Distortion Surface: old measurement, new approach

According to Tyll's measurements, up to 200-300 in the bass:

https://www.stereophile.com/images/ifmeasure/SennheiserHD595.pdf

https://www.stereophile.com/images/ifmeasure/SennheiserHD598.pdf

Tyll's Clear peak is a little lower than Bob's, probably because the measurements are on different 'heads' (different pressure-chamber effects on the driver resonance):

https://www.stereophile.com/images/FocalClearsnA1BRQE000007.pdf

 

Yeesh, so the 595 goes from 50 to roughly 225, ~4.5x

Clear goes from 60 to 500, ~8.3x.

I wonder if there are any others that have a higher max/min ratio.

I forgot I actually have the Innerfidelity allgraphs 2016 pdf here: https://www.dropbox.com/sh/d7a9pho6d71doan/AAAXJBpws7IjS4ln1WoKU0RMa?dl=0

Buuuut there's a thousand pages there to sift through.

 

The Clear is the highest ratio headphone I've seen (Tyll's measurement shows about 6x). A lot of multi-BA IEMs have pretty high ratios, and because their lows are commonly very low, their frequency response distortions are more extreme because of the way the voltage divider formula works with amp output impedance.

I've snagged that from you before, it's very handy to have. The online version is friendlier to navigate though:

https://www.stereophile.com/content/innerfidelity-headphone-measurements

 

No, I don't have one of those in my lab collection. Fortunately it appears that others are able to provide the requested information.

I should have noted that my measurements were free air.

 

Distortion in bass is less pronounced at lower SPLs which makes sense, but it seems like those two peaks at around 2.2kHz and ~5kHz are steady. Not artefacts?

Edit: I almost guessed the same about why midrange THD was creeping upwards but then I actually remembered to read the caption haha. At least I guessed right

Lemme guess how to interpret: notable bass distortion especially at high volumes but generally remarkably clean all throughout the rest of the audible spectrum save for a couple of peaks that could translate to a slightly grittier treble presentation? Basically: pretty dammed impressive.


Curious how a fully stock HD600 compares here.

Also, possibly idiotic suggestion but how much of a pain would it be to have individual surfaces for D2, D3, and D4 with each surface being translucent and superimposed on the same axes?

Might be interesting, like how you could glean bass quality on e.g. the Campfire Andromeda not being the best since D3 takes over the entire lower end with D2 underneath etc. Or maybe sticking to traditional "slices" at set SPLs.

 

It would be interesting so see that JAR600 THD surface decomposed into individual surfaces for D2, D3 and D4 through D10.

Conjecture: If most of the rise in distortion in the low frequency were due to D2, and if D4 through D10 were found to remain very low, that would be a really great result.

 

This is excellent work. What tools are you using?
ARTA for measurements?
??? for visualizations?
??? for automation from data collection to visualization?

 

These displays are very cool and look useful for comparing distortion components between headphones, but when presenting measurements of individual headphones I hope you'll still show individual SPL slices with all the harmonics on each. IMO the latter makes for easier assessment of how the different harmonics relate to each other and how that relationship changes with SPL. I found that really educational when you showed it for BA vs DD IEMs.

 

Fantastic guys
What a wonderful way to actually 'see' distortion vs frequency at the real world levels we listen.
Equally, being able to see the influence of 2nd, 3rd, 4th & higher harmonics is 'possibly' key to a components sound signature?
With different scales applies to source, amp, headphones, speakers.
Brilliant work... Thank you.

 

2. I gave this some consideration years ago but came to the conclusion that it is good to know the sins of commission that will be added to the particular frequency being measured. After awhile it became more intuitive.

 

0.02: distortion in terms of absolute SPL (A-weighted?) makes more sense definitely; percentage is a "neat" thing and pretty well established at this point but makes it hard to see the scale of things in a tangible way.

You experimented with presenting harmonic distortion AT the frequencies that they manifested at but I personally somehow had more trouble with those.

Seems more intuitive (imo) to keep with the older way for this one, e.g. "oh distortion is high under 200Hz AND it's primarily D3 so it's a wooly muffled mess (MDR-Z1R)" or "huh the Klipsch HP-3 has a weird sort of THD spike around 3kHz, mainly D2 but with some upper harmonics, coinciding with a CSD ridge around there— maybe that's why the treble is a bit less refined?"