Audio Measurements on a Headfi Budget

[anli]

Hi!

Also have got the coupler from AE. A seller also has sent two mic calibration files, and one of them goes with 1/24 smoothing applied. Can not imagine the way such file can be useful.

Any ideas?

 

[csglinux]

Hi!

Also have got the coupler from AE. A seller also has sent two mic calibration files, and one of them goes with 1/24 smoothing applied. Can not imagine the way such file can be useful.

Any ideas?

Nothing intrinsically wrong with a 1/24th octave smoothing - it depends what it was smoothing.
If your coupler has significant narrowband peaks and troughs that need correcting, it's not likely that any calibration file is going to be universally applicable. At least, not very accurately.

There is a way to find out. We have a little project ongoing to help folks like you calibrate clone couplers to the GRAS standard: https://www.head-fi.org/threads/ety...r-ears-and-your-couplers.908512/post-17008210

 

[anli]

Nothing intrinsically wrong with a 1/24th octave smoothing - it depends what it was smoothing.
If your coupler has significant narrowband peaks and troughs that need correcting, it's not likely that any calibration file is going to be universally applicable. At least, not very accurately.

There is a way to find out. We have a little project ongoing to help folks like you calibrate clone couplers to the GRAS standard: https://www.head-fi.org/threads/ety...r-ears-and-your-couplers.908512/post-17008210

Yes, smoothing variant of mic calibrations file can be useful as long as it exists. Just I don't understand those use cases when calibration file with smoothing is preferable in comparison with raw calibration file.

 

[gordonli]

Yes, smoothing variant of mic calibrations file can be useful as long as it exists. Just I don't understand those use cases when calibration file with smoothing is preferable in comparison with raw calibration file.

one reason I can think of is it may have been calibrated in batches (though I dont know the process behind batch calibration but it happens like with umik1), so as cglinux said, a narrowband calibration may be less accurate than smoothed

 

[anli]

one reason I can think of is it may have been calibrated in batches (though I dont know the process behind batch calibration but it happens like with umik1), so as cglinux said, a narrowband calibration may be less accurate than smoothed

It is interesting. The thing is the seller has wrote do not use 1/24 smoothed variant as long as I don't understand why to use it, as far as deviation can be more on wrong use.

 

[csglinux]

Yes, smoothing variant of mic calibrations file can be useful as long as it exists. Just I don't understand those use cases when calibration file with smoothing is preferable in comparison with raw calibration file.

IMHO, smoothing is always preferable. Let's say you've compared measurements from an identical IEM. Your coupler gives a resonance peak at 7 kHz and the GRAS RA0045 gives a peak at 7.25 kHz for an identical insertion depth. If you just take the difference between those two curves and apply it as your calibration file, every other headphone you ever measure is going to present with a trough at 7 kHz and a peak at 7.25 kHz.

I think what @gordonli is alluding to is that you'll typically create a calibration file from an average over multiple headphones. An average of deltas from multiple headphone measurements is bound to be somewhat uneven. You really want to compensate for just the average error in the coupler, otherwise every measurement you take is going to have that unevenness in it. If your coupler is very different from the GRAS standard, there's no hope of finding a calibration file that will perfectly match GRAS results for all headphones measured. The hope is, if your coupler is close enough, with an averaged and smoothed calibration curve, your results will at least be reasonably close for any future measurements.

 

[csglinux]

It is interesting. The thing is the seller has wrote do not use 1/24 smoothed variant as long as I don't understand why to use it, as far as deviation can be more on wrong use.

In my experience, you'll find the calibration files offered by AE and Taobao sellers always fall into one of two categories: 1) completely useless or 2) non-existent.

 

[anli]

In my experience, you'll find the calibration files offered by AE and Taobao sellers always fall into one of two categories: 1) completely useless or 2) non-existent.

Probably you are right. At any case resulting FRs look similar to those found around. So these files don't seem to be too harmful.

 

[csglinux]

Probably you are right. At any case resulting FRs look similar to those found around. So these files don't seem to be too harmful.

The test would be to compare your measurements, with and without the calibration file, against those from a GRAS coupler. Your measurements could easily be off several dB here and there. That matters if you want to compare your measurements against those from external databases, because these errors could be larger than the differences between two headphone models.

If you don't plan on making external comparisons and you only want to see the rough shape of the curve for internal comparison purposes, then you don't need to worry about calibration.

 

[anli]

The test would be to compare your measurements, with and without the calibration file, against those from a GRAS coupler. Your measurements could easily be off several dB here and there. That matters if you want to compare your measurements against those from external databases, because these errors could be larger than the differences between two headphone models.

If you don't plan on making external comparisons and you only want to see the rough shape of the curve for internal comparison purposes, then you don't need to worry about calibration.

Absolutely agree. Unfortunately there isn't anybody around me interested in measurements. It there a place where measurement maniacs share their mdat files?

 

[csglinux]

Absolutely agree. Unfortunately there isn't anybody around me interested in measurements. It there a place where measurement maniacs share their mdat files?

Ooh yes - I can put you in touch with many such maniacs I will pm you...

 

[losber]

https://www.head-fi.org/threads/home-made-iems.430688/post-17616954
Can anyone identify what microphone is he using for measurements?
I used Google lens to translate subtitles on the video. He said it costs about 3000 yen. So it probably can be a cheap electret mic, right?
Here are few quotes from the video:
"
It's cheap, about 3000 yen
If you want to achieve accuracy, you will need more than 20000 yen for the microphone and interface for measurement.

"
Attach the silicon tube to the tip like this.
to match the length of the ear canal.
The inner diameter is about 6~8mm, please match it with the thickness of the canal.
"

Can anyone identify what microphone he is using for measurements?

 

[losber]

I think that mic is something like Clippy EM272 or similar based on Primo mic capsule. The post https://www.head-fi.org/threads/audio-measurements-on-a-headfi-budget.893084/post-15306776 and the link https://drbadphil.com/tag/primo-em172 led me to that conclusion.

​

 

[losber]

What do you think about this coupler https://www.aliexpress.com/item/1005001641119247.html (type 3)?
Is it too bad for $40?

 

[losber]

So, I've received this coupler and a cheap USB sound card.
This sound card (or similar looking) is based on CM6206-lx audio controller and is offering with other coupler https://www.aliexpress.us/item/2255800731956095.html
so I bought it separated with lower price.
The coupler is good, as I expected, the serial number is engraved on the coupler and matches the name of the calibration file that the seller sent me.

After reading a few notes about sound cards on the CM6206/CM6206-LX and studying the datasheets for the controller, it became clear that I have a very budget card. The reference circuit (in barely readable quality) is only for CM6206 v1.7 datasheet, where it is recommended to install ferrite beads on the power lines from the USB port, separate analog and digital ground, and connect the grounds through a ferrite bead. This sound card has nothing at all from these recommendations.
Power supply according to the datasheet is 5V, but in the circuit from the datasheet it is used through a 5V-to-4V converter. In my case, the ground is united and I cannot easily separate it to analog and digital. There were no beads, no places for them (there are known boards with separated ground and unsoldered elements), the power traced through a 3.3V converter, and because of this, the VBIAS pin for the microphone was not 2.25V, as according to the datasheet, but about 1.26V.
I removed the voltage converter 1117 and shorted 1117 traces with ferrite bead to take 5V power to CM6206-LX.
I cut the trace from the 5V USB contact and also soldered the ferrite bead into the gap.
I untied the first pin of the chip, according to the datasheet, logical 1 turns off the chip (power down mode), I did not understand why they used it here at all.
The major modification was replacing capacitors in MIC IN and FRONT OUT circuits. I used this article for reference https://tehnoblog.org/c-media-cm620...ass-extension-low-frequency-response-upgrade/

I watched this video https://www.youtube.com/watch?v=ylEHwV9RtFs and made few test measurements for TRN BA15 to make sure I made those modifications correctly.
Here are my graphs after all modifications.
Output device CM6206-LX sound card


Output device HiBy R3Pro2022


There are known graphs for reference https://eplv.squig.link/?share=TRN_BA15
https://melatonin.squig.link/?share=TRN_BA15