10 KHz Square wave output?

[boraomega]

Many people are using square wave signal for testing and estimation of amps performances. But factor that is seldom taken into account or stated is rise and fall times of the used square wave signal, which is of outmost importance. Many amps will looks pretty good with, say, 1uS rise and fall time square, with minimal or even completely absent ringing, but try it with 20nS rise time square and you will se a hell of a different picture. Slew rate of such 20nS signal is somewhere about 300V/uS allowing you to test all but fastest amps. Try adding this shaper after your sine signal generator, for it will produce square wave from sine with right about 20nS of rising and falling time.

[fab]

Hi boraomega

I do not understand how a real audio input signal could produce such a square wave that fast in the amp. Could you explain.

thanks

[AndrewT]

Hi,
I'm no expert and still learning but you are right. No conventional audio amp will ever reproduce that fast an input signal.
The real point is that you are injecting a test signal to help you determine that the amp is operating correctly.
Those very fast edges really stress the electronics and if they are not quite right, odd looking traces appear on the oscilloscope.
Finally your RF filter on the input will round off the the corners of the square wave by quite effectively slowing down the rise time of the test signal. In fact this could be an argument for easing the RF filter for testing & then returning to the longer time constant for ordinary (audio) use.

[boraomega]

FAB, you are partially right. Average real audio input signal is never that fast and you will find that somewhere about 8V/uS is all the speed you need to be more than "fast enough" with usual average output power (say... 10-30W) used in room listening (higher power amps needs higher Slew Rate figure to maintain quality and capabilities). But don't forget that in a good number of high quality recordings very fast transients exists which "requests" much better Slew Rate figure than 8V/uS to be reproduced faithfully. If amp's Slew Rate figure is not "up to the job", it will cause something known as "slew induced distortion".

AndrewT already gave you second part of explanation.

[Duo]

This has become a very interesting thread about slew rate. I like what's being said.

Definitely, I'm glad someone mentioned the nyquist theory on sound cards. That's such a true thing.

As far as slew induced distortion, I've listened to a great many amplifiers in which that occurs at high power level.

I should do some more serious research on this.

[amplifierguru]

Square wave testing gives much information about an amplifier's bandwidth and stability in one simple visual test as has been said, but the test gear needs to have a capability beyond the amp unity gain bandwidth of NFB closure.

You need to disable any input filtering when tuning so the full bandwidth of the amplifier is stimulated by the harmonics of the sq wave. Then you can tune compensation C's and evaluate reactive load behaviour. Then the input filter (usually 100KHz +) should be reconnected to remove unwanted HF garbage and RF pickup.

The slewing rate of an amplifier only needs to exceed that of the highest frequency expected at full power - usually 20KHz which would be 2u7S for a 30W amp and double for a 4 times, 120W. Anything above this band should be filtered out but a simple 6dB passive input filter below about 100KHz will start to rollof the 20KHz. Having said that the distortion of an amp generally rises dramatically approaching the slewing rate limit with high level modulation of current usually at the Vas, so for low THD through to 20KHz it a good measure to have a higher slew rate.

I have designed amplifiers which have an infinite slew factor where the input (and feedback) filters cut in at 12 dB/octave before the amp slew rate is reached - so it cannot be slewed with normal signal levels. Rather acadaemic with 22KHz CDs.

Greg

[mura]

Hello and thanks again for all inputs,

So if my amplifier has 110V rail voltage to what extent should i drive the amplifier with a 10K Sq wave?

[amplifierguru]

Hi Mura,

If you are testing the amp for the first time, start off with a small square wave with a , say, 1 MHz BW - a generator with sq wave output to 100KHz should do this. 2.83Vrms is OK. Look for square and smooth rising and falling then try 1uF. See if ringing is well damped.

Then crank it up to say 10Vrms do the same tests. It's still only 12.5W. Should handle that easily and well damped.

Nows the time to adjust any Ccomps for smoothest rise time i.e. quickest rise/fall without overshoots. Then try 2uF.

It should instill confidence at this stage, before trying full power SINEWAVE test at 10KHz. Ensure no triangulation or fuzzy waveform indicating oscillations AT ANY POINT on waveform. Check no zero crossing notches.

Good luck.
Greg