That is correct. Thus to the designer the objective is to keep all "solid state" elements out of saturation. In those old days if someone wanted it really fast he had only one choice: emitter coupled logic. This always operated in linear region no saturation.
Last edited:
Do not confuse slew rate and rise time.
Slew rate is a parameter of a NON linear amplifier behavior when its input is over driven. The response has typically a roughly constant slope.
Rise time is a parameter of a linear amplifier when it is not pushed into a non linear behavior. The response is typically a variable slope that start steeper and ends flatter.
There is no rule to compare one another. It doesn't make sens to use SR with some derating to estimate linear speed. They can be close or far away, it all depends of the detailed design. Slew rate, a non linear parameter, is useless in audio.
Another aspect of "How fast do you need" is how fast audio signals are ?
How fast a 100W on 8 ohm amplifier will really see in real audio signals.
I doubt we need 100W at 20Khz. Tweeters and ears would not survive it anyway.
Slew rate is a parameter of a NON linear amplifier behavior when its input is over driven. The response has typically a roughly constant slope.
Rise time is a parameter of a linear amplifier when it is not pushed into a non linear behavior. The response is typically a variable slope that start steeper and ends flatter.
There is no rule to compare one another. It doesn't make sens to use SR with some derating to estimate linear speed. They can be close or far away, it all depends of the detailed design. Slew rate, a non linear parameter, is useless in audio.
Another aspect of "How fast do you need" is how fast audio signals are ?
How fast a 100W on 8 ohm amplifier will really see in real audio signals.
I doubt we need 100W at 20Khz. Tweeters and ears would not survive it anyway.
The reason for rise time instead of slew rate is just comparability. With step response measurements of speakers musical instruments....one cannot overdrive a speaker. As long as magnetic induction is constant the voice coil exerts a mechanical force F=B*L*I what the mechanical system's response to the step is ..fast Fourier etc. In either system we are in linear mode of operation
I think you are exaggerating a bit, although I agree with you that a 10 kHz full-power THD test is more useful than slew rate limiting values. When you stay far below the slew rate limit, you can at least be sure that there will not be any gross distortion due to slewing or due to sub-slewing TIM or SID. But indeed, how quickly distortion drops with increasing x when you stay a factor of x below the slew rate limit is very dependent on the amplifier design.
To play "Deutsche Marschmuzik - Einzug der Gladiatoren" at full volume, you need to be able to process a full volume 8 kHz sine wave with little distortion.
I do not know about all amplifiers, but in the case I looked close at slew rate, there is no need to stay far below the slew rate limit.
This case is a very simple amp based on a high voltage op amp. A 100W / 8 ohm amp. It does distort at full power for frequencies higher than 12Khz because of slewing. I am sure it is from slewing because of the amplitude frequency relation where distortion starts raising.
This slew rate limit is quite below the datasheet SR in the op amp datasheet.
The slew rate limit I am talking about here is the linear behavior limit, not the datasheet's SR.
The raise of distortion comes in very sharply this is why I say there is no need to stay far below the slew rate limit. A bit below is enough for no more distortion than at lower amplitudes. This is pretty much as sharp as distortion that comes in when clipping.
That 4 to 5 V/us figure is strange, the LM4780 datasheet specifies 8 V/us minimum, 19 V/us typical. Could it be that it isn't slew rate limiting anymore when you reach 90 % of the step?
In any case, the datasheet also specifies good full-power distortion figures at 10 kHz and higher, so the LM4780 should have no problem with Einzug der Gladiatoren.
My simple view is based on the fact that if you cascade amplifier stages then the cumulative effect is to give the lowest bandwidth. So if a disk is played with 20kHz BW then a 20kHz amplifier is going to seriously limit its performance. Therefore I tend to design to 10x BW (200kHz) which is possible. For 100W amplifier into 8 ohms that means 50V/us, but I can't tell much (any?) difference with 30V/us. Not tested anything less, since 30V/us is quite easy to achieve too.
I don't think speaker systems for 100W will put 100W into the treble unit. The crossover networks usually attenuate the HF - depending on the tweeter design of course.
I don't think speaker systems for 100W will put 100W into the treble unit. The crossover networks usually attenuate the HF - depending on the tweeter design of course.
ok but is BW full power BW or "half power"? To put in perspective....a tube amp is internally pretty much faster than semiconductor but the transformer for 200 kHz full power BW is a luxury you cannot afford. I am not an expert in tube amps but learned the Williamson amp had been equipped with a rather sophisticated transformer but its power BW ...i guess certainly not above 20 to 50 kHz at a very high price. And yet there are few if any complaints that much less sophisticated tube amps limit performance of standard compact disk in any way. I think the 30V/us answers the quest for "fastest audio amp".
You make a good point about transformers. I'm not sure a 200kHz wide-band transformer is possible (unless using ferrite - then LF disappears) let alone a luxury. Sowter's best designs, last time I looked, were maxing at around 100kHz, only better with mumetal; higher power units were generally 40-80kHz. It does appear, though, that tube amps seem to deliver the goods without too much trouble.
Some figures from the maximum rise rate of a sine signal.
Vpeak x 2 Pi x Frequency.
For a 100 W in 8 ohms, Vp is 40V
At 12 KHz Max rise rate is 3V/μS
At 20 KHz Max rise rate is 5V/μS
Such an amplifier that is not capable of this will distort at full power, but there is no need for margins or for some insidious effect of slew rate.
Vpeak x 2 Pi x Frequency.
For a 100 W in 8 ohms, Vp is 40V
At 12 KHz Max rise rate is 3V/μS
At 20 KHz Max rise rate is 5V/μS
Such an amplifier that is not capable of this will distort at full power, but there is no need for margins or for some insidious effect of slew rate.
First of see my posts about Slew rate definition.
To get at the non linear behavior slew rate as given in op amp datasheets. Simply input to the amp a large amplitude square signal and see the output signal.
To get at the linear max rise rate, I input to the amp sine signals of various amplitudes and frequencies and see the THD. The idea is to see at what amplitude x frequency there is a sharp THD increase.
At low amplitude x frequency THD is ok, at high amplitude x frequency THD is bad, at some value it goes from ok to bad, that is the point I look for.
To get at the non linear behavior slew rate as given in op amp datasheets. Simply input to the amp a large amplitude square signal and see the output signal.
To get at the linear max rise rate, I input to the amp sine signals of various amplitudes and frequencies and see the THD. The idea is to see at what amplitude x frequency there is a sharp THD increase.
At low amplitude x frequency THD is ok, at high amplitude x frequency THD is bad, at some value it goes from ok to bad, that is the point I look for.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.