the schematic is only to show the principle not the details. In practice
an RC shunt across the secondary side of each transformer is necessary
plus an Rc across each diode. of course the electrolytics ( i have a bunch of huge Sprague computer/military grade ) are shunted with foil capacitors.
This way the far end where the two groud rails intersect to reference ground
is almost noise-free. Its not only switching diodes the power line has lots of noise. it is very advisable to run preamps , RIAA amps and such on accus.
I cannot see why a complementary wilson should be worse than folded cascode. You will have to use complementary types in all circumstances but even the best pairs will have differences in Ft, hFE, Cre, Cob, Cib, Early voltage. So at that point i do not agree with you.
The version you have drawn is clever in the way that you use folded cascode to acces the basis instead of emitter of the input pair.
If you choose folded cascode for the VAS, i would prefer a current source instead of the emitter resistor.
Sonnya this is more or less design philosophy. I try to keep the number of variables a minimum. I need not control what is not there. The secret of very simple topologies which quite often "sound better" than complicated is simple: they are simple. Wilson mirrors are described for DC only by a set of 5 non linear equations and these don't tell much about dynamical characteristics. I don't believe a simple simulator such as LTspice can simulate Wilsons. In general as Hiraga had noted replacing resistors with CS changes the harmonics spectrum in an undesirable manner such that odds are greater than even harmonics. Imho in terms of subjective sonic quality it doesn't matter at all whether the thd is 1% or 0.01% what matters is that the harmonics spectrum stays relatively constant to changes of load with frequency and to output amplitude.
Or put it otherwise I do not hear what I see but I might see what I hear.
Or put it otherwise I do not hear what I see but I might see what I hear.
As i see it, it is all a matter of taste. You prefer simple, and i don't care if it simple or advanced as long as it does the job.
Take for example the SSA amp in the first version. The idle current in every stage was floating all over the place. We replaced the simple bias generator made of resistors with CCS. That was keeping everything in place. Some prefer it simple, but in this case simple "is not good enough!" in my world.
The TSSA is way better as the design is changed. But they are also more complex than my amps in the last version.
I use the LTspice as a tool of guidance, and until now it has not let me down. I actually has Hyperlinx Analog, LineSim, BoardSim and Thermal in my PADS ES suite package. They are only a tool for guidance and they do not replace real life measurements.
But for the tools LTspice, Hyperlynx or what they are called you would proberly have been building hundreds of amps. before getting to the results.
Hiragas papers are good no doubt about it, but he did not have the tools we have today. And maybe he had made som changes and come to another conclusion. I dont say it is the case but he proberly would have.
It is a fact that the the footprint the distortion leaves has an impact on how we perceive the "painting" the system delivers us.
Some might like the high distortion that valve amps produce. I am not fan of it as i like it uncoloured. Actually Pix from Sweden which build my Mirand sold it as he wanted experiment with valve amps. He told me that the Valve amp produced a sweet sound but it was not as precise and uncoloured as the MIRAND A1. He does like both types of amps but it is down to the taste of the person listining to it.
Back to my designs. The actually measures as they simulate within a margin.
I do not want to tell you how you should design. But i think my knowledge for using the wilson current is very well proven both in simulation and in real life.. Not in 2 hours of work but several month!
I welcome design idears or changes but it should be documented. Or at least why it is done and not because we think so.... If we think so, it is better to say "should we try this instead?".
It is not a closed thread, dont get me wrong. But please say why you have changed it instead of saying it is stupid to use the "wilson current" mirrors.
Take for example the SSA amp in the first version. The idle current in every stage was floating all over the place. We replaced the simple bias generator made of resistors with CCS. That was keeping everything in place. Some prefer it simple, but in this case simple "is not good enough!" in my world.
The TSSA is way better as the design is changed. But they are also more complex than my amps in the last version.
I use the LTspice as a tool of guidance, and until now it has not let me down. I actually has Hyperlinx Analog, LineSim, BoardSim and Thermal in my PADS ES suite package. They are only a tool for guidance and they do not replace real life measurements.
But for the tools LTspice, Hyperlynx or what they are called you would proberly have been building hundreds of amps. before getting to the results.
Hiragas papers are good no doubt about it, but he did not have the tools we have today. And maybe he had made som changes and come to another conclusion. I dont say it is the case but he proberly would have.
It is a fact that the the footprint the distortion leaves has an impact on how we perceive the "painting" the system delivers us.
Some might like the high distortion that valve amps produce. I am not fan of it as i like it uncoloured. Actually Pix from Sweden which build my Mirand sold it as he wanted experiment with valve amps. He told me that the Valve amp produced a sweet sound but it was not as precise and uncoloured as the MIRAND A1. He does like both types of amps but it is down to the taste of the person listining to it.
Back to my designs. The actually measures as they simulate within a margin.
I do not want to tell you how you should design. But i think my knowledge for using the wilson current is very well proven both in simulation and in real life.. Not in 2 hours of work but several month!
I welcome design idears or changes but it should be documented. Or at least why it is done and not because we think so.... If we think so, it is better to say "should we try this instead?".
It is not a closed thread, dont get me wrong. But please say why you have changed it instead of saying it is stupid to use the "wilson current" mirrors.
See the difference to my version. There is 30dB less on second and third harmonic.
That is an awsome chart, Sonnya. Which circuit was it? Possibility is that the mosfet lacks sufficient drive to not sound boring. If it doesn't then this is a must build amplifier.
It is a remarkably good amplifier you have there, Sonnya.
I hope more people can find it and enjoy building it
Here is the latest schematic posted by Sonnya, attached
I hope more people can find it and enjoy building it
Here is the latest schematic posted by Sonnya, attached
Thx Sonny,
Yes, the TSSA made me realise that a few things in my SE CFB design needed some attention - but those things are more or less fixed now so I can do an preliminary comparison - final comparison can only be done when both amps have optimised PSUs - and that may take a few months.
I'll post it over in TSSA builders thread soon.
Yes, the TSSA made me realise that a few things in my SE CFB design needed some attention - but those things are more or less fixed now so I can do an preliminary comparison - final comparison can only be done when both amps have optimised PSUs - and that may take a few months.
I'll post it over in TSSA builders thread soon.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.