Burst Response! HD800, SR-207, HD650

Hah, heck if I know, all that stands out to me at 1500Hz is that initial attack is eager and that swing back towards rest position is protracted. Not sure if those warbles are related to the resonance in cup chambers somehow, add to the romanticisation of the sound. Curious what's keeping the trace from going back over middle point with that second little bump.

500Hz is surprisingly... free of features? Scary good compared to other cans even, basically textbook perfect unless I'm very mistaken about all this.

 

Ahh, gotcha. Was focusing on the wrong frequencies, haha! Fricking hell, headphones with a built in whammy bar.

Dumb question time, how is it shown here that there's a phase shift? Might have been mentioned elsewhere but if so, missed it. What really stands out to me at 4800 besides the tail end is how there seems to be a bit of "tightening" in response between the 4-6th cycles before coming back up to level. Not sure what might be causing that.

 

Yup, I didn't know to read them that way. Thanks Marv! Now going back to check the other ones out haha.

 

These burst responses are certainly an interesting measurement, it seems like a very easy to understand way to visualize things that are not visualized in other measurements.
Ill have to learn more about all these types of measurements.

I wonder if I could do something on types of acoustic measurements for my senior capstone project next year... would certainly be interesting. Ill have to see. Would have access to all the nice equipment and anechoic chamber though.

 

That's really weird. Maybe it's the other way around that high frequency ringing and overshoot from the starting of the signal makes it reach the peak earlier than normal. Here I overlaid the 4800Hz over the 1500Hz burst. Aside from the first cycle peaking earlier than the other one it seems fine to me.

Maybe we're also seeing some distortion here*, although I find it hard to believe that less than 10% distortion is visible by looking at the waveform directly. It could cause a shifting of the peaks in the waveform however, as seen here:
https://www.stereophile.com/images/archivesart/310arc.fig4.jpg or here:
https://www.passdiy.com/projects/images/content/distortion2.png

Did you visualize them so that it normalizes for the start of the waveform? I'm wondering if 4800Hz itself might be delayed vs other frequencies, although the 50Hz would be delayed far more. Actually VituixCAD lets us check that from the FR...
EDIT: According to VituixCAD 4800Hz is delayed by about 0.05ms relative to 1500Hz for your RS2 measurement here. That's about 1/4th of the 4800Hz cycle. 50Hz is delayed by 3ms, though, so I assume you normalized the start for the visualization.
*I think it may look slightly tilted towards the right, but I could be imagining things.
FWIW, here's the GD I got from the FR measurement: https://i.imgur.com/wy219Mr.png

 

Yup, I've thought of creating FR graphs in the future where the soft flat plate coupler respones gets overlaid past 5/6kHz on top of my in-ear measurements. Makes you realize much more easily which peaks are caused by the headphone itself and which are caused by interactions between the headphone and the ear.
The other (interesting) option would be to DIY a dummy head with absorptive ears, but which still have pinna gain. Some of the denser foams might work for that.

Maybe our ears end up hearing it that way, but I don't think 'frequency shift' is what's happening here***. Let me try to explain how I thought it happens.

At the start of the signal the 4800Hz tone and also a small step gets triggered.* However because 4800Hz is some 10dB lower relative to the nearby peaks (2.2, 4 and 10kHz) which will get triggered by the "starting" of the signal, this is more obvious compared to other headphones.** Similar story with a burst at 4kHz on an HD800 on a FPC.
*(Signal is time-limited so it cannot be bandlimited. Since it's not a real step with infinite slope but rather the start of the sine, the bandwidth is smaller than infinite, but the nearby peaks will likely have not insignificant output.)

Since 4800Hz is a bit delayed compared to other frequencies (almost 1/4th of its own cycle vs 1500Hz, see above) and since the impulse has a much faster rise-time we end up with a peak in the output that is a bit earlier than we would expect for a 4800Hz sine.

Something like this:

The output (green) would be a superposition of the 4800Hz sine/burst and the blue curve.
Note that I've made it look like the "impulse" has a similar frequency as the 4800Hz sine. It's probably a bit lower - there's a peak at 4kHz and around 6kHz there's less GD than at 4.8kHz, so it would start earlier. Both of those could play a major role here. It's very unlikely the "added" signal would be as neat looking as this. It's really just a superposition of all the other frequencies aswell. Not a periodic signal.

**Take the worst-case example. A system with almost no output at 4800Hz. The starting of the 4800Hz tone would still trigger a band limited version of its impulse or step response centered around that 4800Hz center frequency. After that decays we would be left with almost no signal and at the end we would see the same impulse, but inverted. Or at least that's what I think.

It's still delayed, just less delayed vs 4800Hz. I think that it looks weird on an FFT is because the FFT is only an approximation aswell.
***Our ears are just a weird FFT-like device after all.

 

@Hands isn't that just the same thing with a dB dBFS scale instead of percentage?

 

Serious got it above, it's just displaying as either percentage of total (the sine waves) or dBFS [decibels relative to full scale] (the doodoo bloops). COULD BE WRONG, but reason for percentage looking like sine waves is probably just that the trace indicates "swing" of diaphragm relative to rest state while the dBFS is... how much noise there is— stuff coloured in is noise, background is absence.