The RMAA (RightMark Audio Analyzer) Source and Audio Device Measurement Thread [Overview of Measurements in Post #3, Tutorial in Post #2]

[hakuzen]

thank you very much and forgive my total ignorance.
i'm going to start reading from all your links, thanks!

 

[castleofargh]

don't ever apologize for trying to learn. remember that the huge majority of audiophiles aren't even where you are right now. and I'm not trying to boost your ego, it's the sad reality.
so many people will tell you how a device lacks subs or is warm, etc, as if it was a constant characteristic of the source. and half of those won't even mention the headphone/IEM used to listen because they have no clue that the load and the amp section will interact. they will see a flat measurement unloaded and think that measurements are wrong, because they ear it warmer with their IEM. and they'll blame the source because what they know is so limited they can't even think to look for the proper correlation. as a direct result, many will rather doubt measurements(they don't know how to read anyway) than themselves.
some noticed from experience that different headphones didn't seem to react the same way to a given source, but many don't understand why so they call it synergy. and then blame synergy for everything they don't understand like it's the magical 6th dimension of electricity that explains it all.

 
you on the other hand, from looking at frequency graphs loaded and unloaded in this topic, you've already seen how the impedance of the DAP/amp can change the signature of some IEMs. and how Chris and others use that knowledge to determine if a device is low impedance or not by using specific IEMs with strong reaction to that phenomenon. 
and you've also seen how the impedance of the IEM can form a high pass filter with some caps at the output of the DAP/amp, and roll off the low frequencies in direct relation to the impedance of what IEM/resistor we plugged into the amp.
in practice, that can be very useful knowledge to decide on a sound system. more so with IEMs and portable gears where damping ratios aren't always what they should.
 
 
PS: @Chris I keep on half derailing the topic because you have such a tidy and up to date first page. and understanding what happens IMO goes with understanding how to do those measurements which was one of your goals. but it's your topic, if you wish to keep it clean with only measurements and FR talks, just say so and I can remove what you don't want here.

 

[HiFiChris]

 
PS: @Chris I keep on half derailing the topic because you have such a tidy and up to date first page. and understanding what happens IMO goes with understanding how to do those measurements which was one of your goals. but it's your topic, if you wish to keep it clean with only measurements and FR talks, just say so and I can remove what you don't want here.

 
No problem, mate! 

 
 
As I've made a well-structured front page (that I could improve by sorting the measured devices in alphabetical order though), sometimes some "off-topic" is welcome as long as it doesn't fill multiple pages with only OT and not RMAA content. 

 
 
Speaking of that, I should really finally get around measuring my Nexus 7 2nd generation tablet but am too lazy to load the RMAA test file on it. I haven't even ever connected a headphone to it yet, at least not to my knowledge.
Its predecessor had a horribly high output impedance.

 

[Samueru Sama]

   
 
Speaking of that, I should really finally get around measuring my Nexus 7 2nd generation tablet but am too lazy to load the RMAA test file on it. I haven't even ever connected a headphone to it yet, at least not to my knowledge.
Its predecessor had a horribly high output impedance.

Archimago's measured a 33 Ohm output impedance in the Nexus 7 2nd gen at 1 KHz, though he didn't test it at lower frequencies..

 

[HiFiChris]

@Samueru Sama
 
Thanks!

So it is still really bad for IEMs and headphones with non-flat impedance response, which is not that surprising though.
 

I should still do some Triple.Fi 10-loaded RMAA measurements nonetheless, along with my iPods.

 

[Samueru Sama]

Today I tested two Blu phones. A Blu advance 4.0 L2 and a Blu Studio 7.0.
 
Output impedance:
 
@1 KHz
Advance 4.0 L2 = 65 Ohm.
Studio 7.0 = 88 Ohm.
 
@120 Hz
Advance 4.0 L2 = 110 Ohm.
Studio 7.0 = 122 Ohm.
 
This means both have output caps.
 
So I tested them with rightmark, with pure resistive loads: http://prnt.sc/dc9eno
 
Didn't bother to test the 7.0 with a 32 Ohm load, it already has a terrible roll-off with a 320 Ohm load.

 

[hakuzen]

about smartphones and tablets..
 
iPad Mini 2
 
unloaded and loaded (Pai Audio MR3)
 

 
 
Measures (@ 1KHz, 0dBFS, no clipping), using DMM:
Output impedance: 1.1Ω
Max power: 0.935Vrms(33Ω).

 

[hakuzen]

Xiaomi RedMi Note 2
 
unloaded and loaded (Pai Audio MR3) (note different scale):                          same, activating Mi Sound Enhancer (for general in-ear):
 
   
 
Measures (@ 1KHz, 0dBFS, no clipping, distortion<1%), using DMM:
Output impedance: 17.45Ω (thanks to @bartzky).
Max power: 0.368Vrms(33Ω); it would accept more voltage than this, but next tick volume (maximum volume, 1.12Vrms unloaded, 0.733Vrms@33Ω) does clip.

 

[HiFiChris]

Thanks a bunch, @hakuzen, I've added both to the overview! 

 

[Samueru Sama]

  Xiaomi RedMi Note 2
 
unloaded and loaded (Pai Audio MR3) (note different scale):                                            same, activating Mi Sound Enhancer (for general in-ear):
 
   

 
It would be great if you could report the output impedance, though you need resistors.
 
A way to measure output impedance is to use the calibration test in rightmark, create a 0 dBFS tone, hit "recording only" on rightmark, play the tone and increase the volume until it says -1 dB, then add the resistor and check how much it drops, if it drops to say -4 dB with 32 Ohm resistors (make sure that the device being tested doesn't clip when you add the load), it it means there's a -3 dB drop with that load, then use this:
 
1 Volt * 10^(-3 dB drop/20) = 0.7079 Volt. I.E. 1 Volt is the unloaded voltage and the result is the loaded one, you can use any voltage you want for the unloaded voltage, it will give the same result.
 
And the rest is simply using the output impedance formula: (Vunloaded - Vloaded ) / Vloaded * Load.= Output impedance. Or just use this to get R1: http://www.raltron.com/cust/tools/voltage_divider.asp
 
Input voltage = Unloaded voltage.
Output voltage = Loaded voltage.
R2 = load.
 
That example gives an output impedance of 13.2 Ohms.
 
 
It's not 100% mandatory to use resistors, the problem with headphones is that you have to know their impedance at 1 KHz, and even measurements of the same headphone by third parties show difference impedances at the same frequency.
 
And to know if there are output coupling caps you could either repeat the process at say 60 Hz or just run the rightmark test with the load, like you discovered before with the cheap soundcard.
 
I use a 3.5mm cable with three alligator clips: http://prnt.sc/dcadq5
 
And if you have a decent multimeter, you can measure the max output power of the thing at "X" load, just check the spectrum in rightmark until you see that distortion jumps or you just reached 0 dBFS (max volume), however there's a an important detail, you gotta make sure that it isn't your ADC that is clipping. There's a maximum input voltage for those things and is usually not specified. My A58M-E for example, it can take signals up to 1.2 Vrms, more clips it.
 
To fix the issue you have to add a voltage divider before the ADC and after the load!! This diagram explains it (It says speaker amp but it can be any headphone driver you want):
 
http://image.prntscr.com/image/861e8b1257c54f8b84fcf6acb22fe651.png
 
Notice that R3 and R1 aren't needed if you know their input impedance of your PC line in, if it is say 50K ohm, with a R4 and R6 being 50K, there will be a 2x attenuation at the input, and the higher R4 and R6 are, the more attenuation you get. Also notice that there's no ground connection between the amp output and the PC line in, because there's already a ground connection between the line out and line in of your PC (or most have) adding that extra connection will only create a ground loop, though if the amplifiers input and output ground are isolated you will need to do that connection.

 

[bartzky]

It would be great if you could report the output impedance, though you need resistors.

17.45 Ohms @ 1 kHz - measured with a DMM

 

[hakuzen]

(i've separated xDuoo X3 from Benjie S5's post, to have individual posts, and because more info added for X3)
 
xDuoo X3
 
Rockboxed version, 48KHz:                                                                     Rockboxed version, 44.1KHz:
  
 
Stock firmware version, 48KHz:            Stock vs Rockbox, loaded with Pai Audio MR3, 48KHz (big format to distinguish the ripples in rockbox version; click on the image):
    
 
It is a very good impedance response with MR3. But check what it measured with iBasso IT03 (dunno the reason for this impedance ripple below 200Hz):
 

 
Measures (@ 1KHz, 0dBFS), using DMM:
Output impedance: phones out, <0.1Ω; line out, 200Ω.
Max. output (not clipping / "low" distortion):
          - phones out: high gain, 2.5Vrms (@64Ω, or unloaded); low gain, 0.47Vrms
          - line out: 1.5Vrms
 
There is some hiss with MR3. More than in dx90, d14, and mojo (in order, dx90 is the quietest), but less than benjie s5, for example.

 

[HiFiChris]

Shinrico SHD5:
 
 
Line Out (RCA):
 

 
Very convenient: one gets a 100-step volume control with the analogue output, so the SHD5 can be used without a separate pre-amp.
 
 
Digital Out (COAX) into Leckerton UHA-6S.MKII:


 
 
 
/edit:  Moved it to "Other" in the third post. While it is technically a DAP, it is a stationary audio player and I guess most people assume battery-powered devices when they see "DAP" in the overview.

 

[pila405]

I don't wish to be unthankful, because I do appreciate the effort taken in order to measure the different devices. This is very helpful.
But there seems to be a discrepancy between your measurements of the Chord Mojo and measurements from Stereophile Ohm Image: 
http://www.stereophile.com/content/chord-electronics-mojo-da-headphone-amplifier-measurements#h1qZF5tVfa55H9kL.97
http://ohm-image.net/data/audio/rmaa-chord-mojo-24-bit

How come? Because the difference is quite significant. 

 

[HiFiChris]

  I don't wish to be unthankful, because I do appreciate the effort taken in order to measure the different devices. This is very helpful.
But there seems to be a discrepancy between your measurements of the Chord Mojo and measurements from Stereophile Ohm Image: 
http://www.stereophile.com/content/chord-electronics-mojo-da-headphone-amplifier-measurements#h1qZF5tVfa55H9kL.97
http://ohm-image.net/data/audio/rmaa-chord-mojo-24-bit

How come? Because the difference is quite significant. 

 
I don't see the "significant difference/discrepancy"..

Stereophile's measurement is unloaded/taken with a high impedance load.

I don't see the difference between my and Ohm Image's measurement - the unloaded and high impedance loaded measurements show the same FR, and the in-ears show a different deviation because, well, they're obviously different in-ears that have got a different impedance response.


/edit:

Fwiw: I think it's obvious by looking at a couple of my graphs, however for those who haven't noticed it: My sound card has got a steep low pass filter around 17-18 kHz, so obviously the number right at 20 kHz is not the actual deviation - my results are "only" accurate up to 18 kHz.