would not be able to tell ?? So I wonder whether the difference in a JLH is as great as one thinks. For the experiments I did, the SE version is slightly warmer, but only slightly. I liked the transparent sound of the bridged more.
The above does not apply of course to Aleph, as the Aleph current source is basically a current multiplier.
Is my interpretation correct, or did I miss something ?
Patrick
Indeed. And I do know people who prefer a not-too-well-matched single-ended JLH to bridged, or an Aleph to Aleph-X.
But then I thought I used the word "prefer" rather than better, as I knew these things are very subjective. : )
Referring back to my previous post #1186. Not having any simulation software, I tried calculating the open loop response of the output stage of the JLH with measured transistor characteristics in an Excel sheet, and observed that ideally matched transistors with a perfect phase splitter driver stage will actually cancel all even harmonics at the loudspeaker output (in the single ended configuration). The dynamic current contribution from both transistors would vary according to output voltage (or rather output current, as I assumed a pure resistive load of 4 ohms), but the loudspeaker would not be able to tell ?? So I wonder whether the difference in a JLH is as great as one thinks. For the experiments I did, the SE version is slightly warmer, but only slightly. I liked the transparent sound of the bridged more.
The above does not apply of course to Aleph, as the Aleph current source is basically a current multiplier.
Is my interpretation correct, or did I miss something ?
Patrick
Discussion
Open for discussion, no hate-mail please:
- This nifty but not yet perfect DP-input should allready lower
overall THD to the lower 10th under every condition.
- With Rx, the Dc-offset point can be reached right into the
ballpark.
- MJ11016 should be a very, very good replacement for the
MJ15003.
Who dares to tell us the different in quality and sound?
Open for discussion, no hate-mail please:
- This nifty but not yet perfect DP-input should allready lower
overall THD to the lower 10th under every condition.
- With Rx, the Dc-offset point can be reached right into the
ballpark.
- MJ11016 should be a very, very good replacement for the
MJ15003.
Who dares to tell us the different in quality and sound?
Attachments
> I seem to have missed something....
Yes. You missed the picture.
> what DP-input ?
Differential Pair. See also e.g. Post #1167 from yours truely.
> MJ11016 should be a very, very good replacement for the MJ15003.
Can you give your reasons ?
Purely from measurement, I would prefer MJ21194 -- much more constant hfe, and not too high bandwidth as Sanken 2SC2922. But not built anything yet, so no subjective comments.
Patrick
Yes. You missed the picture.
> what DP-input ?
Differential Pair. See also e.g. Post #1167 from yours truely.
> MJ11016 should be a very, very good replacement for the MJ15003.
Can you give your reasons ?
Purely from measurement, I would prefer MJ21194 -- much more constant hfe, and not too high bandwidth as Sanken 2SC2922. But not built anything yet, so no subjective comments.
Patrick
MJ110016
> MJ11016 should be a very, very good replacement for the >MJ15003.
>Can you give your reasons ?
>Purely from measurement, I would prefer MJ21194 -- much more >constant hfe, and not too high bandwidth as Sanken 2SC2922. >But not built anything yet, so no subjective comments.
The behaviour in simulation for every circuit, I put my hands on was just outstanding.
My amp is not finished yet, so I have to depend on a second opinion or somebody who allready tried.
Michael
> MJ11016 should be a very, very good replacement for the >MJ15003.
>Can you give your reasons ?
>Purely from measurement, I would prefer MJ21194 -- much more >constant hfe, and not too high bandwidth as Sanken 2SC2922. >But not built anything yet, so no subjective comments.
The behaviour in simulation for every circuit, I put my hands on was just outstanding.
My amp is not finished yet, so I have to depend on a second opinion or somebody who allready tried.
Michael
Hi Michael,
I'm running un-mirrored input differential myself at the moment.
Notes;-
If your psu is not regulated you can get hum through that tail resistor. Put another resistor in seies with the tail resistor (choose to drop about 0.6V), connect a smoothing capacitor, and fit diodes as in my circuit to prevent power-up thump. Result - silent power-up (with dual rails) and background.
Watch out for instability due to the extra stage of rf phase change - this does not show up adequately through simulation. I cured this by fitting 10nF across the feedback base-emitter differential transistor and adding the rf input filtering. The input bandwidth is still wide open but not troubled by clicks etc. No other stabilisation components needed here.
I have made many A-B comparisons between single input and differential pairs over the years, and I prefer the improved output zero control. If differential pair usage increases the open loop gain then the distortion is reduced and the loudspeaker damping increased; both aspects are worthwhile, but not always audible when compared to a single input. Where the differential input really scores though is when you turn the volume up, because the amplifier remains much more stable and does not make the loudspeaker cone 'breathe' the same as when input bias shifts with the non-linear driver current draw.
Good luck with your construction.
Cheers ......... Graham.
I'm running un-mirrored input differential myself at the moment.
Notes;-
If your psu is not regulated you can get hum through that tail resistor. Put another resistor in seies with the tail resistor (choose to drop about 0.6V), connect a smoothing capacitor, and fit diodes as in my circuit to prevent power-up thump. Result - silent power-up (with dual rails) and background.
Watch out for instability due to the extra stage of rf phase change - this does not show up adequately through simulation. I cured this by fitting 10nF across the feedback base-emitter differential transistor and adding the rf input filtering. The input bandwidth is still wide open but not troubled by clicks etc. No other stabilisation components needed here.
I have made many A-B comparisons between single input and differential pairs over the years, and I prefer the improved output zero control. If differential pair usage increases the open loop gain then the distortion is reduced and the loudspeaker damping increased; both aspects are worthwhile, but not always audible when compared to a single input. Where the differential input really scores though is when you turn the volume up, because the amplifier remains much more stable and does not make the loudspeaker cone 'breathe' the same as when input bias shifts with the non-linear driver current draw.
Good luck with your construction.
Cheers ......... Graham.
Attachments
Graham, thanks for your input, which does help a lot.
The final goal will be a current-mirror, but first I have to learn how to calculate these thing and more important, how it will be affect the rest of the JLH and if the JLH has to be modified, what I really don´t want to do.
Cheers
Michael
The final goal will be a current-mirror, but first I have to learn how to calculate these thing and more important, how it will be affect the rest of the JLH and if the JLH has to be modified, what I really don´t want to do.
Cheers
Michael
Mj11016
I had a look at the datasheet out of curiosity today.
MJ11016 is a NPN darlington with hfe up to 1000 and bandwidth of 4 MHz. Which is great.
The only problem I have with it is that the hfe varies between 0.3A and 3A by more than a factor of 6. Not exactly very good linearity.
My naive understanding of circuits like JLH or Aleph-X is that you want to maximise open loop linearity as much as you can, and only do minimum correction by NFB.
In which case I would argue for MJ21194 due to intrinsic linearity, and make up for the 10 times lower hfe (in comparison) with high hfe devices in the second stage.
Perhaps Graham would care to shed some light on the discussion ??
Cheers,
Patrick
I had a look at the datasheet out of curiosity today.
MJ11016 is a NPN darlington with hfe up to 1000 and bandwidth of 4 MHz. Which is great.
The only problem I have with it is that the hfe varies between 0.3A and 3A by more than a factor of 6. Not exactly very good linearity.
My naive understanding of circuits like JLH or Aleph-X is that you want to maximise open loop linearity as much as you can, and only do minimum correction by NFB.
In which case I would argue for MJ21194 due to intrinsic linearity, and make up for the 10 times lower hfe (in comparison) with high hfe devices in the second stage.
Perhaps Graham would care to shed some light on the discussion ??
Cheers,
Patrick