Headphone CSD waterfall plots - Page 22

Quote:

Originally Posted by LFF 

LOL!

 

Even more tempting now....

 

I bought an HD650 for the 3rd or 4th time today (yes, as in I've bought them and sold them that many times!).  I was already considering it and then the graphs in this thread put me over the edge.  I think I need headphoneus anonymous. 

Quote:

Originally Posted by purrin 

I don't consider the treble roll-off or more precisely the slight downward tilt as "recessed", but rather as something necessary for headphones which have the drivers blasting directly into the ears. 

Well.  These plots and my recent discovery of Isone (binaural speaker and room simulator for HPs) have kind of thrown a wrench in the way I gauge hifi & head-fi.  I've noticed that with Isone calibrated to sound like reference monitors 2 meters away in an anechoic room all my usual tracks sound warmer.  In essence it confirms what you say above and what I have suspected since I started using Isone, warm headphones are neutral HPs & neutral speakers sound like warm HPs.  

Quote:

Originally Posted by purrin 


Just a clarification from my notes. "some bass" what meant to mean "some decent/good bass." Hmm, very strange. The sample I auditioned did not sound like a Grado at all!? It most certainly did not sound forward. Instead it sounded like someone shoved the 2kHz lever up 10db on a cheap 5-band EQ. It was truly a horror. LFF said that I had them on for no more than two minutes, but I was actually ready to take them off after 15 seconds. I just didn't want to be impolite (LFF didn't give me any clue of what he thought about them), and also I wanted to check out a few more tracks just to make sure that I wasn't insane. I don't think my actual words were "messed up", they were a more colorful version of that phrase.
On the CSD (see overhead version below). What it says is that the membrane of the driver is vibrating like a drum at 2kHz and that this behavior has lasted for 10ms and is still continuing. This is one of the worst looking plots I've every measured; it's not even close,especially if you take into account the spotty FR at t=0. You can even see the severe gaps (at least -25db down) at ~7.5kHz and ~9.5kHz.

It sounds like we are talking about two completely different headphones. That would be some severe variation in builds.

Mine sound about 3-4dB up from 1.6Khz - 3.15Khz, and maybe about 3dB down at 8Khz (in comparison to test tones immediately surrounding the dip).

Not sure if this will help to uncover some mysteries.

We can see the FR variation and what the impedance curve looks like as well.  It seems it might be quite easy to hear a wide variety of T1 signatures depending on the variables.

http://www.innerfidelity.com/images/BeyerdynamicT1.pdf

http://www.innerfidelity.com/images/BeyerdynamicT1SN3964.pdf

Quote:

Originally Posted by purrin 

Tyll's measurements also include the effect of the headphone enclosures. You can say there is a peak at 9kHz on Tyll's FR graph, but you cannot say for certain there is ringing (or at least ringing caused of the driver.) IMO, the fleshy supercomputer that we have between our ears is very good at filtering out enclosure effects and hear what the drivers are actually doing. This is why the CSD plots are a useful supplement to Tyll's FR data.

Sorry about my ignorance, but how is that possible that your setup is immune to enclosure resonances? I mean, how the microphone distinguishes between an enclosure or a driver resonance? I can imagine that driver resonance may sound quite a bit nastier, but the microphone is something that cannot really understand the difference between a driver or an enclosure resonance.

About HE-500 I agree it is probably a case resonance. If it were a driver thing, it would probably be apparent on impedance graphs. Impedance graphs for all the orthos are pretty perfect, so one can conclude that there are no driver resonance in orthos? Maybe this is the main reason for the superior sound quality of ortho technology.

As a side note, I want to congratulate you, Purrin for this new approach about sound quality that you has so generously revealed to us. Resonances are very important to music reproduction. Indeed, music is more or less a harmonious dance of carefully tuned resonances generated from all the instruments!

I have found, in my small experience as a head-fier that Frequency Response alone is very limited. The brain can (over time) compensate frequency response differences (I would say as much as about whole 5 dB!). I have been experimenting with digital (linear phase) equalisation and have found that my brain, given enough time - about a week or so-, ‘acclimatizes’ itself to changes I do.  

As I understand resonances (enclosure or driver) are minimum phase phenomena. Do you thing they can be compensated with minimum phase filtering?

Did I see a CSD plot for the SR-007  in this thread?

 I thought I did, but then I used the "Search This Thread" tool on "007" and that didn't result in any SR-007 CSD plots.

Did the SR-007 get measured?

Quote:

Originally Posted by purrin 

 

 

No totally immune, just greatly reduced. It's a simulated free air measurement or "anechoic plate" that I'm using. In other words, the sound wave is not reflected between the surface (of dummy head) back to the inside of the cup. Instead it absorbed or passed through. I'm not giving away my secrets just yet. Only four people in the world (Tyll is one of them) have seen my setup.

 

 

Electrical impedance doesn't correlate with mechanical resonances associated with middle to high frequency ringing. If I did, there would be no need for me to take these measurements! (I would just look at impedance graphs all day.) Not all orthos are perfect. The LCD2 sample I measured had ringing at 7kHz and 9Khz. Also this is the HE300 (dynamic driver) we are talking about, not the HE500. I hope I didn't erroneously called label it the HE500! (Going back to check.)

 

It's entirely possible it's case resonance given the design - how of the dynamic driver is placed in the shell. However, I can't make excuses for it because of the following:

1. None of the other headphones, even those with similar circular cup designs exhibit this behavior;
2. The CSD plot exhibits what I heard. These were my impressions of the headphone before they were measured: Hmm. Wierd. Really midrangy. Some bass. Doesn't seem extended up top. Unbalanced tonally. A midrange bump at 1kHz. Some ringing around 3kHz? Not a bad spot for ringing to occur, but it's still there. Nothing offensive up top. Maybe some very slight ringing at two spots in the treble? This headphone just sounds messed up. Lacks resolution. Not fun to listen to. Taking them off.

 

You can use a low-pass or notch filter. 

 

 

 

 

Thank you for you response. I was talking about HE-500, not 300, resonance as seen to Tyll's frequency and impulse response graphs.

Quote:

Originally Posted by purrin 

No totally immune, just greatly reduced. It's a simulated free air measurement or "anechoic plate" that I'm using. In other words, the sound wave is not reflected between the surface (of dummy head) back to the inside of the cup. Instead it absorbed or passed through. I'm not giving away my secrets just yet. Only four people in the world (Tyll is one of them) have seen my setup.

Hi Marv, if you want to keep the sauce secret, you should stop using the word "anechoic" as it gives some serious hint ;).

Quote:

Originally Posted by Tyll Hertsens 

He can use my data, but my current methods of data acquisition seem to provide rather murky CSD plots. Slowly but surely looking into it.

 

Arnaud has a little conversion routine for my spreadsheets  Maybe he'll post a coupe here just for a look-see.

 

Purrin, I really admire the work you're doing dude.  Lots of information in those plots.  Thanks mang!

As Tyll mentioned, he graciously made some of his raw data available for me to extract CSD curves (see below). I have indeed created some macros in Excel to automate the data loading / reduction / and exporting of graphs. But as Tyll mentioned, there is potentially some issue with artificial dampening of the data and I hope Tyll will find some time to revisit this at some point (problem is that you'll need the headphones again to get some measurements done...). 

 

What drove me to doing this work for Tyll was reading about headphone CSDs from this very thread and then helping Marv to create his CSD processing routine. While I think Marv is going truly an excellent work and really got his measurement skills well tuned, I also very much believe people need to be careful with interpreting Marv's results. In particular, most result published are so called "anechoic" plate (e.g. ~100% absorbing of incidence waves), which is nowhere near the actual operating conditions of the headphone. Marv has clearly expressed that he was trying to make driver related resonances standout rather than those acoustic resonances in the earcup cavity but lately this CSDs are trying to be interpreted as faithful renditions of the headphone ringing, which is not quite right.

 

IMHO, every resonance (be it driver or acoustic one) should be looked at even though the CSD graphs can turn up ugly as a result. At the very least, Marv should post more of the following comparisons (http://www.head-fi.org/t/566929/headphone-csd-waterfall-plots/150#post_7767415 ):

 

 

Black: "Rigid Plate" Left Channel

Blue: "Rigid Plate" Right Channel

Yellow: "Anechoic Plate" Left Channel 

Red: "Anechoic Plate" Right Channel 

 

 

Quote:

Originally Posted by milosz 

Did I see a CSD plot for the SR-007  in this thread?

 

 I thought I did, but then I used the "Search This Thread" tool on "007" and that didn't result in any SR-007 CSD plots.

 

Did the SR-007 get measured?

I loose track of what I post where and you also probably actually saw the results and then forgot the thread ;). I posted some results here ( http://www.head-fi.org/t/572968/modeling-headphones-acoustic-performance#post_7774290 ) and here's the whole picture for the 007 vs. 009 comparison. Note that I am using InnerFidelity's background image because it is Tyll's data and he should be credited accordingly. But, just to keep it clear: this data was not officially acknowledged by Inner Fidelity and you will not find it on their website. It is just my shot at extracting CSDs from Tyll's data and although I could verify that the CSD for the whole impulse data matches the traditional FRFs from Tyll, I can't guarantee the results.

First a comparison of the classical magnitude response function to make sure my CSD post-processing in Excel was ok:

Next a comparison of SR007A and SR009 magnitude response functions (equalized for the dummy head response in typical listening environment, normalized by max SPL):

Finally, the CSDs extracted from the same (single position) impulse response data (equalized for the dummy head response in typical listening environment, normalized by max SPL):

Quote:

Originally Posted by arnaud 

 

As Tyll mentioned, he graciously made some of his raw data available for me to extract CSD curves (see below). I have indeed created some macros in Excel to automate the data loading / reduction / and exporting of graphs. But as Tyll mentioned, there is potentially some issue with artificial dampening of the data and I hope Tyll will find some time to revisit this at some point (problem is that you'll need the headphones again to get some measurements done...). 

 

 

IMHO, every resonance (be it driver or acoustic one) should be looked at even though the CSD graphs can turn up ugly as a result.

 Note that I am using InnerFidelity's background image because it is Tyll's data and he should be credited accordingly. But, just to keep it clear: this data was not officially acknowledged by Inner Fidelity and you will not find it on their website. It is just my shot at extracting CSDs from Tyll's data and although I could verify that the CSD for the whole impulse data matches the traditional FRFs from Tyll, I can't guarantee the results.

 

 

I would like to ask you if you kindly could do the same with HE-500, using Tyll's data to extract CSD curves. I know that Tyll expressed his consent in post 310. It would be very interesting to compare HE-500 CSD plots with LCD-2, especially after Tyll's latest praise of HE-500. 

It would probably be more useful if this thread was just Purrin's measurements because you can't compare Tyll's to Purrin's, as they're done with different methodology in different rooms on different equipment with different humidity and different foods cooked in the house, different cleaners used in the kitchen and on clothes, and different flora and fauna in the area, etc... :)

I may not be as software capable as you guys. That said, allow me to contribute my 2 cents.

I have been trained as a mastering engineer and continue to train my ears all the time - going on almost 12 years now. I wouldn't say I have perfect golden ears but they serve me well. I have noticed that all of the faults I hear with a headphone actually show up on Purrin's graphs more so than on any other measurement graphs I have seen. I find them to be reflective of what I hear and what I perceive in the sonic signatures of headphones.

I never tell Purrin my thoughts on a headphone before he listens to it as I don't want to influence his thoughts nor reinforce or weaken any bias already present, if any. After his listening, we share comments and I'm surprised how much in agreement we usually are. AFAIK, Purrin doesn't have the formal ear training I do but his aural perception is very accurate. This is confirmed with his measurements.

If you search in some threads, I commented about the problems with the HE-300 before they were measured. Purrin, AFAIK, didn't read those comments and he also commented on the exact same problems when he heard them. The measurements confirmed it.

Anyway, the point I am making is that I feel that Purrin's measurement methodology accurately portrays exactly what I believe I am hearing. I honestly feel that no other measurements I have seen accurately portray what I am hearing. They get close but not as close as Purrin's.