phase_accurate said:
The output stage can be assumed to possess (for practical purposes), a virtually infinite slew rate....
Mains-borne RF should be taken care of nicely by your amps PSRR and decoupling/ grounding arrangements.....provided this is done properly...
Your main problem therefore, is minimising RF voltage swing across the minor-loop compensation capacitor or network....
...so if 200 kHz always is desired as power bandwidth according to Mr. Jung this formula is valid:
SR = 2*pi*f*SQR(2*Z*P)
SR = 2*pi*200*10E3*SQR(2*Z*P)
2*pi*200000*SQR(2*8*100) = 50.26 V/us (John Curl is right here
)
f = Frequency
SR = Slew rate
P = Power
Z = Speaker impedance.
As a side note Mr Hugh Dean (AKSA) claims that required power bandwitdth is much less, about 30 -50 kHz so I'll gather there are different opinons about this.
SR = 2*pi*f*SQR(2*Z*P)
SR = 2*pi*200*10E3*SQR(2*Z*P)
2*pi*200000*SQR(2*8*100) = 50.26 V/us (John Curl is right here
)f = Frequency
SR = Slew rate
P = Power
Z = Speaker impedance.
As a side note Mr Hugh Dean (AKSA) claims that required power bandwitdth is much less, about 30 -50 kHz so I'll gather there are different opinons about this.
Per-Anders, I base my recommendation for slew rate on what I personally measured 25 years ago with mis-tracking phono cartridges. They can really SPIT! I have seen measurable output at 500KHZ!
For the record, we have found that, in general, that a high speed power amp has many sonic advantages. It is best to keep them darn fast, if possible.
For the record, we have found that, in general, that a high speed power amp has many sonic advantages. It is best to keep them darn fast, if possible.
Just for clarity, EXACTLY 100W into 8 ohms would imply only 40V/us, but I have never designed an amp in the last 30 years with a slew rate of under 100V/us and I never will. It is just too compromising of the rest of the design.
Early 70's amps had slew rates of about 10-25V/us for the most part and they really sounded lousy. The first amp that I listened to in my own system, that really stood out, was the Otala power amp, which has 100V/us. I still have one today, in my office, and it still sounds great.
There are several approaches to understanding worst case slew rate. One way is actually measuring it. I have used a high speed memory oscilloscope to do this. Another way is to attempt to calculate it from available bandwidth, which is usually much more than 20KHz.
Many, not used this kind of audio measurement, might think that transients start at 0V and go either + or - , but actually the transient can start from the bottom of the waveform, go through 0, and end at the top of the waveform in 10us or so. This coupled with a 5 times multiplier to keep the slew rate mechanism from adding distortion, as it is not usually an abrupt limiting mechanism, then gives 40V/us absolute minimum and 50V/us as a reasonable minimum. As I said, I would still design at 100V/us or more, just because it is easy enough to do. I suspect that other factors such as open loop bandwidth, input linearity, etc, are equal contributors to audio improvement, but these mechanisms are what creates low slew rate in the first place, so reduction in dynamic phase distortion or PIM is a bonus.
Early 70's amps had slew rates of about 10-25V/us for the most part and they really sounded lousy. The first amp that I listened to in my own system, that really stood out, was the Otala power amp, which has 100V/us. I still have one today, in my office, and it still sounds great.
There are several approaches to understanding worst case slew rate. One way is actually measuring it. I have used a high speed memory oscilloscope to do this. Another way is to attempt to calculate it from available bandwidth, which is usually much more than 20KHz.
Many, not used this kind of audio measurement, might think that transients start at 0V and go either + or - , but actually the transient can start from the bottom of the waveform, go through 0, and end at the top of the waveform in 10us or so. This coupled with a 5 times multiplier to keep the slew rate mechanism from adding distortion, as it is not usually an abrupt limiting mechanism, then gives 40V/us absolute minimum and 50V/us as a reasonable minimum. As I said, I would still design at 100V/us or more, just because it is easy enough to do. I suspect that other factors such as open loop bandwidth, input linearity, etc, are equal contributors to audio improvement, but these mechanisms are what creates low slew rate in the first place, so reduction in dynamic phase distortion or PIM is a bonus.
mikeks said:
Still wrong i am afraid....
Mike, those kind of posts like above have we seen many times from you. The thread here is pretty technical and you are not moving the debate forward at bit with such posts.
If you feel Mr. Curl is wrong and you obviously know the answer, would you please enlighten us. If no, just avoid these "you are wrong, I am right" posts.
Moderator Per-Anders is speaking!
Plain me is wondering how your formula looks like, Mike. You gave a reference to a National Leraning brief with the exact formula which I have come up with. This is wrong according to you. Still you mention this document as correct. I'm not following you and I'm sure nobody else is.
Re: How to change slewrate?
1 A very fast input stage with rather low gain
2 An emitter follower mayby
3 High current in the VAS stage
4 A second emitter follower feeding the output stage and eliminating load of the VAS stage.
5 Output stage with fast transistors
If you really want speed without working too hard, why don't you switch over to current feedback? You can get 500 kHz, or 1 MHz just like that! I got approx. 22 MHz in my headphone amp and my latest has 100 MHz power bandwidth!! I might mention though that there are other problems related to current feedback amps but they are generally faster.
Examples of current feedback amps can be found here, here, here and here,
If you increase the current in the VAS stage (the stage before the output stage) and/or decrease the frequency compensation capacitor which you probably have, then you increase the speed.... but you do that you must also pay attention to the input stage also. Everything is connected and related to each other so if you really want to make a high slew rate amp you must pay attention to the whole amp.lumanauw said:
1 A very fast input stage with rather low gain
2 An emitter follower mayby
3 High current in the VAS stage
4 A second emitter follower feeding the output stage and eliminating load of the VAS stage.
5 Output stage with fast transistors
If you really want speed without working too hard, why don't you switch over to current feedback? You can get 500 kHz, or 1 MHz just like that! I got approx. 22 MHz in my headphone amp and my latest has 100 MHz power bandwidth!! I might mention though that there are other problems related to current feedback amps but they are generally faster.
Examples of current feedback amps can be found here, here, here and here,
I agree that tubes don't have this kind of slew rate. Tubes also have different distortion mechanisms than solid state, AND tubes have less negative feedback. They may not even slew rate at all, I can't be sure that they do. Of course, you don't often see a 50V rms or so output, like I like to design into solid state, out of many tube amps, either. What do tubes do? Do they rise-time limit? Do they clip their input stages? Who can answer this?
John, your answer is quite diplomatic . Many times here was talked about mechanism of distortion and mostly it was about distortion on " border " of hearability. On tube PA is distortion hearable in absolute majority of cases and people still like them. Why ? It can't to be explain by absence slew rate limits in input stage .
. Many times here was talked about mechanism of distortion and mostly it was about distortion on " border " of hearability. On tube PA is distortion hearable in absolute majority of cases and people still like them. Why ? It can't to be explain by absence slew rate limits in input stage .What about symetrical SR? I believe there are 2 thoughts on this; one says that as long as the SR is large enough it doesn't matter if the + excursion and - excursion have the same SR or not, the other says that they both need to be the same.
I would think that if the SR was different for the +/- then the 1/Fb would be different with increasing freq, potentially affecting the sound.
Is this not one of the reasons why full complimentary circuits came about? To attempt to have an equal SR for both sides of 0.
I would think that if the SR was different for the +/- then the 1/Fb would be different with increasing freq, potentially affecting the sound.
Is this not one of the reasons why full complimentary circuits came about? To attempt to have an equal SR for both sides of 0.
Did you use a sqarewave with as c;ose to an instantaneous rise as the simulation software will allow? In oder to get a SR figure that represents the best the cicuit can do you have to use a stimulous that is more demanding than the circuit's capability -- otherwise all it shows is the slew rate of the stimulous not of the circuit. If this is obvious, sorry - it just reflects the error I made first time I tried to usethe "d(...)" function.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.