HifiBerry DAC2 HD Review (Rpi HAT)

[AnalogSteph]

@amirm Was the dynamic range measurement done relative to -10 dBFS or what? Otherwise, 105 vs. 118 dB(A) would be one heck of a lot of difference for the same DAC. The former would be completely laughable for a PCM1796, the latter about in line with what an Asus Xonar D2 gets out of it.

I suspect that the analog supply may be relatively low, making the NE5532 opamp somewhat unhappy near 0 dBFS. (Remember, the part has some issues with common-mode distortion, and we're seeing a lot of H2 here.) Maybe the switch-mode circuitry seen is an inverter to generate -5 V from +5 V... it would be sufficient for 2 Vrms output even with traditional non-RRO parts (in fact, many Sony CD players used to employ exactly these supply voltages), and the DAC will hard-clip at 0 dBFS anyway.

 

[itsikhefez]

The problem with USB is that until the RPi4, all previous version of the RPi does not have a full USB implementation in hardware. USB audio needs isochronous transfers every frame (125 microseconds), and it has to be scheduled in software, and that's really hard to do. So you end up having a lot of popping/clicking noises whenever a frame is missed.

The HAT on the other hand uses GPIO pins to pump out I2S signal driven from a hardware clock, so you won't have timing issues. So HAT is only way to reliably get audio out of RPi3 or earlier.

I've been using this HAT for a while with a Raspberry Pi2, through a DIY preamp and a Benchmark AHB2.
Recently added a DAC3L on top of the AHB2 and have been using USB from the Pi instead of the Hifiberry.
I have no pops/clicks/noises or anything whatsoever ...
Software I am using is piCorePlayer client with Logitech Media Server on a dedicated PC.

P.S. I wanted it to look fancy so put it in a nice case with panels and a VFD screen
P.S.2 The picture is outdated, it was from when I was using a Digi+ and not DAC2 HD.

 

[m8o]

Hmm, just bought a DAC2 (non HD) for a build for a friend; the prior DAC+ Pro measure fairly well. Was considering getting this HD too for myself but didnt, unrelated to seeing this as I hadn't seen it yet. Guess I'm returning to Allo Boss v1.2 after this.

Why oh why won't Allo or ApplePi produce another run of the Katana or ApplePi HATs, respectively?!

 

[AnalogSteph]

Except for jitter (which should be audibly unconcerning either way), the DAC+ Pro is at best equal and usually markedly worse than the DAC2 HD (dynamic range in particular appears to be nowhere even close). Allo Boss can scrounge together about 102 dB of dynamic range by the looks of it, which while about 6 dB better still falls short of the DAC2 HD (which is at least 4 dB better again). It is to be hoped that the DAC2 Pro gets a bit closer to PCM5122 datasheet specs than the DAC+ Pro did.

Interestingly, Archimago's results are wildly different for both DAC2 HD and DAC+ Pro... perhaps an issue of power supply used for the Pi or something? There has to be a reason why the results obtained by the boss are so far off the mark.

BTW, going by the DAC+ Pro IMD+N(f) sweep, I'd be looking for some crummy ceramic capacitors performing lowpass filtering in its (very short) analog signal path... that's a lot of 3rd harmonic up there.

 

[iucoen]

I
Recently added a DAC3L on top of the AHB2 and have been using USB from the Pi instead of the Hifiberry.
I have no pops/clicks/noises or anything whatsoever ...

See: https://github.com/raspberrypi/linux/issues/2215

You might be luck that your DAC does not require isochronous transfer every frame. All of my DACs use XMOS and they all do.

 

[abdo123]

Interestingly, Archimago's results are wildly different

Where? Seems pretty identical to me

 

[pugs]

^ That set of graphs is only for the DAC2 HD, not the +pro. Just different dBFS settings

 

[somebodyelse]

See: https://github.com/raspberrypi/linux/issues/2215

You might be luck that your DAC does not require isochronous transfer every frame. All of my DACs use XMOS and they all do.

Whether you get the pops/clicks is heavily software dependent. I've never managed to reproduce it with piCorePlayer. I could reliably reproduce it with the same hardware using Raspbian lite running BruteFIR. Pin BruteFIR to a single cpu core and it goes away again. This is with an XMOS-based Focusrite Forte.

 

[audio1234]

To avoid any misinterpretations or doubts, it would be really nice if amir could redo the measurements avoiding roon and based on a local playback via the filesystem.

 

[SAC_2]

Is there any point to these vs just connecting to a regular DAC via USB? I am streaming from a Pi to a dirt cheap Amazon DAC temporarily and it sounds pretty damn good to me. I imagine the Pi feeding my E30 will sound amazing.

Setup is key as well. My setup is in my kitchen so odds are extremely high that the weak SINAD wouldn't even matter in the context of refrigerator/oven fan noise. Either way this is money better spent elsewhere it seems.

 

[tw99]

Is there any point to these vs just connecting to a regular DAC via USB? I am streaming from a Pi to a dirt cheap Amazon DAC temporarily and it sounds pretty damn good to me. I imagine the Pi feeding my E30 will sound amazing.

Unless you end up with a ground loop, no. Pi4 plus a decent USB dac is capable of being far better than anyone's hearing.

 

[pugs]

To avoid any misinterpretations or doubts, it would be really nice if amir could redo the measurements avoiding roon and based on a local playback via the filesystem.

Agreed, get roon out of the equation and take away any factors that a may skew results. Saying this as archimago results are more inline with the spec of the dac, something isn't adding up

 

[laudio]

The ZX-81 was one of my very first computers. And the wretched 16k RAM expansion pack that if you even breathed, it would make bad contact and the computer would crash.

Of course it overheated in Australia with a tiny flag heatsink on the 7805 in a mostly sealed plastic case. And that membrane keyboard- what a joy (not).

Same here. I remember soldering the kit together back in the day because I couldn't afford an actual assembled unit. $99. Messed it up putting it together and then corrected it.

Think I still have it plus the flaky 16K expansion kit in the basement somewhere. Best thing about the Sinclair was it enabled learning assembly language (Z80) (at least for me), on the cheap. Worth far more than the $99 in the long haul.

 

[abdo123]

^ That set of graphs is only for the DAC2 HD, not the +pro. Just different dBFS settings

???

This is the DAC 2 HD thread why would i discuss the + pro?

 

[pugs]

???

This is the DAC 2 HD thread why would i discuss the + pro?

I think cause he was talking about two different dacs, just the way it reads and your quote appeared to reference that both were the same. Sorry if misread that, no harm done.

 

[amirm]

@amirm Was the dynamic range measurement done relative to -10 dBFS or what?

Pretty sure it was 0 dBFS. You all need to ask me these questions when the review is fresh in my mind.

 

[amirm]

To avoid any misinterpretations or doubts, it would be really nice if amir could redo the measurements avoiding roon and based on a local playback via the filesystem.

Why do you keep saying this? His results are the same as mine as you just post:

See THD+N of -92/94 dB? This is what Archimago got:

He shows -92 dB which is what I got on the worse channel. My better channel shows -94 dB per above. Second harmonic is identical to his as well.

 

[User24376t34]

I wouldnt trust Roon to deliver any decent sound quality to a hifiberry board. When I tried Roon the sound quality was quite bad to my DAC+ XLR compared to playing lossless music directly, ie HifiberryOS.
Although Roon telling me a full lossless signal chain was present.
Roon couldn´t or wouldn´t help.
I suggest to make the test again without Roon

 

[m8o]

A bit OT, but maybe a PSA for others who have not looked at the 'About' page of hifiberry.... am I the only ignoramus who didnt realize when buying from their [hifiberry's] website it ships from Switzerland? I went that route over Amazon Prime for better prices (a key factor in all RPi projects) and not everything I wanted was available there. The time to deliver however is affecting my plans and time to deliver the project I'm building for a friend. So, just an fyi here for any others in the USA unaware of this.

 

[voodooless]

I wouldnt trust Roon to deliver any decent sound quality to a hifiberry board. When I tried Roon the sound quality was quite bad to my DAC+ XLR compared to playing lossless music directly, ie HifiberryOS.
Although Roon telling me a full lossless signal chain was present.
Roon couldn´t or wouldn´t help.
I suggest to make the test again without Roon

I think you have the burden of proof the wrong way around. @amirm get's the same results as another independent source that did not use Roon. So I'd say that it's up to you to show measurements that proof that Roon has problems since you're making the claim.