E1DA 9038D Review (portable DAC & Amp)

[staticV3]

@alter4 That's the newest version.

 

[amirm]

I'm pretty sure there is one.

Then your argument that my analyzer runs wideband but your audio gear doesn't due to simple RC filter doesn't hold.

 

[amirm]

Very important point there. It's not that DACs or amps must not emit any ultrasonics, they just have to stay within legal limits.

And where is the certifications that this device has?

 

[amirm]

It's clear that a balanced DAC output will be measured with balanced in

Not so. I measure both outputs of many such DACs. Difference is tiny -- usually 3 dB is so.

 

[alter4]

Thanks! I got my device with typeC-typeC wire but I would like to charge my android phone (at least have dedicated power for the 9038D to less intensive discharge) and enjoy music the same time. So might be someone could bring me light:
1) Does something like this work for it?
2) Should I avoid any "power noise" from my included Samsung power adapter?

 

[staticV3]

Thanks! I got my device with typeC-typeC wire but I would like to charge my android phone (at least have dedicated power for the 9038D to less intensive discharge) and enjoy music the same time. So might be someone could bring me light:
1) Does something like this work for it?
2) Should I avoid any "power noise" from my included Samsung power adapter?

It is possible to supply data to the 9038D from your phone while supplying power from an external source like a phone charger or power bank, therefore preserving your phone's battery life.

There is one big problem however:
Most of these USB "Y-splitters" have data and power lines connecting USB host and slave. That means that they will still work even without an external power source plugged in.
If you use such a Y-splitter to connect your phone, the 9038D, and a power bank for example to each other, then one of three things can happen:

• The phone will continue supplying power to the DAC as always, with the power bank doing nothing
• The power bank will correctly supply the DAC, preserving your phone's battery life
• Your phone will see the power bank as another power consumer, charging it in addition to powering the DAC, depleting its battery even faster. This happened to me using a OnePlus/Samsung phone and an Anker/Ravpower battery.

To avoid this, you need to use a Y-splitter without a 5V connection between USB host and slave. This can be done either by opening up the Y-splitter
and cutting the red wire inside, or by buying a splitter made specifically for this purpose.
The only splitter I know of with this configuration is sold by Yinlumei on Taobao. Here's the link: https://m.intl.taobao.com/detail/detail.html?id=587754912509
and a picture (the link only shows the connectors for some reason):


It is also possible to supply power externally to both the DAC and your phone at the same time. However, you will need a phone with support for USB power delivery, a USB hub with power delivery passthrough, and a power delivery charger. No janky mods required.

 

[amirm]

To continue this thought: what is the difference between a balanced headphone and a loudspeaker?

Answer: slew rate.

APx555 has wideband for a reason. If we were to always filter what goes into it, why on earth would they design it with 2 MHz bandwidth? They don't. The system does have a weak point which is called slew rate limiting. Switching class D amps have very high slew rate in the frequency range of their amps. The switching swings from rail to rail so has incredibly high slew rate, causing problems for the front end of the AP analyzer. For this reason, I use AES-17 (wideband) filter. A headphone amp with the same switching frequency would have far lower voltage swings so has much lower slew rate so doesn't cause the same type of issue for AP. Indeed, I have measured hundreds of DACs with switching power supplies or USB power input, none of which have show the type of problems you see when AP slew rate limits.

Note that the AP AES filter has very wide bandwidth:

Mine is rated for "40 kHz" but that only means that it is within AES spec in that region. As you see, it lets through the input almost unchanged until about 100 kHz. So it will not solve any issues with ultrasonic frequencies up to that region.

Finally, there is a cost to these filters. Here is the specs for my AES filter:

Distortion
< -110 dB harmonic (measured at 70 Vpp, 1 kHz)
< -100 dB IMD (at 70 Vpp with 18 kHz and 20 kHz dual-tone test signal)

This is not much of an issue for power amplifiers. But for DACs? It can degrade their performance to their much higher performance and cause its own set of arguments.

 

[capslock]

Not so. I measure both outputs of many such DACs. Difference is tiny -- usually 3 dB is so.

My point was you in all likelyhood use the balanced in to measure XLR out? And probably single ended in to measure RCA out?

 

[capslock]

Answer: slew rate.

APx555 has wideband for a reason. If we were to always filter what goes into it, why on earth would they design it with 2 MHz bandwidth? They don't. The system does have a weak point which is called slew rate limiting. Switching class D amps have very high slew rate in the frequency range of their amps. The switching swings from rail to rail so has incredibly high slew rate, causing problems for the front end of the AP analyzer. For this reason, I use AES-17 (wideband) filter. A headphone amp with the same switching frequency would have far lower voltage swings so has much lower slew rate so doesn't cause the same type of issue for AP. Indeed, I have measured hundreds of DACs with switching power supplies or USB power input, none of which have show the type of problems you see when AP slew rate limits.

Note that the AP AES filter has very wide bandwidth:

Mine is rated for "40 kHz" but that only means that it is within AES spec in that region. As you see, it lets through the input almost unchanged until about 100 kHz. So it will not solve any issues with ultrasonic frequencies up to that region.

Finally, there is a cost to these filters. Here is the specs for my AES filter:

Distortion
< -110 dB harmonic (measured at 70 Vpp, 1 kHz)
< -100 dB IMD (at 70 Vpp with 18 kHz and 20 kHz dual-tone test signal)

This is not much of an issue for power amplifiers. But for DACs? It can degrade their performance to their much higher performance and cause its own set of arguments.

I wasn't talking about headphone amps vs power amps, I was talking literally about (balanced) headphones and (passive) loudspeakers. Both couldn't care less whether they are connected to a single ended or BTL amp and whether on of their terminals is grounded. (Of course, never connect a regular TRS wired headphone to a BTL amp.)

So you likely measure BTL power amps with balanced inputs? Do you use single ended for single ended amps?

But to follow your filter tangent, it was very interesting to learn about slew rate limitations in the AP. I think the reason they have a wide open preamp that the analyze has a 16 bit, high sample rate mode. But why didn't they design a simple RC lowpass that is automatically activated when in 24 bit mode? I wasn't thinking of slewing limitations, and I am not even sure they are going to be more severe with a class D amp than with a budget DAC simply because I'm sure the input is attenuated when measuring signals of more than a few volts. My thinking was a about RFI:
https://www.analog.com/media/en/training-seminars/tutorials/MT-096.pdf

RFI will not only cause offset variation in very LF measurement situations, it will also cause HD in audio frequency amplification. Maybe not enough to show up in a 16 bit measurement, though.

Lastly, the AES-17 must be better than specified, or we wouldn't have seen those beautiful results of the nc400 and 1ET400.

 

[IVX]

capslock,
that's why I add LPF right on the first stage of my Cosmos ADC front-end, as any audio device, actually, always contains input RC.
I don't call the distortion induced by HF input noise "rectification" because the rectification just a particular case, a general reason of that distortions has two components:
a) real-world opamps inherently nonlinear
b) real-world opamps have much less loop-gain at high frequencies

 

[NM156]

It's not that DACs or amps must not emit any ultrasonics, they just have to stay within legal limits.

Does the ultrasonic noise from this device stay within legal limits when it drives Amir's AP inputs crazy?

 

[IVX]

NM156, 100%
To be honest, I see no problem with APs bad HF immunity in that test. At least that test have detected my 9038D 1.5MHz noise at one channel, hence, the test exposed something interesting. The fact that 2-3mV 1.5MHz noise isn't audible and doesn't produce distortions and will not turn headphones to the ash is another topic.
The wrong result is DR and THD+N at levels close to 0dbfs caused by trivial ground loop common-mode noise. It is not really principal to use ONLY XLRs, BNC is ok but how did mention RME, you have to isolate DUT to eliminate GND loop noise affects the DUT noise measurement. XLR with CMRR >100db just can help you to solve that task a lot easier than isolation and/or searching for the best GND points. 9038D got the result isn't crazy wrong vs Meizu HiFi DAC pro, as an example, which actually has 127db(130db(A)) DR but not 103db, 16x times error.

 

[NM156]

E1DA 9038D Measurements (Headphone Amp)
Power is everything in headphone amps so let's see how much we have at 300 ohm:

I feel the urge to ask again because I don't have an answer yet: why this anomaly on Ch2 below 0.5mW?
The sudden difference between the two channels is a whopping 10dB.
IVX explained the reason for the general channel imbalance (additional noise in only one channel due to the placement of the SMPS on the PCB), but not for this anomaly.
I would really like to know.

 

[IVX]

NM156, it seems we use completely different languages, I did explain that difference in THD+N at 300mW by 1.5MHz noise presented at one of the channels.

 

[NM156]

NM156, it seems we use completely different languages, I did explain that difference in THD+N at 300mW by 1.5MHz noise presented at one of the channels.

OK, let me clarify:

In my first post #148 about this issue, I made a mistake and wrote 0.5mV instead of 0.5mW. I actually meant watts (for power).
In my post #162 I corrected this. Sorry if this typo confused you.

Now, if I look at the graph in my post #253 then I see that the curve for channel 2 (light red) suddenly (coming from right) increases (compared to the red curve for channel 1) at about 0.5mW power.
You write about THD+N at 300mW, I don't know why, I don't see 300mW for 300 Ohm in the graph, just a maximum of 20.21mW.

If I understand you correctly, you are saying that the sudden increase is due to the AP, that is, it is just a measurement problem.
In other words, if the AP had an input LP filter (at least for channel 2), this increase would not be there.
Is this correct?

 

[IVX]

NM156, right, and I have posted the result with such 70kHz LPF+notch, as you can see there is no any anomaly, all are smooth and THD+N -118.2db(0.000123%) one ch, 117db another.
PS: sorry if I did wrongly understand something too, I keep working and quite hard, so I have seconds to reply here and read not really carefully

 

[DeepSpace57]

As an end-user, I can ask what the heck the stand-alone DACs like gustard X16 are not influenced by noise through Amir`s test equipment, then.

The answer could simply be they are still better than these dongles. However;

Can we notice this?

 

[IVX]

gustard X16 has internal volume higher than 9038D >205x times, not even mentioned how is Meizu smaller You can check Hidiz S8 as well, it is the same CS43131 DAC as Meizu HiFi but suddenly Amir's setup got it clean even in unbalanced BNC inputs, I have no idea how it happened. GND loop is such kind of thing, never steady stable, a good idea to fix it once, and forget.

 

[Haider]

Received mine yesterday, hooked it up to my PC & iPhone and sounds great with my Sennheiser PX100 and Amazon HD streaming. On a par with headphone output on the Marantz PM6006 amp. Have to do some critical listening to see if Chi-Fi beats Vietcong-Fi...

 

[AudioJester]

Can I ask about the android app and "programming" the dongle. If I change settings in the app and then connect the dongle to a windows pc - are the settings still changed? Or do the changes only hold when connected to an Android device running the app?