The reason is simple: the amp will oscillate! If you have at least three different stages, input stage, voltage amplification stage, output stage. The bandwidths of all these stages must be carefully matched if you are going to apply feedback. When you ahve control over this you will get a slew rate. If you for some reason aren't pleased with the slew rate you have to tweak the design.
Re: Re: Hi Moderator Senor
Hi MIKE,
Yeah I agree with you about the basecurrent which is absolutely higher for slow BJT's because the majority carrires require a lot more pushup to avoid the sagging of high frequency current requirements, thats why their is aneed of lotmore base current.
regards,
ampman
MikeB said:
Hi MIKE,
Yeah I agree with you about the basecurrent which is absolutely higher for slow BJT's because the majority carrires require a lot more pushup to avoid the sagging of high frequency current requirements, thats why their is aneed of lotmore base current.
regards,
ampman
I'm not triggering semantics here, but why does feedback comes to base of differential (voltage feedback) usually slower than feedback comes to emitors (current feedback)? What's the difference in mechanism of feedback in those 2 that makes one of them faster?
Hi, Mikeks,
Why are you saying putting resistor accross capacitor is BAD IDEA?
Hi, Mikeks,
I just read about putting 220K parrarel with Cdom in VAS to lower OLgain.
Why are you saying putting resistor accross capacitor is BAD IDEA?
I'm afraid you try sidestep the output stage matter. Amp oscillates because output stage limits the linear SR and VAS has to overdrive it due to NFB. Careful matching if I understand it correct boils down to making sure each subsequent stage is faster than previous. What about no-NFB amp?peranders said:
I'm still confused because as I see it, output devices can have unlimited SR for practical purposes only once - then you swap them..
When I think of SR, it's current change rate onto fixed load. For current to change in transistors, all sorts of capacitance has to be overcome before there is even drive current left. Already that sets some limit to SR of devices unless VAS is ideal voltage source (in which case - who needs output stage?). Consider IRFP250, its Ciss is 2-5nF, with 100ohm resistor in series at gate that alone limits SR to some earthly value (RC). No VAS has zero output impedance.
Important factor here is that this RC effect depends on the load current, so max output power makes huge difference.
BJTs require excessive base current, that could exceed their SOA. On other hand, HFE drops to 1 at some SR, thus output stage looses its purpose - all the current is loaded onto VAS. And I wonder if darlingtons help there. Perhaps driving tran of darlington needs to be faster then.
Someone said here that output stage becomes a switch at some limit. I understand that this is so because trans have some intrinsic delay of carrier reorganization, output is delayed after which avalanche like increase of output occurs, approaching switch like behaviour. Also depends on load current.
From other angle, when VAS gives high SR, output SR departs from linear. That alone gives SID. Then its a matter of how much is tolerable. I assume that its acceptable level of SID that sets the required minimum for SR of the amp, requiring hefty difference compared to calculated minimum based on PBW. Therefore my understanding is that arguing about audible bandwidth is a wrong issue to discuss completely.
In all, I can't seem to get it how comes output device parameters do not directly limit how far you can push SR of the amp. By all means they do, as they face the hardest load conditions of the amp.
I won't say" go and read some books" but wimms you must know how feedback in detail works. The output stage is much, much faster than the VAS stage. I'm afraid you are totally wrong about the cause of oscillation.wimms said:
A NFB amp is a different thing but I don't think the topic is about that.
Let me also point out that the thread was about amp_man's rather fantastic statement of slew rate needs for PA amps. Amp_man hasn't commened it so I'll guess it stands for him and isn't arguable.
I think we have established what the minimum need is at least.
peranders said:
Per-Anders,
I have a weak memory that I red somewhere else that the output stage is the slowest stage, but one can not be sure.
How have you come to the conclusion that the VAS is slower than the output stage, how did you study the case?
Have you meassured in a same design with the VAS cascoded and without cascode?
Have you meassured the VAS without an output stage connected to it or with?
If VAS was connected together with the output stage, how did the output look like(BJT, FET's, etc...)?
Was there a feedback capacitor or capacitor load connected from VAS?
How have you meassured that the output stage is slower?
With what transistors (eg. BJT, MOSFET's etc) in VAS, output stage and eventual cascodes?
etc. etc.
Regards Michael
peranders said:
The amp will oscillate if you have feedback that turns into positive feedback, which can happen because of phase shifts in the amp reach 180 degrees for some frequency. If at that frequency the gain is still higher than one, the amp generates its own input signal, i.e. it oscillates. It has nothing directly to do with the number of stages, but there is a connection because more stages generally give more phase shift.
The positive feedback can also occur because of feedback around a single stage, for instance in a fet stage from drain to gate.
Generally, one does not match the stages in this respect, but compensates one stage "dominantly" so that it rolls off the gain before the dreaded 180 degrees point.
The expression "you have a slew rate" doesn't mean anything on itself. An input signal has a slew rate, and if the amp can't keep up with it, it goes into slew rate limiting.
Since slew rate limiting only can occur in amp stages that exhibit large voltage swings, in a run-of-the-mill amp it will be the Vas stage that is sensitive to it.
Funny enough, the easiest way to do dominant compensation (see above) to keep the amp stable is with a cap from C to B on the Vas. And there goes your slew rate capability....
The question of the output stage slew rate is interesting, I have never seen that mentioned. But remember that the root cause of slew rate limiting is the limited current available to charge the miller cap and, if used, the compensation cap. In output stages, there is plenty of current available to charge anything, and if more current is needed for low resistance loads, the driver will pump enough base current (or gate charge current) into it to make it slew fast enough. I'm not saying it is impossible, after all *everything* has it's limit (except stupidity), but I think it is orders of magnitude less a problem than the voltage stage slew limitations.
Jan Didden
Output stage influence on Slew Rate
The output stage has a significant influence on slew rate, You work back from the current necessary to drive the load at the rate of change necessary to meet the slew requirement. If its pure resistance the arithmetic is pretty straight forward. But at 500 V/uS 1000 pF of speaker cable is significant. (I may have misplaced a decimal in my head but I think its .5A!)
Then you must provide enough current gain through the output and driver (and of course, VAS) to drive that load current.
Here is where Powerfets become interesting for two reasons. First they need less drive that typical large bipolars at these rates of change. They still need a lot though. Second, you need to stop the drive at some point, and FETs are much easier to turn off than bipolar transistors.
However if the VAS can't slew fast enough its all academic. Calulate the amp for the peak slew rate, and if you are fanatic, make every stage up to the final output operate in class A. It becomes obvious why good fast amps will never be energy star rated.
-Demian
The output stage has a significant influence on slew rate, You work back from the current necessary to drive the load at the rate of change necessary to meet the slew requirement. If its pure resistance the arithmetic is pretty straight forward. But at 500 V/uS 1000 pF of speaker cable is significant. (I may have misplaced a decimal in my head but I think its .5A!)
Then you must provide enough current gain through the output and driver (and of course, VAS) to drive that load current.
Here is where Powerfets become interesting for two reasons. First they need less drive that typical large bipolars at these rates of change. They still need a lot though. Second, you need to stop the drive at some point, and FETs are much easier to turn off than bipolar transistors.
However if the VAS can't slew fast enough its all academic. Calulate the amp for the peak slew rate, and if you are fanatic, make every stage up to the final output operate in class A. It becomes obvious why good fast amps will never be energy star rated.
-Demian
comments
hi Moderator Senor,
Do you really have established what the minimum need is at least.
Would you be kind enough to tell us what type of criteria are you following towards the estimation of necessary slewrate requirement in Professional environment, Where amplifier is located at 20 to 80 meters away from the speakers and still driving the reactive loads along with full power[sometimes a bit overloaded].
Regards,
kanwar
peranders said:
hi Moderator Senor,
Do you really have established what the minimum need is at least.
Would you be kind enough to tell us what type of criteria are you following towards the estimation of necessary slewrate requirement in Professional environment, Where amplifier is located at 20 to 80 meters away from the speakers and still driving the reactive loads along with full power[sometimes a bit overloaded].
Regards,
kanwar
I think we have come to some conclusion but I have a hard time to melt your theories, hysterical fast amp, 80 meters cable (=8-15 nF), rather large capactive load.
Somehow I have a hunch of that using an amp with 200-500 kHz bandwidth for bass etc seems somewhat crazy of the reason that the combination gets unstable easily.
Somehow I have a hunch of that using an amp with 200-500 kHz bandwidth for bass etc seems somewhat crazy of the reason that the combination gets unstable easily.
Folks, I don't know if I can make better sense of everything that you are talking about, but slew rate is a LIMITING CONDITION that you don't want to approach. Do you need 500V/us? Couldn't hurt! Can you get away with 10V/us? Maybe with a bass amp, even a midrange amp, but not a full range amp, for best solid state audio fidelity.
Now, do some output devices limit slew rate? Yes
Do some input topologies limit slew rate? Yes
Today, with improved devices, the output stage should not limit the slew rate to any significant degree. However, 30 years ago, this was not the case.
However, the most significant limitation to slew rate is both the amplifier gain bandwidth, and the transconductance of the input stage. It can be shown that transistors have so much transconductance, that compensating for this lowers the slew rate below what is acceptable, in most cases.
When you try to make a fast amp, you usually also make it more sensitive to capacitance on the output. This can make an amp oscillate. Protecting an amp from output capacitance with an output coil can compromise the overall sonic quality of a power amp. This is why I choose to designing about 100V/us into my typical designs, today, even though I have designed faster amps in the past. I can remove the output coil, and still remain stable.
I hope that this helps.
Now, do some output devices limit slew rate? Yes
Do some input topologies limit slew rate? Yes
Today, with improved devices, the output stage should not limit the slew rate to any significant degree. However, 30 years ago, this was not the case.
However, the most significant limitation to slew rate is both the amplifier gain bandwidth, and the transconductance of the input stage. It can be shown that transistors have so much transconductance, that compensating for this lowers the slew rate below what is acceptable, in most cases.
When you try to make a fast amp, you usually also make it more sensitive to capacitance on the output. This can make an amp oscillate. Protecting an amp from output capacitance with an output coil can compromise the overall sonic quality of a power amp. This is why I choose to designing about 100V/us into my typical designs, today, even though I have designed faster amps in the past. I can remove the output coil, and still remain stable.
I hope that this helps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.