[Otto Motor]

My progress report - figuring the issue out using the known **** earphone:

Summary for SchiitFulla amp/****:

• At 85 dB, there is an effective 2 dB channel imbalance at 1 kHz.
• This imbalance disappears at 94 dB.
• The imbalance widens substantially at 75 dB.
• Channel matching between the earpieces is good when measuring both on the same channel.

Summary for audioquest dragonfly amp/****:

• Excellent channel matching at 85 dB.

Implications
In other words, the earphone in question needs more volume with the Schiit -- or another amp.

Why all this? I don't know. I'll get an annotated pdf ready for download documenting my steps.

 

[Brooko]

Whats the output impedance on the Dragonfly? The Fulla is ~ 0.5 ohm. What are the specs on the **** - especially impedance and sensitivity? Does it use crossovers?

 

[csglinux]

Yeah, I could try the Shaning M0!

I love the M0! Would be curious to see if this cures your channel-imbalance issue. The M0 should be plenty good enough (power and low z-out) to measure IEMs.

 

[castleofargh]

Great idea! But this means swapping earpieces, tricky with MMCX connectors, as they tend to die.

oh don't do that. you need a solution for IEMs that won't have detachable cables anyway, so I suggest to make a little something to place between the amp and the IEM's cable. when I started I was using a male to male short and using small croco clips to connect the right rings on the Jack. it was a little messy and I feared that over time the croco clips might ruin the gold platting on the Jack. so I ended up getting my soldering iron and making myself an ugly little thing with a switch between channels. the output is always only the left channel of my amp and the switch connect to either left or right ring on the IEM's Jack.
but I guess it's not as easy as it seems for people who don't already have a shitload of wires, plugs and components lying around from all my previous failed projects. also I suck at this so I always end up making monsters.I was so jealous when I saw... I think it was @hakuzen showing a pic of a nice clean impedance measurement wiring for REW with a little box and everything clean. mine looks like a blob ^_^.

Are any precautions necessary with regards to E12's rather high noise level when measuring very efficient IEMs? Or is it safe to assume the noise to be uncolored enough to not affect measurements in any meaningful way?

for frequency response alone, you could have Skrillex playing 10dB below the sine sweep and you'd be fine. we only have to make sure that the signal is always louder than the noise, and it usually will(I don't want to imagine an amp with an output noise louder than a full scale or almost full scale test signal ). in practice ambient noise in the room is way more of an issue. at least it is for me. can't go too loud because BA drivers just can't, and my mic isn't the right type for that job, but I can't go quiet because my room even with the fanless computer is always close to 35/40dB SPL and in the day, I'm swamped by sub frequencies. at least vampires need to be invited in.

 

[castleofargh]

My progress report - figuring the issue out using the known **** earphone:

Summary for SchiitFulla amp/****:

• At 85 dB, there is an effective 2 dB channel imbalance at 1 kHz.
• This imbalance disappears at 94 dB.
• The imbalance widens substantially at 75 dB.
• Channel matching between the earpieces is good when measuring both on the same channel.

Summary for audioquest dragonfly amp/****:

• Excellent channel matching at 85 dB.

Implications
In other words, the earphone in question needs more volume with the Schiit -- or another amp.

Why all this? I don't know. I'll get an annotated pdf ready for download documenting my steps.

this looks like a good old (but massive) volume knob problem to me. do you get the same thing if you change the level of the test signal output digitally in REW(the stuff at -3dB)?
if I'm correct your issue isn't so much with impedance curve, but with how sensitive the IEMs are. because of their specific sensitivity they make you move the volume knob in a particular area where the problem really is. also most impedance's variation responses with the amp wouldn't show a steady change at all frequencies. so that aspect isn't the main culprit anymore, IMO.



oh and in general, to avoid channel imbalance:
- avoid something with a lot of gain, even unity gain can mean around 2V full scale with most non portable DACs. which is a lot for most IEMs. little gain means that you'll be able to stay away from first third of the volume knob's course where imbalance is more likely to increase.
- get some amp with what they call "digital control" of the volume. I hate that name because we're still fully in the analog domain, but I've seen it used many times. this usually results in outstanding channel matching, the downside being increased noise and a noise that tends to increase when you raise the volume. so it's not very audiophile overall, but it gives very accurate volume setting when we look for that. I have this on my portable amp, along with a negative gain switch, and that little thing has been a dear friend to many of my IEMs.
- settle somewhere around... IDK 0.5V so most IEMs can reach 90dB SPL or more. try to find a sweet spot where the imbalance is lowest(if it's not possible try to find another place but mind the full scale voltage if it could damage your IEMs). and then just stick to that and adjust the signal output in REW at a digital level. it's not an amazing solution in general, but for frequency response measurements it will be very fine. just don't end up with your amp set so that you need -80dB in REW. because it's always cool to have a few bits left to create sine waves ^_^.

 

[Otto Motor]

Whats the output impedance on the Dragonfly? The Fulla is ~ 0.5 ohm. What are the specs on the **** - especially impedance and sensitivity? Does it use crossovers?

The dragonfly is 0.3 ohm.

The ****:
Impedance: 32 Ω
Sensitivity: 110 dB/mW
I'd assume it uses crossovers.
Details: https://www.audioreviews.org/****-****-review/

I documented my measurements, comparisons with amps etc. (a 4-5 hr journey this afternoon) in this pdf:
Download link (same as below): https://we.tl/t-EPPlWSWMmH

The question is: is my Fulla faulty or does it simply not harmonize with such iems? Volume knob problem? I'll write to Schiit and threaten them with a review of their stuff [update: unfair and therefore non-acceptable statement, this is likely a normal thing...should not write after three pints].

From the beginning, the volume know caused a crackling sound -- when I approached the Schiitty customer support, they told me that was normal for small potentiometers [update: totally unrelated to the imbalance; non-acceptable statement].

Thank you very much everybody...I am presently not too receptive as I needed a few Friday evening pints after this.

 

[Brooko]

The dragonfly is 0.3 ohm.
The question is: is my Fulla faulty or does it simply not harmonize with such iems? Volume knob problem? I'll write to Schiit and threaten them with a review of their stuff.

From the beginning, the volume know caused a crackling sound -- when I approached the Schiitty customer support, they told me that was normal for small potentiometers.

Thank you very much everybody...I am presently not too receptive as I needed a few Friday evening pints after this.

Thanks - don't know enough about the Fulla to comment. Some pots suffer channel imbalance at the low end (common issue with an analogue pot). But Schiit know how to make good gear. And also remember that the Fulla isn't designed specifically for what you're trying to get it to do. Upshot - if its not doing the job, egt a cheap adc/dac which will.

I'd also be careful with this sort of throw-away comment

I'll write to Schiit and threaten them with a review of their stuff

From everything I've seen about Schiit - they care about their gear, their customers, and they know good design. I've listened to some of their products and if it wasn't so heavily marked up in NZ, I'd probably buy the Modi R2R, Magni, Vali, Loki combo - and just call it a day

Do some research. Get a fit for purpose set-up, and calibrate it. Then you know where you are.

 

[Brooko]

Just read up on the Fulla - Schiit even advise that it can happen:
https://www.schiit.com/guides/amp-problems

Upshot - its probably the amp low volume channel balance. Use the Dragonfly.

 

[Otto Motor]

Just read up on the Fulla - Schiit even advise that it can happen:
https://www.schiit.com/guides/amp-problems

Upshot - its probably the amp low volume channel balance. Use the Dragonfly.

Thanks Brooko,
It is no problem generating a new calibration file for the Dragonfly. I updated my previous pdf with a few more earphone measurements. The imbalance appears to be a problem mainly at lower volumes, that's my conclusion, too.

Download link
https://we.tl/t-EPPlWSWMmH

This is in accordance with Schiit's statements (your link above). I had initially referred to the issue as equipment incompatibility but was talked into having a faulty Fulla. I have never used the Fulla for listening to earphones but only with full-sized headphones, and never had any problems.

I'll submit my data to Schiit and expect that they'll confirm their "amp problems" statement. Good enough to know that all this is normal.

Problem solved. Thanks everybody.

 

[CoiL]

we need an amplifier because a DAC is simply not designed to handle headphone impedance and current demand from such loads. a DAC "expects" at least a few thousand ohm connected to its output.
second problem, a DAC is likely to have maybe 100 or 200ohm at its output. so imagine you plug into that a multidriver IEM with an impedance going very low at some frequencies, that could often result in several dB change in the FR compared to using the same IEM into any DAP or amp with around 1ohm or less at the output. so not using an amp here results in what you wish to avoid, and most likely in ways more significant than when using even a rather "colored" amp.

I don’t have an E12, so unfortunately I can’t tell. The under 3ohm output impedance shouldn’t mess with crossovers in a low impedance multi-BA. And realistically the noise floor shouldn’t be an issue. You’re measuring sine wave ouputs at approx 85dB, so any noise floor will be masked.

this looks like a good old (but massive) volume knob problem to me. do you get the same thing if you change the level of the test signal output digitally in REW(the stuff at -3dB)?
if I'm correct your issue isn't so much with impedance curve, but with how sensitive the IEMs are. because of their specific sensitivity they make you move the volume knob in a particular area where the problem really is. also most impedance's variation responses with the amp wouldn't show a steady change at all frequencies. so that aspect isn't the main culprit anymore, IMO.



oh and in general, to avoid channel imbalance:
- avoid something with a lot of gain, even unity gain can mean around 2V full scale with most non portable DACs. which is a lot for most IEMs. little gain means that you'll be able to stay away from first third of the volume knob's course where imbalance is more likely to increase.
- get some amp with what they call "digital control" of the volume. I hate that name because we're still fully in the analog domain, but I've seen it used many times. this usually results in outstanding channel matching, the downside being increased noise and a noise that tends to increase when you raise the volume. so it's not very audiophile overall, but it gives very accurate volume setting when we look for that. I have this on my portable amp, along with a negative gain switch, and that little thing has been a dear friend to many of my IEMs.
- settle somewhere around... IDK 0.5V so most IEMs can reach 90dB SPL or more. try to find a sweet spot where the imbalance is lowest(if it's not possible try to find another place but mind the full scale voltage if it could damage your IEMs). and then just stick to that and adjust the signal output in REW at a digital level. it's not an amazing solution in general, but for frequency response measurements it will be very fine. just don't end up with your amp set so that you need -80dB in REW. because it's always cool to have a few bits left to create sine waves ^_^.

So, when I`m measuring with iMM-6 @:
* DX50 (hardware modified, with ADA4841-2YRZ DAC differential opamp (30mA output) ) from LO @ 225 vol level (digital)
* Audiotool SPL offset @ -4.5
* measuring @ ~90dB SPL
* 24-bit 48kHz FLAC pink noise @ FLAC-only decoding software.
* with DAC differential opamp decoupling capacitor and LO output coupling capacitors in chain
* LO specifications:

Frequency response: 20Hz~20KHz +/-0.2dB.
S/N: -109dB +/-3dB.
THD+N: 0.003%.
Output level: 1.5V rms (1kHz 0dB)

Spoiler: DX50 schematic

...I`m doing it wrong and should use HO & low gain?

 

[castleofargh]

So, when I`m measuring with iMM-6 @:
* DX50 (hardware modified, with ADA4841-2YRZ DAC differential opamp (30mA output) ) from LO @ 225 vol level (digital)
* Audiotool SPL offset @ -4.5
* measuring @ ~90dB SPL
* 24-bit 48kHz FLAC pink noise @ FLAC-only decoding software.
* with DAC differential opamp decoupling capacitor and LO output coupling capacitors in chain
* LO specifications:

Spoiler: DX50 schematic

...I`m doing it wrong and should use HO & low gain?

IMO, yes.
you measure something under specific conditions and get a result from those conditions. that is not wrong in itself, but I'm going to assume that your desire is to measure conditions close to real life usage for most people. and that's clearly not a LO straight into an IEM. now of course if that's how you use your IEMs on this DAP(why?????) then you're right to test that for yourself. as always, we end up having to discuss your own intent when doing a measurement.

2 side questions:
-when you say 90dB SPL, you have found a way to calibrate that more or less properly by using some known reference? or is it just what the app shows by default and we can only pray that it's not too far off?
- I don't remember if the DX50 shows problematic low freq roll off because of caps. have you tried to see if you had a significant increase in low frequency roll off when using lower and lower impedance IEMs(maybe compared to measurements from other people)?

 

[CoiL]

assume that your desire is to measure conditions close to real life usage for most people. and that's clearly not a LO straight into an IEM.

I was thinking another way around... ppl have different gear... different sounding DAC`s and different opamps... giving anyway different end result. That`s why I thought using simply DAC is most "uncolored" SQ.
I`m using HO. LO is only for external speaker systems analog feed.

-when you say 90dB SPL, you have found a way to calibrate that more or less properly by using some known reference? or is it just what the app shows by default and we can only pray that it's not too far off?

I measured some of my IEMs (with different impedances) with and external SPL meter (don`t know model, got it briefly for loan at work) and adjusted average vol level according to that from DX50. Should be "accurate" enough.

 

[castleofargh]

I get the idea but in practice you're opening the door to atypical variations based on impedance response because the LO is probably up to 100 or 150ohm. meaning that each IEM has the potential to create it's very own deviation from the more typical response it would have had from a random headphone out somewhere around 1ohm. and you create that serious inconsistency in the name of consistency to avoid what usually would be minor and very smooth difference in the overall frequency response between DAPs or amps.

 

[CoiL]

LO is probably up to 100 or 150ohm.

I couldn`t find out DX50 LO impedance from internet. I know HO is <0.5 Ohm. Would like to know, but how? I need some dummy resistor & multimeter?

Anyway, seems I`m back to drawing board... UH OH

 

[castleofargh]

I couldn`t find out DX50 LO impedance from internet. I know HO is <0.5 Ohm. Would like to know, but how? I need some dummy resistor & multimeter?

Anyway, seems I`m back to drawing board... UH OH

I just googled and found a familiar website with a very familiar Tachikoma giving answers to all the good children. in hindsight I should have started there instead of asking you a bunch of stuff and trying to guess what's what.

Did some basic measurements as promised, and here are the results: headphone-out output impedance is 0.7ohm, line-out output impedance is 124ohm. Low gain headphone-out max voltage 1.079Vrms, mid gain 1.75Vrms, high gain 2.655Vrms . Line-out FR cure is flat. Headphone-out FR curve has some minor bass roll-off due to output capacitors. With a 47ohm dummy load, it is -0.75dB @ 20Hz with corner frequency @ 8Hz. Output capacitance is estimated to be around 400uF. Purely on RMAA result, LO looks cleaner than HO (which is expected of course), though both don't measured quite as good as X3, especially on IMD. Since they are still well below the 1% threshold, it doesn't seem to be of any major concern. Output power is quite good, comparable to most standalone portable amps.