Audio Measurements on a Headfi Budget

[csglinux]

Is this still the main thread to discuss reasonably accurate and not outrageously expensive DIY measurement rig setup?

I'll say yes only because I don't know of any other threads discussing reasonably-priced measurement hardware. (I'm not talking about DIY couplers that you can make out of vinyl tubing, etc. I know that's been discussed elsewhere, but a totally build-your-own coupler approach makes it ridiculously challenging to compare measurements from different rigs.)

(I ordered a coupler & pinnae yesterday and hoping to be able to contribute to measurements). If so, I plan on documenting and sharing what I learn getting my rig setup here =).

Appreciate the enthusiasm and any future contributions

Also started thinking about how to crowd source measurements in a responsible way. Should that be discussed here or in another thread?

That could be the logical conclusion of all this. If there were enough enthusiastic measurement peeps out there willing to contribute to a larger community-supported database I'd be happy to try and get it hosted somewhere like HypetheSonics.com. But there are some issues we'd all need to work through first:

1) Coupler/mic variation. There are some very good clone couplers out there, but in general, you're going to get a larger variation between clone couplers than you would between GRAS/B&K/Larson Davis, etc., couplers.
Some clone couplers come with mic calibration files, but even if they do, they seem to be of fairly poor quality. You could use this tour kit to calibrate a 711-clone rig against a GRAS RA0045: https://www.head-fi.org/threads/ety...phone-for-your-ears-and-your-couplers.908512/ (@earfonia recently did this. It's a bit of work, but his results are now in excellent agreement with the RA0045 across a broad range of IEMs.)

2) Measurement procedure/preferences would have to be common. For example - foam tips or silicone? Small, medium or large tips? There are a surprising number of people out there who still believe that measurements aren't accurate unless you see an 8 kHz resonance peak in the graph. Some insist on always adjusting insertion depth to give an 8 kHz resonance peak; others only sometimes. Many a good database has been essentially ruined by nonsense like this, because you then can't make a fair comparison between IEMs designed for shallow vs deep insertion.

Dumping everybody's data into one giant database would be the easy part. The challenge would be ensuring that what we'd be comparing would be consistent from rig to rig. It would only take a few bad contributions to warp all the statistics and make comparisons meaningless or, at best, uncertain. I'm happy to pursue this if there's an interest, but a global database would mean contributors needing to do a bit of leg-work first and potentially getting vetted or screened to ensure their data would be of sufficiently-good quality, and that they'd be agreeable to following a common procedure. I suspect that last part could be challenging

 

[aldarrin]

2) Measurement procedure/preferences would have to be common. For example - foam tips or silicone? Small, medium or large tips? There are a surprising number of people out there who still believe that measurements aren't accurate unless you see an 8 kHz resonance peak in the graph. Some insist on always adjusting insertion depth to give an 8 kHz resonance peak; others only sometimes. Many a good database has been essentially ruined by nonsense like this, because you then can't make a fair comparison between IEMs designed for shallow vs deep insertion.

My thoughts into the matter went as far as this:

1. Database everything BUT every entry needs to be from a verified account. If someone starts dumping "bad" data, their entries can be hidden and/or deleted.
2. We'd need a few "standard" devices. Say before we make your measurements published for on ear headphones you need to submit a measurement for a KPH30i or similar cheap but common on ear. Same for over ear (HD6XX would be a good candidate, but I'd prefer the standard to be cheaper, say something under $100). Same for IEMs (seems like we've settled on ER2SE for this). So new accounts would need to upload measurements of a very well known device before they can upload anything else. I think it makes sense to have separate permissions for on ear, over ear, and IEM measurements as they have different equipment needs.
3. Methodology. As an analyst by trade, I always prefer lots of clean data, but if I can work with just lots of data. I'm imagining a form that gets filled out with each measurement would suffice (autofill text would make subsequent uploads quicker). That way we can capture not only the make & model of the device being measured, but also insertion depth, tips / pads used, description of any mods. Then we can see how different tips / pads affect the FR.
4. Differing equipment. I'd prefer each measurement account have a profile that clearly shows each device used in the signal chain for their measurement system (e.g. REW=> DAC => AMP => [object being measured] => pinnae => plate => coupler => mic => conditioner => ADC => REW). And for critical devices in the chain, a more robust description than "I use an IEC711 coupler", but that it's an IEC711 coupler and a link to the item listing on the store they bought it from and a picture of that exact device. Over time I think it'll be apparent which devices contribute to better or worse measurements. Accounts would need to get their measurement profile approved before we accept any measurements.
5. Account process in summary: request account, confirm account with email, create measurement profile, measurement profile approved by admins / mods, submit measurement report for standard IEM / OE / on ear, measurement report approved, user allowed to submit measurement reports for that class of device. If a user starts loading bad data, that data is first hidden and possibly purged.

As for viewing the measurements, I assumed it look similar to Crinacle's system. Select device class, select make, select model but add modifiers that are also selectable: mods done, pads / tips used / pad/tip condition / model year or version (e.g. 2019 Sundara or 2020 Sundara), etc.

 

[earfonia]

My thoughts into the matter went as far as this:

1. Database everything BUT every entry needs to be from a verified account. If someone starts dumping "bad" data, their entries can be hidden and/or deleted.
2. We'd need a few "standard" devices. Say before we make your measurements published for on ear headphones you need to submit a measurement for a KPH30i or similar cheap but common on ear. Same for over ear (HD6XX would be a good candidate, but I'd prefer the standard to be cheaper, say something under $100). Same for IEMs (seems like we've settled on ER2SE for this). So new accounts would need to upload measurements of a very well known device before they can upload anything else. I think it makes sense to have separate permissions for on ear, over ear, and IEM measurements as they have different equipment needs.
3. Methodology. As an analyst by trade, I always prefer lots of clean data, but if I can work with just lots of data. I'm imagining a form that gets filled out with each measurement would suffice (autofill text would make subsequent uploads quicker). That way we can capture not only the make & model of the device being measured, but also insertion depth, tips / pads used, description of any mods. Then we can see how different tips / pads affect the FR.
4. Differing equipment. I'd prefer each measurement account have a profile that clearly shows each device used in the signal chain for their measurement system (e.g. REW=> DAC => AMP => [object being measured] => pinnae => plate => coupler => mic => conditioner => ADC => REW). And for critical devices in the chain, a more robust description than "I use an IEC711 coupler", but that it's an IEC711 coupler and a link to the item listing on the store they bought it from and a picture of that exact device. Over time I think it'll be apparent which devices contribute to better or worse measurements. Accounts would need to get their measurement profile approved before we accept any measurements.
5. Account process in summary: request account, confirm account with email, create measurement profile, measurement profile approved by admins / mods, submit measurement report for standard IEM / OE / on ear, measurement report approved, user allowed to submit measurement reports for that class of device. If a user starts loading bad data, that data is first hidden and possibly purged.

As for viewing the measurements, I assumed it look similar to Crinacle's system. Select device class, select make, select model but add modifiers that are also selectable: mods done, pads / tips used / pad/tip condition / model year or version (e.g. 2019 Sundara or 2020 Sundara), etc.

I agree with you. Measurement results alone is not meaningful and can actually be bad data (misleading info) without those details like measurement equipment, setup, and methodology. Especially the methodology, it is not all specified in international standards like:

IEC 60268-7 https://webstore.iec.ch/publication/1226
IEC 60318-4 https://webstore.iec.ch/publication/1445

For example, what is the recommended ear tip to be used for the measurement? Silicone, foam, medium size, small size? All of those affect the measurement. And it has to be clearly mentioned on the measurement result. And old recommendation like the 120 ohm headphone amplifier output impedance is no longer relevant these days as we prefer to see measurement result from a very low headphone amp output impedance.

I learnt a lot from @csglinux about IEM measurement, and now I'm drafting my measurement methodology and FR target based on the current 711 coupler that I use. I have 2 couplers, both are clone couplers, and one of them as mentioned by @csglinux above is actually pretty good. I bought the good coupler from Taobao (with the help from a friend from China). It is also available in AliExpress which is easier for us to buy, but a bit more expensive.

Here is the link of my IEC 60318-4 compliant coupler if anyone interested. It is ICP / CCLD based microphone, so need CCLD driver for the microphone amplifier. But some DIY trick might work as well by lowering the 48V phantom power from the XLR input of the audio interface. I use the B&K 1704 for my coupler. I spent about 2 weeks to create the calibration file for it, to calibrate it to @csglinux GRAS RA0045 using the IEMs in the Etymotic ER2SE tour.

Taobao link:

https://item.taobao.com/item.htm?spm=a1z0d.6639537.1997196601.15.4c197484slCmuU&id=542357654840

Choose option / type : 4

the seller has the 2nd hand B&K 1704, but not listed in his shop. Need to talk / chat with him if we want to include it.



AliExpress link:

https://www.aliexpress.com/item/1005001640688144.html?spm=2114.12010615.8148356.28.4cca10efsdJQn7

Choose type : 4

B&K1704:

https://www.aliexpress.com/item/1005001640607614.html?spm=a2g0o.cart.0.0.58123c00USo6Hj&mp=1

 

[csglinux]

All good suggestions guys. BTW, I can personally recommend that coupler of @earfonia's. It's very good. Accurate mic too.

I completely agree about the need to document the rig in detail, but I might be coming at @aldarrin's question from a slightly different perspective.

IMHO, the usefulness of a collaboration like this would be to have one giant meta-database. There's nothing to stop individuals from hosting their own databases. Folks can easily do that today - Marshall Lochbaum's software (https://github.com/mlochbaum/CrinGraph) is free and easy to use and is already being used by a dozen or so folks, including Crinacle. But multiple separate databases don't allow us to compare, rank, score, search by profile, eq one headphone to another, etc., across all measurements. That only makes sense if all measurements follow the same procedure with sufficiently-compatible equipment. For example, the couplers would needs to be of the same type (all well-calibrated), all driving sources would need a very low z-out, any pinnas should also be identical if we're comparing over/on-ear or earbud measurements. If we have to document differences in equipment, that would seem to suggest differences are potentially significant, and arguably, that would mean they shouldn't be put together on the same database.

 

[aldarrin]

Here's what I have on order so far:

SPL calibrator:
https://www.aliexpress.com/item/32894240716.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

Coupler, Type 1:
https://www.aliexpress.com/item/4000073069259.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

Artificial pinna w/ plate:
https://www.aliexpress.com/item/1005001530169539.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

Sound card (ADC): (probably not needed, but it was recommended on the coupler's listing and was only $26).
https://www.aliexpress.com/item/4000072992146.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

UMIK-2 (for sure to be used to calibrate the SPL of the measurement system w/ the SPL calibrator, not sure if I want to replace the mic that comes with the coupler)

Equipment on hand:
Monoprice Liquid Platinum, A90, D90, ADI-2 Pro BE, MiniDSP Ears (lol), a bunch of headphones, a few IEMs.

I'm not an IEM expert by any stretch, but I think we'd want to have standard measurements (the one's used to validate the accuracy of each reviewer / measurer)... starting with the the methodology you guys used in the tour. If the website can have a forum, admins can discuss and vote on more specific standards for measurements. But, besides what constitutes a standard measurement, I think there's goodness in seeing measurements for shallow v deep insertion, silicone v foam tips, etc (as long as it's clearly listed, maybe we could ask for pictures of the device on the measurement rig).

Yeah, I'm a fan of collecting as much data as possible as long as it's relatively simple for the more technical members of the community to determine what is / isn't a valid measurement and the public / less technical folks are not exposed to the bad data. Like we don't need a full public ASR v Abyss feud over one dude's measurements (although this sort of discussion within the community could be very useful for improving measurements, but probably done only between people already vetted by the system / providing good measurements).

 

[earfonia]

Here's what I have on order so far:

SPL calibrator:
https://www.aliexpress.com/item/32894240716.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

Coupler, Type 1:
https://www.aliexpress.com/item/4000073069259.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

Artificial pinna w/ plate:
https://www.aliexpress.com/item/1005001530169539.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

Sound card (ADC): (probably not needed, but it was recommended on the coupler's listing and was only $26).
https://www.aliexpress.com/item/4000072992146.html?spm=a2g0s.9042311.0.0.5d144c4dPcHePv

I found that we do need the SPL calibrator every time we are going to start a measurement session. After few days sometime I saw around 0.1 dB shift of REW SPL reading from previous calibration. Not always, but having the SPL calibrator is handy to make sure that the SPL reading stays accurate.

After some measurement of ER4XR, I found out that my ND9B SPL calibrator output is around 94.5 dB SPL, not the 94.0 dB SPL as I expected previously. So now I calibrated my coupler SPL reading to 93.6 dB SPL using the ND9B. From previous observation the correction factor on the SPL loss in the coupler (due to mic distance and coupler volume) is around 0.9 dB SPL, so if the ND9B output is exactly 94 dB SPL, I calibrate my coupler to 93.1 dB SPL. But if the ND9B output is 94.5 dB SPL, then I calibrate the coupler to 93.6 dB SPL.

Explanation here:


The coupler shown in the YouTube video is my 1st coupler, similar to the one in your order list. My 2nd coupler is better in both FR and THD measurement. But the 1st coupler can be calibrated for better FR measurement. But THD measurement is always showing higher THD result than my 2nd coupler.

1st coupler type: E610A SN: 20201 - Mic requires common 3-5 Volt plugin power.
2nd coupler type: E610A SN: 20229 - 1/2 inch ICP Mic requires ICP driver with 4 mA constant current at around ~ 24V.

 

[aldarrin]

I found that we do need the SPL calibrator every time we are going to start a measurement session. After few days sometime I saw around 0.1 dB shift of REW SPL reading from previous calibration. Not always, but having the SPL calibrator is handy to make sure that the SPL reading stays accurate.

After some measurement of ER4XR, I found out that my ND9B SPL calibrator output is around 94.5 dB SPL, not the 94.0 dB SPL as I expected previously. So now I calibrated my coupler SPL reading to 93.6 dB SPL using the ND9B. From previous observation the correction factor on the SPL loss in the coupler (due to mic distance and coupler volume) is around 0.9 dB SPL, so if the ND9B output is exactly 94 dB SPL, I calibrate my coupler to 93.1 dB SPL. But if the ND9B output is 94.5 dB SPL, then I calibrate the coupler to 93.6 dB SPL.

Explanation here:


The coupler shown in the YouTube video is my 1st coupler, similar to the one in your order list. My 2nd coupler is better in both FR and THD measurement. But the 1st coupler can be calibrated for better FR measurement. But THD measurement is always showing higher THD result than my 2nd coupler.

1st coupler type: E610A SN: 20201 - Mic requires common 3-5 Volt plugin power.
2nd coupler type: E610A SN: 20229 - 1/2 inch ICP Mic requires ICP driver with 4 mA constant current at around ~ 24V.

Edit: just to note, that video is the reason I ordered the SPL calibrator =). It was very useful!

Are the THD measurements consistently off? Like does it always show say +3 db extra distortion or extra distortion at certain frequencies? Or is the difference random?

 

[earfonia]

Edit: just to note, that video is the reason I ordered the SPL calibrator =). It was very useful!

Are the THD measurements consistently off? Like does it always show say +3 db extra distortion or extra distortion at certain frequencies? Or is the difference random?

I will share with you some measurement. No it is not that bad. It is consistently higher than the ICP based coupler, probably around 0.5% - 1.0% higher. I haven't really calculate the difference. When I have time this week or next I will share some measurement comparison.

 

[earfonia]

@aldarrin here are the comparisons between couplers that I have currently, plus the DIY Earthworks M23 coupler belong to @csglinux.

IEC 60318-4 couplers tested:

1. IEC 60318-4 plinth coupler with ½ inch ICP prepolarized microphone. Model E610A SN: 20229
2. IEC 60318-4 plinth coupler with plug-in-powered (±5V) microphone. Model E610A SN: 20201
3. Earthworks M23 + E610A coupler SN: 20494
4. MiniDSP UMIK-1 + E610A coupler SN: 30168

The coupler No. 2 is similar to the coupler that you bought recently. Mine calibrated to coupler No. 1, therefore the FR measurement result looks pretty close.

I measured 2 IEMs: Audio-Technica CKN50 and BLON BL-05S

Frequency Response Measurement of Audio-Technica CKN-50 (using small large bore ear tip):

Frequency Response Measurement of BLON BL-05S (using small generic 3.8mm bore ear tip):

The more interesting part is the THD measurements:

Audio-Technica CKN-50 THD Measurements:







BLON BL-05S THD Measurements:







For those who are interested to check the mdat file, you can download from here:
https://drive.google.com/drive/folders/1DSNnRCZxLW1_JPqs9TnIIz80AVRDhhwL?usp=sharing

 

[yeboyi]

Anyone know how can i measure headphones on my head ? I don't wanna buy head and ears. I also wanna know the hrtf of my own ear/head. Is it possible ? I think youtuber Oluv does that. Can anyone help me how he does that and if it's accurate ?

 

[csglinux]

Anyone know how can i measure headphones on my head ? I don't wanna buy head and ears. I also wanna know the hrtf of my own ear/head. Is it possible ? I think youtuber Oluv does that. Can anyone help me how he does that and if it's accurate ?

Yes, you can do it. It's necessary to get your own HRTF for the best possible spatial surround-sound effects from in-ears. This is what's done for the Smyth Realizer and Impulsifer. A couple of caveats - the in-ear mics aren't perfect, because the presence of the mics change the effective response of your ears. It's fair for over-ears. For in-ears, I don't know of any suitable mics that are small enough to fit deep enough in your ear canals and still allow an IEM to be inserted normally. That sounds too painful to even try. The guy that probably has best knowledge of these in-ear mics is Jaakko Pasanen (@jaakkopasanen). IMHO, he's the smartest guy to have ever graced headfi. And I'm hoping that bit of flattery will elicit a response from him He's the author of autoEQ and Impulsifer. We had some discussions about the mics needed for Impulsifer a few years back, but things may have changed since then, so I'm not sure what his current top mic recommendations would be. But to emphasize again, I think this is only going to work reasonably for over-ears, not in-ears. You can put mics inside IEMs (we've built some for that purpose), but they change the FR quite notably compared to that measured from a typical coupler mic.

 

[yeboyi]

Yes, you can do it. It's necessary to get your own HRTF for the best possible spatial surround-sound effects from in-ears. This is what's done for the Smyth Realizer and Impulsifer. A couple of caveats - the in-ear mics aren't perfect, because the presence of the mics change the effective response of your ears. It's fair for over-ears. For in-ears, I don't know of any suitable mics that are small enough to fit deep enough in your ear canals and still allow an IEM to be inserted normally. That sounds too painful to even try. The guy that probably has best knowledge of these in-ear mics is Jaakko Pasanen (@jaakkopasanen). IMHO, he's the smartest guy to have ever graced headfi. And I'm hoping that bit of flattery will elicit a response from him He's the author of autoEQ and Impulsifer. We had some discussions about the mics needed for Impulsifer a few years back, but things may have changed since then, so I'm not sure what his current top mic recommendations would be. But to emphasize again, I think this is only going to work reasonably for over-ears, not in-ears. You can put mics inside IEMs (we've built some for that purpose), but they change the FR quite notably compared to that measured from a typical coupler mic.

Yeah i'm not interested in in ears. I'm over ear headphone guy. Need a budget good in ear mics to measure the response of the over ear headphones. Compare them to each other, compare them to harman curve if it's possible, use it for eq etc..

 

[gordonli]

Sorry if this is violating forum etiquette, but I'm interested in buying one of these aliexpress 711 couplers with mic, and I couldnt find whether their mic calibration files come with a sensitivity factor or not, because I'm just as interested in finding out my listening SPL, as I am in measuring my IEM frequency responses. @TonySunshine @earfonia @TLDRonin @csglinux @aldarrin can any of you enlighten please?

like these ones:
https://www.aliexpress.com/item/400...b16576165176222788eae48!12000022787612855!sea

https://www.aliexpress.com/item/400...b16576165176222788eae48!12000023512553277!sea

If not, I guess the next easiest option would be to buy one of the couplers without mic, and put it onto a Umik-1 to get a ballpark SPL.

And then super hardcore would be earfonia's SPL calibration, but I would need to buy like another 3 pieces of equipment

 

[csglinux]

Hi @gordonli, I would recommend you try to join this tour:

https://www.head-fi.org/threads/ety...r-ears-and-your-couplers.908512/post-17008210

@earfonia put some gadgets in the kit to calibrate SPL, but more important than that, you can use the kit to calibrate your coupler/mic combo to match the GRAS standard. Most clones don't come with calibration curves, and those that do aren't reliable. If you don't properly calibrate a clone coupler, its measurements can't be compared to anything else. Stand-alone databases are perfectly fine, but there's no need to even use a clone coupler for that - vinyl tubes or vibro veritas couplers will show the trend. (Not the exact same trend as your ear, but then no coupler will exactly match that anyway.)

If you've gone to the trouble of acquiring a clone coupler, you'd want to calibrate it so it's comparable with others - most notably, you'd want to be able to put your measurement data against standard curves like Harman. You can, of course, generate your own preference curve (which every man and his dog is doing these days, using a statistically-irrelevant sample size of one), but these curves are useless for external comparisons if you've no idea how much your mic is rolling off at the extremes - which clone couplers all have a tendency to do. I think this is one reason why all the new preference targets you see from reviewers these days have less sub-bass than the Harman target. Their mics are all rolling off in the sub-bass, so the SPL they're hearing there is higher than their coupler is reporting. The same is true in the treble, but to a slightly lesser extent.

If not, I guess the next easiest option would be to buy one of the couplers without mic, and put it onto a Umik-1 to get a ballpark SPL.

No, no, no, no, no. Don't do it! 1) That calibration would be hugely unreliable anyway, and 2) you should buy a coupler with a mic and never ever remove the mic. If you do, you'll change the effective air volume in the coupler. I've tried stuff like this and it's really hard to get it right. Just buy a decent clone coupler (with mic) and calibrate it.

 

[gordonli]

@csglinux Thanks for all the advice. That tour looks amazing, though a smidge beyond my ambitions at moment, wish you all the best with it.