Yes , my "slew rate failure"
amp outperforms the tested 100v/us one in both HF THD and bandwidth.Sounds a little better , too. Both exceed what is needed for full power 100k operation , so it (better slew) does not matter. The "faster" one is a lowly EF2 .. another impediment to performance at high current.OS
cbdb,
Most of what you think seems correct but you have the wrong idea about what "slew rate" means.
EVERY time-changing voltage ALWAYS has "a slew rate", which is measured in volts per unit of time at the output, even if it's only changing at 1 volt per year.
Saying "slew rate" says NOTHING, NADA, ZILCH about linearity.
Would a 1 volt per year slew rate necessarily mean that something is operating in its "non-linear" region? "Not necessarily".
I guess that you must be thinking of "slew-rate limited".
When people talk about the "MAXIMUM slew rate" spec for an amplifier, they are referring to the fastest slew rate for which an amplifier does _NOT_ go into any non-linear region.
So an amp that has a max slew rate spec of 40 V/us would still be operating as linearly as it usually did and sound fine if the signal never made its output change at more than 40 V/us. But if you pushed that same amp to 50 V/us, THEN it might get into non-linear regions and the sound could be bad.
Tom
cbdb,
Not necessarily. Slew rate and bandwidth are only very-loosely related to each other (if at all).
If you wanted to try to relate slew rate and bandwidth, you would have to pick an amplitude, first; probably the maximum output amplitude, since that would be a worst-case condition for the slew rate. Even then, I don't think you can relate them in a very meaningful way (see below).
With a single frequency (sinusoid waveform), the maximum slew rate at one amplitude would be different than the maximum slew rate at a different amplitude. A larger amplitude at the same frequency would have to have a larger maximum slew rate (which, by the way, would always occur at the zero-crossing point, for a sinusoid waveform). See post # 92.
The bandwidth's upper limit, on the other hand, only tells us where (above what frequency) the gain can no longer be sustained and has dropped by 3 dB, which, for a fixed input amplitude, would be when the output amplitude has fallen by 3 dB.
I think it's impossible to really relate bandwidth and max slew rate, very well, since the "bandwidth" definition doesn't specify that we still have to be in a linear operating mode at the -3 dB point.
Getting past all of that, if you look at my equation for the maximum slew rate for a sine wave (given any frequency and amplitude), in post # 92, you will see that the slew rate for audio frequencies (for a single frequency, anyway) gets NOWHERE NEAR the max slew rate specs of the amps in question. BUT, solving backwards for "max frequency while still operating linearly", given the max slew rate spec and the maximum output amplitude, MIGHT, I think, tell us the highest frequency that each amp could perform well at, i.e. still staying linear, with the worst-case maximum output amplitude. So maybe that could be some measure of the "speed" of an amplifier(?).
I would be tempted to guess that an amp that could go "faster", well, could also do "slower" better (i.e. more accurately). But that's just a guess.
Tom
Last edited:
You can't "push" it. 40 is max ... period. At 250khz,the output level attenuates by not only the input LP filter (which I Have set below this point) , but by the slew, as well. When this happens , THD rockets and the waveform even looks asymmetrical(or non-sinusoidal) . This same effect happens at about 600k on the faster 100v/us LIN topology amp.
Ps - ridiculous .... we are not after an AM radio broadcast amp !!
No , but it retains the lowest THD to the highest frequency of the 2 ... by far.
OS
Last edited:
It is not necessarily right because fast video opamps (made for hf signals) are getting hot with nf-signals. Possible temperatur drift could affect everything.
But we are drifting here with the topic away.
Let us focus on how an opamp influences or even creates a wide soundstage.
Post #113 gives a hint:
OK. Maybe not THAT amp, but one that didn't have an input filter. Anyway, I think I was only trying to make the point that if the "max slew rate" spec was somehow exceeded, then "bad things" could happen, but below that slew rate it should still be operating linearly.
Maybe. Don't know. But I have been assuming (and have seen no good-enough reason to change that assumption) that I am much more interested in using opamps (and other amps) in ways that result in the most-accurate reproduction of the source, at the final output, so that I can experience whatever soundstage is inherent in the source material, as accurately as possible, i.e. whatever was intended to be given to us by the artists and the recording engineers.
Last edited:
To elaborate
1 below is fast , can reach 500k ... but with .05% THD. measured 100v slew into a 8R resistive load on CRO.
2 below is "slow" can't reach 500k, but can go to 150k with no thd increase.
It might be more of a topology/total loop gain effect at work here ???
(1 is 115 db OLG and 2 is only 60DB.)
.
OS
1 below is fast , can reach 500k ... but with .05% THD. measured 100v slew into a 8R resistive load on CRO.
2 below is "slow" can't reach 500k, but can go to 150k with no thd increase.
It might be more of a topology/total loop gain effect at work here ???
(1 is 115 db OLG and 2 is only 60DB.)
It would just naturally attenuate (I simulated without the filter) , whether that is bad
. OS
Attachments
Last edited:
the number has to be normalized for output V swing - typical estimates by amplifier designers as conservative goals use 50-200 KHz full amplitude sine wave zero crossing slew rate as their estimate
160 KHz is conveniently ~ 1 Mrad/s giving ~ 1 V/us /Vpk
when actual music signals have been looked at their slew rate is much less - some suggest most music spectra plots show ~3-5KHz "power bandwidth"
as I mentioned earlier there really aren't any commercial music distribution formats with >100 KHz power bandwidth - some analog media do have "small signal" bandwidth approaching this - but records can't be cut, tracked at full amplitude at high frequencies, tape heads roll off from several mechanisms - analog tapes have a AC bias that must be filtered out with frequencies ranging from 80-200KHz where you want large attenuation, not just the corner frequency
apropos THD
Here I show the result of a calculus of the RMS error computation of a complex signal (or more real 'music' approximation). The calculation process is not elegant but practical and easy to understand, it consist in a discrete sweeping over the BW (small steps better) and calculating the fourier amplitudes (more harmonics better) of the input signal*gain and the output signal, then calculating the RMS error of each harmonic and integrating RMS errors. It is a massive computational consuming process, so I show only 6 results (the input sine waves combination) and the final result of the integration. Interestingly the final error (unwanted harmonic content for the whole input material) is about 2 orders of magnitude higher than the worst THD at that frequency. Given that -40db RMS levels of distortion maybe audible (with no masking) I conclude that THD's higher than 0.01% should 'change' music nature (timbre) in a way more or less pleasant but audible.
Here I show the result of a calculus of the RMS error computation of a complex signal (or more real 'music' approximation). The calculation process is not elegant but practical and easy to understand, it consist in a discrete sweeping over the BW (small steps better) and calculating the fourier amplitudes (more harmonics better) of the input signal*gain and the output signal, then calculating the RMS error of each harmonic and integrating RMS errors. It is a massive computational consuming process, so I show only 6 results (the input sine waves combination) and the final result of the integration. Interestingly the final error (unwanted harmonic content for the whole input material) is about 2 orders of magnitude higher than the worst THD at that frequency. Given that -40db RMS levels of distortion maybe audible (with no masking) I conclude that THD's higher than 0.01% should 'change' music nature (timbre) in a way more or less pleasant but audible.
Attachments
Textbook definition of slew rate is the maximum rate of change of a voltage. What you're defining is just rate of change.
Big confusion
Slew rate is the maximum rate of change of a signal.
The terms I was using appertain to amplifier signal handling ability. In this context, the various signal processing theorems and transforms useful in telecommunications or for radar signal analysis can safely be disregarded.
Slew rate is the maximum rate of change of a signal.
The terms I was using appertain to amplifier signal handling ability. In this context, the various signal processing theorems and transforms useful in telecommunications or for radar signal analysis can safely be disregarded.
there's no need to get hung up on semantics - we use words in different contexts, understood by willing participants who want to communicate instead of trying to "score points"
Slew rate does come from the idea of a velocity limit most likely earliest use was for naval guns which couldn't be turned faster than a limit given by the available power
its use/definition probably entered the EE world in a big way during WWII due to trying to build analog computers for antiaircraft guns, bomb sights
in the EE world talking about the slew rate of a particular signal is pretty unambiguously referring to the maximum time derivative of the waveform
you can even use both EE senses of "slew rate" in the same sentence and most engineers will not even notice the ambiguity
the slew rate limit (modifier often omitted) of an amplifier refers to the maximum time rate of change of its output in the limit of a much higher slew rate input signal
"power bandwidth" refers to the highest frequency sine wave having the maximum undistorted amplitude where the amplifier's slew rate limit starts causing added distortion as the frequency increases
Slew rate does come from the idea of a velocity limit most likely earliest use was for naval guns which couldn't be turned faster than a limit given by the available power
its use/definition probably entered the EE world in a big way during WWII due to trying to build analog computers for antiaircraft guns, bomb sights
in the EE world talking about the slew rate of a particular signal is pretty unambiguously referring to the maximum time derivative of the waveform
you can even use both EE senses of "slew rate" in the same sentence and most engineers will not even notice the ambiguity
the slew rate limit (modifier often omitted) of an amplifier refers to the maximum time rate of change of its output in the limit of a much higher slew rate input signal
"power bandwidth" refers to the highest frequency sine wave having the maximum undistorted amplitude where the amplifier's slew rate limit starts causing added distortion as the frequency increases
Last edited:
On the slew rate subject I would agree that going overboard is not necessary. As pointed out above, there is not much energy in normal music above about 5kHz. I think the trick here is to ensure that the amplifier bandwidth is set not by it's slew rate limitation, but by it's front end filter, or, the source signal. So if your amp has a 40V/uS slew rate, ensure through the use of the appropriate filter that any signal coming into the amp does not exceed this value. This of course is not rocket science, and is well documented.
My Ovation amp is 100V/uS, but the frond end filter is set to 160kHz. I measured the rise time with the filter disabled at 1us.
My Ovation amp is 100V/uS, but the frond end filter is set to 160kHz. I measured the rise time with the filter disabled at 1us.
Agree to filter the input, even if it is not real life, step response stimulus is another test that every amp should pass ...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.