A few random comments: You can't speed up slew rate by adding feedback. It's an intrinsic limitation of the voltage-gain and output stages. Reproducing music doesn't require much slew rate at the amplifier output, because the high-frequency amplitudes are low. A 40Vrms sine wave at 20KHz requires a slew rate of 7V/us, but no one listens to such signals and that is at least an order of magnitude beyond what music requires. However, slew rate of the voltage-gain stage is important for overcoming crossover notch distortion from class AB output stages. No amplifier needs to be able to make nice fast square wave edges at frequencies above ~7KHz (or even above ~4KHz if the listener is a senior citizen). No one can hear the difference between a 7KHz sine wave and a 7KHz square wave; neither can anyone hear harmonic distortion of signals that high.
Colorado dreaming from your beautiful terrace and those mountain views of the Rockies and Flatirons. I own the BHK's Preamp and 250 Int Amp and I am very happy with the musicality and "Slew" rate. Thanks Paul!
Good modestly priced audio amplifiers can accurately reproduce a square wave in the audible range of frequencies. Far more important is the loudspeakers ability to handle transients. Most box/cone drivers have less than perfect transient response compared to the amplifier, so choice of speaker along with room acoustics will swamp any amplifier characteristics. Electrostatic speakers have a very light and fast diaphragm and can actually produce a reasonable square wave and so have much lower distortion compared to a box speaker.
This goes back to the discussions of loudness and IMPULSE accuracy to the original sound. It seems that the playback gets closer to the real deal when the volume goes high for many high end systems. That's what I hear. This also gets into the reason why my class A triode wired amps sounded good to me. I just needed to turn up the volume. All that information floats in the recording only when amped appropriately. The dealer blew out tweeters in speakers back then as I demoed systems due to the 1st order crossover and the need to crank it to make an impression. Solid state amps simply couldn't do it at any volume.
For an open loop amp (i.e. no feedback) the high frequency bandwidth is easy to calculate. It is f(max) = SlewRate/(6.28 Vmax). So for a preamp with a slew rate of 100 volts/microsecond operating at +4 dBu, the max voltage is 2.45 and thus the bandwidth is 6.5 MHz (wow!). But, for a 1500 watt power amp (into an 8 ohm load) with the same SR of 100 V/us, the bandwidth is 'only' 103 kHz (still quite respectable). The real problem with slew rate is when feedback is introduced: The output signal is limited by the slew rate, so when we feed that signal back to the differential input, its signal will be different from the input and thus the difference signal will no longer resemble the input which will result in some nasty distortion. We have a name for this, it's called Slew Induced Distortion, or SID.
You're talking about full-power bandwidth, which is a different concept than the amplifier's linear bandwidth. The problem with slew rate exists regardless of feedback.
@@marianneoelund2940 That's absolutely correct. As you know, this is a difficult concept to explain to most subscribers without losing them. Increasing negative feedback increases bandwidth and decreases gain and effective output impedance as well, not too mention it has a profound impact on SID, TIM and THD, but that's a lot for non-engineers to grasp.
Love your videos, Paul. Thank your for giving of yourself in creating these valuable lessons; your thoughtful demeanor and humorous style are delightful to enjoy and alway brighten my day, and your apparent kind Humanity is reflected well.
Paul has a great personality and his videos are fun to watch. However, they also create work for the real engineers, who are often dismayed at his blunders and feel compelled to spend time correcting them. Example: His claim that adding feedback to an amplifier can improve its slew rate, is as valid as saying that adding a cruise control unit to your car can increase its rate of acceleration.
@@marianneoelund2940 - point taken, Marianne; "the numbers don't lie". However, it appears that Paul is doing SOMETHING right. I am certain that Antonio Stradivarius had his critics who disagreed with the sound of his violins, and yet, history has proven that Stradivarius had "the right stuff". To this very day, even tho they possess the original blueprints and records, rocket scientists and engineers STILL cannot reproduce the historic Saturn V rocket engines, because all of the talented "old school" engineers and technicians have died with their knowledge. Whenever engineers tell me that something is "impossible, ridiculous" I find myself musing over an article I'd read years ago explaining that according to "the numbers, Bumble Bees can NOT fly". Or, in the words of Mark Twain "They didn't know it was impossible, so they did it". Thanks for your response. Kind Regards, Mark PatrioticMillionaires .org
@@modemmark421 Thankfully, we don't need to reproduce engines like the F1, because we have better solutions today. Still, if it ever became necessary, and sufficient funds were available, it could be done; it just takes time. If we did it once starting from scratch, we can do it again. And the space shuttle RS-25 main engine is nothing to sneeze at either, with more than 400K lb of thrust.
@@marianneoelund2940 Yes, I am glad to see that we are in agreement; just as we "old timers" inspired those like the new rocket scientists, it appears that you are on track to improve on Paul's contributions to our world, as well. Thank you, in advance. I am boundlessly hopeful and breathlessly inspired by your generation... "curiosity and an open mind are great travel companions". I wish you a Buddha-ful Life.
Slew rate and damping are no longer problems for modern day amplifiers of any moderately reasonable caliber at all. Transient challenges relate to tweeter types and abilities.
Great explanation of a moote term. I never fully understood the term previously. Thank You Paul! (..and, if I lived in a home such as yours, I would only go to work once or, maybe, twice a week, just to steer the ship in the right direction!..)
@Fat Rat LOL!!.. Well *Fat Rat*, yes indeed! For any wife to 'loose' a husband to another woman is bad enough but, to loose him for a pair of speakers?..!!! : -0
If you want accurate transient response get studio monitors. They are powered and bi amped specifically for accurate transient response. The woofer and tweeter have there own amp (bi amped) so the distance between the amp and speaker is very short. I recommend cheap krk rokit 8's or yamaha hs 8's for getting into good speakers. You will be blown away by the sound quality for the price and no need to buy amps. Studio monitors always go overlooked in the hifi world for some odd reason even tho they often sound the best.
I wonder why feedback to fix the slew-rate issue sounds "miserable"? I mean it might be a band-aid (I have no idea, I'm not an electrical engineer), but if it fixes the problem, what other problem does it cause to make it sound miserable?
Feedback can't improve slew rate. Adding feedback does not cause "miserable" sound. Feedback reduces distortion, but if the amplifier runs out of slew rate, it introduces an *uncorrectable error* and feedback loses some of its authority to cancel the distortion.
Transient response isn't entirely related to an amplifier ... it's related to EVERY single stage along the line ! Starting from the recording ( especially with any compression .. MP3 stuff and all the rest ) the slew rates of the stages within the amplifier have to matched as well so that faster stages don't precede later stages .. the power supplies must be good enough by virtue of the quality of the components really. Speakers ultimately play out the total result in the end ... suffice to say low mass drivers are the only way to go here ..
Thanks, Paul. Why do manufacturers not mention anything about slew rate or transient response in their specifications or descriptions? I just looked for this information on the websites for Emotiva, McIntosh, Bryston, FM Acoustics, and PS Audio. None mentioned this for any of their products. Is this factor found in some other specification that would be given?
Hare deLune Yes, engineers measure a lot more data than marketing folks are promoting and often those data makes a much more valuable judgement of the audio fidelity. Some times it’s a bit idiotic, like an amp with a THD advertised at 0.01% connected to a speaker that has a THD of 1%, where nobody talks THD 🤔
Slew rate started getting into the mainstream back in the 70s when it was discovered that amplifiers that used a lot of feedback created something that became called Transient Intermodulation distortion and the fix was to use devices with greater slew rates and less feedback. Once that became the norm, there was little reason to talk about it or include in the specifications.
Stewart Markley: Thank YU KIND SIR. You have nice way of getting to the point with all the necessary information and not 1 unnecessary worde. THUMBBE SUPS!
1) Feed Back cannot improve slew rate, on the contrary, Feedback always reduces the Open loop Amp given slew rate. The Higher the feedback loop Gain is (open loop amp gain X Feedback Gain) the lower the Slew rate obtained (Total amp gain is reduced, Frequency response is increased but slew rate is decreased, distortion is decreased, Damping factor increases). 2) The higher the Gain feedback loop is, the less stable the amplifier become. the most unstable situation is for unity gain amplifier. this is the reason for most liner OP AMP to add a very low frequency roll off open loop gain characteristic.
Ideally a power amp just multiplies the input voltage to output voltage no matter whatever speaker impedance load. Great amps do that fairly precisely within the audible frequency range but your speakers are where things get messy and the transient behavior is not only about cones starting motion fast enough but even more importantly stopping fast enough not having ringing decays.
@@marianneoelund2940 I'm not sure what you mean by a "voltage source"? Actually the ideal power amplifier is a source of amplified voltage that doesn't get impacted by the speaker load impedance (which translates to ideally an infinite damping factor also). A speaker load is usually not close to a simple constant resistance and great amps can keep the same output voltage even at the frequencies where the speaker impedance gets to the lowest peaks, which for some speakers can be below 4 Ohms. Of course the ability of keeping constant voltage at low impedance translates into ability to drive the needed extra current (and thus power).
Yes, however by using a heftier power supply you can improved, also depends on the load on the particular circuit and how is the circuit is design. In the case of a single BGT driving an LED though a 1 Kohm resistor, for example, connected to a low impedance scope between the collector and emitter of the transistor you will see a near perfect wave. Of course all depends on the transistor you choose, however if your goal is instantaneous raise and fall you need to pick just the right transistor. One can be the MJE15032G wich switches at 30 MHz but there are way faster reaching 700 GHz. All depends on reaching the point of diminishing returns.
The trick for fast BJT switching is to keep it out of saturation. The 30MHz parameter for MJE15032G is its fT, which is current gain-bandwidth product. That doesn't mean it can switch at 30MHz in practical circuits.
@@marianneoelund2940 Ok, true, However at 30 MHz is way adobe the human sound band with. and a transistor like the one mentioned goes a long way to improve the 'Transients' Are you from Scandinavia? I detect Scandy on your last name. No offence at all!
@@mtabernig Yes, it's actually Øelund. I kept my maiden name and my father is from Denmark. One very fast amplifier design is the Onkyo A-9150 which is supposed to achieve 500V/us. Input stage uses 4 pairs of 2SA1312/2SC3324 in a symmetrical arrangement (NAD uses similar input topology). Voltage gain stage is KTA1360/KTC3423 with 2SA1200/2SC2880 drivers. Output drivers are again KTA1360/KTC3423 but they have current boost provided by TTA004B/TTC004B. Output transistors are 2SA1962/2SC5242. It might be interesting to compare the KTA/KTC and TTA/TTC part specs to the MJE device you suggested.
@@marianneoelund2940 Yes, it would be. I am an old engineer that started my carrier with tube amplifiers and never got convinced that solid state was superior. When in the early times the TO package was the best power transistor had nothing to claim from a tube like thre 6V6 or 6L6 on push-pull. those were the days. My late wife name was Karen Lund, also Danish . Nice to make your acquaintance. mtabernig@socal.rr.com
@@marianneoelund2940 Impressing amplifier and plenty of power for a home theater. Very nice. Was it built in Japan? or they had a Japanese version and an American version?
Why TUBES have a very slow slew rate compared to a modern solid-state amplifier? It is it because of the numerous negative feedback, etc...? Tube amplifier with low slew rate sounds very lifelike compared to an amplifier with transistors and a fairly high slew rate?
This is a good point, in fact I'm not even sure if you can see slew rate limiting in tube amplifiers, because their modest bandwidth might keep you from getting there. The tubes themselves can produce very high slew rates, but the output transformers can't pass such fast edges.
That's just it. In order to have a meaningful measurement of slew rate (and some others) you would have to disconnect the amplifier's feedback network (if any) which no one is going to publish. Many amplifiers would not work at all in such a state.
No, you do not need to disconnect the feedback in order to measure slew rate. Paul was completely wrong about improving the slew rate of an amplifier by adding feedback; it can't do that. Neither was he right in claiming that slew rates of 100V/us or higher are needed for transient response with music signals. What he calls a "really bad" slew rate of 1-2V/us is sufficient to reproduce an extremely sharp sound (like ceramic dishes banging together) at full power from a 200W/channel amplifier.
Dear Paul, I’m intrigued about your mention of square waves. I learned from my engineering studies that a 10kHz square wave is made up of a 10kHz sine wave (the “fundamental”) plus lots of higher order harmonics. In theory, an infinitely steep square wave would contain infinitely high harmonics (ie. where the wave switches instantaneously from full negative to full positive amplitude). Put it another way - if you want to be able to render a “perfect square wave”, your amp would need to have an infinitely high frequency response. However, the latter never happens, and neither do our ears have the ability to hear it. So, my point is: Aren’t you simply describing an amp with a higher frequency response? ie. The higher the frequency response, then the steeper looking is the resulting “square wave”?
Slew rate limiting is a separate effect from edge risetime. The limited frequency response of an amplifier produces a certain rise- and fall-time for the edge of a square wave, which is essentially a constant independent of amplitude. As a decent approximation, Tr or Tf is 0.35/BW. The edges have the shape of a first-order decay, for example a step input that takes the amplifier output from -10V to +10V would fit the function V(t) = 10 - 20*exp(-t*2*pi*Fc), where Fc is the amplifier's rolloff frequency (bandwidth). For example, an amplifier with 50KHz BW has a Tr or Tf of 7usec (measured from 10% to 90% of the step edge), and that remains 7usec independent of step amplitude - until the amplitude becomes large enough to encounter the dreaded slew rate limit. If you push an amplifier to its slew rate limit, then the step edge takes on a linear ramp shape and the overall waveform is trapezoidal. The rise and fall times will increase as the amplitude is increased further. While the output is slewing, the amplifier has fallen behind tracking the input and is playing "catch up."
Marianne, if we take Paul's very *worst-case* numbers (which he states in his video): ie. a 50volts swing; @ 1 volt per microsecond slew rate (ie. taking 50 micro-seconds to get there) - Q: In your subjective opinion, above what audio input frequency do you think the average adult human would detect that the amp is struggling to "catch-up"? (Or would any resulting artefacts be outside our normal hearing range -- ie. anything above 15kHz)?
@@nitram419 In the case of 200W/chan class amplifiers, the most difficult audio signal I've found so far (clattering ceramic dishes) would require 2V/us slew rate to reproduce cleanly at full power. An amplifier that size with only 1V/us slew rate could produce audible distortion in some special cases, but again only if pushed to high power. Slew-induced distortion will occur at frequencies above 3KHz at 200W (8 ohm load). But at a lower volume of 50W, distortion doesn't happen until 6KHz is exceeded, and now we're getting into a region where the distortion products are challenging the human hearing range; no one can hear symmetric distortion of a 7KHz or higher sine wave. But there is another consideration besides direct slew-induced distortion. The slew rate limit is usually set at the voltage gain stage. The ability of the amplifier to correct for transient distortion such as crossover notch distortion is improved by having higher slew rate capability in that stage. For that reason, slew rates 3-5 times higher than needed for full-power sine waves at high audio frequencies could be justified.
@@marianneoelund2940 Ok thanks Marianne. Good to know that one has to be at the extremes -- both in terms of the chosen audio material and the amp's gain level.
@PS Audio: Hey, Paul. That looks to be one lovely home you have there! On this subject I'd like to inquire on your opinion of bridged amplifiers, which are said to have better slew rate compared to their single stereo iteration. Most seem to feel bridging amps brings with it a list of technically specified cons that overshadows the pros (if any?) that may follow. Myself I'm positively inclined towards the bridged amp constellations I've heard, because to my ears it usually sounds better this way, even though it's heading me on a collision course with what appears to be theoretical consensus! This question may have been better suited to one of your video blogs, if singled out, but hopefully you'll consider my question here instead.
There is absolutely nothing wrong with bridged configurations, provided each amp is capable of handling the load impedance it sees (half of the speaker impedance).
Almost no amplifier today will have difficulty with transients as an effect of inadequate slew rate. Transient response issues arise more from an amplifier's deficiencies with reactive loads, but much more important than that, is the speaker's transient response.
I feel like I've had "Transient Responce" and the Damping Factor mixed up for quite some time. Do they go hand in hand when we are talking about amplification or am I way off?
@Fat Rat he's a bit prickly. He doesn't like sharing his money or his time with others. Paul and others with like minded politics would not approve of Mr Numan's selfish Thatcher supporting ways. ;)
"Audiophile, How Do I?" That channel name that I just made up is from a new trend starting on RU-vid that I kind of like. This could have been called that. Maybe Paul could cover some of the more basic things like SHOWING how to do things like connecting cables the best way or other really basic things where even people who may believe they are fairly knowledgeable may be missing something and not even know it. Hmmm, got myself thinking. Dammit!😄 Stay tuned.
"...even people who may believe they are fairly knowledgeable may be missing something and not even know it." With that, my friend, you have hit upon a truth more profound within the audiophile community than you can possibly imagine! The problem then becomes, that you can lead a horse to water, but you can't make it drink. 😔
@Fat Rat I was thinking in terms of things like Why VTA Matters, how to make your vintage gear last longer, how to not rip your jacks out with high end cables, how to tell which speaker has your amp pissed off stuff like that. Simple for most of us here and maybe even explained elsewhere but with little to no working knowledge.
@Fat Rat Because people are resistant to change. Because sometimes new information can be 'too radical' for someone who is comfortable with the way things are. Because different people (including audiophiles) have different priorities. I'll give you one small example: If you want *the* absolute *best* sound quality in anything from a listenable system on up, minimalism is the only way to go. There is absolute proof that simple, direct, short signal paths sound better than anything else. Now, just try convincing a person who relies on an EQ or DSP in their system that their sound will improve if they give that stuff up.
Yes, it's very good. But for music, 500V/us capability is never used, so it's rather a waste. If I were you, I'd hook that amp up to a transmitting antenna, and go on-air with it.
@@Enemji Music doesn't move anywhere near that fast. A 30W/chan amplifier only needs a slew rate of 4V/us to produce a full-power 20KHz sine wave, which is the most you ever need because a 20KHz square wave sounds the same as a sine wave.
Why don’t speakers have a slew rating? It would seem to me that almost all electronics will have a better slew rating than (any) speaker, so, if this is true, then the slew rating coming out of the amplifier is meaningless if the speaker can’t match the squareness of the wave.
hey paul ,couldnt help but look at your floor boards and how the gaps dont line up.both panel ends should have been cut at the correct same angle and then all of the gaps would have aligned up.
Let's quit worrying about square waves and start looking at actual music. What's the measured rise time of a cymbal strike? A snare drum whack? What's the fastest transient made by a real musical instrument? I'll bet that it's a lot slower than we think. On this measured basis, what slew rate do we need, through the entire system, to reproduce it accurately?
Worst case is the slew rate you need to produce a 20KHz sine wave at the amplifier's full power. You don't need square-wave capability at all, for frequencies ~7KHz or higher. For a 200W/chan amplifier, 7V/us is sufficient. Real music signals require less than that. Paul's claim that designers should have 100V/us or more as a goal, is overkill.
@@marianneoelund2940 Many years ago Nelson Pass wrote about this issue, and actually measured out the slew rate needed to reproduce a cymbal crash using the fastest phono pickup of the time: the Panasonic strain gauge cartridge. I believe Sao Win used a modified one for his turntable and cartridge system. Nelson got very surprising results similar to what you wrote, which varied according to the power amp output. It was a shocking outcome during the height of the Great Slew Rate Wars.
@@darthbubba866 Brings back some memories. Mechanical slew rate at the phono needle is another potential issue, and if not adequate will cause mis-track distortion that sounds like clipping. I had some LPs back in those days that required a higher force setting for the stylus. Today we fortunately have much more capable sources, and in my case I've been experimenting with live sounds rather than recorded sounds, so I often have spectral content over 30KHz included.
@@darthbubba866 Testing progress with high-energy sound sources: [Setup - Shure SM81 mic, Mackie 802VLZ4 mixer, Tektronix MSO4104 scope in high res. mode at 1Msa/sec. Data is scaled up to just touch +/- 56V peak to fit voltage swing of a 200W/ch (8 ohms) amp.] I've found the most energetic sources are small glass or china cups, struck with a hard object. Fastest edges I've recorded would require 4.3V/usec from the power amp at threshold of clipping. Signal energy is concentrated at 2, 5, 9, 13.5, 18, 19.5, 21 and 23KHz, with weak harmonics appearing over 40KHz. Decay time constant is about 20-25msec. But this would be a level that is dangerous to ears and tweeters; I certainly wouldn't want to be in the room with a signal like this going into the speakers. I also looked at sound from the top notes of my piano, but didn't find anything that would require more than 2.2V/usec from the power amp. With the 4th of July coming up, I might have an opportunity to check some small explosive devices.
@@marianneoelund2940 Impressive test setup! I'd never have thought to use a small glass or china cup for a fast attack sound with supersonic overtones. Fourth of July, firecrackers, hmmm... 😁
The question should be how fast of a rise time is needed for music to be reproduced accurately. Also the slew rate needed is dependent on the voltage level of the device, less for preamplifiers and more for power amplifiers.
@blackened 1 No, rise time only states the time to go from 10% to 90% of a waveform, it does not include the voltage or current factor which slew rate includes.
@@StewartMarkley And more importantly, rise time is a linear phenomenon which is amplitude invariant, while slew rate effects are highly amplitude-dependent.
For a 200W per channel amplifier, slew rates of 1-2V/us are adequate for any conceivable music signal, in spite of what Paul believes. It only takes 7V/us for it to reproduce full-power 20KHz sine waves, but (I hope) no one listens to those. But higher slew rates can be beneficial to allow the amplifier to correct crossover distortion more accurately.
Paul, could you please harness your admin power and delete all my comments from your channel. i'm sick of those internet witchhunters keep pestering me even though i have been absent from this channel for like half a year already, these sick people just won't stop, as if they had been holding a grudge against me since an eternity and a half ago. just delete them, they'd just trigger more lunatics down the line.
Which Nyquist are you referring to? The sampling theorem applies to converting between continuous- and discrete-time domains. The stability criterion has much to do with good amplifier design (although not related to slew rates), but the audiophile community is blissfully unaware of it, so far.
@@marianneoelund2940 The Niquest quantum sampling theorem cannot differentiate wave form, only zero cross in the time domain. So correctly according to Paul the delta at cross is what we are trying to measure. Nyquest definds no difference between square triangle saw tooth trapezoid and sinusoidal near upper bandwidth of a audio amp.
Class D amps are an interesting case. It doesn't depend on output stage speed, as that's typically capable of switching at more than 2000V/us, but there can still be a slew rate limit after the output filter, caused by the design of the control circuitry. I've encountered that issue with some switch-mode power supplies that recover slowly when a load suddenly turns on, because an integrator in the control path has gone into saturation. With class AB designs in the 200W/chan range, a slew rate of 20-40V/us is plenty to ensure that all of the potential benefits are realized. Any more than that, is just marketing fluff.
HELLO mine fredns! the PRESCENCE on Marshall amp was on the Fenders before it. I saw a video wher they sai dits a feedback in the POWER AMP part of the guitar amp. NOT the preamp. Is this the article you were intereste din... The presence control, on the other hand, is part of the power amp section of your amp. Technically, it operates as a “high-frequency shelving boost,” which is a fancy way of saying that it can increase high frequencies but not decrease them. You can turn high frequencies up but not turn them down. THANKES YOU AND THUMBBE SUPPES!!!
Transients on the amp and preamp are simple compared to what happens to transients during recording/reproduction. This is a principle reason DSD kills PCM.
Transient response? Music is filled with transients. Think of the opening blat of a trumpet or the pluck of a string. That first transition from quiet to loud is called a transient response. The better able a piece of electronics can handle the rapid response from zero to whatever is the transient.
and then you feed that perfect transient into a speaker with a 2nd or higher order passive crossover which then destroys that perfect transient from your amp . Makes perfect sense ....
I want some of your THINGS so this message if about HOPE. - m. *TAX DOLLAR TECHNOLOGY* Created (1. APPLE COMPUTER (2. MICROSOFT (3. AMAZON (4. FACEBOOK. fk 401k crap. I WANT MY TECHNOLOGY Investment UBI... *worth TRILLIONS!*
