Do many amps pass the square wave test?
square wave in = square wave out
Or is the test meaningful, correlating with the ability to reproduce music (complex sine waves) well?
Generally tube amps do not pass square waves (SW) well but many play music well.
How does the (SS but somewhat tubelike mids) of Alephs fare on the ‘SW test’?
square wave in = square wave out
Or is the test meaningful, correlating with the ability to reproduce music (complex sine waves) well?
Generally tube amps do not pass square waves (SW) well but many play music well.
How does the (SS but somewhat tubelike mids) of Alephs fare on the ‘SW test’?
rick57 said:
I don't think that the shape of square wave means the quality of the reproduced music.
I had a square wave in-out test for my new amp recently (see the thread named, Monolithic SuperSymmetry with Current Feedback). The test was predominantly to see the peaking of high frequency and to evaluate and control the margin of amp’s stability.
JH
It is not necessary that you get a rectangular out of your amp like the one leaving your function generator.
But what the average speaker does to a rectangular is much worse than the error introduced by a bandwith limited tube amplifier.
The difference is, that the bandwith limitations of an amplifier (ans speaker !) restrict slew rate and cause tilted top and bottom of the squarewave while a speaker also introduces phase anomalies (i.e the responses of the tweeter, midrange and woofer arrive in single file).
Phase (i.e. temporal behaviour) is important for the perception of sound direction.
Look at it this way: A rectangular is quite a simple shape. If a speaker is not even able to reproduce this one properly - how in all the world could it reproduce music, which is a lot more complex ?
As always: It is important that all properties of a speaker are well balanced. It does not help to have a perfect temporal behaviour at the cost of a lot of other response anomalies.
Regards
Charles
But what the average speaker does to a rectangular is much worse than the error introduced by a bandwith limited tube amplifier.
The difference is, that the bandwith limitations of an amplifier (ans speaker !) restrict slew rate and cause tilted top and bottom of the squarewave while a speaker also introduces phase anomalies (i.e the responses of the tweeter, midrange and woofer arrive in single file).
Phase (i.e. temporal behaviour) is important for the perception of sound direction.
Look at it this way: A rectangular is quite a simple shape. If a speaker is not even able to reproduce this one properly - how in all the world could it reproduce music, which is a lot more complex ?
As always: It is important that all properties of a speaker are well balanced. It does not help to have a perfect temporal behaviour at the cost of a lot of other response anomalies.
Regards
Charles
Below I repost what I already did within another thread.
It is the response to a squarewave of 200 Hz and 10 kHz, reproduced by a system with lower and upper rolloff of 12 dB/Octave and no phase anomalities inbetween.
For the lower signal frequency the effect of the lower cutoff frequency predominates (tilted top and bottom), for the higher signal frequency it's the upper cutoff that has the largest impact on the waveshape (finite slew-rate).
But you can clearly see that there are none of the typical multiple transients, at the beginning of each half cycle, that are typical for most multiway systems.
Regards
Charles
It is the response to a squarewave of 200 Hz and 10 kHz, reproduced by a system with lower and upper rolloff of 12 dB/Octave and no phase anomalities inbetween.
For the lower signal frequency the effect of the lower cutoff frequency predominates (tilted top and bottom), for the higher signal frequency it's the upper cutoff that has the largest impact on the waveshape (finite slew-rate).
But you can clearly see that there are none of the typical multiple transients, at the beginning of each half cycle, that are typical for most multiway systems.
Regards
Charles
square waves
I don't think square waves are good to judge the sound reproduction quality of an amp. But there are quite usefull to test the stability margin (which has an influence on the sound quality of course).
The amp should pass a square wave like the 2nd curve of phase-accurate's previous post.
The rise time of the edges determines the 3dB freq response point (3dB = about 1/pie the equivalent freq of the rise/fall time).
There should be not more than 10-15% overshoot (that's my target, YMMV). Then what I normally do is load the amp capacitively (from 10.000pF to 1uF) and keep an eye on the square wave response. For poweramps I use some cable between the amp and the cap load to introduce some inductive impedance, because it really isn't fair to ask the amp to cope with a cap directly at its output jacks.
The same with preamps, I use up to 1000pF to simulate (too) long high-cap interconnects.
If in any of these cases the amp oscillates or is marginally stable the square wave response will show very high overshoot and/or oscillatory bursts.
Jan Didden
I don't think square waves are good to judge the sound reproduction quality of an amp. But there are quite usefull to test the stability margin (which has an influence on the sound quality of course).
The amp should pass a square wave like the 2nd curve of phase-accurate's previous post.
The rise time of the edges determines the 3dB freq response point (3dB = about 1/pie the equivalent freq of the rise/fall time).
There should be not more than 10-15% overshoot (that's my target, YMMV). Then what I normally do is load the amp capacitively (from 10.000pF to 1uF) and keep an eye on the square wave response. For poweramps I use some cable between the amp and the cap load to introduce some inductive impedance, because it really isn't fair to ask the amp to cope with a cap directly at its output jacks.
The same with preamps, I use up to 1000pF to simulate (too) long high-cap interconnects.
If in any of these cases the amp oscillates or is marginally stable the square wave response will show very high overshoot and/or oscillatory bursts.
Jan Didden
This has not much to do with gain or delay. Since the topic was mentioned within a speaker thread of Richard as well, I posted the response of a phase accurate two-way speaker system.
The waveshapes are those of a wide fourth-order bandpass. The tilt is caused by the lacking ability to reproduce DC. All waveshapes are damped quite well and the tilt is not to be mistaken as overshoot, though one is led to believe this. The flattened slopes of the rising and falling edges are simply caused by the finite rising and falling speed.
Many amps would be better than this example in terms of HOW MUCH, at least at the lower end (where a very low first order rolloff is easily achieved with SS amps, even DC reproduction, but more difficult with tube amps). But the general SHAPE would be the same, as long as there is no ringing.
But keep in mind most speakers will be worse.....
As Peranders stated, rectangular tests are quite good for determining stability. It is not by accident one of the vavourite test signals for almost any feedback control system !
Regards
Charles
The waveshapes are those of a wide fourth-order bandpass. The tilt is caused by the lacking ability to reproduce DC. All waveshapes are damped quite well and the tilt is not to be mistaken as overshoot, though one is led to believe this. The flattened slopes of the rising and falling edges are simply caused by the finite rising and falling speed.
Many amps would be better than this example in terms of HOW MUCH, at least at the lower end (where a very low first order rolloff is easily achieved with SS amps, even DC reproduction, but more difficult with tube amps). But the general SHAPE would be the same, as long as there is no ringing.
But keep in mind most speakers will be worse.....
As Peranders stated, rectangular tests are quite good for determining stability. It is not by accident one of the vavourite test signals for almost any feedback control system !
Regards
Charles
squarewave and...
With speakers, it is a bit different than with amps or digital.
With speakers there is a problem of aligning two or more sections in parts which in sum give that what was the starting point. It is not a hard task in the term of the amplitude response, but the things become worst when phase come in story. As sad in another thread, assuming linear amplitude response, impulse (transient) response is directly related to the phase response.
Charles, can you give us some more info about the system whose graphs you posted? Thanks in advance.
Pedja
With speakers, it is a bit different than with amps or digital.
With speakers there is a problem of aligning two or more sections in parts which in sum give that what was the starting point. It is not a hard task in the term of the amplitude response, but the things become worst when phase come in story. As sad in another thread, assuming linear amplitude response, impulse (transient) response is directly related to the phase response.
Charles, can you give us some more info about the system whose graphs you posted? Thanks in advance.
Pedja
Square waves are a good test signal. But….
You need to band-limit the input signal between 50 kHz and 100 kHz with at least a 1st order filter. Otherwise if the slew rate of the input signal is too high you will drive the amp partly into limiting (saturation) somewhere. Most time the input stage.
Also a good test signal is a 80% 50 Hz sine with a 20% 1 kHz square wave added. It shows dynamic instability if apparent.
You need to band-limit the input signal between 50 kHz and 100 kHz with at least a 1st order filter. Otherwise if the slew rate of the input signal is too high you will drive the amp partly into limiting (saturation) somewhere. Most time the input stage.
Also a good test signal is a 80% 50 Hz sine with a 20% 1 kHz square wave added. It shows dynamic instability if apparent.
Square wave is a very useful signal to test the amplifier. You can calculate both amplitude and phase characteristics from step response or impulse response by means of Fourier transform, I do not want to get into more details.
High frequency limit (-3dB) can be calculated from rise time tr (10%-90%) as fh = 0.35/tr. It imeans that for tr = 1us fh = 350 kHz. From long time duration of square wave (low repetition frequency) one can see low frequency limit of the amp, it corresponds to exponential decay of square wave top line. Square wave is excellent for testing of stability of the amp together with diffrent kinds of load (capacitive etc.). But you have to use several different repetition frequencies of square wave, say 100Hz, 1kHz and 10kHz.
High frequency limit (-3dB) can be calculated from rise time tr (10%-90%) as fh = 0.35/tr. It imeans that for tr = 1us fh = 350 kHz. From long time duration of square wave (low repetition frequency) one can see low frequency limit of the amp, it corresponds to exponential decay of square wave top line. Square wave is excellent for testing of stability of the amp together with diffrent kinds of load (capacitive etc.). But you have to use several different repetition frequencies of square wave, say 100Hz, 1kHz and 10kHz.
Pedja
The examples shown are from a simulation of a two-way active system using an MSW (Manger Schallwandler).
It was done with a subtractive crossover and equalisation. The equalisation is necessary because the MSW is not that well behaved at it's lower end. More details and discussions can be found within the Manger thread.
I will try another principle for the MSW in the future and the Veroboard prototype of this crossover will find it's final destination in a phase_accurate box with a cheap fullrange (Monacor SPH68X).
Regards
Charles
The examples shown are from a simulation of a two-way active system using an MSW (Manger Schallwandler).
It was done with a subtractive crossover and equalisation. The equalisation is necessary because the MSW is not that well behaved at it's lower end. More details and discussions can be found within the Manger thread.
I will try another principle for the MSW in the future and the Veroboard prototype of this crossover will find it's final destination in a phase_accurate box with a cheap fullrange (Monacor SPH68X).
Regards
Charles
I dont understand this one, any competently modern designed amplifier has either a low pass filter included on the input in order to avoid this phenomena or is otherwise designed so that no saturation can occur.
BTW there is a misunderstanding that tube amps can not be good when reproducing squarewaves, even many transformer coupled amplifiers can be give quite respectful results and OTL amps can compete easily with any transistor amp. My own OTL amp have a rise time of ~ 1 us and no overshoot or ringing with capacitive loads up to 3.3uF.
Using square waves is usually a good method of testing amplifiers for stability and frequency response and is easy to use.
Regards Hans
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.