• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
Let me share a few basic measurements of the Sabaj A1 (new 2022 version) stereo integrated amplifier. The amp is based on the popular Infineon MA12070 chip and I bought mine for $68.99 (total, shipped to EU).

A few pictures of the amp:
Sabaj A1 2022 brick.jpg

Sabaj A1 2022 back.jpg

As you can see the power brick (Sabaj branded, 24V / 6,75A, FCC and CE markings) is bulky and the amp is quite tiny! Due to to small case all controls and ports are relatively difficult to operate as everything is cramped into a small space. The tone pots have a center indent (though not very precisely positioned - as you will see). There is unfortunately a relatively low-quality, imprecise feel to all of the pots, but apart from that build quality feels solid to me.

The tools used for measurements:
  • Topping E50 DAC as signal generator (USB input, RCA outputs), 192kHz sample rate and 32bit depth
  • E1DA Cosmos ADC (2,5mm TRRS Aux input), 192kHz sample rate and 32bit depth
  • Asio4All drivers (192kHz sample rate)
  • REW software for measurements
  • Google Sheets for impedance calculations
  • DIY resistive dummy load
  • Revel M16 loudspeaker for complex load test
All of the measurements were done with the volume pot set for 29dB gain and tone controls set for flattest response (see image above for pot positions), and with amp's RCA inputs.

Let's dive in! :D

1kHz SINAD
Sabaj A1 2022 - RCA input, 1kHz 5W into 4 Ohm.png

SINAD seems to be limited by the rising noise and distortion (with frequency). A few other amps based on the same chip exhibit similar behavior (e.g. SMSL A100 measured by @amirm), and it appears to be related to the implementation of the output filter - i.e. if there is no choke/inductor on the outputs the amp will show this kind of behavior. Still - pretty solid result for such a cheap amp, IMHO!
It is implied that in this kind of implementation a loudspeaker inductance will act as part of the output filter and help with the rising distortion and noise. To see if there is any improvement I measured 1kHz response @ ~150mW into a pure resistive load and a real loudspeaker (Revel M16):
Sabaj A1 2022 - 1kHz 150mW - resistive vs complex load.png

As you can see, with a loudspeaker the result is slightly worse (rather than better) so it seems to me proper output filter implementation is important for MA12070-based amps.

[EDIT 2022-05-22] Multitone response
Here's the multitone response (5W into 4Ohm, one channel driven), for this I used the very nice Multitone Analyzer tool by @pkane:
index.php

The result (10,4 bits) seems very similar to that of the SMSL A100 (I guess that's as expected).

Frequency response and load (in)dependence
Sabaj A1 2022 - Frequency response.png

As we can see there is around -0,7dB droop at 20Hz and around -0,2dB at 20kHz which is not too bad.
Please note that the above response is the flattest I could get, and it is actually with the tone controls set very slightly below the center-indentation. At the center-indent position there is 2,8dB bass boost and 0,3dB treble boost (see next section for more information).

We saw above there is very little load dependency with this amp. Based on the measured voltage with 4 Ohm and 8 Ohm resistive loads I calculated and plotted the output impedance:
Sabaj A1 2022 - Output impedance.png

Output impedance appears to be around 0,06 Ohm, which equals to an 8Ohm damping factor of about 125, and it appears to be pretty constant with frequency. All of this means that the amp's frequency response should be pretty stable against any kind of load.
To test this I measured the frequency response when the amp is driving a complex load of a real loudspeaker (Revel M16), and I compared the result to an amp with a very high output impedance (Denon RCD-N9 Ceol):
Sabaj A1 2022 vs Denon Ceol RCD-N9 - Frequency response into a complex load.png

Please note that both amp's responses are normalized to the response when driving a pure resistive 4Ohm load. As you can see the Sabaj A1 2022 frequency response deviates from flat by about 0,1dB when driving this loudspeaker, while Denon deviations are >1dB with the same loudspeaker.

Tone controls
As I mentioned above, the center indentation of the tone controls is not very precise and doesn't give the flattest response. Here's a plot showing the range of the controls:
Sabaj A1 2022 - Tone controls range.png


Power into 4Ohms (20kHz BW, one channel driven)
The 80W power rating by the manufacturer seems overly optimistic, as with just one channel driven I got to only about 64W RMS at 10% THD into my 4,1Ohm resistive dummy load.
Sabaj A1 2022 - Left Ch, THD+N vs level, 4Ohm.png

Sabaj A1 2022 - Right Ch, THD+N vs level, 4Ohm.png


Power into 8Ohms (20kHz BW, one channel driven)
The 40W manufacturer rated power is almost reached (I measure 37W) when driving my 8,2Ohm resistive dummy load, if I allow 10% THD (again, this is only one channel driven).
Sabaj A1 2022 - Left Ch, THD+N vs level, 8Ohm.png
Sabaj A1 2022 - Right Ch, THD+N vs level, 8Ohm.png


There's not a lot of power in this tiny amp, but the results seem comparable to other amps with the same chip.

Power vs distortion into 4Ohms (45kHz BW, one channel driven)
new Sabaj A1 2022 - THD+N vs level (45kHz BW).png


Here we see the typical rise of distortion with frequency, from 5kHz onward.

Conclusion
All-in-all I'd say this is a very nice little amp given the low price. Performance is solid and in line with many other similar amps with the same chip. A few implementations seems to perform slightly better (e.g. Loxije A30 and Topping PA3s - perhaps due to the output filter implementation?), but those seem to to also cost more.

Listening to the amp I thought it sounded good - or should I say transparent - just as most other amps I listened to. :D
The power was in my opinion more than enough to listen to my Revel M16 at 2,3m distance in a 18m2 open living room - though it is a residential building so I can't really blast the music anyway. :)

[EDIT 2022-05-22] Updated power graphs with correct scaling for the x-axis.
[EDIT 2022-05-24] Updated output impedance and 45kHz BW THD vs power plot with equivalent, but nicer-looking versions. Note that additional BT measurements are available in post #5 (only SBC is supported).
 
Last edited:

DSJR

Major Contributor
Joined
Jan 27, 2020
Messages
1,735
Likes
2,180
Location
Suffolk Coastal, UK
Sorry to keep repeating this, but this would have been a pretty well excellent performance by low priced 1980's amplifier standards.. (19 + 20khz IMD was often in the -60's and not put down back then)

Keeping with the vibe in the sentence above, does anyone have an original surviving low cost vintage NAD 3020 or Rotel 820 (did these in tweaked simplified 'BX' form ever come to the US?) or maybe a Yamaha AX 300 or Denon PMA 250 that they could send in for testing, purely to show if and by how much these low cost baby amps differ from the best beer-budget fare some of us started out with back then. I'm frankly stunned these things work at all to be honest :D
 
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
Thanks for the measurements. Promoted to home page.
Thanks a lot! :)

In the meantime I measured also the multitone response (5W into 4Ohm, one channel driven), for this I used the very nice Multitone Analyzer tool by @pkane:
1653208231340.png

The result (10,4 bits) seems very similar to that of the SMSL A100 (I guess that's as expected).
 
Last edited:
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
Regarding BT support - the manual states AptX is supported (page 10), but I can't switch to it with my Android phone (Samsung S8) - it always reverts back to SBC.

So I used the Bluetooth Tweaker app on my PC and ran the codec information tool, this is what I got:
Code:
[CODEC#1 supported by device]
CODEC Type: SBC, Sampling Frequency: 16/32/44.1/48kHz, Channel Mode: Mono/Dual Channel/Stereo/Joint Stereo, Block Length: 4/8/12/16, Subbands: 4/8, Allocation Method: SNR/Loudness, Min/Max Bitpool: 2/53

[CODEC#2 supported by device]
CODEC Type: SBC, Sampling Frequency: 16/32/44.1/48kHz, Channel Mode: Mono/Dual Channel/Stereo/Joint Stereo, Block Length: 4/8/12/16, Subbands: 4/8, Allocation Method: SNR/Loudness, Min/Max Bitpool: 2/53

[CODEC selected by Windows]
CODEC Type: SBC, Sampling Frequency: 44.1kHz, Channel Mode: Joint Stereo, Block Length: 16, Subbands: 8, Allocation Method: Loudness, Min/Max Bitpool: 2/53
That seems to suggest only SBC is supported. @peterwen could you perhaps comment? Thanks!

For completeness, here's the 1kHz tone streamed from my phone via BT SBC to the amp:
1653213081520.png

If I stream from my PC I get slightly different results (main difference is in noise level and type - these might be measurement artifacts due to sample rate, grounding, clocking etc.):
1653213670360.png
 

JohnPM

Active Member
Technical Expert
Joined
Apr 9, 2018
Messages
239
Likes
655
Location
UK
Remember to change the "Ref resistance for dBW" figure in the RTA appearance settings to match your load for correct scaling of the dBW or W graph axes.
 
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
Remember to change the "Ref resistance for dBW" figure in the RTA appearance settings to match your load for correct scaling of the dBW or W graph axes.
Hello John - thanks, however I did do this. In the spectrum window I get correct RMS values, but in distortion view REW seems to present peak values instead of RMS (just for the x-axis) for both voltage and power. I can share more details privately - was thinking about reporting it anyway.
 

JohnPM

Active Member
Technical Expert
Joined
Apr 9, 2018
Messages
239
Likes
655
Location
UK
Hello John - thanks, however I did do this. In the spectrum window I get correct RMS values, but in distortion view REW seems to present peak values instead of RMS (just for the x-axis) for both voltage and power. I can share more details privately - was thinking about reporting it anyway.
If the "Full scale sine rms is 0 dBFS" View preference was selected when making the measurements but deselected when subsequently viewing them that would shift voltages by sqrt(2) and powers by 2.
 

mdsimon2

Major Contributor
Forum Donor
Joined
Oct 20, 2020
Messages
1,483
Likes
1,885
Location
Detroit, MI
@dominikz, how did you get those wide band power sweeps all on the same plot? Whenever I run stepped sine power sweeps the only place I can view the results is in the distortion tab which does not allow me to view multiple measurements at once. For other measurements I can use overlays to do this but the power sweeps are always grayed out.

Michael
 
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
If the "Full scale sine rms is 0 dBFS" View preference was selected when making the measurements but deselected when subsequently viewing them that would shift voltages by sqrt(2) and powers by 2.
Thanks for the tip! However I wasn't changing the setting between measurement/viewing so I might have an issue - I sent you a PM with more details.

EDIT: The power plots in post #1 are now updated with the correct scaling on the horizontal axis, to make them easier to read.

@dominikz, how did you get those wide band power sweeps all on the same plot? Whenever I run stepped sine power sweeps the only place I can view the results is in the distortion tab which does not allow me to view multiple measurements at once. For other measurements I can use overlays to do this but the power sweeps are always grayed out.

Michael
Unfortunately there is no easy way that I know of - I was manually overlaying the individual sweeps in MS Paint :D (I used the selection tool with 'transparent' option checked). I just changed the THD+N plot colors before exporting the sweeps.
 
Last edited:
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
That looks interesting, but I'm a bit skeptical of the 15V-19V DC input rating. :confused:
Looking at the MA12070 datasheet that would likely result in quite lower maximum output power: according to chip spec the maximum +26V supply voltage gives about 80W (into 4Ohm @10% THD), +24V about 70W and +19V should give only around 45W.
 

fgfgfg

Member
Joined
Apr 3, 2019
Messages
24
Likes
25
Location
Russia, RND
+24V about 70W and +19V should give only around 45W.
Even at 24V these give only [email protected], I don't think that few watts dropped with 19V would change anything
Also, if we can trust cap rating at photos, this will handle up to maximum rating of 26 (if input mos can), no PSU included anyway, but it's dirt cheap)
 
Last edited:

sseppala

Member
Forum Donor
Joined
Jan 19, 2021
Messages
22
Likes
14
Thank you for the review and especially for the output impedance measurement. I was just recently looking for output impedance specs for MA12070 based amps, but could not find any. I wish Amir would also include them in his reviews. The importance of damping factor has been largely downplayed at this forum, and with passive speakers I agree, but IMO it matters more if you are interested in DIY active speakers, like I am.
 
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
I wish Amir would also include them in his reviews. The importance of damping factor has been largely downplayed at this forum, and with passive speakers I agree, but IMO it matters more if you are interested in DIY active speakers, like I am.
I'd also like to see more measurements of output impedance - in acute cases it can significantly impact frequency response, and as we know humans have relatively high sensitivity to frequency response deviations.

I guess it is not considered a big issue these days because most amps have reasonably low output impedance, and also FR deviations can be easily corrected by EQ (if you have access to EQ and the means to measure amplifier output, that is!).
 

sseppala

Member
Forum Donor
Joined
Jan 19, 2021
Messages
22
Likes
14
I just remembered that a while back I had seen another output impedance measurement for the MA12070 and it was the Topping PA3s. It was measured by Wolf of L7Audiolab and the output impedance was about 0,011Ohm https://www.l7audiolab.com/f/topping-pa3s-prototype/

Considerably better than 0,06Ohm, but it has dual MA12070 setup, hence twice the power and price. It also has balanced inputs but no Bluetooth or tone controls.
 
Last edited:
OP
dominikz

dominikz

Addicted to Fun and Learning
Forum Donor
Joined
Oct 10, 2020
Messages
501
Likes
1,742
As I noted in the first post, there's a lot of similarity in these measurements with SMSL A100 (another amp based on the same MA12070 chip) - so I thought it might be interesting to compare the 45kHz THD+N vs level measurement of the two amps. Therefore I digitized the plot from Amir's SMSL A100 review and added a couple other amps for reference:
1653668300888.png

I find it fascinating how well the SMSL A100 and Sabaj A1 2022 plots overlap :D - even though these are different units and measured on different instruments by different people!

Topping PA3s is based on the same MA12070 chip, but has two of them, a higher voltage power supply and probably an output filter - we can see significantly better performance and higher power output compared to both the Sabaj and SMSL units.

ICEPower 50ASX2SE (from Amir's B&K AV30.2 review) exhibits the least distortion and least variability with frequency, but with similar power to SMSL A100 and Sabaj A1 2022 units.
 
Top Bottom