step response....
Absolutely. More so with a parasitic prone high current stage.
OS
#2 - can even see the damped oscillation of the predriver/ccs node. But the resulting output is rounded.And, just a little overshoot w/capacitor load. "Fine tune" of the main driver basestopper is what makes for stability. a high Ft main driver also makes things easier.
OS
Absolutely. More so with a parasitic prone high current stage.
OS
#2 - can even see the damped oscillation of the predriver/ccs node. But the resulting output is rounded.And, just a little overshoot w/capacitor load. "Fine tune" of the main driver basestopper is what makes for stability.
a high Ft main driver also makes things easier.OS
Attachments
I think you have to seprate parasitics and overall loop behaviour.
If I look att your plots above OS, I believe this is driver/output stage behaviour and you are taming parasitic issues, and adjusting the base stopper is one way to do this. But for slew rate and bandwidth limiting, I think those issues usually need a different approach.
I also absolutely agree that looking at step response behaviour gives a good indication of how the loop is performing.
If I look att your plots above OS, I believe this is driver/output stage behaviour and you are taming parasitic issues, and adjusting the base stopper is one way to do this. But for slew rate and bandwidth limiting, I think those issues usually need a different approach.
I also absolutely agree that looking at step response behaviour gives a good indication of how the loop is performing.
You say a lot of sensible things Tom. If you have two amplifiers, one with a max slew rate of 25V/uS and another with 500V/uS, and you feed both with a signal that has a max slew rate of 5V/uS, both amplifers perform identical in this respect: both will reproduce the signal with no slew-rate limited artifacts.
However, people don't agree how close you can get to the slew rate limit without getting into trouble. That is why many designers take a large safety margin and design a 100W amp with a slew rate of say 50V/uS even if the music signal will never have any content above 5V/uS. Since it is pretty easy and cheap these days to design a 50V/uS amp, the safety factor makes sense.
Quizz: which element in a power amp usually limits the slew rate of the whole amp?
jan didden
Also one must think about that the amplifier needs to slev app 30 times more than the max slope the input signal carries......due to the amplification...to my thinking Slevrate must be somewhere in the neighborhood of 50V/us
device capacitance seems to me as the most limiting slevrate factor...
device capacitance seems to me as the most limiting slevrate factor...
Hi Tom,
All of us have a tendency to be a little sloppy with the use of the term slew rate, often as a matter of convenience. This includes me, even in a couple of places in my book.
In the strict sense, slew rate is not a property of a signal. Rather, it is the maximum rate of change that a device of some sort can deliver. I think historically it may have been associated with the maximum rate that the turret of a military gun could move (rotate).
So, for an amplifier, slew rate is commonly associated with the maximum rate of change of its output voltage, properly deemed its voltage slew rate. It, by itself, is a specification of the amplifier, independent of signal. Power amplifiers also have maximum output current slew rate, which is important, but not often discussed.
A signal does not have a slew rate, it has a rate-of-change, as in volts/us. Where we get sloppy out of convenience is when we say something like "a 1V peak 20 kHz sinusoid has a (peak) slew rate of 1.25 V/us". This is more convenient than saying "a voltage rate-of-change of...". So we get sloppy with the semantics out of convenience. I do it and many others do it, and people usually know what we mean in spite of the incorrect semantics. By the same token, many of us frequently say something like "watts RMS" when we really should be saying something like "average power".
There has indeed often been too much emphasis on slew rate as an amplifier parameter. There are certainly loopholes in its connection to performance. While we know that an amplifier with too little slew rate headroom above the maximum voltage rate-of-change called for by the program material will not sound good in general, we cannot say that having a slew rate headroom of 10:1 will guarantee a good-sounding amplifier or one with very low high-frequency distortion. For example, an amplifier with a high slew rate might be plagued by bad dynamic crossover distortion. Some amplifiers also obtain high slew rate by having some of the circuits in their signal chain go into a sort of class B mode, which may also have unwanted consequences for sound quality.
However, often the legitimate design practices that lead to higher amplifier slew rate do lead to better high-level high-frequency performance - often by reducing peak signal excursions at the input stage.
Cheers,
Bob
No, the number of 5V/uS I quoted was for 20VRMS (28V peak; 50WRMS in 8 ohms) output level.
In most amplifiers, the one component setting the slew rate is the compensation capacitor in combination with the standing current of the input stage.
Let's just assume that you have a differential input stage with say 1mA tail current, and a regular voltage amplifier stage that has a 100pF compensation capacitor. That compensation capacitor has to be charged and discharged during the signal swing to the (alomost) output voltage level. The max current available for that is, you guessed it, that 1mA standing current in the input stage. So the slew rate is determined by the time it takes to charge/discharge a 100pF cap with 1mA.
Slewrate in V/uS is equal to I (amps)/C (uF), so in our example it is 10V/uS if I got my sums right
The device slew rate limitations usually are orders of magnitude higher.
jan didden
Last edited:
Arguing about words is rarely fruitful, unless someone is genuinely confused about meanings.
A signal will have a maximum dV/dt somewhere in its waveform. An amplifier will have a maximum dV/dt it can deliver (which is usually assumed to be a constant value, although conceivably it could be signal/memory dependent). If the latter is sufficiently greater than the former, then the amplifier can deliver the signal without slew rate limiting. Are we agreed?
A signal will have a maximum dV/dt somewhere in its waveform. An amplifier will have a maximum dV/dt it can deliver (which is usually assumed to be a constant value, although conceivably it could be signal/memory dependent). If the latter is sufficiently greater than the former, then the amplifier can deliver the signal without slew rate limiting. Are we agreed?
Just my 12 cents - never try to suppress 2nd and 3rd harmonics (by strong feedback) to negligible level at expense of growth of high order harmonics 5th and higher (above level of 2nd and 3rd). Never, and regardless absolute value of them. If you did that, you would get perfectly sterile, fatigue sound.
a pretty empty comment to me - how many times do we have to refer to the Baxandal/Scroggie results with the additional urging of people to read the whole article, look at all of the results, extended predictions with high feedback? or point to the discussions, sims in Bob's feedback thread?
we don't design circuits with deliberate 5th, 7th harmonics absent others, above 20 dB negative feedback doesn't give peculiar cancellations that could be "tuned" - it reduces all visible orders of harmonics in the output and feedback above 20dB keeps on reducing all visible harmonics as the feedback factor increases
or remind people that the “harmonic growth” results are equally valid for degeneration, implicit “100% local feedback of followers – you are using “strong” negative feedback when you brag that you design circuit to be linear 1st
A signal will have a maximum dV/dt somewhere in its waveform. An amplifier will have a maximum dV/dt it can deliver (which is usually assumed to be a constant value, although conceivably it could be signal/memory dependent). If the latter is sufficiently greater than the former, then the amplifier can deliver the signal without slew rate limiting.
A question :
If the former is greater than the later, can we consider that the amp enters in a saturate state during the time this occurs (and maybe a little longer due to recovery) ?
A question :
If the former is greater than the later, can we consider that the amp enters in a saturate state during the time this occurs (and maybe a little longer due to recovery) ?
That would depend on the circuit. In most cases the max slew rate of the amp is limited by the current charging a capacitance. It all depends on what is producing the current, and what any negative feedback (if present) tries to do. I guess it also depends on what you mean by saturation. I don't think there is a generic answer to your question.
Bob,
Sorry, I do realize that there's no sense in beating a dead horse (except, of course, for the sheer joy of it <grin>), but...
You are probably correct as usual, about the accepted use of the terms. But it still doesn't make sense to me to think that the term "slew rate" should be understood to mean "maximum slew rate".
It makes more sense to me to think of slew rate as a rate of change: Slewing is synonymous with changing. So a slew rate is a change rate, or rate-of-change. So slew rate should not mean maximum slew rate. (So then I also see no reason that the term slew rate could not be applied to any variable's rate of change.)
Anyway, I only brought it up, originally, because I thought that using "slew rate" to mean "non-linear operation", as somone was saying, did not make sense. And I thought you even mentioned something to the same effect. But if "slew rate" actually does mean "maximum slew rate", then the poster who said it meant non-linear was actually very close to being correct, since anything infinitesimally more than the "slew rate" spec would give a non-linear output vs input characteristic.
Tom
Thats wrong, unless the slew rate test has changed. A very fast full level step input drives most amps into there non linear region until the feed back network has time to "catch up". In an LTP input amp the + in side goes to 1v while the - in stays at zero, this overloads the LTP input which is designed to stay below 1v/OLG.
Sorry,what I meant was that the max slew rate of an amp is measured while the amp is its non linear mode (more of a switch than an amp), does that make more sense?
By saturation due to a slew-rate limit, I mean what happens when there is no transmission of a modulated signal from the input to the ouput.
If, as in the above case, the capacitor is too slowly charged, the voltage difference between the inverting and non-inverting inputs of the amp increases so much that the input stage goes in overload and the negative feedback does not anymore control the output stage.
I have been accustomed to think about slew-rate by default in the same way, always precising "limit" or "maximum" when required with transient signals.
As I said, it depends on what you mean by saturation. Given your definition, then slew rate limiting necessarily involves saturation.forr said:
I tend to think of saturation as something like what happens to a BJT when the collector drops so low that the base-collector junction becomes forward biassed. Given that, slew rate limiting does not involve saturation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.