Sabaj A20d 2023 DAC & HP Amp Review

[renaudrenaud]

Nice review! The comparison with Rossini dCS is really fun!
I just bought this dac.

I guess people who buy dCS products are not concerned by Sabaj. There is no social representation with this brand.

 

[frabor]

I assume the usa version comes with a remote. A trigger would have added less than $2 to hte overall cost yet I would have made a welcome addition

 

[AudioSceptic]

Not sure I like the front handles. They are not pretty and clutter the small face, not very useful, and I’m sure add to the cost some.

I agree. What use are they?

 

[AudioSceptic]

This is a review and detailed measurements of the 2023 edition of Sabaj A20d balanced DAC and headphone amplifier. It was sent to me by the company and costs US $420.
View attachment 307923
As with last generation, the A20 comes in gorgeous packaging that completely sets it aside from its competitors. The display is ultra clear and very responsive and I like the large volume level indicator. Back panel shows what you expect:
View attachment 307924
The labels were a bit hard to read but that is a one-time aggravation.

The 2023 revision switches from ESS DAC to AKM AK4499EX.

Sabaj A20d 2023 DAC Measurements
When I first started to test the unit, I was getting high harmonic distortion, sinking expected SINAD by some 15 dB. It took a bit of looking but then found four "sound color" settings with all but the first one causing such distortion. I turned that off to get proper results:
View attachment 307925

This is excellent performance, easily landing the A20d in our top 20 best DACs ever measured:
View attachment 307926
View attachment 307927

Unbalanced RCA performance is almost as good:
View attachment 307928

Output was actually higher than the above and performance improves a bit if you let it go to the max:
View attachment 307929

You need exceptional noise performance to land in our top 20 list and A20d naturally delivers:
View attachment 307930

Linearity is perfect:
View attachment 307931

Multitone distortion is vanishingly low:
View attachment 307932

In order to make our tests more comparable to what is published elsewhere, I added a test of 50 Hz tone driving into a very low impedance of 600 ohm as used by Stereophile:
View attachment 307933

As a way of comparison, here is how the dCS Rossini Apex ($32,800) performed in the same test:

The A20d beats the worst case harmonic of Rossini by 7 dB yet costs nearly 80 times less!

IMD performance is excellent:
View attachment 307934

Jitter test shows a bit of internally generates spurious tones but otherwise is excellent:
View attachment 307935

We have the usual set of filters:
View attachment 307936View attachment 307937

Here is our wideband noise+distortion vs frequency:
View attachment 307938

Sabaj A20d 2023 Headphone Amplifier Measurements
We have both 1/4 inch and 4.4 mm jacks but both convey the same signal so let's go with the latter:
View attachment 307939
View attachment 307940

These are extremely clean with good bit of power to drive just about any headphone. Despite lack of ultra low (negative) gain, noise performance is excellent at just 50 mv:

View attachment 307941

View attachment 307942

Conclusions
The 2023 revision of A20d brings excellence in every category from look and feel to performance across the board. It leaves nothing to complain about.

I am happy to recommend the Sabaj A20d DAC and headphone amplifier.

Manufacturer Specifications

Name	Sabaj A20d 2023
Input	USB. Optical, Coaxial. Bluetooth
Output	6.35mm/4.4mm headphone, RCA. XLR
Output level	RCA 2.3Vrms. XLR 4.6Vrms
HPA power	2W*2(160) 1W*2(322)
HPA gain	LOW OdB. HIGH +8dB
THD+N	HPA 0.0001% (-120dB) . DAC 0.00006% (-123dB)
Dynamic range	RCA 126dB. XLR 131dB
SNR	RCA 126dB. XLR 131dB
Output impedance	HPA NEAR 0Q, DAC 1000
Bluetooth	BT 5.0 (support apt-X HD, LDAC, apt-X, AAC, SBC)
USB transmission	Asynchronization
USB compatibility	Windows 7, 8, 8.1. 10, 11, Mac OS X 10.6 later, Linux
Bit width	USB 1bit~ 32bit, Optical/Coaxial 16bit~ 24bit
Sampling rate	USB 44.1~ 768kHz DSD64-512, Optical/Coaxial 44.1~ 192kHz DoP64
Power Consumption	<20W
Size	200×60×210 (W×H×D)
Weight	1500

----------
As always, questions, comments, recommendations, etc. are welcome. Click here if you have some audio gear you want me to test.

Any donations are much appreciated using: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/

So, multiple filters but *none* gives full attenuation at 22 kHz.

 

[MZKM]

As a way of comparison, here is how the dCS Rossini Apex ($32,800) performed in the same test:

The A20d beats the worst case harmonic of Rossini by 7 dB yet costs nearly 80 times less

You don’t get it Amir, the Sabaj has its 3rd harmonic higher than its 2nd, so even though it’s past -100dB its obviously garbage.

But yeah, for ~$450 with tax this product is nice combo unit with no audible faults.

 

[gvl]

I agree. What use are they?

Likely their main purpose is to make their equipment look different.

 

[Ken1951]

When you say "it does not compare to dCS" I don't know what you mean. In what way is in not comparable? Can you clarify this statement? Thanks

This is what I don't get - someone doesn't care that it sounds the same and would pay $32K more - for the same sound? To me, that's just cray-cray!

 

[NTK]

So, multiple filters but *none* gives full attenuation at 22 kHz.

The target upper frequency range of audio DAC has always been flat to 20 kHz. With 44.1 kHz sampling rate, Nyquist frequency is 22.05 kHz, therefore the aliasing artifacts above 22.05 kHz are "reflected back" with the mirroring point at this frequency. The aliased component reflected to 20 kHz is thus 22.05 + (22.05 - 20) = 24.1 kHz. This is the reason stop band is specified at 24.1 kHz, because only signals above this frequency are aliased into the pass band.

For example, here is the page from the AK4499 datasheet. (I am only using AKM as an example here. Last I checked, every other audio DAC IC manufacturer does the same.) At fs = 44.1 kHz, passband (PB) is 0.4535 × fs = 19.99935 kHz, and stop band (SB) = 0.546 × fs = 24.0786 kHz.

 

[GiBo61]

Thanks Sabaj for sending another product to Amir for review. Now could you please send the A30a to him, so that all the owners and fans of it on this forum can finally find out whether it is as good as we think it is?

I second that!

 

[usersky]

116mW into 300ohm is almost fine by me. The rest is inaudible so this to me seems lower than a Topping DX3pro. The funny ears do not make for the price diff.

 

[ObjectiveSubjectivist]

@amirm can we ask for adding 50hz measurement as well as 19+20khz IMD graph for every Dac from now on?
It would be nice to compare some of the results with those from stereophile.

 

[polmuaddib]

When you say "it does not compare to dCS" I don't know what you mean. In what way is in not comparable? Can you clarify this statement? Thanks

I was refering to industrial design in a way that dCS looks more monumental, beautiful, luxurious...
In a way that if I made in a week what dCS costs, every week, I would rather buy dCS then Sabaj, nevermind that Sabaj measures slightly better. After 100 SINAD, who cares....

I am not trying to be disrespectful, I do aplaud Sabaj for great engeneering, but as I said, it seems worthless to bother to get 1 dB better SINAD then the device before...

Thought I was clear, sorry.

 

[AudioSceptic]

The target upper frequency range of audio DAC has always been flat to 20 kHz. With 44.1 kHz sampling rate, Nyquist frequency is 22.05 kHz, therefore the aliasing artifacts above 22.05 kHz are "reflected back" with the mirroring point at this frequency. The aliased component reflected to 20 kHz is thus 22.05 + (22.05 - 20) = 24.1 kHz. This is the reason stop band is specified at 24.1 kHz, because only signals above this frequency are aliased into the pass band.

For example, here is the page from the AK4499 datasheet. (I am only using AKM as an example here. Last I checked, every other audio DAC IC manufacturer does the same.) At fs = 44.1 kHz, passband (PB) is 0.4535 × fs = 19.99935 kHz, and stop band (SB) = 0.546 × fs = 24.0786 kHz.

I refer you to Amir's review here <https://www.audiosciencereview.com/...hord-mojo-2-review-portable-dac-hp-amp.34160/>,
"You get the full audio band to nearly 22 kHz and then excellent attenuation. This is what I hope every DAC chip/company implements instead of the lazy ones we see where cut off is at 24 kHz."

Also see any reviews of devices using ESS DACs.

 

[AudioSceptic]

I was refering to industrial design in a way that dCS looks more monumental, beautiful, luxurious...
In a way that if I made in a week what dCS costs, every week, I would rather buy dCS then Sabaj, nevermind that Sabaj measures slightly better. After 100 SINAD, who cares....

I am not trying to be disrespectful, I do aplaud Sabaj for great engeneering, but as I said, it seems worthless to bother to get 1 dB better SINAD then the device before...

Thought I was clear, sorry.

How much worse would the dSC have to be before you would consider it overpriced?

 

[Zensō]

Trivial, I know, but what’s with all the random fonts on the unit. Looks like a school concert flyer.

Agreed. Being a graphic designer by trade, when performance and price are comparable typography becomes a differentiator for me. The design work on this case is pretty awful to be honest.

 

[TK750]

I see that you said that this was "Sabaj A20d balanced DAC and headphone amplifier" but I don't see how one would connect balanced headphones to the amp- are the 1/4" and 1/8" jacks 4-conductor balanced connectors?

It's not an 1/8 (3.5mm) jack but a 4.4mm jack which is almost always TRRRS balanced headphone in consumer audio, sometimes referred to as pentaconn after the brand name. Not to be confused with bantam (also 4.4mm) in pro audio which is usually TRS mono, fortunately the male end looks quite different in terms of shape and obviously fewer poles.

Amir did mention this in the review but perhaps you missed it, he also pointed out that it is provided for convenience rather than providing a 'true' balanced output (1/4 inch and 4.4mm giving the same output)

 

[NHL99]

And how did it sound, a straight wire with gain?

 

[polmuaddib]

How much worse would the dSC have to be before you would consider it overpriced?

I don't know it's exact price, but I know it's overpriced already.

 

[JDS]

THat's not the case. The noise floor in my FFT is lower and hence shows the extra harmonics. Stereophile tests simply hide those.

Even if it were the case, has anyone ever successfully detected such differences in a proper double blind test when the distortion products are >110dB down?

 

[GXAlan]

I was refering to industrial design in a way that dCS looks more monumental, beautiful, luxurious...
In a way that if I made in a week what dCS costs, every week, I would rather buy dCS then Sabaj, nevermind that Sabaj measures slightly better. After 100 SINAD, who cares....

I am not trying to be disrespectful, I do aplaud Sabaj for great engeneering, but as I said, it seems worthless to bother to get 1 dB better SINAD then the device before...

Thought I was clear, sorry.

I own Marantz reference gear even though it just uses a quartet of HypeX NC500OEM. So I definitely believe in paying for ergonomics, aesthetics, and support…

I would challenge you to take a different perspective.

Why upgrade to the latest dCS if you already own a previous generation dCS that hit 100 dB SINAD? It’s past the point of audibility. Its not worthless for ESS/AKM to push the limits of their tech the same way dCS can push the limits of their tech.

The dCS Rossini is an incredible product but it’s like buying a mechanical watch that is operating at 11 Hz. It will not be better than a routine quartz watch but the fact that it is very impressive to see how the architecture works.

"In all dCS DACs, the Ring DAC architecture is discrete and analogue. It consists of a resistor array, a voltage regulation bus, and a buffer/output gain stage that connects to the preamplifier or power amplifier…

…There are 48 current sources within the Ring DAC, all of which produce an equal amount of current. The Field Programmable Gate Array (FPGA)–controlled nature of the Ring DAC allows the sources to be turned on and off in such a way that any component value errors are averaged out over time. Firing the same bit three times on the Ring DAC might give one output slightly high, the next slightly low, the next somewhere in the middle, as opposed to outputting the sample slightly high every time or slightly low every time (as seen in a Ladder DAC, for example).

…The Ring DAC process may be thought of as decorrelating errors. Background noise (an uncorrelated error, one which is not linked to the audio signal itself) is very prevalent in nature, whereas artificial distortion (a correlated error) is not. This results in the Ring DAC having class-leading distortion performance, particularly at lower signal levels. This means that more fine musical detail can be resolved and heard.”

It’s a clever idea since it takes the idea of averaging the way that an ESS DAC does, but it also adds a bit of temporal noise reduction, not unlike the video world. It will be interesting to see if someone would be willing to send in a dCS unit to Amir to test if any of these claims are true.

However, the Sabaj is like high precision quartz. As chip manufacturing has improved from the era of the original 1987 Ring DAC concept where the fastest CPU was a 286, it is now possible for chips from AKM and ESS to do what “$30000” of discrete parts needs.

This is a real accomplishment from the engineers as AKM and ESS.

The rationale for comparing Sabaj and dCS is to continue to push the dCS further in the next round. Why stop at 48 current sources? Why not do 96 or 192 sources? The prospective owner of dCS shouldn’t assume that they are getting the most accurate DAC, they are purchasing an incredible sounding DAC built on anachronistic technologies and done with a beautiful fit/finish.

Meanwhile, these ultra DACs are like buying a Grand Seiko high precision quartz or a Citizen Caliber 0100 or even a 262 kHz Bulova. It’s not going to be any better than a budget quartz in actual practice but it’s super cool to have something like that.

In the watch world, there are normal people. They wear a watch they like and that’s it. That’s the audience that uses their smartphone for telling time. No worries.

Audiophiles are watchphiles. It’s only the insecure who criticize the choices of others. Some people love the crazy mechanisms that are less accurate than a quartz, some people like dive watches even though they don’t dive, or go for ultra accurate quartz products.

There is a lot of room for preference and opinion. What you don’t want is false advertising. If a watch advertises a depth rating, it better do it. If a mechanical watch says it’s more accurate than anything else out there, it’s fair to point out quartz beats it.

What ASR and measures are doing are acting as a Timegrapher. It’s objectively telling you how accurate the product is.

Deciding how much performance you want and if there are other criteria besides accuracy that you are willing to pay for is personal preference — but it’s good to see the shared appreciation for audio hardware which any ordinary person wouldn’t appreciate.