Square wave test

Hi!

Today I went to do some measurements on a Soundcraft 6000 console my friend bough. First I ran some sine waves to test distortions and pink noise for transfer curve plot. It all performed well. Excellent actually, since distortion was very low.

But then I ran a square wave from my interface into channel line in and monitored direct out from the channel.
It was not a square wave any more! Check the picture I attached.

On the right is a wave that is being sent to the channel, and on the left is a measured result. Obviously different.

But all other measurements are great and this console actually performs very clean and undistorted!

Tell me is this normal or I did something wrong?
Shouldn't the resulting waveform be also flat-top?

I tried on many channels and the waveform measured is allways the same!

That's the HPF on the desk (you won't expect to DC path trough the strip) (too small the plots to do any math) but for a single CR filter you would expect something like that in a low freq square, try with a higher (1k maybe) and that effect will be smaller.

Let's say that I see a 10% of the value going down, so ln(0.9) will be about -0.1 which means you have the period of your square 1/10th of the time constant, about 10Hz pole, so -3dB point of your strip (with test stuff around included) will be at 10Hz. just rough math because I can't see the relation in 1 division of the scope. What did your freq response measure? where was the low end -3dB point? If you go too high you may the rise time of the square wave to be longer. Both effects are because of the limited band of an audio signal path. In any case, trust your freq/phase response better, You may want to see an step response more than a square wave response, it will say a lot about impulse response of the desk (without putting an impulse which harder to get and dangerous if big enough)

JS

Some tilt is normal...

But square wave are not naturally occurring signals so use judgement when testing with them.

JR

Thank you Joaquins and JR for your responses!

I didn't test it with a higher freq. square tone. I forgot to do it!
While I guessed that this is normal to some extent, just as JR confirmed, I couldn't help wondering why.
So from what I understand this is HPF from the DC blocking cap and RC filter on the input and output stage of the strip (and probably on the DAC and ADC of my audio interface that I used for measuring), right? I didn't test the loopback response on my audio interface to see to what extent it occurs without having channel strip inserted into the path.

But my measures of the freq. response and phase are all good. And I'm not worried about this test. I was just wondering why it happens, and Joaquins explained it well!

Here are the freq. and phase measurements.
Vertical resolution of the freq. response is +/- 1 db only so low and high roll off is actually only about 0.15db at the extremes.


Luka

Hi,

What kind of software do you use ?  (multiscope ?)

Thanks,
Cheers,

For the squarewave test I used Cubase's built in Multiscope plugin, and for other measurements I've used Smaart 7

A 0.15dB at 20Hz gives a -3dB at 4Hz, so my so roughly estimation of 10Hz isn't way off and you may have two or more poles from another part of the chain. Remember the transfer function is gutted from Smaart which is compensating your converters, there is no problem with that, and may probably came from the converters. You should do the square at your converters only or use another soft that gives you the transient or transitory response (I don't know how you spell it) as fuzz measure for mac, don't know for windows.

JS

In Smaart software there's a impulse response measurement, but I didn't use it. This is probably the transient response test you mention.
I was using borrowed laptop with Smaart from a friend who uses it for live gigs and I'm not too familiar with all the measures it has, but next time I'll try to run some Impulse tests to see how it performs.
And I'll definitely try to run a squarewave from output to input of the audio interface! It didn't cross my mind to do that while I was at it!

Try those and we'll see

JS

Square waves are fine for testing individual components like transformers. However for testing products that have any frequency shaping like filters or EQ will mangle square waves mking the measurement meaningless.

Chuckles said:

Square waves are fine for testing individual components like transformers. However for testing products that have any frequency shaping like filters or EQ will mangle square waves mking the measurement meaningless.

Click to expand...

While not completely meaningless, phase shift that alters the harmonic relationship between the many partials that make up a square wave can complicate simple visual observations. Two square waves with similar energy content (and sound) can look dramatically different on a scope image.

Square waves for testing audio paths will generally contain higher frequency content than real music, so experience and judgement is required to properly evaluate results.

IMO it's a good idea to use sine waves for testing until you understand the idiosyncrasies of square wave testing. .  If you have experience and a good handle on how they behave you can make quick judgements when comparing multiple similar paths (like in a console)  using that one signal source.  From observation of a well behaved path you can interpret HF and LF response from how the square wave looks.

JR