Many years ago I think I remember one UK reviewer saying slew rates were like an arms race . Curious to know more a figure of about 6 V/uS was said to be enough for 100 W if real music . Anything above that a bonus or silly ? Having so many times read of greater slew rates I feel certain there is more to this story . A home cinema guy said he measured 1.3 W for dialogue ( 93 dB efficiency speaker ) , 250 W for the spaceship flying in . We covered symmetry the other day which I suspect was approved of ? Any ideas on minimum slew rates for 100 W ?
This link says about slew rates near the bottom of the page . By coincidence I have built nearly all these at some time having asked myself the same questions . Glad he thought it worth documenting .
Elliott Sound Products - Audio Power Amplifier Design Guidelines
One guy showed a 1 kHz square-wave and how to use it for evolution .He then said do not use a speaker for the test as the tweeter will fry . Too obvious ? Well it does say even that simple test has no musical reality .
This link says about slew rates near the bottom of the page . By coincidence I have built nearly all these at some time having asked myself the same questions . Glad he thought it worth documenting .
Elliott Sound Products - Audio Power Amplifier Design Guidelines
One guy showed a 1 kHz square-wave and how to use it for evolution .He then said do not use a speaker for the test as the tweeter will fry . Too obvious ? Well it does say even that simple test has no musical reality .
Last edited:
A simple way of looking at it is to consider a full power 20KHz sine wave. This has a period of 50uS, so to go from max -ve to max +ve rails takes 25uS. 25 x 6V/uS = 150, volts peak to peak, 75V rails. An amp with such could probably do near 400W into 8ohms; so I reckon we're safe ... 
Frank
Frank
Last edited:
You know, both Walt Jung and Matti Otala beat this topic to death with REAL measurements more than 1/3 century ago. It was found that 50V/us was necessary for a 100W amplifier, to be relatively SURE that TIM (SID) would not be generated.
What many here STILL seem to forget is 20KHz may be the highest frequency considered important in audio listening, BUT the working bandwidth goes on and on in many situations.
For example, for phono cartridge mistracking, 200KHz is pretty normal with MC cartridges, 500KHz has been measured. Where is your 6V/us THEN?
What many here STILL seem to forget is 20KHz may be the highest frequency considered important in audio listening, BUT the working bandwidth goes on and on in many situations.
For example, for phono cartridge mistracking, 200KHz is pretty normal with MC cartridges, 500KHz has been measured. Where is your 6V/us THEN?
Last edited:
Of course, it is also to be considered that future digital will have more effective bandwidth, with 24/192, 24/96, and SACD already here. You would be surprised how well these sources can fit the TIM profile, making 50V/us for a power amp, but only 5V/us for a preamp, pretty much a necessary spec for highest quality audio.
John do you have any references for that, papers or measurement reports where we can read up on it?
Edit: we xposted, you posted just above me, thanks
jan
My related question is, assuming the amplifier slew rate is fast enough to track the input, then is any additional slew rate margin a direct indicator of reduced SID.
It just seems to me this distortion is from larger nonlinear current swings in the input stage, so it could be reduced by reducing the current swing or making it more linear. If in the process this typically results in an increased measured slew rate, then maybe one can use this as in indicator but it still seems indirect to me.
Direct correlation to slew rate would imply that CFA type amplifiers would be automatically superior in this area (which may be true I don't know) but I dont even see a fundamental between a CFA and VFA if both are operating within the same i-v portion of the input transistors for the audio input.
Thanks (not sure I should have posted this, though crowd around here)
-Antonio
It just seems to me this distortion is from larger nonlinear current swings in the input stage, so it could be reduced by reducing the current swing or making it more linear. If in the process this typically results in an increased measured slew rate, then maybe one can use this as in indicator but it still seems indirect to me.
Direct correlation to slew rate would imply that CFA type amplifiers would be automatically superior in this area (which may be true I don't know) but I dont even see a fundamental between a CFA and VFA if both are operating within the same i-v portion of the input transistors for the audio input.
Thanks (not sure I should have posted this, though crowd around here)
-Antonio
Two questions . Ignoring the feedback argument which seems strong . Can we usefully apply input bandwidth limiting ? My by ear tests were inconclusive as I feel my amplifier has no real problems . If I deliberately made a low slew rate amplifier would some sensible input filtering transform it ? Theory suggests it should . As John said an MC pick up might generate 200 kHz output . I could imagine a 4 pole passive input filter is required ? These are things I know well so should not ask . Still very interesting to read the views of others as often the right thing's are done for the wrong reasons .
LTP degeneration . If we have the re of the Long Tail Pair set at 50 R is that less good than lets say 39 R external ( RE ) and 11R internal ( re ) . My instinct is to say the eternal is a real resistor and therefore better . I have always assumed 50 R ( per emitter ) in all my designs . It seems to be a good compromise . I found the inclusion of the resistor or cascode in the CCS very interesting ( tail ) . Almost a free lunch .
LTP degeneration . If we have the re of the Long Tail Pair set at 50 R is that less good than lets say 39 R external ( RE ) and 11R internal ( re ) . My instinct is to say the eternal is a real resistor and therefore better . I have always assumed 50 R ( per emitter ) in all my designs . It seems to be a good compromise . I found the inclusion of the resistor or cascode in the CCS very interesting ( tail ) . Almost a free lunch .
There, you see, Nige? Exactly what I've been telling you all along.
And you ask me why I think Beritish made amps sound slow and sometimes quite dead - band limiting is useful only way up there at or above 200 kHz. Most British amps use band limiting both down below at or below 20 Hz and up above from 50 kHz or so. The logic being "There's nothing there that we want." Not quite so.
I apologise for coming in so late to a subtopic I sort of initiated, but this forum has a wild way of sometimes NOT alerting me to new posts, and it has been dead silent for the last 6 days or so. I came in today to ask what happened, somebody die, or something, and I find you eager little beavers are at it full force.
John, thans for the hint, GK7 thanks for the link. I downloaded it so I can study it later, I don't want to just run through it.
From what I gathered from the discussion here, it seems that the criteria suggested in the Nationl Seminvonductor Audio cookbook, 0.5 V/uS per PEAK volt output is, as I took it even then, too conservative. Right from the start, I assumed double that value, i.e. 1 V/uS per peak output volt, or 40 V/uS for a nominally 100 W/8 Ohms. More can't hurt, unless it compromises something else.
I sort of knew H/K got it right when they made HK 680 integrated amps, nominally delivering 85/130 W 8/4 Ohms, but giving it a slew rate of 80 V/uS. Richard Miller, who is credited as the chief of its design team, obviously agrees with Otala, Jung & Curl. Hardly surprising, given that H/K employed Otala for some 6 or so years.
On a purely personal level, I am very pleased with this turn of events, I have always thought and spoken out loud that H/K is the one company to watch among all mass producing manufacturers, they were always at least one cut, and often more, in front of the rest of the pack. Citation XX is still thought by many to be one of the best power amps ever made by anyone, but unfortunately, few were made and it's really rare. I never even saw one live.
Hallelujah and amin!
As ever since the beginning of time, it all boils down to use your head when designing it and get it right from the design stage, later added cure-alls will only ruin it.
Be balanced, favor nothing over the whole, and for the love of God, do not save on your output stage, be generous. Power supplies are assumed to be top notch.
Too bad Thorsten is not here, most of this is right up his alley. I've been talking to him quite a lot, and much of what I've seen here I also heard from him.
Thank God, common sense and good reasoning is still alive and well.
Honestly Frank, I don't find it strange at all. You can find the notion documented in Otala's papers published by IEEE in 1976 or 1977, not sure which.
As John quite rightly pointed out, Matti Otala and Walt Jung beat the subject to death.
I wish I could send you a copy of Otala's initial paper, but unfortunately, I no longer have a photostat copy of it, perhaps some good soul here has a link to it?
I guarantee it's WELL worth reading (or your money back
).Let's discuss now who first started using Ohm's law in audio amplifiers.
No.... don't tell me somebody is still using that ancient waffle?
I suspect it was Lee de Forest who first used Ohm / Kirchhoff . It deserves an answer .
I came to the occlusion that input slew limiting required a filter far more sophisticated that is commonly used if to be of any use . For example a 4 pole Butterworth with 35 KHz cutoff frequency would be about -0.05 dB @ 20 kHz , - 3dB @ 35 kHz and -36 db @100 KHz . If this could be a offered in a preamp as an additional output then perhaps it has value ? NAD had the less filtered output called LAB .
I came to the occlusion that input slew limiting required a filter far more sophisticated that is commonly used if to be of any use . For example a 4 pole Butterworth with 35 KHz cutoff frequency would be about -0.05 dB @ 20 kHz , - 3dB @ 35 kHz and -36 db @100 KHz . If this could be a offered in a preamp as an additional output then perhaps it has value ? NAD had the less filtered output called LAB .
Honestly Frank, I don't find it strange at all. You can find the notion documented in Otala's papers published by IEEE in 1976 or 1977, not sure which.
As John quite rightly pointed out, Matti Otala and Walt Jung beat the subject to death.
I wish I could send you a copy of Otala's initial paper, but unfortunately, I no longer have a photostat copy of it, perhaps some good soul here has a link to it?
I guarantee it's WELL worth reading (or your money back
actually Cordell did a lot to beat to death Otala's specific prescription for avoiding TIM/PIM with "flat loop gain" – with better explanation, Cordell built hardware both dedicated test hardware and low PIM but high feedback amp
CordellAudio.com - A MOSFET Power Amplifier with Error Correction
both Walt Jung and Marshall Leach were early Otala enthusiasts - later came to the conclusion that you don't have to use Otala's "flat loop gain" prescription - high feedback, sloping loop gains can still have vanishingly low TIM/PIM
perhaps Otala deserves credit for sparking the renewed slew analysis effort in audio amps but his actual recommendation, initial model have been shown lacking
Last edited:
JCX, I am 59 years old. For the last 25 years, I have witnessed and in some cases participated in oh-so-many pro- and anti-Otala forums, I lost count. I am definitely NOT about to get myself into another Cordell vs Otala debate.
I do not claim or think that Otala got it all right, or even just got it all. Ditto for Cordell, or for that matter, anybody. I acknowledge that there are many roads which lead to Rome, that there is no one single way to do things, but at least several, or even many.
I believe in what Otala has said (the part I know of) because it has served me well and because I choose to do so. But this in no way means Otala is perfect, or that there are no others who may also have got it right, just used a different method. I don't know, I haven't researched them all, simply because of the old rule - if it ain't broke, don't fix it.
And I do have a beef with the extreme Otala camp people, who have somehow come to the conclusion that one should use no global NFB at all if possible. Otala never said "Thou shalt not use any global feedback". He simply said that global NFB should be used to sort of iron out the already clean drip-dry shirt, use it as icing on the cake. Using no global feedback usually means letting your output stage work with some error, although I have seen schemes to wrap the output stage inside it own local NFB loop. My feeling is that this is not worth the time and trouble, and my experience tells me that the best sounding (to me) amplifiers WILL have a global NFB somewhere in the region of 17 ... 26 dB, an open loop FULL power bandwidth of no less than 50 kHz, etc.
So, please, let's leave it at that.
- Status
- Not open for further replies.