i did a few simulation of pavel s amp, and here are the few findings..
the amp seems to work quite well for a prototype, it need some
compensation and some work to suppress ill effects ,notably a bad
behaviour when coming from a clipping, although the phenomenon is
only evident when using a lightly overloaded square wave..
also, there s stability issues as soon as 100nF is in paralele with a load,
8R in my simulations..
the slew rate that was was claimed is real, but contrary to mr Curl
supposition, the amp is not overcompensated, to the contrary, it need
far more compensation to work properly, thus the fast slew rate is
not achievable in practice...
distorsion figures are quite good, but still in the usual range
of this topology, be it with fets or bjts...
so no breakthrough,but surely, with some tweakings and addings, it could be a great amp, and no doubt a significant improvement of the JC3..
here the fourier analysis at 1 and 10Khz , +29dbU output signal,
i used a classical lag compensation for the measures..
the amp seems to work quite well for a prototype, it need some
compensation and some work to suppress ill effects ,notably a bad
behaviour when coming from a clipping, although the phenomenon is
only evident when using a lightly overloaded square wave..
also, there s stability issues as soon as 100nF is in paralele with a load,
8R in my simulations..
the slew rate that was was claimed is real, but contrary to mr Curl
supposition, the amp is not overcompensated, to the contrary, it need
far more compensation to work properly, thus the fast slew rate is
not achievable in practice...
distorsion figures are quite good, but still in the usual range
of this topology, be it with fets or bjts...
so no breakthrough,but surely, with some tweakings and addings, it could be a great amp, and no doubt a significant improvement of the JC3..
here the fourier analysis at 1 and 10Khz , +29dbU output signal,
i used a classical lag compensation for the measures..
Hi wahab,
I appreciate your simulations, but i have real world amp and its measurements. It is neither overcompensated, nor undercompensated, it is just about right compensated. And it has no stability issues with 100nF cap load. It has these issues in simulation (not in a built sample), so probabaly models are far from reality.
On the other hand, I understand that those who do not build ask for simulation perfection.
Regards,
I appreciate your simulations, but i have real world amp and its measurements. It is neither overcompensated, nor undercompensated, it is just about right compensated. And it has no stability issues with 100nF cap load. It has these issues in simulation (not in a built sample), so probabaly models are far from reality.
On the other hand, I understand that those who do not build ask for simulation perfection.
Regards,
Last edited:
thanks for the efforts. a few questions if i may:
- can you say more about the device models used in you simulations and your simulator settings?
- what specifically lead you to conclude the amp is insufficiently compensated? i would think if this was the case, some clues would be seen in several posts of the actual amp pma built, for examples, the square wave tests in post 14 (http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier-2.html#post1993728) and phase plots in post 62 (http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier-7.html#post1995208)
- post 12 (http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier-2.html#post1993716) of the actual amp performance doesn't suggest evidence of the clipping recovery concerns, but i guess high level square or pulse waves might be better for such a test. can you say more about the conditions that caused the concerns you have regarding clipping recovery (what signal levels, what loads, etc...)?
thx,
mlloyd1
- can you say more about the device models used in you simulations and your simulator settings?
- what specifically lead you to conclude the amp is insufficiently compensated? i would think if this was the case, some clues would be seen in several posts of the actual amp pma built, for examples, the square wave tests in post 14 (http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier-2.html#post1993728) and phase plots in post 62 (http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier-7.html#post1995208)
- post 12 (http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier-2.html#post1993716) of the actual amp performance doesn't suggest evidence of the clipping recovery concerns, but i guess high level square or pulse waves might be better for such a test. can you say more about the conditions that caused the concerns you have regarding clipping recovery (what signal levels, what loads, etc...)?
thx,
mlloyd1
Last edited:
hi , pavel,
you are quite right, anyway , i saw the real world scope image
that you published, and at first look, they seems very good, indeed..
on the first hand, you are surely agree that we can t build every
amp that is published just for further verification, it would be too
time consuming..no doubt the simulators are somewhat tricky, but my
first sentence was to say that it worked quite good at first try,
and trust me, all the published design do not work well at the start in
simulators, which is quite indicating about the designs implementations..
it s enough to have an idea of the designs doability..
best regards,
wahab
hi, mlloyd
i used pavel s choices for the cascoded differentials and vas, and
different devices for the final stage :
2SA1837/2SC4793 as drivers, which i think are faster than those used in
the original drawing..
2SA1302/2SC3281 as output power devices, which are also faster
than the one in the original design...true that they are no more
produced, although toshiba replaced them with almost identicals
2SA1943/2SC5200...
as you say, and as noted in my post, i used a high level 10Khz square wave
so that the amp would enter clipping slightly....
also, i did use using half steps, in both directions, negative and positive..
anyway, i did simulate to see if the amp in the published form is doable,
and as the results suggest, it s yes, as the simulations proved very encouraging...
best regards,
wahab
yes, pavel,
although i used other devices only for the final stage,
it s true that the global balance of the amp is shifted,
but as it also well worked is a good clue about how
seriously it was designed...
anyway, if i had to go for a new design, i would
choose this one...
i must admit than although i m new at DIY ,i m old,
and i know your ideas and realizations for years, i did read
extensively your articles..too bad there is not more..
regards,
wahab
Hallo PMA, here on diyaudio is another interesting topology - the so called "Circlomos". first results promiss good quality and after optimizing and pure Class-A ultimate quality. Please send your comment there:
http://www.diyaudio.com/forums/solid-state/154388-its-cheap-its-n-its-dirty-its-circlomos.html
Speed and stability, and sound quality resulting from both
.To me, run for lowest THD is exaggerated by many, on expense of lower speed, unsymmetrical behaviour of rising edge and falling edge, hard limitation, and necessity of additional elements for stability like output coils. This is the tax for many circuit stages and race for lowest THD. I do not want to advise anyone which topology to use, on the other hand I not necessarily need to accept advices of my competitors
.
PMA... just quickly "off topic" and probably no quick answer really
Do you find that different topologies have a "characteristic" type of sound quality.
I am thinking of the classic long tailed pair input stage for example vs a single ended input topology.
"Different" in that specs or numbers can't really explain it.
So that amps using one basic topology, although they may differ significantly in sonics, they still have that "family" type sound... if that makes sense.
Do you find that different topologies have a "characteristic" type of sound quality.
I am thinking of the classic long tailed pair input stage for example vs a single ended input topology.
"Different" in that specs or numbers can't really explain it.
So that amps using one basic topology, although they may differ significantly in sonics, they still have that "family" type sound... if that makes sense.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.