This is a detailed review and measurements of Sabaj Da3 DAC and portable headphone amplifier. For comparison, I selected the Dragonfly USB stick DAC and headphone amplifier. I purchased both units myself. As of this writing, the Da3 retails for USD $120 from Amazon including prime shipping. The Dragonfly black is a bit cheaper at USD $100 with Prime shipping.
From functionality and usability point of view the Sabaj Da3 leaves the dragonfly in the dust. Its OLED display while small, shows sample rate, output connection type and volume level. Volume level is achieved with up/down controls on the unit. The Dragonfly has only an LED which changes colors. Decoding what it means requires the enigma code book of the world war one. The Da3 also has "balanced" and unbalanced headphone outputs. For this review I only tested the latter.
Physically the Sabaj Da3 is larger as you can see from a picture of both connected to my laptop:
The Da3 is clearly larger but it is quite light. It is thicker than a cell phone but half the width. It has some kind of rubber silicone covering its entire back which makes it pretty sticky on the table. The headphone cable has to work hard to pull it off there as a result.
From aesthetic point of view though, the sharp corners are hard on the hand and ugly to boot.
Both are plug-and play in Windows 10 creators edition. I was also able to "gen up" the asio interface on both using ASIO4ALL library for measurements.
Speaking of measurements, I am sure you all are anxious to see them. These are performed using my new Audio Precision Analyzer (APx555). I am still fine tuning its operation so don't directly compare these results to tests done with my previous analyzer.
Measurements
Let's start with our "dashboard" view of each unit. First, Sabaj Da3:
We see a nice, nearly 2 volt RMS output into my high impedance analyzer load. Distortion is about 0.003% with full amplitude digital signal at 1 kHz.
SINAD is a new measure: it shows what amount of the signal is distortion and noise free. We see about 90 dB of usable dynamic range that is free of noise and distortion.
Now the Dragonfly black:
Right away we see quite worse performance. Output drops to 1.2 volts RMS. Distortion increases to 0.02% or more than an order of magnitude higher. And usable dynamic range (SINAD) is 72 dB or so. That is quite a gap in performance in all respects. But let's drill down some more.
Here is jitter and noise using our J-test signal with Da3 level matched to that of Dragonfly:
The ticks at 1 kHz cadence are likely the USB package traffic noise which they both share.
Noise though, is lower in the Da3 (in red) at all but the extreme of our frequency range.
THD+N vs frequency shows Da3's advantage once again:
If you want to know the spectrum of the distortion, here it is:
We see that the Dragonfly has significantly more second and third harmonics than Da3. The third harmonic on Dragonfly rises to just -70 dBr -- pretty high.
Now for everyone's favorite, linearity test:
This test is still under refinement. But for now, the Da3 surprisingly almost matches the loopback performance of my Audio Precision analyzer! While ostensibly it has a "18 bit" resolution based on my 0.1 dB deviation metric, it remains well behaved up to -120 dB meaning it has essentially 20 dB of linearity.
The Dragonfly delivers good performance but deviations while small, start fairly early and hence its "15 bit" score. However, it remains well behaved to better than -110 dB so it has no trouble delivering good performance on 16 bit CD material and then some.
Where the Dragonfly beats the Da3 is in output impedance:
At 1.8 ohms, the Da3 output impedance is still acceptably low but not quite as low as Dragonfly at 0.6 ohm.
Looking at output voltage (and hence power) at load impedances of 33 to 300 ohm we see this for Sabaj Da3 (and single instance for Dragonfly for comparison -- it should say 0.45 volts, not millvolts):
The Sabaj Da3 produces predictable performance for all loads with onset of clipping starting at 1 volt output almost regardless of load. So to compute the power, simply divide 1 by your load impedance. I have done the math for 33 ohm resulting in 30 milliwatts of power. I think the manufacturer spec is 50 milliwatt which you can get if allow more distortion than the minimum.
The Dragonfly on the other hand clips sooner and at higher distortion+noise level.
With balanced output likely producing even higher levels, the Da3 has very competent output drive capability for a USB powered unit.
Finally let's look at channel tracking as the volume is reduced from max to min:
Deviation is essentially 0 dB until the end -- a superb performance. What is more, the volume notches are glitch-free unlike some other units I have tested. The last setting causes a mute and hence the sharp drop at the end. Well done Sabaj!
Not shown is the frequency response which was down just 0.2 dB at 20 kHz (flat to 20 Hz at the low end). So no worries there.
Power Consumption
There is of course no free lunch. The higher output power comes at the cost of more power consumption. The Da3 uses about 150 milliamp while idle and rises to about 180 milliamp when playing 44 kHz content. The Dragonfly on the other hand sips power at just 30 milliamps.
Subjective Listening
I don't have a ton to offer here as I am anxious to get this review out due to high demand.
I did listen to my power hungry Sennheiser HD-650s with the Da3 using unbalanced output and it seemed to have sufficient drive for it toward the end of its volume scale. Sonically I did not detect any faults or signatures.
Conclusion
I must say, I was biased against this product going in. It looks a bit homely and the name is funny sounding. But the Da3 delivers where it counts. It has excellent measured performance easily besting the Audioquest Dragonfly Black. It is almost fault-free in every respect. Put it in a more sexy package and Sabaj could sell this for hundreds of dollars!
Recommended!
As always, questions, comments, corrections, etc. are welcome.
-----
If you like this review, please consider donating funds for these types of hardware purchases using Patreon (https://www.patreon.com/audiosciencereview), or upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
From functionality and usability point of view the Sabaj Da3 leaves the dragonfly in the dust. Its OLED display while small, shows sample rate, output connection type and volume level. Volume level is achieved with up/down controls on the unit. The Dragonfly has only an LED which changes colors. Decoding what it means requires the enigma code book of the world war one. The Da3 also has "balanced" and unbalanced headphone outputs. For this review I only tested the latter.
Physically the Sabaj Da3 is larger as you can see from a picture of both connected to my laptop:
The Da3 is clearly larger but it is quite light. It is thicker than a cell phone but half the width. It has some kind of rubber silicone covering its entire back which makes it pretty sticky on the table. The headphone cable has to work hard to pull it off there as a result.
From aesthetic point of view though, the sharp corners are hard on the hand and ugly to boot.
Both are plug-and play in Windows 10 creators edition. I was also able to "gen up" the asio interface on both using ASIO4ALL library for measurements.
Speaking of measurements, I am sure you all are anxious to see them. These are performed using my new Audio Precision Analyzer (APx555). I am still fine tuning its operation so don't directly compare these results to tests done with my previous analyzer.
Measurements
Let's start with our "dashboard" view of each unit. First, Sabaj Da3:
We see a nice, nearly 2 volt RMS output into my high impedance analyzer load. Distortion is about 0.003% with full amplitude digital signal at 1 kHz.
SINAD is a new measure: it shows what amount of the signal is distortion and noise free. We see about 90 dB of usable dynamic range that is free of noise and distortion.
Now the Dragonfly black:
Right away we see quite worse performance. Output drops to 1.2 volts RMS. Distortion increases to 0.02% or more than an order of magnitude higher. And usable dynamic range (SINAD) is 72 dB or so. That is quite a gap in performance in all respects. But let's drill down some more.
Here is jitter and noise using our J-test signal with Da3 level matched to that of Dragonfly:
The ticks at 1 kHz cadence are likely the USB package traffic noise which they both share.
Noise though, is lower in the Da3 (in red) at all but the extreme of our frequency range.
THD+N vs frequency shows Da3's advantage once again:
If you want to know the spectrum of the distortion, here it is:
We see that the Dragonfly has significantly more second and third harmonics than Da3. The third harmonic on Dragonfly rises to just -70 dBr -- pretty high.
Now for everyone's favorite, linearity test:
This test is still under refinement. But for now, the Da3 surprisingly almost matches the loopback performance of my Audio Precision analyzer! While ostensibly it has a "18 bit" resolution based on my 0.1 dB deviation metric, it remains well behaved up to -120 dB meaning it has essentially 20 dB of linearity.
The Dragonfly delivers good performance but deviations while small, start fairly early and hence its "15 bit" score. However, it remains well behaved to better than -110 dB so it has no trouble delivering good performance on 16 bit CD material and then some.
Where the Dragonfly beats the Da3 is in output impedance:
At 1.8 ohms, the Da3 output impedance is still acceptably low but not quite as low as Dragonfly at 0.6 ohm.
Looking at output voltage (and hence power) at load impedances of 33 to 300 ohm we see this for Sabaj Da3 (and single instance for Dragonfly for comparison -- it should say 0.45 volts, not millvolts):
The Sabaj Da3 produces predictable performance for all loads with onset of clipping starting at 1 volt output almost regardless of load. So to compute the power, simply divide 1 by your load impedance. I have done the math for 33 ohm resulting in 30 milliwatts of power. I think the manufacturer spec is 50 milliwatt which you can get if allow more distortion than the minimum.
The Dragonfly on the other hand clips sooner and at higher distortion+noise level.
With balanced output likely producing even higher levels, the Da3 has very competent output drive capability for a USB powered unit.
Finally let's look at channel tracking as the volume is reduced from max to min:
Deviation is essentially 0 dB until the end -- a superb performance. What is more, the volume notches are glitch-free unlike some other units I have tested. The last setting causes a mute and hence the sharp drop at the end. Well done Sabaj!
Not shown is the frequency response which was down just 0.2 dB at 20 kHz (flat to 20 Hz at the low end). So no worries there.
Power Consumption
There is of course no free lunch. The higher output power comes at the cost of more power consumption. The Da3 uses about 150 milliamp while idle and rises to about 180 milliamp when playing 44 kHz content. The Dragonfly on the other hand sips power at just 30 milliamps.
Subjective Listening
I don't have a ton to offer here as I am anxious to get this review out due to high demand.
I did listen to my power hungry Sennheiser HD-650s with the Da3 using unbalanced output and it seemed to have sufficient drive for it toward the end of its volume scale. Sonically I did not detect any faults or signatures.Conclusion
I must say, I was biased against this product going in. It looks a bit homely and the name is funny sounding.
But the Da3 delivers where it counts. It has excellent measured performance easily besting the Audioquest Dragonfly Black. It is almost fault-free in every respect. Put it in a more sexy package and Sabaj could sell this for hundreds of dollars!Recommended!
As always, questions, comments, corrections, etc. are welcome.
-----
If you like this review, please consider donating funds for these types of hardware purchases using Patreon (https://www.patreon.com/audiosciencereview), or upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
Last edited:
