Loxjie D30 DAC and Headphone Amplifier Review

[Propheticus]

I'd like to understand how this is a ground loop when:
- it happens without any input attached (no s/pdif nor usb).
- it disappears when the powerline adapter, which is plugged into a different wall socket 2m away, is removed.*

Do you mean an internal ground loop or via the headphone plug?

I guess it's a matter of trying out one of those RMI/RFI filtered extension leads. So I ordered one.


*: Yes, I also tried plugging it in to the extension lead which is connected to the socket on the powerline adapter. Sadly it does not appear to filter its own noise from this output. Yet another reason to avoid powerline network solutions if possible... performance was already mediocre with pings to my router of avg 10ms and peaks to ~30ms.

 

[Toku]

I'd like to understand how this is a ground loop when:
- it happens without any input attached (no s/pdif nor usb).
- it disappears when the powerline adapter, which is plugged into a different wall socket 2m away, is removed.*

Do you mean an internal ground loop or via the headphone plug?

I guess it's a matter of trying out one of those RMI/RFI filtered extension leads. So I ordered one.


*: Yes, I also tried plugging it in to the extension lead which is connected to the socket on the powerline adapter. Sadly it does not appear to filter its own noise from this output. Yet another reason to avoid powerline network solutions if possible... performance was already mediocre with pings to my router of avg 10ms and peaks to ~30ms.

The noise of your A30 is not due to the ground loop because the noise is generated when nothing other than the AC line is connected.
The cause is that the power line net adapter superimposes strong pulse noise on the power line, which invades the power supply of the D30.
To prevent this, either stop using the powerline net adapters or insert a noise filter into the D30's powerline.
Powerline net adapters are also allowed in Japan, where I live, but they are rarely used due to these noise issues.

 

[Propheticus]

Thanks, exactly what I thought. The filtered extension lead arrives on Monday. Fingers crossed.
At the same time looking at alternatives for the powerline adapters... not really happy with them anyway. Perhaps a mesh solution like the TP-link deco sets.

 

[somebodyelse]

Any local radio amateurs will be happy if you get rid of the powerline networking. It's arguable whether it should be allowed in the EU at all as it causes exactly the problems the EMC directive was intended to prevent, even if somehow the adapters make it through the testing process.

 

[Labjr]

Radio amateurs think their interference is your problem!

 

[Propheticus]

Update time. The delivery was rather quick and the extension lead with filter arrived today instead of Monday.
Sadly the interference was softer, but not quite gone. Using another outlet away from the adapter reduced interference a bit further. Yet, it was still audible during silence / soft passages in music.
Then I tried changing the powerline setting from MiMo to SiSo mode... success; silence! Apparently in single connection mode, some problematic frequency bands are not used.
I'll still be looking at alternatives to powerline, but for the time being this solves the issue. The lower theoretical max bandwidth of SiSo mode doesn't really bother me.


First Experience:
ASIO4all is unstable with this DAC, 10ms buffer offset is glitch free > stop application / start application and suddenly it glitches and wont stop until a buffer offset of 20ms is set. Buffer size did not change anything. WASAPI event mode is also glitchy on higher sampling rates like 96kHz. The (legacy) safer push mode works fine.
Then I found out the Loxjie site has XMOS drivers for this D30 including ASIO... not mentioned in the manual anywhere. It adds 16- and 24-bit output format options to the sound control panel, instead of 32-bit only with generic Microsoft drivers. The ASIO drivers also work perfectly.

Had some fun listening to music using Foobar2000. Subjective, unmeasured, biased, non-blind comparison to onboard audio: sounds good to me. High-hats/sibilance seems 'softer' and not so painfully harsh. Also bass feels like it has more oomph. I know I could be telling myself a lot, but I think it's telling I like turning up the volume more than before. That used to become harsh and painful to listen to pretty quickly.

 

[Propheticus]

I wondered how it compared to my onboard audio so I fired up RMAA.
I used the mic in of my Asus motherboard with a Realtek ALC1150 codec Asus dubbed SupremeFX.
Because I have no good way to determine output voltage, I tried to compare apples to apples by picking a fixed mic level. I determined this mic level by going pretty high on the D30's volume, but safely below clipping distortion.
During calibrating the level in RMAA to -1dB, this corresponded to a volume of 40/60 on the D30 and a mic level of 78% with 0dB boost. Going above 45 (with lower mic% to reach -1dB) resulted in distortion/clipping. My normal listening happens around a volume setting of 15 into 30 Ohm / 100dB@1mW Fidelio X2 headphones.
The onboard headphone out had to be set to 78 on the Windows volume slider to match the same -1dB level. Normal listening used to happen at ~16.

12dB more noise from the D30, THD+N% is much better though.
Dynamic range is not reaching 96dB, clearly caused by a higher noise floor.

D30 rolls off low end -1db at 24Hz and -3db at 13Hz, where onboard does this around 10 and 7Hz.
This was surprising to me, because while listening to the D30 I seem to 'feel' more bass. As if there is actual sub-bass that is lacking/missing while listening via onboard. I switched back and forth (level matched by ear, forgive me) and the difference in bass is clearly there.

Looking at individual graphs, onboard shows a L/R channel imbalance of 0.5dB, the D30 none.

THD+N, 1Khz @-3dB. It's clear that 2nd and 3rd harmonics are better on the D30 but overall noise floor is higher.

Split in separate graphs. Left = Onboard, Right = D30. D30 has Less harmonics but overall the plot looks more 'jittery' or noisy.

Stereo crosstalk looks a lot worse.

All in all a mixed bag of results. I'm aware this measuring holds no candle to what Amir does. Still, I hope my method is comparing apples to apples and does show a fair comparison.

 

[ninetylol]

Spoiler: Quote

I wondered how it compared to my onboard audio so I fired up RMAA.
I used the mic in of my Asus motherboard with a Realtek ALC1150 codec Asus dubbed SupremeFX.
Because I have no good way to determine output voltage, I tried to compare apples to apples by picking a fixed mic level. I determined this mic level by going pretty high on the D30's volume, but safely below clipping distortion.
During calibrating the level in RMAA to -1dB, this corresponded to a volume of 40/60 on the D30 and a mic level of 78% with 0dB boost. Going above 45 (with lower mic% to reach -1dB) resulted in distortion/clipping. My normal listening happens around a volume setting of 15 into 30 Ohm / 100dB@1mW Fidelio X2 headphones.
The onboard headphone out had to be set to 78 on the Windows volume slider to match the same -1dB level. Normal listening used to happen at ~16.

View attachment 95557
12dB more noise from the D30, THD+N% is much better though.
Dynamic range is not reaching 96dB, clearly caused by a higher noise floor.

View attachment 95561
D30 rolls off low end -1db at 24Hz and -3db at 13Hz, where onboard does this around 10 and 7Hz.
This was surprising to me, because while listening to the D30 I seem to 'feel' more bass. As if there is actual sub-bass that is lacking/missing while listening via onboard. I switched back and forth (level matched by ear, forgive me) and the difference in bass is clearly there.

Looking at individual graphs, onboard shows a L/R channel imbalance of 0.5dB, the D30 none.

View attachment 95566
THD+N, 1Khz @-3dB. It's clear that 2nd and 3rd harmonics are better on the D30 but overall noise floor is higher.

View attachment 95568
Split in separate graphs. Left = Onboard, Right = D30. D30 has Less harmonics but overall the plot looks more 'jittery' or noisy.

View attachment 95567
Stereo crosstalk looks a lot worse.

All in all a mixed bag of results. I'm aware this measuring holds no candle to what Amir does. Still, I hope my method is comparing apples to apples and does show a fair comparison

This is the headphone out from the Loxjie? Doesnt surprise me. If it would be rca out I would.

 

[Propheticus]

It's indeed the headphone out. I'm using it as an headphone amp and compared it to the headphone out of the motherboard.

 

[escape2]

In RCA line-out mode, does this D30 show volume/attenuation on a -dB scale?

 

[Propheticus]

In RCA line-out mode, does this D30 show volume/attenuation on a -dB scale?

No, in the variable mode it shows volume on a scale of 0 to 60. 60 corresponds to a voltage of (up to) 3.0V, fixed mode is 2.2V and DDSD mode is 2.0V.
How many dB this corresponds with depends on your (power)amps input sensitivity / gain, right?

 

[poppy]

Those who have this unit, how is the remote? Good range, non buggy, easy to use?

 

[escape2]

No, in the variable mode it shows volume on a scale of 0 to 60. 60 corresponds to a voltage of (up to) 3.0V, fixed mode is 2.2V and DDSD mode is 2.0V.
How many dB this corresponds with depends on your (power)amps input sensitivity / gain, right?

Any idea what volume level 2.0V output corresponds to?

 

[Propheticus]

Those who have this unit, how is the remote? Good range, non buggy, easy to use?

Popped in 2x AAA to test for you. Worked fine up to a range of ~6m. (max distance I tested). This is already more than the manual's 5m (16ft)
The EQ (filters) and Mode (Headphone/PRE out) buttons have to pressed a second time before actually changing the setting. The other options work on a single press.
You can navigate the menu using the cursor buttons; The Centre button opens the menu, Right or the Centre button selects a menu item, Left goes a menu level back/exits.
The volume buttons bottom right do nothing, volume is changed using the "level" buttons on the bottom left.

Any idea what volume level 2.0V output corresponds to?

Why not use the fixed DDSD mode if you're after 2.0V output?
I've got no way to measure. Assuming the volume is logarithmic, I'd expect a volume setting of ~53/60 to be 2.0V-ish.

 

[escape2]

Why not use the fixed DDSD mode if you're after 2.0V output?
I've got no way to measure. Assuming the volume is logarithmic, I'd expect a volume setting of ~53/60 to be 2.0V-ish.

Thanks.
I am not after 2.0V output. I would need the pre-amp function of the D30, if I were to get it. I just want to make sure I don't push too much voltage to the amp, in case that's a bad thing.

 

[Propheticus]

Thanks.
I am not after 2.0V output. I would need the pre-amp function of the D30, if I were to get it. I just want to make sure I don't push too much voltage to the amp, in case that's a bad thing.

Turn the power AMP gain up enough so that a volume of 45 on the pre is at "I'll get in trouble with the neighbours" levels and you're safe

On a more serious note: Isn't that what a gain setting is for? Start with a low gain. If that is too low to get to adequately loud listening levels on full pre volume, pick a medium gain. When you know the power amps input sensitivity and available gain settings, (calculating) a more to the point answer is possible.

 

[escape2]

Do these various filters make any audible difference? Which one results in least amount of post-processing?

 

[Propheticus]

Audible...not to me. Just go with one of the sharp ones or low dispersion.

• Sharp looks like a zero-phase filter with evenly distributed pre- and post-ringing,
• Short sharp like a minimum-phase filter with only post-ringing,
• Low dispersion is something in between with more post than pre-ringing.

I use "short sharp" (penultimate in that graphic; short delay sharp roll-off), because to me it looks unnatural to have pre-ringing. An echo before the actual impulse doesn't happen in nature.
Extra argument could be that masking is stronger after a pulse than it is before, which theoretically make the pre-ringing have a higher chance of becoming audible. Following this reasoning either short sharp or low dispersion is the 'best' option.
Yet, I did not really hear differences between the filters while listening to music. Maybe it take synthetic samples specifically created/selected to hear differences...or my hearing is simply meh (33y old).

 

[escape2]

My D30 just arrived from Amazon US. It came with two power cords, and neither one of them has a US plug. Ugh...

 

[Labjr]

My D30 just arrived from Amazon US. It came with two power cords, and neither one of them has a US plug. Ugh...

Such a rare power cord. I hope you can find one.