E1DA 9038D Review (portable DAC & Amp)

[capslock]

Thanks Ivan! I am curious about the lower current because some user here said his D got pretty hot with headphones whereas his S stayed cool.

 

[staticV3]

The difference won't be that large. Without headphones plugged in, and while playing 44.1/48 content, I measured 503mW power consumption from the D and 454mW from the 9038S.
Since the S has higher voltage and current capabilities than the D, it's easy to see that the overall efficiency is better there, but the difference isn't that big.

 

[DeepSpace57]

@IVX Hi ivan, have you shipped the latest version of 9038SG3 to @amirm ?

 

[amirm]

@amirm, Your PC HF noise isn't AC main 60Hz, there are from hundreds kHz and up to hundreds MHz, and 1uF at such frequencies is shorter than your USB cable(any 1uf MLCC has 10mOhm of ESR typically and 1-10MHz simply shorted by 1uF to the AP chassis, hence, to its GND wire, next few meters to your PC GND, and returns with USB cable back).

I thought you were talking about mains leakage. Not general USB noise. The analyzer input is band limited to 22.4 kHz in testing so not sure what hundreds of kHz and MHz has to do with the measurements I show.

 

[amirm]

sure, if you use a DAC when lay in a grounded tub full of water, you have a slightly similar GND loop. Normally, you are well isolated from the GND, and doesn't matter grounded PC in use or battery-powered Android. Hence, the most realistic method of measurement is isolated USB source, or isolated AP, or correctly grounded entire setup. As a criterion of how far your AP+DUT setup from an ideal, you can compare that with a battery-powered USB source(Android) as a nearly ideal sample. Using the unbalanced input without isolation is the worst case ever.

Plenty of people use computers with these dongles, specifically laptops. So the source is not always battery powered. When used on Android and such, the OS stack messes with the samples with default making any such refinements in measurement moot anyway. In addition, desktop DAC+Amps are subject to the same condition in their headphone out. They don't seem to be in need of creating battery operated setups to test them.

What I can't figure out is the source of complaints. Your device measured very well and I gave it high marks. What kept it from being the best of the best was clipping at 0 dBFS. That is what needs to be sorted out. Nothing else needed attention and certainly any argument. You can state that you think performance would be better with battery and that would be that.

 

[amirm]

To be very clear, here is the dashboard:

The limiting in performance is third harmonic at about -113 dB or so. Add just a tiny amount of noise in the analyzer and the dongle and you get down to 111 dB SINAD as shown. Yes, there is some low level noise but that is the "grass" well below the harmonics at -140 dB. No way is that our concern, nor is it shown to be the reason distortion is what it is.

Also, as noted much earlier in the review, I allowed the dongle to be tested at 2.7 volts rather than 2 volts. That should have helped it get slightly better performance in SNR/DNR, offsetting the small amount of noise bleed into the device (from whatever source).

Really, I have never seen a manufacturer be so unhappy with what is a great review. And high praise of the designer.

People are getting confused thinking any and all measurements including testing speakers is faulty. Nothing like that is remotely the case.

 

[Rottmannash]

To be very clear, here is the dashboard:

The limiting in performance is third harmonic at about -113 dB or so. Add just a tiny amount of noise in the analyzer and the dongle and you get down to 111 dB SINAD as shown. Yes, there is some low level noise but that is the "grass" well below the harmonics at -140 dB. No way is that our concern, nor is it shown to be the reason distortion is what it is.

Also, as noted much earlier in the review, I allowed the dongle to be tested at 2.7 volts rather than 2 volts. That should have helped it get slightly better performance in SNR/DNR, offsetting the small amount of noise bleed into the device (from whatever source).

Really, I have never seen a manufacturer be so unhappy with what is a great review. And high praise of the designer.

People are getting confused thinking any and all measurements including testing speakers is faulty. Nothing like that is remotely the case.

I can imagine he's like a Dad with a newborn child-perfect and beyond reproach. I just ordered one and I am certain it will sound great so no worries.

 

[IVX]

What I can't figure out is the source of complaints.

it is not about my product but about scientific principles. The truth has priority #1, not a high mark, and not money. Your setup shows its own noise, I already a few times said what needs to change in that setup to turn the science On.

 

[MC_RME]

I did that directly to the 3.5mm plug with 6mm PLB traces, as you can see on my pic, AP's chassis tied with one more croc-clip to the USB-C shield. That's it, that setup has minimal difference vs isolated Android.

To rest this question: your pic shows that you do not connect shield to the DUT but take it from the AP (via XLR pin 1). Thanks for the info, I will try that as well. So far I (like most others, IMHO) only used the unbalanced input in unbalanced cases, and stopped noise artefacts by USB isolation, battery power supplies and specific grounding arrangements. Will be interesting to see if your method is easier and still comparable and fully compatible. Will still require to solder some special cables for various purposes, though.

 

[IVX]

MC_RME, furthermore, I use USB isolator for AP's USB interface to get almost 100% match to Android result on PC. So, XLR with >100db CMRR(AP has 2 pots for each channel to adjust max of CMRR, one resistance balance, another variable capacitor for HF CMRR) is not a panacea too, at least not for 100.00% but perhaps for 97% for sure.

 

[amirm]

it is not about my product but about scientific principles. The truth has priority #1, not a high mark, and not money. Your setup shows its own noise, I already a few times said what needs to change in that setup to turn the science On.

What science? You haven't shown the impact of the issue on the measurements. The main impairments are 0 dBFS clipping and third harmonic.

 

[IVX]

amirm, I stopped that discussion. I said all that I think about the situation and I have nothing to add. Remember, people believe you.

 

[Hifihedgehog]

I bought a 9038D and like it a lot. However, the response I am seeing here from IVX is highly irregular for him. I don’t know what it is, but I would recommend to him to get it settled. Normally, he is pretty easy going and very open when people give constructive feedback. People know him enough to know where he is coming from and his decency and integrity. But for people on the outside looking in, this might be a bad look. Honestly, if I were him, I would take @Amir’s feedback to heart and apply the fixes he needs to rectify the clipping. If I were him, I would be pleased to see Amir raving about my product and giving it his recommendation. I don’t know if he is feeling salty about the S9 getting the upper hand in the measurements, but Amir’s test setup doesn’t lie about the clipping. The best choice he can make is to learn from this valuable experience. He can be grateful for the opportunity to become better than ever before. It is because Amir was thoughtful enough to tell him his honest observations kindly and positively instead of telling him only what he may have wanted to hear. In fact, after perhaps a dozen responses from Amir to answer other concerns, Amir has continued to be patient with him. Take my advice, IVX: drop it and move on. You are hurting yourself and your reputation by keeping this going.

 

[IVX]

I don’t know what it is, but I would recommend

As I said, I stopped to discuss Amir's test setup. It is not my business but I know a lot of engineers and they know what it is

 

[Veri]

I bought a 9038D and like it a lot. However, the response I am seeing here from IVX is highly irregular for him. I don’t know what it is, but I would recommend to him to get it settled. Normally, he is pretty easy going and very open when people give constructive feedback. People know him enough to know where he is coming from and his decency and integrity. But for people on the outside looking in, this might be a bad look. Honestly, if I were him, I would take @Amir’s feedback to heart and apply the fixes he needs to rectify the clipping. If I were him, I would be pleased to see Amir raving about my product and giving it his recommendation. I don’t know if he is feeling salty about the S9 getting the upper hand in the measurements, but Amir’s test setup doesn’t lie about the clipping. [...]

It's not about the clipping though. IVX is saying, when using a slightly different set-up to measure the 9038d direct output, with better CMRR the overall output will look cleaner (so third harmonic will also be lower supposedly, and overall SINAD might increase by at least few dB). That is from his own testing. As Amirm measured now, it shows the 9038d more in a line-out use case to another device and not as a straight output to, like @MC_RME worded it nicely, "a floating, earth-free headphone." which will perform (even) better than what is currently shown here. Wiring it a little differently to the APx, you could maybe simulate that.. but it's yet another measurement step, I guess it won't happen

In any case Amir has shown the 9038d measures very well. IVX is simply saying from his own testing, in real-world application it might perform even better due to the reasons mentioned.

 

[capslock]

It's not about the clipping though. IVX is saying, when using a slightly different set-up to measure the 9038d direct output, with better CMRR the overall output will look cleaner (so third harmonic will also be lower supposedly, and overall SINAD might increase by at least few dB). That is from his own testing. As Amirm measured now, it shows the 9038d more in a line-out use case to another device and not as a straight output to, like @MC_RME worded it nicely, "a floating, earth-free headphone." which will perform (even) better than what is currently shown here. Wiring it a little differently to the APx, you could maybe simulate that.. but it's yet another measurement step, I guess it won't happen

In any case Amir has shown the 9038d measures very well. IVX is simply saying from his own testing, in real-world application it might perform even better due to the reasons mentioned.

Concur with Veri. 3rd harmonic merits more discussion. When I pointed out that SINAD in Amir's measurement was limited by - 114 dB 3rd harmonic (post #88), MC_RME kindly provided a link to measurements with and without USB isolator. That showed an increase in even harmonics that cannot plausibly explained by noise being integrated in FFT bins (see post #114).

So that 3rd HD spike could well be an isolation issue (I would have tought RF rectification would indeed cause odd harmonics). It could also be due to output level being too close too clipping. We can't tell from Amir's measurement.

Ivan's said that he never measured worse than -120 dB 3rd on any 9038D. So at what level was this? Full scale?

Is the clipping a question of firmware, hardware or USB bus voltage? If bus voltage, there could be clipping at 4.8 V but not at 5.0 V.

 

[capslock]

Since we cannot rule out that grounding issues will give us harmonics, how about about doing a dashboard for unbal and another for bal?

This way we preserve backwards compatibility and users can pick which one is relevant to their application. As has been said before, headphone users couldn't care less about grounding and should look at the balanced results. Measurements enthusiasts looking for a signal generator should be using balanced in anyway.

And for plain old enjoyable listening, anybody fanatic enough to go looking for the best DAC around should also make sure they have a preamp with balanced input, so they have more use for the unbal measurement, too.

On a side note, what USB isolators can be recommended? I suppose those $10 isolators which are good for 12 MBit/s or full speed are not sufficient, we need high speed?

 

[julian_hughes]

For people thinking of using this with smartphone it's worth noting two really important drawbacks not touched on in the review:

Power consumption is HUGE for a dongle. At idle it consumes 95 mA, and in use hundreds. This *will* have a very obvious effect on battery life. Most dongles designed for use with phones achieve standby power consumption a third or half that, and correspondingly lower consumption in use.

Noise from LTE: the dongle becomes noisy with 4G & 5G.

This looks like a great product for use with a laptop or PC but impractical or disappointing for pairing with a phone.

 

[jannek]

This is nothing the dongle can be blamed for... Being powerful enough to drive full size cans doesn't work without current. I could imagine a situation on a business trip when charging the phone is possible and 4G/5G not relevant... On a plane or in a hotel room for example. Usage of a y-adapter for supplying energy from a powerbank or similar was also mentioned.

 

[julian_hughes]

This is nothing the dongle can be blamed for... Being powerful enough to drive full size cans doesn't work without current. I could imagine a situation on a business trip when charging the phone is possible and 4G/5G not relevant... On a plane or in a hotel room for example. Usage of a y-adapter for supplying energy from a powerbank or similar was also mentioned.

Other dongles including products tested here manage to have much lower idle consumption but still achieve output suitable for full size headphones, while also completely suppressing LTE noise. I may be wrong but I don't think you can simultaneously charge a phone from a powerbank while the phone supplies power to external devices such as a dongle. My Sony definitely switches USB modes and can't do this. Perhaps other brands or newer type C implementations can do this?

I do have a similarly current hungry dongle, the LG HiFi Plus (the original LG manufactured one, not the odd DIY/improvised device tested by Amir). It just kills the battery of a smartphone, very obviously impacting screen time available on a charge. It sounds beautiful, can drive my HiFiMan Planars but is simply impractical as a cellphone accessory. It's great for a laptop though. It does completely suppress LTE noise btw. In the end I settled on using a Musiland MU1 dongle because it has much less impact on battery life and its sound quality is more than good enough for mobile use in the city or while doing stuff at home. It also totally suppresses all LTE noise and can drive full size headphones reasonably well. It draws about 30 mA when idle and certainly does not draw a constant 400 or 500 mA in use like the E1DA models.


I've owned and used a variety of different portable amps and USB dongles and experience has shown me that real life issues can matter at least as much, or even more, than great benchmarks, because in tests not everything that matters gets considered and assessed. I use my "second best" dongle in preference to my "best" dongle because as soon as the host is a cell phone other factors come into play. My favourite tiny portable amp, an ancient Topping NX-1, became less appealing way back when I switched my smartphone to a 4G model and discovered it blasted all that LTE noise right into my ears! It's still useful with other products of course.

I'm sure the E1DA products sound really excellent. I'm not offering negative assessment of them but thought it worth pointing out a couple of things that didn't get a mention in the tests & review but which do have a really big impact on user experience, usefulness and satisfaction in the real world beyond the test bench.