squarewaves for headphone measurements is it useful ?

[solderdude]

But whats the purpose of that? We already know the transducer is bandwidth limited. A simple sweep tells us that and characterises it in a more meaningful and

I already explained the purpose of the measurements. They are part of a measurement suite and it tells me somethings that other measurements don't as clearly. A simple sweep doesn't tell the whole story.
The results won't be that different if I fed a squarewave that is BW limited to 30kHz.
The goal is NOT to look for perfect squarewaves, they don't exist.
The same way as a squarewave is used to look into the boundaries of a filter which otherwise would not show clearly.
A sweep in a DAC does not tell you anything about its total behaviour (till the nyquist) either.

I am quite sure you won't judge a DAC or amp, tuner or whatever based on a single sweep and call it a day.
You don't measure the squarewave response of an amplifier with a 20kHz bandwidth squarewave either.
I see the value of this measurement. Perhaps you feel the measurement is non conclusive and not needed... please ignore the measurements and have a look at the FR plots only.

 

[March Audio]

What I am trying to establish is just what is it telling you?

 

[Krunok]

I assume he wants a bit more 'spice' in clarity/presence and tighter bass ?

Exactly.

Would recommend to look into the Beyerdynamic COPP (but modified as per my website) or HD58X (Massdrop only or via people not liking it)
Maybe DT1990 (with EQ) but there are more in my recommended list which is loosely based on my preference which, looking at the plot of Rtings is close to his preference as well.
Maybe the ATH-M50X ? Beyer DT250 ?
Will always remain a personal thing though.

I'll check them out, thank you!

 

[Wombat]

I read a few of your reviews on headphones I've owned or used. Your description of the sound seems dead on to what I hear of them. Especially the various Grados. I could hear in the first two minutes why people talked about them, and also why I couldn't live with them for more than about 5 minutes. I'd used SR325s which were the best of theirs I'd heard. The basic character was still there. It was better than the SR60's and SR80's I owned for a short time, but still the same sort of sound balance.

I've still have some DT880 Pro phones, and your description seems right on those also. I've always found them a bit bright or sharp at high levels, but at more normal levels I could live with it due to their good qualities elsewhere. I'll have to try the tissue paper trick. Reminds me of the T-shirt trick I'd suggest for early Thiel speakers. Cut a circle from a cotton t-shirt a bit larger than the tweeter and put it inside the cloth in front of the tweeter.

The t-shirt filter seems to be very severe. I am only comparing it to the subjective use of tissue paper over the Yamaha NS10M tweeter.

 

[solderdude]

W I am trying to establish is just what is it telling you?

How the headphone behaves after the membrane has moved position.
Could easily include a step response but don't want prolonged DC on a headphone driver.
Is there any other way to test for this other than either using lots of sine bursts at different frequencies ?

 

[March Audio]

How the headphone behaves after the membrane has moved position.
Could easily include a step response but don't want prolonged DC on a headphone driver.
Is there any other way to test for this other than either using lots of sine bursts at different frequencies ?

Thats fair enough, but there is no point in applying signals that will never be encountered.

 

[Wombat]

There comes a point where all of this difficulty goes beyond a difference in a hobby/interest and becomes obsession(anal).

 

[solderdude]

This thread was a side track of another thread... it may continue here... and ultimately end.

 

[Wombat]

In simple terms: https://www.sweetwater.com/insync/square-wave-response/

 

[solderdude]

In simple terms: https://www.sweetwater.com/insync/square-wave-response/

Rise time: The time required for the signal amplitude to change from 10 to 90 percent of the total square-wave amplitude.

Sadly this does not come to full fruition in my measurements. I suspect the mic is to blaim and perhaps limited BW of the mic amp.
On my measurements I cannot see risetime differences between the different (and fast) drivers, possibly due to the mic not reacting fast enough to a transient. (suspicion) but slower headphones are discriminated.

Overshoot: The amount by which the peak of the square wave exceeds its normal positive or negative amplitude. Large overshoots indicate peaking or excessive phase shifts in the high-frequency response of the device under test. Overshoot normally occurs only on the leading edges of a square wave.

This is easily spotted, just as undershoot and says a lot about any emphasis in the upper treble. This isn't as clearly visible in FR plots as those are log and scope shots are linear.

Ringing: The tendency of band-limited square waves to oscillate on the peaks. This is largely a subjective measurement with the results stated as the severity of ringing. The greater the high-frequency peaking, the greater the ringing.

Very visible here and again, due to the linear measurement shows something else than CSD. You can 'see' frequencies by counting periods in a certain time frame

Tilt: A measure of low-frequency behavior. As low frequencies are filtered, phase shifts are introduced which cause the leading edge of the square wave to rise and the trailing edge to fall at more of an angle. This produces a tilt to the top and bottom of the square wave. The tilt is usually expressed as a percentage of the peak amplitude of the square wave.

Very visible in both the 40Hz and 440Hz squarewaves.

Slew Rate: A measure of how fast a signal changes from one instantaneous value to another. The ideal square wave changes from one amplitude extreme to the other instantly. Practical devices cannot keep up with this transition and will often have a maximum speed with which they can change. This speed limitation will result in a tilted straight-line portion of the square wave edge.

Because of this I not only look at the shape of the waveform (especially the impulse) but also show it next to the signal which in itself gives a lot more info than just the squarewaves one sees at other headphone sites, which on top of this also show ringing of used steep filters as well as ear canal resonances.

 

[solderdude]

Thats fair enough, but there is no point in applying signals that will never be encountered.

The point is to, as instantly/fast enough (as possible) to move the membrane and stop it on a dime. For this only step, a squarewave or needle pulse is needed.
Just thought of another way but requires editing afterwards.
Gate some white noise and look how the membrane behaves when the signal is removed.
Maybe, someday I will build this, not so difficult to do.

For now I use a squarewave to 'move' the membrane's position.
All headphones are tested at the same SPL but due to efficiency differences at different power levels.

I welcome any other ideas for measuring aspects of headphones I haven't thought of, or considered well enough.

 

[March Audio]

The point is to, as instantly/fast enough (as possible) to move the membrane and stop it on a dime. For this only step, a squarewave or needle pulse is needed.
Just thought of another way but requires editing afterwards.
Gate some white noise and look how the membrane behaves when the signal is removed.
Maybe, someday I will build this, not so difficult to do.

For now I use a squarewave to 'move' the membrane's position.
All headphones are tested at the same SPL but due to efficiency differences at different power levels.

But the point I have been making is that the highest frequency signal it will encounter is typically 22.05 kHz. Thats as fast as it needs to move.

 

[solderdude]

Only true for RBCD , hires stuff will have higher frequencies. Also filterless NOS DAC's and slow filters will have signals far above the intended range. For DAC's we test well above the filter response and those signals ultimately will also be applied to BW limited transducers.
So testing higher frequencies seems to be realistic.
We both know square-waves are not realistic. They are a test signal to show the response of the DUT and test its extremes.

The test is to explore the resonances of the headphone. For that I want it to move as fast as possible. Ultimately my mic/pre-amp has limits that are reached before those of the fastest headphones. I am not looking for those boundaries though. I could not care less if it moved more rapidly than I can hear. What I am after is the resonances in the audible range.
I split this in 3 'bands' 40Hz (can show LF extension below 40Hz), 440Hz and 5kHz (the needle pulse is half a period of a 5kHz squarewave).
The squarewave excites more frequencies at the same time than sine waves.
Only gated white noise could do 'better' but don't expect much differences in the response.

I do agree that the HF content of music signals does not reach the same SPL as that of the test signal.

 

[Krunok]

I do agree that the HF content of music signals does not reach the same SPL as that of the test signal.

Isn't that a little bit of an understatement? I would expect HF signals to be much lower than the rest of the signal..

 

[pwjazz]

Thanks. That sounds good. We Germans tend to like listening to the Dutch language anyway.

As a speaker of German and English, I find Dutch frustrating because it sounds both extremely familiar yet unintelligible to me.

 

[maxxevv]

Rise time: The time required for the signal amplitude to change from 10 to 90 percent of the total square-wave amplitude.

Sadly this does not come to full fruition in my measurements. I suspect the mic is to blaim and perhaps limited BW of the mic amp.
On my measurements I cannot see risetime differences between the different (and fast) drivers, possibly due to the mic not reacting fast enough to a transient. (suspicion) but slower headphones are discriminated.

Overshoot: The amount by which the peak of the square wave exceeds its normal positive or negative amplitude. Large overshoots indicate peaking or excessive phase shifts in the high-frequency response of the device under test. Overshoot normally occurs only on the leading edges of a square wave.

This is easily spotted, just as undershoot and says a lot about any emphasis in the upper treble. This isn't as clearly visible in FR plots as those are log and scope shots are linear.

Ringing: The tendency of band-limited square waves to oscillate on the peaks. This is largely a subjective measurement with the results stated as the severity of ringing. The greater the high-frequency peaking, the greater the ringing.

Very visible here and again, due to the linear measurement shows something else than CSD. You can 'see' frequencies by counting periods in a certain time frame

Tilt: A measure of low-frequency behavior. As low frequencies are filtered, phase shifts are introduced which cause the leading edge of the square wave to rise and the trailing edge to fall at more of an angle. This produces a tilt to the top and bottom of the square wave. The tilt is usually expressed as a percentage of the peak amplitude of the square wave.

Very visible in both the 40Hz and 440Hz squarewaves.

Slew Rate: A measure of how fast a signal changes from one instantaneous value to another. The ideal square wave changes from one amplitude extreme to the other instantly. Practical devices cannot keep up with this transition and will often have a maximum speed with which they can change. This speed limitation will result in a tilted straight-line portion of the square wave edge.

Because of this I not only look at the shape of the waveform (especially the impulse) but also show it next to the signal which in itself gives a lot more info than just the squarewaves one sees at other headphone sites, which on top of this also show ringing of used steep filters as well as ear canal resonances.

So if I'm interpreting these correctly, if applied at specific frequency ranges, it would help explain / identify things such as sibilance and 'muddiness', 'rolled-off' and stuff of the sort ?

 

[Krunok]

As a speaker of German and English, I find Dutch frustrating because it sounds both extremely familiar yet unintelligible to me.

LOL - how true!

 

[solderdude]

Isn't that a little bit of an understatement? I would expect HF signals to be much lower than the rest of the signal..

As in 80 dB lower...

For digital files a squarewave is an illegal signal but for analog it is not.. Agreed it does not exist in nature but is not illegal nor should it be BW limted is the point I am getting at. It also shows ringing quite well but only about 20dB down or so.
It has its limits but is useful to me.

 

[solderdude]

So if I'm interpreting these correctly, if applied at specific frequency ranges, it would help explain / identify things such as sibilance and 'muddiness', 'rolled-off' and stuff of the sort ?

Yes, that's what I can see in those signals as well as in 'step response' in REW.
Sometimes the FR looks pretty 'flat' but still shows up as a hump in those plots.

 

[March Audio]

Only true for RBCD , hires stuff will have higher frequencies. Also filterless NOS DAC's and slow filters will have signals far above the intended range. For DAC's we test well above the filter response and those signals ultimately will also be applied to BW limited transducers.
So testing higher frequencies seems to be realistic.
We both know square-waves are not realistic. They are a test signal to show the response of the DUT and test its extremes.

The test is to explore the resonances of the headphone. For that I want it to move as fast as possible. ultimately my mic/pre-amp has limits that are reached before those of the fastest headphones. I am not looking for those boundaries though. I could not care less if it moved more rapidly than I can hear. What I am after is the resonances in the audible range.
I split this in 3 'bands' 40Hz (can show LF extension below 40Hz) 440Hz and 5kHz (the needle pulse is half a period of a 5kHz squarewave.
The squarewave excites more freuencies at the same time.
Only gated white noise could do 'better' but don't expect much differences in the response.

I do agree that the HF content of music signals does not reach the same SPL as that of the test signal.

Well I have examined many High rez files and few of them show anything significant other than noise above about 24 Khz. An example Talking Heads Track

A sweep can be used to look for the headphone resonances and get the impulse response. REW software works this way and will do this.

As for NOS dacs, well if you are interested in noise that shouldnt be there ......