Yes. Jung is where I got my original response.
Thanks Jan for posting the link to the SID articles from Walt Jung.
This is my conclusion after reading all 4 parts:
- Any SID will be detected by THD measurements (with high bandwidth up to 100kHz or so). Hence, there is no point in measuring slew rate in a finalized amp - unless for marketing reasons. Maybe it is helpful during development...
- The relation he found between slew rate and THD performace applies to op amps which were available in the late 70‘s, there were some exceptions though. This does not necessarily apply to modern design and certainly not to power amplifiers. E.g. Doug Self identified the Class-B output stage as the limiting factor in a blameless amp with a negligible contribution of input stage/VAS to overall THD.
So what is the reason to optimize amps for slew rate rather than THD? Are there further distortion mechanism revealed after this series of articles? If so, I would be very interested in this!
This is my conclusion after reading all 4 parts:
- Any SID will be detected by THD measurements (with high bandwidth up to 100kHz or so). Hence, there is no point in measuring slew rate in a finalized amp - unless for marketing reasons. Maybe it is helpful during development...
- The relation he found between slew rate and THD performace applies to op amps which were available in the late 70‘s, there were some exceptions though. This does not necessarily apply to modern design and certainly not to power amplifiers. E.g. Doug Self identified the Class-B output stage as the limiting factor in a blameless amp with a negligible contribution of input stage/VAS to overall THD.
So what is the reason to optimize amps for slew rate rather than THD? Are there further distortion mechanism revealed after this series of articles? If so, I would be very interested in this!
In my experience, competently designed power amps are optimized for BOTH slew rate and THD. Including the power amps found in Bob Cordell's textbook and Douglas Self's textbook.
Optimizing for THD includes output stage selection, output stage biasing (to minimize crossover distortion), and topology selection. Notice how both Cordell and Self show ever increasing improvements in distortion as the topology is improved and the output stage is polished.
Optimizing for slew rate is a fairly straightforward process, described by James E. Solomon in a paper that is very frequently cited here on diyAudio, of balancing gm versus Ibias versus Ccomp. Both Cordell and Self describe it in detail. These modifications are more or less independent of the optimizations carried out to minimize THD.
And that's what most designers choose to do: optimize both THD and slew rate.
Optimizing for THD includes output stage selection, output stage biasing (to minimize crossover distortion), and topology selection. Notice how both Cordell and Self show ever increasing improvements in distortion as the topology is improved and the output stage is polished.
Optimizing for slew rate is a fairly straightforward process, described by James E. Solomon in a paper that is very frequently cited here on diyAudio, of balancing gm versus Ibias versus Ccomp. Both Cordell and Self describe it in detail. These modifications are more or less independent of the optimizations carried out to minimize THD.
And that's what most designers choose to do: optimize both THD and slew rate.
I read Self thoroughly and he is definitely not promoting slew rate optimization unless for marketing. Not even for slew rate symmetry as pointed out by Jung is important for best THD performance in op amps.
In Cordells book I could also not find a hint to optimize for slew rate.
Do you have a link to the Solomon paper? All I find in the internet deals with op amps where, as it seems, at least in the late 70‘s available devices, the input stage inbalance is the limiting factor for THD and not the output stage as in power amps.
In Cordells book I could also not find a hint to optimize for slew rate.
Do you have a link to the Solomon paper? All I find in the internet deals with op amps where, as it seems, at least in the late 70‘s available devices, the input stage inbalance is the limiting factor for THD and not the output stage as in power amps.
Hmmm, if it's a bad idea to optimize for slew rate,
then perhaps it's a good idea to anti-optimize for slew rate.
This turns out to be pretty easy to do! To cut slew rate in half, merely
_
then perhaps it's a good idea to anti-optimize for slew rate.
This turns out to be pretty easy to do! To cut slew rate in half, merely
(i) reduce the IPS bias current by a factor of two (change R5 to 15K) ;
(ii) keep IPS transconductance same as before (change R15 and R16 to 232 ohms) ;
(iii) keep P1 pole frequency same as before (don't change C1, leave it at 30 pF).
By simply swapping out three resistors, we cut slew rate in half, without changing bandwidth or THD. If big slew rate is bad, maybe small slew rate is good.(ii) keep IPS transconductance same as before (change R15 and R16 to 232 ohms) ;
(iii) keep P1 pole frequency same as before (don't change C1, leave it at 30 pF).
_
Attachments
Speed is not bad. Mathematically it has to be at same speed as the input to the amplifier which might be of any speed. Added delay in rise time is it perceived?
Secondly how fast are our OPS? I have did a small sim where you can see that using a faster ips consider Lichtstark and i have probed the both output from VAS and from the output of the amplifier. Please comment on it.
In my opinion slew is not in first priority. Firstly THD, IMD and then we can consider higher slew. I believe if someone can get better slew without any artifacts then its not a problem but getting other sets of problems with higher slew is not my choice though.
Secondly how fast are our OPS? I have did a small sim where you can see that using a faster ips consider Lichtstark and i have probed the both output from VAS and from the output of the amplifier. Please comment on it.
In my opinion slew is not in first priority. Firstly THD, IMD and then we can consider higher slew. I believe if someone can get better slew without any artifacts then its not a problem but getting other sets of problems with higher slew is not my choice though.
Attachments
Solid State Power Amplifier Supply Part 1 Please check at the bottom of the page:
>
By way of example, a typical commercial value capacitor, rated at 10,000uF/63V and costing some Euro 8-9, will have a speed of 30-40V/uS at best. An equivalent Elna for Audio series black, costing some Euro 15-25, will have a speed in the range of 80-90V/uS, i.e. at the very worst, double the speed of the best case in commercial cap land. A Siemens Sikorel cap, costing some Euro 20-30, will slew at over 100V/uS - but at a price.
>
Here is one thing which really to be taken into consideration AFAIK even capacitors have charging and discharging times. Taken that into consideration like a how fast does the psu capacitors deliver the current into the ops. I believe that also has an effect on how good is slew. I found this link long ago that the author states about the slew rate of the capacitor and he clearly mentions that Sikorel has highest slew rate as much as 100V/usec so in that case lets consider an amplifier circuit is capable of much higher slew as much as 300V/uSec then in that case the psu cap is the bottle neck. Many say that yes we just draw current from the psu caps but even the psu caps has charging and discharging times. So are we taking that into consideration? What I say is that when driving high currents into the load is the slew is constant or getting down because of the PSU speed. How to measure the slew rate of the capacitor though?
>
By way of example, a typical commercial value capacitor, rated at 10,000uF/63V and costing some Euro 8-9, will have a speed of 30-40V/uS at best. An equivalent Elna for Audio series black, costing some Euro 15-25, will have a speed in the range of 80-90V/uS, i.e. at the very worst, double the speed of the best case in commercial cap land. A Siemens Sikorel cap, costing some Euro 20-30, will slew at over 100V/uS - but at a price.
>
Here is one thing which really to be taken into consideration AFAIK even capacitors have charging and discharging times. Taken that into consideration like a how fast does the psu capacitors deliver the current into the ops. I believe that also has an effect on how good is slew. I found this link long ago that the author states about the slew rate of the capacitor and he clearly mentions that Sikorel has highest slew rate as much as 100V/usec so in that case lets consider an amplifier circuit is capable of much higher slew as much as 300V/uSec then in that case the psu cap is the bottle neck. Many say that yes we just draw current from the psu caps but even the psu caps has charging and discharging times. So are we taking that into consideration? What I say is that when driving high currents into the load is the slew is constant or getting down because of the PSU speed. How to measure the slew rate of the capacitor though?
Why would anybody be interested in the voltage slew rate of a supply cap? I don't want the voltage to change; people go through great lengths to design their supply so the voltage is constant under varying loads. There is nothing to voltage slew.
If you expect the cap to supply the load current then you should be interested in the current slew rate: how fast can the cap supply current to a load that demands fast current changes.
The current slew rate is related to the ESL of the cap and the wiring to the load and, to a lesser extend, the ESR of the cap
Jan
If you expect the cap to supply the load current then you should be interested in the current slew rate: how fast can the cap supply current to a load that demands fast current changes.
The current slew rate is related to the ESL of the cap and the wiring to the load and, to a lesser extend, the ESR of the cap
Jan
Yes Jan,
It's the Amps per microsecond, that is the correct way to assess the capacitors.
Hang an inductive cable between the Source of current and the Receiver and one finds that the current prefers to come from the local supply rail decoupling. The PSU just recharges the decoupling as quickly as the inductance will allow.
It's the Amps per microsecond, that is the correct way to assess the capacitors.
Hang an inductive cable between the Source of current and the Receiver and one finds that the current prefers to come from the local supply rail decoupling. The PSU just recharges the decoupling as quickly as the inductance will allow.
how do you calculate the current slew? I believe the current slew should be in the same order for may be 4 times for 4ohm or 8 times less for a 8ohm load consider for the 4ohms as best as we can consider a better realistic scenario so how do you calculate and what are the practical values observed for take an example of 10000uF / 100V cap.
Consider 100uF /60V has 100mOhm ESR so how fast it can give 10A current to the transistor.
This question does make no sense at all. Current speed or slew are a matter of voltage and inductance. ESR limits the peak current - but not speed.
There are very few manufacturers wo specifiy the ESL.
Epcos states 15..20nH for the sicorel, i.e. something like 2cm of blank wire.
Considering the real world wiring, this is neglible in all aspects relevant to audio. Except for audiophools, who judge the audio quality according to the price tag.
Why out of topic? The thread name is 'slew rate' and that is what we are discussing, no?
Jan
Assuming for the measurement of maximum rate, you can assume that the load step is ideal, i.e. change from one value to the other instantaneously.
So you have a voltage on the cap, a lumped impedance (ESL, Lwiring, etc) to the load step. Then you can calculate how long it takes for the current to change from what is was with the old load to the value with the new load. Divide by time and you have the current slew rate of the cap+wiring.
It's a simple equation, but I can't remember it right now. I believe the rule of thumb is about 5 times the Rload/L value, in seconds, for the current to get to the final value, but again not sure. Anybody has it ready?
Jan
Yes good point. It must be maximized. Does the ESR in a cap limit the maximum rate of change of the charge/discharge current?
Jan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.