[Nak Man]

Today I was measuring a tweeter for my old maggie and wondered what kind of FR readings I would get using this mic in front of phones. Put it about 1 cm in front of 595 cups and was extremely surprised to see smooth response compared to regular speaker readings ! Four hours and 16 charts later, I was glad that despite this simplistic method some of the charts reflect my impressions on how those phones sound in real life.

Sound coming from white noise cd -> amp -> phones -> mic -> nak deck -> soundcard through line in. Area below 100hz tend to reflect ambient noise, and bottom part is comparison to old faithful 595. I wish I knew if there's a smoothing function in this program so that they can look as nice as Headroom's and not as jagged ... anyway here's some of those quick and dirty charts:

Default 'build a graph' 650:

Similar sound signature of my 531 with my 595:

Surprisingly smooth 501 .. and there's that edgy sibilant:

Way to go sr80 ! peaky peak ...

Maybe it's better to use sweeps or pink noise but I wasn't trying to achieve what Wolfgang accomplished ... please keep this holiday spirit high as to refrain from flaming ... and post all complaints to moderators.

Cheers.

Edit - Special thanks to GSferrari for hosting the pics for me, hopefully there will be no more act of disappearing pics.

 

[gsferrari]

I dont see any images?

Using Windows XP Pro SP1 + Firefox

 

[Nak Man]

Shoot, they were there a minute ago ... linked to yahoo freebie site. Come back tomorrow.

edit - wth is this ? shown when i tried to access the linked file.

Sorry, this site is temporarily unavailable!
The web site you are trying to access has exceeded its allocated data transfer. Visit our help area for more information.
Access to this site will be restored within an hour. Please try again later.

 

[gsferrari]

dude - email them to me and I will give you the links to use in your post

check your pm for email addy

 

[hugz]

Well i can see them

 

[Nak Man]

Two more charts of favorite phones:

Koss 'bass n treble' SportaPro:

HD580:

I was tempted to guess that shallowing of 580's midband below 1.5k as 'laidbackness', but due to out-of-head measurement, mic positioning can give +/- 2 to 3db differences, large enough to swallow 580/595 midband variance. And of course I can be totally wrong to call laidbackness simply by looking at FR chart. Any moral in this story remains along those with larger than 3db gaps in charts, for example like allover wide differences found here:

Philips 'best bang' 890:

It could be that hp890 operate as 'semi closed', considering the rough similarity of their pattern to those of 271s' and 25sp's (grossly bowled midbass). Or they're not fully burned in yet.

edited to move picture hosting

 

[Nak Man]

Just for benchmark here's my speakers' L vs R chart with added leaf tweeter, using same measurement at 2 meter distance - but with wider scale (50 vs 35db). Really jagged compared to those above but without dips and peaks of phones' acoustical 'correction'. Just ignore the bottom part - those are with original faulty tweeter plus some experiments with subs. I'm still lazy to work on those tilted highs.

Maggie 1.4:

 

[Andrea]

Absolutely interesting, Nak Man -- almost professional looking measurements.

 

[Nak Man]

Thank you for your kind words Andrea !

Those jagged edge made supposedly smooth lines look confusing. =)

 

[BowerR64]

what graph program is this?

 

[Nak Man]

Old version of SpectraLab ... almost won't run on current windows.

 

[Nak Man]

On mysterious 271s .. with what I perceive as some strange effect on mids, but can't really pinpoint what sort of anomaly. Due to closed nature of 271s I used a small cardboard to seal the cups with mic poking through small punched hole. Not quite good seal around mic though. Also re-read 595 (red trace) with same cardboard method. Please ignore the large bass hump on 271s (brown line) as I hold the phones in my hand and even a slight brushing finger caused big jumps on chart.

What observed was a discontinuity or sharp step between 500hz-1k.

Very interesting (for me, at least =) to see how sealed measurement pushes the high end down to reflect more speaker like FR (except for that sharp dip on 4.5k). Worse slope for 595 since its angled driver didn't face mic directly.

Cheers.

 

[WmAx]

The carboard will resonante signficantly, likely causing problem with the measurement. As well probbly the joint/connection between the carboard and microphone. Also, the chamber you have created causes severe resonances/nulls related to the distances in the cavity you created. Also, a repeatable and accurate positition of the mic(a jig is needed) to transducer is needed when making any differences to the headphone then measuring for difference assesmet. Ideally, a non-interfernce(dampened acousticly, efficevitly mechanically inert for purposes of these measurements) couplling device and positioning jig is needed to accurately assess the transducer and rear enclosure properties.

-Chris

Quote:

Originally Posted by Nak Man On mysterious 271s .. with what I perceive as some strange effect on mids, but can't really pinpoint what sort of anomaly. Due to closed nature of 271s I used a small cardboard to seal the cups with mic poking through small punched hole. Not quite good seal around mic though. Also re-read 595 (red trace) with same cardboard method. Please ignore the large bass hump on 271s (brown line) as I hold the phones in my hand and even a slight brushing finger caused big jumps on chart. What observed was a discontinuity or sharp step between 500hz-1k. Very interesting (for me, at least =) to see how sealed measurement pushes the high end down to reflect more speaker like FR (except for that sharp dip on 4.5k). Worse slope for 595 since its angled driver didn't face mic directly. Cheers.

 

[taymat]

Interesting looking charts and to me the hp890 one does show a major problem I immediately became aware of when I got my rs-1s. They're very shrill-I mean the upper midrange/lower treble area can become harsh and unrefined quite easily, especially with brighter sources. Still I can't believe that the graph is so poor, but then they were my first audiophile headphones, I'd been listening to cheap Sony and panasonic junk before. I've never judged a headphone on Frequency response specs, but graphs are a whole different ballgame and I believe they show a taster of any headphones sound. They should become mandatory for manufacturers to show with their products and then people could see in a shop or online, differences in sound quality and tone. I mean nearly all speakers have a 'frequency response' of 40hz-20kz at -10db, but how many cheap speakers can do it +-3db like audiophile kit? If people understood measurements better and manufacturers could be bothered to measure their products better, then the better products would be more likely to be bought and maybe become better. So more graphs please and in fact a graph database would be great, like the one at headphone.com.

 

[Nak Man]

Quote:

Originally Posted by WmAx Ideally, a non-interfernce(dampened acousticly, efficevitly mechanically inert for purposes of these measurements) couplling device and positioning jig is needed to accurately assess the transducer and rear enclosure properties.

Thanks for your input Chris. I usually measure speakers for diy purposes and wondered what kind of trace will I get from out-of-head phones measurement.

Despite surprisingly good flat line up to low treble, sealing the cups as I see from last reading gave huge swing to upper response, hence the need for Alexander the Fake mannequin in-head measurement. I don't think I will elaborate further as those will consume big effort - especially for angled drivers ala sony / 5x5, that would require true ear canal listening direction, plus reflective outer ear surface, to get proper reading.

At this point I'm glad enough to be able to correlate some characters that I hear from those phones with simplistic measurement, at least for straight firing drivers from open cans.