My tentative conclusion from all the SIM's done recently (hundreds, perhaps) is that for SOTA designs 200-300v/usec is more in the ball park.
THx-RNMarsh
THx-RNMarsh
Last edited:
The good doctor Leach (GA TECH) told me that the slew rate of an amp is set at the input. Has anyone here seriously looked at the design of a M. Levinson 33 amplifier. To me it looks like a CFA but has obviously a VFA driver (front end). It's probably against policy to post the schematic without their lawyers shaking my tree. Isn't that we all are striving for, albiet with a little more simplicity. The thing has 20 complementary output transistors. I am not in the same league with you designing guys and I'm just an audiophile and I'm really hesitant about speaking up. Ray
May I ask how did you reach this (tentative) conclusion from all these SIMs? Or was the (tentative) conclusion pre-drawn 😀?
😎🙂
Hi John,
Can you divulge a new, better number than .5-1v/usec or SR for SOTA designs?
THx-Richard
Last edited:
It seems that I have seen slew rate related to the musics absolute slew rate requirement many times. I think a better question seems to be how does the slew rate of an amplifier relate to the maximum needed slew rate as determined by the music requirement. Does the sound improve simply by having an unrelated slew rate in the power amplifier, does a very high slew rate improve the final reproduction of the music or does the slew rate have more to do with the phase response and SOA of the amplifier or is this like having a 1000hp engine in a car to drive to the corner Quicky Mart that is only 400 meters away?
That is the way it has been viewed in the past, also. My experience with VFa and CFA is different..... to get the super low THD at 20KHz/full power, you inevitably end up with a higher SR than that. It is an indirect consequence. I am fairly certain that if one would design primarily for high SR that you would not always end up with a low THD amp. But that shouldnt preclude us from a SR number that better reflects a SOTA design.
But I am now most interested - when more get built from the SIM's here - is the listening experience of these SOTA designs.
THx-RNMarsh
But I am now most interested - when more get built from the SIM's here - is the listening experience of these SOTA designs.
THx-RNMarsh
Last edited:
Richard,
If the result of a low distortion amplifier is a high slew rate then so be it. As you say as long as you can reproduce the highest frequencies with low to almost nonexistent distortion mechanisms isn't that the ultimate goal?
The new Pono music player is already getting into this type of discussion, do we need a bit rate much higher than a good CD or do we need engineers to actually record the information without messing things up in the process? Taking away the dynamic range in order to increase the average sound level is just a stupid use of limiter/compression technology. It doesn't improve the music, it just gets into the loudness wars.
I want to hear about the sound more than about some arbitrary slew rate wars, do these different designs that OS has created sound better due to the extremely low THD numbers or is it the devices themselves and how they are strung together that is really making the difference we are looking for? Does the slew rate over a real requirement have anything to do with it.
If the result of a low distortion amplifier is a high slew rate then so be it. As you say as long as you can reproduce the highest frequencies with low to almost nonexistent distortion mechanisms isn't that the ultimate goal?
The new Pono music player is already getting into this type of discussion, do we need a bit rate much higher than a good CD or do we need engineers to actually record the information without messing things up in the process? Taking away the dynamic range in order to increase the average sound level is just a stupid use of limiter/compression technology. It doesn't improve the music, it just gets into the loudness wars.
I want to hear about the sound more than about some arbitrary slew rate wars, do these different designs that OS has created sound better due to the extremely low THD numbers or is it the devices themselves and how they are strung together that is really making the difference we are looking for? Does the slew rate over a real requirement have anything to do with it.
These are some of the questions that fall out of What is a CFA? And what characteristics does it have? And to what degree are they important? We put numbers and faith in THD, IM, Freq response, dB's, Q etc. SR is another one of those numbers. By itself it will make a diffference at some threshold level. For SOTA that threshold is the question and as nothing exists alone, it is correlated to other factors and their numbers.
If reading thru a spec sheet, one wants to have a 'feel' for those numbers significance in planning your selection.
As JC said, all else equal.... higher SR is better.
FWIW, my comfort zone is at least 2-3 hundred v/sec (for SOTA).
THx-RNMarsh
Last edited:
WOW !! 😱
Where did they get that IPS ?? (below)
Its "smokin' " fast and very easy to work with.
It's VFA .... but I could make it CFA 😀 try em' both....
OS
Attachments
This is an invention of mine. I have not seen a circuit like this before. 😎
I can not estimate how high the potential is ... you have to look.
Have fun with it 😉
Ø-stripper, will it be possible to drive it differentially and then current feedback the one leg of the pair..??
Slew rate corresponds to the RF current limit of the frontend. The higher the slew rate, the further this stage is from clipping with a given RF signal.
With transistors in class A, distortion is roughly proportional to loading, so less loading means less distortion in a frontend. Assuming that this distortion mechanism is not buried in Early distortion, which is almost always the case for 1KHz.
This effect is mashed together with all the other frequency-dependent distortion mechanisms in an amp. For instance you may have an amp that has a very high drive ability for the VAS, resulting in high slew rate, but the capacitance that's loading that stage may be highly nonlinear, for instance directly driving FET gates. So slew rate doesn't directly correlate with HF linearity.
Current on demand is a way to increase slew rate, but not necessarily a way to increase linearity with audio signals, if the only effect is to raise the frontend current limit.
With transistors in class A, distortion is roughly proportional to loading, so less loading means less distortion in a frontend. Assuming that this distortion mechanism is not buried in Early distortion, which is almost always the case for 1KHz.
This effect is mashed together with all the other frequency-dependent distortion mechanisms in an amp. For instance you may have an amp that has a very high drive ability for the VAS, resulting in high slew rate, but the capacitance that's loading that stage may be highly nonlinear, for instance directly driving FET gates. So slew rate doesn't directly correlate with HF linearity.
Current on demand is a way to increase slew rate, but not necessarily a way to increase linearity with audio signals, if the only effect is to raise the frontend current limit.
Hi Richard,
That 300 v/us was gravy, and way more than needed for a 50W amplifier. My point was that it is not always difficult to achieve high slew rates with VFA.
That large slew rate was due in part to the use of MIC compensation, which I wanted to use for other reasons as well (superior PSRR and very compatible with the push-pull VAS I was using).
It is a fair question to ask what distortion level I would have achieved if I had not used MIC and gotten 300V/us. I'm guessing it still would have been almost as good (<0.001% at 20kHz full power), given the use of HEC with MOSFET outputs.
BTW, that amp used a single pair of vertical MOSFETs biased at 150mA. A larger amplifier up to, say 150W, would have used two pair of MOSFETs, each biased at 150mA (or more), yielding a larger class A region and only half the amount of transconductance droop (the primary cause of crossover distortion in MOSFET output stages).
Cheers,
Bob
WOW !! 😱
Where did they get that IPS ?? (below)
Its "smokin' " fast and very easy to work with.
It's VFA .... but I could make it CFA 😀 try em' both....
OS
I would make R15-R18 as high as reasonable, since they improve the current sources and are a cheap gain. As I recall you could even short the current sources this way because the buffer nonlinearities are low and they cancel as well.
I noticed as a CFB amp , it IS your "bog standard" NAD/NX (below).
All the inverting side does is V/I conversion for VFA.
In VFA mode (all 8 devices) , it is a self contained CFB "module" ...
as each V/I feeds current to the other "side". 😎
They match each others performance (VFA/CFA).
PS- I have to research the sansui "diamond drive" ... to see the
OEM setup.
Attachments
Last edited:
Bear in mind that high slew rate stresses all three stages in an amplifier, the IPS, VAS and OPS. It is often in the form of needing to deliver high current to charge capacitances. Current "clipping" or similar shortcomings can reduce achievable slew rate. In traditional Miller-compensated amplifiers, it is often determined by the amount of current the input stage can deliver to charge or discharge a capacitor, often the Miller compensation capacitor.
However, the VAS can slew rate limit as well, if it cannot deliver enough current to drive the VAS output node capacitance and the output stage. This can definitely be the case with a simple Darlington output stage (especially if that output stage does not have really fast transistors). A traditional single-ended VAS is limited in its ability to slew in at least one direction, and can suffer from asymmetrical slew capability. Here is where a push-pull VAS performs much better. It can deliver at least twice as much peak current as a single-ended VAS with the same quiescent current.
Many output stages become stressed when they must deliver a high-rate-of-change output current. Slew rate into a 4-ohm load or even a 2-ohm load will be important. Cross-conduction in an output stage under high di/dt conditions can lead to distortion. In this respect, high-speed output transistors and driver circuits that are good at sucking current out of the base are very helpful.
So, an amplifier that can comfortably deliver high slew rate will usually be further away from these sorts of current stresses on the IPS, VAS and OPS.
Cheers,
Bob
However, the VAS can slew rate limit as well, if it cannot deliver enough current to drive the VAS output node capacitance and the output stage. This can definitely be the case with a simple Darlington output stage (especially if that output stage does not have really fast transistors). A traditional single-ended VAS is limited in its ability to slew in at least one direction, and can suffer from asymmetrical slew capability. Here is where a push-pull VAS performs much better. It can deliver at least twice as much peak current as a single-ended VAS with the same quiescent current.
Many output stages become stressed when they must deliver a high-rate-of-change output current. Slew rate into a 4-ohm load or even a 2-ohm load will be important. Cross-conduction in an output stage under high di/dt conditions can lead to distortion. In this respect, high-speed output transistors and driver circuits that are good at sucking current out of the base are very helpful.
So, an amplifier that can comfortably deliver high slew rate will usually be further away from these sorts of current stresses on the IPS, VAS and OPS.
Cheers,
Bob