Review and Measurements of E1DA 9038S BAL Portable DAC & Amp

[raanany]

I'm using 9038s v2 on

• Sennheiser HD600 (Sensitivity 97 dB, Impedance 300 Ohm)
• Audiosense T800 IEM (Sensitivity 90 dB, Impedance 9.2 Ohm)

Volume level on HD600 in Mode2 performance between 80-100 (-5dB to allow EQ) where 100 is really really loud.
T800 in Mode4 IEM between 50-90 where 100 is unbearably loud.

I don't see a fail of the product for IEM.
And I was positively surprised about the performance mode on Hd600.

@lxlx I never used the word fail and I think it's unfitting.
At 90dBSPL / mw your T800 are 26dB shy of the Campfire Andromeda Gold's 116dBSPL / mw. Also, the impedance ratio of 9.2/7 (average) pushes this further to 27dB (E1DA has practically zero output impedance).

No wonder you're happy. I'd be too.

I just think this DAC misses its target audience. It's a great technical achievement but who's the target customer? I need a small portable DAC for on-the-road and office. In both these cases I use IEM. (At home I have my large Senns too but there, I prefer the comfort of a desktop DAC+AMP). IEMs suffer from such a high voltage gain. They'd benefit from the insanely low noise and low output resistance (which is not required for your HD600). It's just the voltage gain which is a killer for the hyper sensitive IEMs. In case of mine I expect to even hear the noise floor (it would be 28dBSPL).

It's a simple change, have the DAC operate with 2 reference voltage options. One for high impedance big cans and one for crazy IEMs

 

[mgmate]

mgmate
you can find a lot of help here https://discord.gg/ESbQ6A
>1200 E1DA users already!

Thanks!

 

[lxlx]

@raanany I don't think it misses its audience. imho your IEM is just insanely sensitive. My T800 is already picky here, but Andro is even trickier.
HD600 and T800 well driven from the same headphone amp is an achievement.
Maybe Andro is just too much to ask for, from such a device? As @IVX said, there are better alternatives for pure IEM usage. And amirm measured some...

 

[Schackmannen]

@Schackmannen I don’t see where in this analysis you’re taking gain and the resulting quantization noise into account. Lets take an extreme example: Say I have an ideal DAC with zero residual noise but the total gain in my system forces me to use just the upper 4 bits of my signal then my LSB noise (quantization noise) is going to inflict SNR=4x6=24dB. I’m sure you’ll hear degradation. What am I missing in my calculation?

By the way, practically, with the Campfire Andromeda Gold’s sensitivity of 116 dB SPL/mW the admittedly low 3.5 uV residual noise of the E1DA 9038S is still going to give you 28dB SPL which is well above the hearing threshold.

I knew the Andromeda was sensitive but didn't know it was that extremely sensitive, seems like a pretty bad design to me. For the residual noise to produce 0 dB SPL it would have to be only 0.14 uV which is pretty much impossible to achieve right? Seems like no matter what you use the Andromedas with you're bound to have some hiss that's potentially audible. Regarding your example I'm not sure I'm following and what you're trying to show, if the DAC has zero residual noise then no matter how much you increase the gain afterwards there will still not be any noise from the DAC, any noise will be coming from the rest of the chain. So if the gain of your system still only leaves you with 4 bits of dynamic range then you must have a crazy amount of gain for no reason at all?

That is exactly my point Ivan: a pot placed between gain-stage and output buffer will always lower the background noise on IEMs.

A pot can only reduce noise by so much before the pot becomes the limiting factor itself, and that limit doesn't seem to be much lower than what some of the best DACs can achieve. Also IMO it's not really a fair example anyways since most pots have horrible channel balance long before it's even turned all the way down so you can't even take advantage of that. Thinking about it now I'm wondering how @amirm performs the 50 mV SNR test on headphone amps, if its done by turning the pot all the way down and then playing with the input voltage to achieve an output of 50 mV then I'm not sure how useful that test is. Seems to me the most realistic way to do the test would be to turn the pot down as much as possible without exceeding a channel imbalance of say 0.5 dB and then adjusting the input voltage to achieve a 50 mV output. If I had to guess I think a lot of the amps would perform quite a bit worse in that test if it was done this way.

 

[raanany]

I knew the Andromeda was sensitive but didn't know it was that extremely sensitive, seems like a pretty bad design to me. For the residual noise to produce 0 dB SPL it would have to be only 0.14 uV which is pretty much impossible to achieve right? Seems like no matter what you use the Andromedas with you're bound to have some hiss that's potentially audible. Regarding your example I'm not sure I'm following and what you're trying to show, if the DAC has zero residual noise then no matter how much you increase the gain afterwards there will still not be any noise from the DAC, any noise will be coming from the rest of the chain. So if the gain of your system still only leaves you with 4 bits of dynamic range then you must have a crazy amount of gain for no reason at all?


A pot can only reduce noise by so much before the pot becomes the limiting factor itself, and that limit doesn't seem to be much lower than what some of the best DACs can achieve. Also IMO it's not really a fair example anyways since most pots have horrible channel balance long before it's even turned all the way down so you can't even take advantage of that. Thinking about it now I'm wondering how @amirm performs the 50 mV SNR test on headphone amps, if its done by turning the pot all the way down and then playing with the input voltage to achieve an output of 50 mV then I'm not sure how useful that test is. Seems to me the most realistic way to do the test would be to turn the pot down as much as possible without exceeding a channel imbalance of say 0.5 dB and then adjusting the input voltage to achieve a 50 mV output. If I had to guess I think a lot of the amps would perform quite a bit worse in that test if it was done this way.

@Schackmannen about the Andromeda - please note that I'm using the Andromeda-Gold (special edition...) which is even more sensitive. The regular Andromedas are slightly less fussy but still crazy. I never heard the Andromeda but the Andromeda Gold are amazing IEMs. The best I've ever heard and they beat just about anything I've heard except my Sennheiser HD800S. As for my example - it is flawed. I later understood why you were right all along (as I wrote above). As you said, if the DAC+AMP residual noise is lower than 0dbSPL and (important and goes without saying but I missed that) the DAC's LSB quantisation noise (taking into account gain and sensitivity) is lower than 0dbSPL (which is the case with all 24 bit and above DACs) then my example doesn't apply. My example describes a non realistic scenario where the DAC's LSB contributes a signal amplitude which is greater than 0dBSPL. In such a case, if you attenuate digitally to compensate for a fixed high DAC+AMP+IEM gain, the LSB bits that "fall off" and get truncated (or rounded off) are bits you could have enjoyed. I gave the exaggerated example of being left with just 4 bits of signal. My mistake obviously is this assumes unrealistic LSB noise and that if the LSB noise is lower than 0dBSPL then you'll always end up having exactly the amount of bits you need to get you the dynamic range you want and this will never be 4 bits but more like WHATEVER dBSPL you like listening to, divided by 6.

TL-DR - you were right all along

 

[trl]

Thanks @trl but the iFi Ear Buddy has a too high output impedance. IEMs often have a very frequency dependant impedance (considerably lower in the bass region) so you must use a very low AMP output resistance (such as E1DA) if you wish to get a flat response.

Just purchased an EarBuddy today and I get a flat freq. response from 20 Ohms IEMs (+2 Ohms the cable), despite the DC output resistance of the EarBuddy of 4 Ohms. Also tested with 30 Hz sines too, not just with regular music.

 

[raanany]

Just purchased an EarBuddy today and I get a flat freq. response from 20 Ohms IEMs (+2 Ohms the cable), despite the DC output resistance of the EarBuddy of 4 Ohms. Also tested with 30 Hz sines too, not just with regular music.

@trl congrats on the new purchase. I'm glad it works for you. For me however, there's a big difference between 20 and 7 ohm (nominal) and 3 ohm (bass) of the Andromeda Gold. Check out the web for resistor divider to see the math (it's basically R_iem / (R_iem + R_amp) ). You'll need to square this value to get the power ratio or just 20*log10 of this voltage ratio. If you do the math for bass (3ohm) versus nominal (7ohm) versus treble (12-14 ohm) impedances you'll that the Andromeda loses quite a few dBs in the bass region. As I said, that's a cool solution for regular IEMs.

 

[trl]

Just realized now that we're speaking about some IEMs with the below specs:
- SENSITIVITY: 116 dB/mW (or 137.55 dB/V SPL)
- IMPEDANCE: 7 ohms @ 1 KHz

I guess a sub-Ohms amplifier impedance would be indeed required, although your math doesn't actually reflects the freq. response of these IEMs nor their actual impedance or I misunderstand your calculations. Here's how two different IEMs can measure differently, regarding their output impedance (not their DC resistance): https://www.innerfidelity.com/images/SennheiserMomentumInEar.pdf vs. https://www.innerfidelity.com/images/InEarStageDriverSD2.pdf. First one if perfectly flat across audio spectrum, while the other one oscillates between 18 Ohms to 165 Ohms.

For such very low impedance IEMs even an unity-gain desktop headamp will struggle to not induce hiss-noise. I'm also not sure what portable devices will be able to drive them well too, but these are definitely complicated to drive IEMs due to their extreme sensitivity and low-impedance.

 

[Sinquito]

Just realized now that we're speaking about some IEMs with the below specs:
- SENSITIVITY: 116 dB/mW (or 137.55 dB/V SPL)
- IMPEDANCE: 7 ohms @ 1 KHz

I guess a sub-Ohms amplifier impedance would be indeed required, although your math doesn't actually reflects the freq. response of these IEMs nor their actual impedance or I misunderstand your calculations. Here's how two different IEMs can measure differently, regarding their output impedance (not their DC resistance): https://www.innerfidelity.com/images/SennheiserMomentumInEar.pdf vs. https://www.innerfidelity.com/images/InEarStageDriverSD2.pdf. First one if perfectly flat across audio spectrum, while the other one oscillates between 18 Ohms to 165 Ohms.

For such very low impedance IEMs even an unity-gain desktop headamp will struggle to not induce hiss-noise. I'm also not sure what portable devices will be able to drive them well too, but these are definitely complicated to drive IEMs due to their extreme sensitivity and low-impedance.

Wouldn't the noise floor of the amplifier (or DAC) cause the hissing in low impedance IEMs? As far as I know, the difference in output impedance of an amplifier would only change the Frequency Response (how it sounds) of the headphone/IEM.

Surely, somebody more knowledgeable will chime in.

 

[raanany]

Wouldn't the noise floor of the amplifier (or DAC) cause the hissing in low impedance IEMs? As far as I know, the difference in output impedance of an amplifier would only change the Frequency Response (how it sounds) of the headphone/IEM.

Surely, somebody more knowledgeable will chime in.

Allow me to try to explain: Given a headphone's amp noise floor (to get it, take the full scale 0dBFS RMS output voltage and divide by 10^(SNR[dB]/20) ), the amount of power going into the headphones (or IEM) is Vnoise^2 / R where Vnoise is the RMS voltage of the noise as calculated above and R is the headphone impedance in Ohms. Multiply this result by 1000 to get the power in mW as most headphones sensitivity figures are normalised to 1mW. Take the power in mW and multiply by 10^(sensitivity/10) where sensitivity is the headphone's sensitivity given in dbSPL/mW. If the result is higher than 1 then the noise is coming into your ears at higher than 0dbSPL and you may hear it if you have good ears and a quiet surroundings.

 

[Sinquito]

Allow me to try to explain: Given a headphone's amp noise floor (to get it, take the full scale 0dBFS RMS output voltage and divide by 10^(SNR[dB]/20) ), the amount of power going into the headphones (or IEM) is Vnoise^2 / R where Vnoise is the RMS voltage of the noise as calculated above and R is the headphone impedance in Ohms. Multiply this result by 1000 to get the power in mW as most headphones sensitivity figures are normalised to 1mW. Take the power in mW and multiply by 10^(sensitivity/10) where sensitivity is the headphone's sensitivity given in dbSPL/mW. If the result is higher than 1 then the noise is coming into your ears at higher than 0dbSPL and you may hear it if you have good ears and a quiet surroundings.

I do believe you, but my comment was about trl's post about how high output impedance on the AMP would induce hiss on low impedance IEMs.

I was expecting some explanation as to why high output impedance would not induce hiss but instead change the frequency response of a low impedance IEM (due to the damping factor) and that hiss is (AFAIK) induce by high noise floor in the AMP/DAC.

 

[raanany]

I do believe you, but my comment was about trl's post about how high output impedance on the AMP would induce hiss on low impedance IEMs.

I was expecting some explanation as to why high output impedance would not induce hiss but instead change the frequency response of a low impedance IEM (due to the damping factor) and that hiss is (AFAIK) induce by high noise floor in the AMP/DAC.

I think the contrary is correct. A high AMP output impedance (all else being equal) will reduce hiss because the output signal and noise will be reduced when compared to low output impedance.

 

[trl]

Wouldn't the noise floor of the amplifier (or DAC) cause the hissing in low impedance IEMs? As far as I know, the difference in output impedance of an amplifier would only change the Frequency Response (how it sounds) of the headphone/IEM.

The sensitivity of headphones is what makes them more or less quiet, but of course, the noise generated by all stages from inside an amplifier is more important. So, it's like you said: a higher output impedance of an amplifier may change freq. response of low-impedance headphones, while a lower output impedance of the headphones usually translates in a higher sensitivity, hence a higher risk of hearing hiss. This is why I recommended EarBuddy for >16 Ohms sensitive headphones.

 

[staticV3]

Hidizs S8 Review (vs Meizu Pro & 9038S)
https://docs.google.com/document/d/1KrTsZ2fhESXIHy7vgufu57WVuu1OzBOFaYlflbYK7hQ/edit?usp=sharing

 

[tifune]

Anyone have an issue w 9038S making some sort of "digital hissing" sound? I can hear it both on my IEMs and studio monitors. It comes and goes at random, almost like radio interference but even at low volume it's a little painful. Tried different cables (balanced) and sources but it always shows up after a short time

 

[mgmate]

Anyone have an issue w 9038S making some sort of "digital hissing" sound? I can hear it both on my IEMs and studio monitors. It comes and goes at random, almost like radio interference but even at low volume it's a little painful. Tried different cables (balanced) and sources but it always shows up after a short time

Not me on Sennheiser HD 660S

 

[Veri]

almost like radio interference but even at low volume it's a little painful

Could be EMI interference..?

 

[tifune]

Could be EMI interference..?

That was suggested to me in the Discord; next time I catch it I'm going to try airplane mode just confirm. If it is EMI, is there anything I can do aside from airplane mode?

 

[Veri]

That was suggested to me in the Discord; next time I catch it I'm going to try airplane mode just confirm. If it is EMI, is there anything I can do aside from airplane mode?

Besides finding the source and removing it, I'm not sure.. :/

 

[Nicosea]

Anyone have an issue w 9038S making some sort of "digital hissing" sound? I can hear it both on my IEMs and studio monitors. It comes and goes at random, almost like radio interference but even at low volume it's a little painful. Tried different cables (balanced) and sources but it always shows up after a short time

I experienced that on Windows 10. I had to go to the audio settings and select 24bits and 96khz for the DAC (it was on 16bit and 44khz).