BMF
contributing
Posts: 99
|
Post by BMF on May 6, 2013 17:38:54 GMT
That's what we need... (the info ) These are substantially different from WM61A and thus need a different compensation. They cannot be substituted directly in any case if they are used for measurements. Do-able with the filter PCB though. a 5dB lift needs to be compensated (centered around 9kHz) and a 3-4dB dip (centered around 17kHz) needs to be lifted. This is possible by using the 2 filter sections. figure III on John Connovers pages: www.johncon.com/john/wm61a/shows the plot from the WM61A (mind the scales !) It differs from the plot of Panasonic themselves who claim it is flat: www.panasonic.com/industrial/components/pdf/em06_wm61_a_b_dne.pdfnicely flat to 5kHz (as are most electret capsules) I'd like to see John Conover's graph at 10 dB resolution and the other graphs at 0.5 dB resolution.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on May 6, 2013 18:21:53 GMT
That's the biggest problems with all graphs out there... including headphone graphs. The scales differ and how big the 'steps' are differ. Everyone uses their own scale.... I once took it upon myself to scale a few different plots from various sites of the ATH-M50 as they all seemed to differ quite much. Turned out the differences were mainly between the 'old' and newer (2012) version and some that appeared very different (due to different scales) suddenly became more correlating. My own measurements (of Ian's old M50) correlate very well with the 'old' versions in the plots above. diyaudioheaven.wordpress.com/headphones/measurements/ (second one from the top) BUT because the scales differ yet again they SEEM very different. To me they sound like they measure... bassy, warm mids, seemingly 'enough' highs and lack of 'air' (presence above 15kHz)... again this is for the old version, the 2012 version is much more balanced. a question of stretching the plots in something like 'paint' or similar programs till the scales are the same in vertical and horizontal planes. Basically the WM61 peaks about +3dB at 15kHz where the new ones peak about +5dN at 9kHz and drop down -3dB again at around 17kHz.
|
|
Rabbit
Administrator
Posts: 7,087
|
Post by Rabbit on May 8, 2013 19:32:24 GMT
The way the scale is shown, they seem to really tail off badly at the top end but the scale is altered as well so 20KHz should be much further to the right, so that tail off would be less steep looking and the zig zags wouldn't be as sharp as they look.
Variations between headphone samples/different type of dummy head and mics are going to give slightly different results as well.
What we need is someone with an anechoic chamber for a head and a graph that extends to 20KHz in the same proportions as the rest of the frequency range.
The variations do make these graphs quite difficult to interpret.
|
|
proid
quite active
Posts: 113
|
Post by proid on Sept 14, 2013 12:32:07 GMT
I am open to changing the BOM to build a better mousetrap. Do you have one you would recommend, with a schematic? I'm using 3.5 mm TRS jacks and RCA plugs. The upper schematic is for usage with non-polar capacitors. The bottom one is for usage with much cheaper electrolytic caps. Beware of the polarity of the used electrolytic capacitors. For the jack plug in the schematic the output is connected to the R channel (ring) but you can also connect it to the tip (Left channel) OR both channels (L + R). Some mic inputs have the R (ring) and ground connected. In that case the signal must go to the tip connection and the ground is the same. NOTE: The mic connection that must be connected to ground has a small trace from the pad to the metal casing. The connectors shown are simply suggestions as to how to wire the most common connectors. You should connect only the connector type you need to use, so not both connectors. 6k8 = 6.8k (just a different way of notation) I dont have some parts exactly like in the schematic so i changed 2.2k to 2k, 100uf/16v to 220uf/16v, 2.2uf to 4.7uf, is that ok?
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on Sept 14, 2013 12:56:58 GMT
Yup that'l work as well.
You do have to add compensation curves in the analysis software though.
Easiest program to use is REW and it also allows compensation curves. I don't use that function as I have that built into the pre-amp.
|
|
proid
quite active
Posts: 113
|
Post by proid on Sept 14, 2013 14:19:42 GMT
I have done the power supply but it didn't work. I plug the output of measurement rig to mic in of my computer soundcard, turn it on with a 9v battery then i try to record sound with sound recorder of window but there are no signal.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on Sept 14, 2013 18:06:00 GMT
With a 9V battery connected.
black lead of multimeter on ground (GND) red lead on + 4.7 uF what voltage (DC voltage) do you get ? red lead on + 220uF what voltage (DC voltage) do you get ? red lead on 6k8 (9V battery side) what voltage (DC voltage) do you get ?
|
|
proid
quite active
Posts: 113
|
Post by proid on Sept 14, 2013 22:39:25 GMT
Here is my measurement: red lead on + 4.7 uF: 0.15v red lead on + 220uF: 6.8V red lead on 6k8 9.2V
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on Sept 15, 2013 9:24:34 GMT
Hmmm skipper ... I fear your alleged 2k0 resistor is a secretly a 20k resistor in disguise (read the above as if it would be narrated by Kowalski)
|
|
proid
quite active
Posts: 113
|
Post by proid on Sept 15, 2013 12:36:45 GMT
Hmmm skipper ... I fear your alleged 2k0 resistor is a secretly a 20k resistor in disguise (read the above as if it would be narrated by Kowalski) I double check it again and the 2k resistor in fact are 2 1k resistors which have color code: brown black black brown brown so in seri, they are 2k resistor, mistake must come from elsewhere.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on Sept 15, 2013 12:51:32 GMT
Hmm skipper... that seems very odd to me. Let me do the math... 2 resistors and the mic are all in series and so the voltage divides and the current in all components is the same. There is 2.4V across the 6k8 resistor. This can only be so when there is a current flowing of 350uA. when a current of 350uA flows through a 2k resistor there would be a 0.7V drop across it leaving a voltage over the mic of 3.1V But the mic has 0.15V across it acc to your measurements. If a 350uA current would flow through a 20k resistor there would be around 7V over it and that's what you are measuring (6.8V). Another possibility is that the FET inside the mic is (almost) shorted and the 0.15V is the result of that. In that case the voltage division across the resistors would differ, there would be 2.0V on + of the 220uF cap. Consider the possibility that 1 brown band in brown black black brown brown (1,000 Ohm - 1%) may be a very dark red (have seen this lots of times) in that case it may well be brown black black very dark red brown (10,000 Ohm - 1%) get out the multimeter and measure the resistors is my advice.
|
|
proid
quite active
Posts: 113
|
Post by proid on Sept 18, 2013 14:14:03 GMT
@frans: I remeasure and at + 4.7 uF it's not 0.15v but 5V,i measure 2k resistor too and it was exactly 2k but i still can't record sound using this mic.
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on Sept 18, 2013 15:54:41 GMT
What's the voltage on + 220uF now ? It cannot be 6.8V any more as 5V over the mic would mean 5.9V on +220uF (450uA). At that current there should be 2V over the mic (if it is genuine WM61). First 0.15V and now 5V points to a defective mic capsule.
|
|
proid
quite active
Posts: 113
|
Post by proid on Sept 18, 2013 17:36:09 GMT
I figured out the problem, Frans, it didn't solder a leg properly so it didn't work. Now i can measure but the frequency from 7Khz look not right, here is the graph of a stock Evo with foam removed: Will calibrate fix this?
|
|
solderdude
Administrator
measureutternutter
Posts: 4,881
|
Post by solderdude on Sept 18, 2013 19:18:05 GMT
The microphone has to be on a sealed baffle and surrounded by material that absorbs sound in the same way as skin.
perhaps cell-foam rubber with a thin leather over it.
The seal has to be good, don't use styrofoam heads or similar shapes that can 'leak' sound.
For REW plots you can set graph limits like 10Hz and 22kHz.
Also you need to calibrate the test rig first.
You have to connect the used headphone amp through a voltage divider so you get roughly the same voltage as the mic will provide and then run the calibration program. This will yield a calibration curve that ensures what comes out of the amp and into the mic amp is 'flat'.
After that calibration is done you need to add to that the compensation curve that belongs to the used mic and add that to the calibration curve (no idea how to do that as I can skip all this).
The results you will be getting after this will be much closer to the truth but also will not be really accurate (I know mine aren't really accurate in a reference kind of way but do come quite close)
In any case you can keep track of what mods have done.
|
|