Read again what you originally posted. The slew rate requirement for audio was never under discussion.
Guys, many here are spouting half-truths about slew rate and negative feedback. You HAVE to have lots of feedback in order to overdrive the input stage. If you have modest feedback, it is often possible to rise-time limit, even without an input filter, and real, input related slew rate will be eliminated by a rather slow rise-time that does not overdrive the input stage. This is how many tube and mid sixties audio equipment were designed, and it can still sound pretty good, maybe better than most IC's.
The first IC's were gain bandwidth limited and this made for a very low slew rate. It is readily defined by the math. Later IC's 'can' have much more gain bandwidth and this raises the slew rate proportionally. Of course, other topologies might give an even better slew rate for a given gain bandwidth. Still, slew rate alone does not give 'everything' that we need, once the slew rate went beyond 5V/us for preamps. There are other factors at work, and that is the challenge, what are they?
The first IC's were gain bandwidth limited and this made for a very low slew rate. It is readily defined by the math. Later IC's 'can' have much more gain bandwidth and this raises the slew rate proportionally. Of course, other topologies might give an even better slew rate for a given gain bandwidth. Still, slew rate alone does not give 'everything' that we need, once the slew rate went beyond 5V/us for preamps. There are other factors at work, and that is the challenge, what are they?
Waly (5336)
Huh? There is absolutely no relationship between the negative feedback and the slew rate. Negative feedback (global, local, you name it) does not affect the amplifier slew rate in any way, shape or form.
Okay, please explain the need for the slew rate of 1000 or more times higher than the real rate of rise. Baksandal never found in real signals slew rate higher than 0.15 V/us
Best regards
Petr
Huh? There is absolutely no relationship between the negative feedback and the slew rate. Negative feedback (global, local, you name it) does not affect the amplifier slew rate in any way, shape or form.
Okay, please explain the need for the slew rate of 1000 or more times higher than the real rate of rise. Baksandal never found in real signals slew rate higher than 0.15 V/us
Best regards
Petr
Small signal: BW=0.35/Tr where Tr is the rise time and BW is the bandwidth. For a 350KHz bandwidth, the rise time is 1uS. The bandwidth is usually defined by the input filter (for avoiding RFI ingress), however without the input filter the bandwidth could be much larger (and therefore the rise time could be much lower). What is important is that the rise time is independent on the signal level; a 1V output and a 40V output will always have the same rise time of 1uS (assuming no large signal slew rate is occurring).
Large signal: assume a Miller compensation cap of 100pF charging with 4mA from a LTP up to 40V. The slew rate is 4mA/100pF=40V/uS and the time required to reach 40V is also 1uS. The slew rate is essentially dependent on the signal level. A 40V signal will slew in 1uS, while a 1V signal will slew in only 25nS.
From the above example it is obvious that, for a square wave signal, at low output levels the rise time will dominate the response in the time domain, while at high output levels both the rise time and the slew rate will dominate the time domain response.
Of course, the above example is arbitrary, results can be very different depending on the topology, compensation method, etc... What is important to always recall is that rise time has nothing to do with distortions, while the slew rate may induce "slewing distortions" if the amplifier slew rate is lower than 2*PI*F*V where F is the signal frequency and V is the peak voltage. For a 20KHz sine signal of 40V peak, the required slew rate is about 5V/us. It is now obvious why choosing 1V/us for each peak output volt is about x8 the minimum required value, definitely making the slewing distortions a non issue.
So, bottom line, if a 100W/8ohm amplifier has a 40V/us slew rate, forget about any slewing distortions. Everything over that is what Mr. Cordell calls "gravy". Running after hundreds of V/uS slew rate doesn't make any sense, and therefore all the noise around CFAs in audio (if such an animal exists, which I have already proven it is not, for all theoretical and practical reasons they are all VFAs) is simply marketing nonsense.
Last edited:
You nailed it ! PM was the issue. When I tried to make it "fly" (uber slew) ,
it (the "H") ran out of margin. 🙁
OS
Add Mr. Curl to the list of people not understanding the difference between slew rate and rise time/bandwidth. "The first IC's were gain bandwidth limited and this made for a very low slew rate." is absolutely wrong, the limited gain-bandwidth has (directly) nothing to do with the slew rate. One can easily build an amplifier of x10 gain and 10Hz bandwidth (GBW=100Hz), while having 100V/uS slew rate. Or an amplifier of x10 gain and 10MHz bandwidth (GBW=100MHz), with a slew rate of the same 100V/uS. Or 10V/us... or whatever you like.
Waly, I did original research on slew rate almost 40 years ago. I also have measured the fastest slew rate from audio sources, and it is faster than what Baxandall found. Of course, I know the difference between slew rate and rise-time, do you?
Can one actually make a very low thd power amp (at 20KHz) and have a SR of .15v/usec? Without input filters. Lets make one and listen.
THx-RNMarsh
THx-RNMarsh
Last edited:
mistracking phono cart are not just "audio" sources
and for more irony - almost all vinyl was cut with ~ 5kHz power bandwith (aka slew rate limit frequency) with 50 kHz cutter head mechanical and drive roll off
many classic recordings used large area condenser mics with <20 kHz roll off
today's hot, close miced recordings with 25-50 kHz mics really only came to the fore with digital audio recording, CD as consumer media
and for more irony - almost all vinyl was cut with ~ 5kHz power bandwith (aka slew rate limit frequency) with 50 kHz cutter head mechanical and drive roll off
many classic recordings used large area condenser mics with <20 kHz roll off
today's hot, close miced recordings with 25-50 kHz mics really only came to the fore with digital audio recording, CD as consumer media
Last edited:
Good idea. Then we can see if any sonic advantages of high slew rate have to do with actual slewing, or are just the result of side-effects of high slew-rate topology. I can think of a few topologies that may be able to achieve this:
1: Error-corrected LTP input where the LTP is the only stage to saturate on slewing. EC is needed to give high gain*linearity at low bias current which limits the slew rate
2: Highly degenerated LTP with antiparallel diodes across the bases which limit slew rate. Needs a high-gain VAS and multi-pole compensation for low distortion.
3: Folded cascode VAS where the cascode bias current can be variably less than the drive bias, so the VAS becomes the slew limiting stage.
That would be impossible (unless you would specify the THD at very low frequency). Such an amplifier would have gross slewing distortions, not only (most likely) audible, but also the THD will be largely affected.
Slewing distortions are not something that escapes from other non linear distortion measurement results.
Last edited:
Waly, it is not any more impossible than it is to make an ultra-low THD diode clipper using bootstraps to eliminate the capacitance and increase the VI slope of the diodes. Similar techniques are used. Distortion probably does suffer though.
It is impossible. Slewing distortion will grossly affect the THD and frequencies (or amplitudes) equal or larger than those predicted by SR=2*PI*F*V. You may either choose low frequencies, or low output levels, but you can't have both high without affecting the THD.
So you either specify 100W/8ohm of "ultra low THD" at under 500Hz, or you specify "ultra low THD" at 20KHz and under 0.1W/8ohm (0.8Veff) output.
Well, I didn't think Marsh meant to set the slew limit below what would be required for audio. I thought he simply meant to make an amp which would could be set to high or low slew rate, but that would have the same small-signal behavior either way. Perhaps the number he gave was too low, but I didn't check that.
It would be possible to have an ultra low 20KHz THD 100W/8ohm amp, with slew rates between (say) 50V/us and 200V/us. I would bet that nobody will ever be able to distinguish such amps in a double blind test.
Now the subjective, anti-DBT, golden eared crowd will activate and jump in, claiming that DBT tests are irrelevant for the sound quality. I'll leave Mr. JC and the rest of the high end audio business people to care about these, I personally couldn't care less of what they think.
Whoops, wrong picture: This should be the right one. Now, this is 'boiler-plate' engineering design that has been taught for the last 45 years or more. Solomon and Willy Sansen learned it from the same professors as I did, back then.
Attachments
Last edited:
I would agree with you, if everything else was same. I doubt it, IMO there would be more differences, than only and only in slew rate, for such two amplifiers.