Measurements and Review of SMSL iDEA and AudioQuest Dragonfly Black DACs

[amirm]

This is a detailed measurement and comparison of SMSL iDEA and AudioQuest Dragonfly Black DAC and headphone amplifiers. Both of these are diminutive devices in the form of "USB thumb drives." The SMSL iDEA is $86 with prime shipping from Amazon. The Dragonfly Black will cost a few more ice cream bars at $99 with prime shipping.

Note that connectivity is different with the SMSL iDEA having a micro USB jack whereas the Dragonfly comes with a normal USB plug.

The SMSL iDEA was kindly loaned to me from a member whereas I purchased the Dragonfly Black last fall (2017) through Amazon.

The SMSL iDEA comes in a slick aluminum package. Due to strength of material they are able to make it much thinner than the Dragonfly Black DAC. It also has volume controls which the Dragonfly lacks. Alas, every time you plug it in, it reverts to a lower volume so be mindful of that if you are doing any comparisons or measurements.

The SMSL iDEA supports many more formats going up to whopping 768 kHz in PCM and DSD 512. The Dragonfly Black lives in the yesterday with support limited to 96 kHz. Indeed the SMSL iDEA beats some desktop DACs with its extensive format support. Here is what Roon reports for SMSL iDEA:

I should say that I tried to play DSD with Roon through SMSL and all I got was noise or clicks. Not sure where the issue is.

Dragonfly Black DAC is limited to 96 kHz as mentioned above:

In operation the SMSL iDEA warms up just a tad but nothing to be concerned about. The Dragonfly on the other hand is completely cool to touch.

I did however run into a serious problem with the Dragonfly DAC. If I connect or disconnect the headphone to it while it was playing, it would mute its output and I could not get sound from it unless I plugged and unplugged it! It didn't do it all the time though. Seems like it has a protection circuit that kicks in but then does not reset.

OK, let's dig into measurements and see what we find. Please read my tutorial on my setup and how these tests are conducted: https://www.audiosciencereview.com/forum/index.php?threads/understanding-audio-measurements.2351/. Naturally all of these tests are performed using USB input. I have expanded my USB tests some but they still are not as complete as the ones that operate over S/PDIF.

Measurements
Let's start easy with frequency response measurements:

Frequency response flatness is fine with the SMSL iDEA dropping just a hair at 20 kHz while the AudioQuest Dragonfly Black peaking just a bit.

What is significant however is the much higher output of SMSL iDEA by over 4 dB. This was clearly audible and significant advantage for SMSL iDEA. Here is the output impedance using a 33 ohm dummy load at 1 kHz (corrected from the first draft of the review):

I will provide subjective impressions as a result of this later in the review.

Jumping to jitter and noise we get:

Unfortunately neither matches the ultra clean output of desktop DACs. Various noise spikes are visible in both. The difference between the main tone at 12 kHz and highest peak was the same for both DACs: 120 dB.

However the SMSL iDEA pulled way ahead in signal to noise ratio with having 4 dB higher output and 6 dB lower noise level.

Let's look at harmonic distortion of a full amplitude 1 kHz tone:

This is a clear win for SMSL iDEA with higher signal to noise ratio and significantly less harmonic distortion.

This gets reflected just the same in THD+Noise versus level:

Past -10 dB both of them have more distortion. Problem is with Dragonfly you would be operating it at its full output given the anemic drive so that problem is even worse there. So another clear win for SMSL iDEA. It distortion and noise here is excellent matching desktop products.

As another confirmation we can look at reproduction of a very low level, -90 dB sine wave (24 bit format, 44.1 kHz):

The SMSL iDEA stomps the Audioquest Dragonfly black by showing a very good sine wave. The output of the Dragonfly Black on the other hand is lower, and marred by lots of noise.

Overall the SMSL iDEA wins the measurement race.

Subjective Listening
I tested both units with three of my headphones: Sennheiser HD650, HiFiMan HE400i, and Grado SR 60e. If levels are not matched, the SMSL iDEA easily blows away the Dragonfly Black. With HD650 especially, the Dragonfly Black simply has insufficient drive. It produces anemic and too low of a level to be usable for my ears. On the other hand, the much more efficient Grado SR60e works well with it. So keeping the advantage of SMSL iDEA output level makes this an easy win for it.

Level matching is challenging as my one means was using Roon volume control. And that unfortunately is not very exact. Getting it close most of the time the two units sounded similar. In a couple of instances I thought the Audioquest had slightly better high frequency but then the darn thing died in the middle of AB switching and I gave up.

Conclusions
The Audioquest Dragonfly Black despite its higher price loses to SMSL iDEA. Given the much more extensive format support, better measurements, and higher output level, my recommendation is easily for SMSL iDEA.

Sadly that is the one I have to return.

Edit: here are the power consumption figures while playing a 0 dBFs 1 kHz tone at 48 kHz sampling:

Dragonfly Black: 0.05 amp no headphone, 0.06 amp Sennheiser HD650 headphone

SMSL iDEA: 0.13 amp no headphone, 0.13 with Sennheiser HD650 headphone

----
If you like this review, please consider donating funds for these types of hardware purchase 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).

 

[Johnny2Bad]

I see that you include your conclusions in the text of your screen grabs for the measurements.

It isn't a big deal, but I find it disconcerting. If someone can't be bothered to read the text, they shouldn't be seeking measurement results, in my opinion. On the other hand it takes away from the Professionalism of the review. Graphs should stand alone, everything you need is there (identifying the DUT is of course fine).

 

[amirm]

Graphs are routinely linked to from other forums and discussions. Without the text in them, that explanation would get completely lost.

Regardless, I can't stand graphs that have no insight in them. The reason you don't see what I do elsewhere is that it takes a lot of work to put the annotations on the graphs as opposed to text under them. It takes me good 2 to 3 hours of work to get it all done, formatted, etc. ready for publication here. I do it because it provides great insight to vast majority of our readers who don't understand the underlying science and engineering in them.

Anyway you are stuck with them as long as I am the chef here.

 

[stalepie]

The output resistance differs from Archimago's measurements;
https://archimago.blogspot.com/2017/05/measurements-smsl-idea-usb-dac.html

 

[PuX]

I sort of expected Dragonfly Black measurements to be pretty bad, Red however should be much better.
It has a better DAC chip, amp and it has 2.1V output.

this is all I could find:
https://www.stereophile.com/content...lack-usb-da-headphone-amplifiers-measurements
and as far as I can tell, all charts look good except the sinewave, which seems to be noisy.

 

[Dimitri]

as long as I am the chef here.

Some graphs mention the website , some don't. If it were my sweat creating all this, I'd have a watermark right across the graph including the year . ( diagonaly from bottom left to top right in case you were wondering

 

[bhigh]

This is a detailed measurement and comparison of SMSL iDEA and AudioQuest Dragonfly Black DAC and headphone amplifiers. Both of these are diminutive devices in the form of "USB thumb drives."

Amir -- Could you add the charts for the Fiio K1 as well? It's a similar form factor at ~ $40 from Amazon.

 

[Dimitri]

If someone can't be bothered to read the text, they shouldn't be seeking measurement results, in my opinion.

In a perfect world, or even an academic environment or academic publication perhaps the above is true, albeit annoying. (I don't like it when in page 32 there is a reference to figure 19-d that happens to have landed on page 33. )

In the real world, posting non annotated graphs is an invitation for someone stealing the work with no credit and / or adding their own "interpretation".

 

[amirm]

The output resistance differs from Archimago's measurements;
https://archimago.blogspot.com/2017/05/measurements-smsl-idea-usb-dac.html

Ok, I dug into this some. I think I see the issue.

I made my measurements at maximum level. Here is what it looks like there with and without load:

As we see it is clipping with 32 ohm load. Archmiago used 20 ohm which would make this even worse.

I then dialed down to where there is no clipping and now the loaded output is very similar to unloaded:

Now the output impedance measures 0.4 ohms.

I was careful in looking at the waveform when I first started but then I stopped doing it. My bad. I will correct the original post. This also explains the difference in some of my previous measurements if the device was clipping with 33 ohm load.

Thanks for bringing this to my attention.

 

[Thomas savage]

Ok, I dug into this some. I think I see the issue.

I made my measurements at maximum level. Here is what it looks like there with and without load:

View attachment 10963

As we see it is clipping with 32 ohm load. Archmiago used 20 ohm which would make this even worse.

I then dialed down to where there is no clipping and now the loaded output is very similar to unloaded:

View attachment 10965

Now the output impedance measures 0.4 ohms.

I was careful in looking at the waveform when I first started but then I stopped doing it. My bad. I will correct the original post. This also explains the difference in some of my previous measurements if the device was clipping with 33 ohm load.

Thanks for bringing this to my attention.

No lobster for you tonight!

I prescribe grits and boiled eggs for a week as punishment for this grave oversight.

 

[amirm]

I know... I know... At least being the King of Audio means there will be no public lashing in the town square!

 

[DonH56]

Unfortunately your fine Steward Sir Thomas has no restrictions against public humiliation so you are not off the hook. Though I was unaware his authority and power extended to dietary restrictions. Quite the horse of a different colour, that.

 

[Thomas savage]

Unfortunately your fine Steward Sir Thomas has no restrictions against public humiliation so you are not off the hook. Though I was unaware his authority and power extended to dietary restrictions. Quite the horse of a different colour, that.

 

[Jimster480]

I see that you include your conclusions in the text of your screen grabs for the measurements.

It isn't a big deal, but I find it disconcerting. If someone can't be bothered to read the text, they shouldn't be seeking measurement results, in my opinion. On the other hand it takes away from the Professionalism of the review. Graphs should stand alone, everything you need is there (identifying the DUT is of course fine).

I think that information being in graphs makes the most sense. It helps people to understand, especially people who aren't as technical or are linked the graph from another location...

Edit: I am a little surprised as I thought that the Dragonfly would be better than it tested...

 

[amirm]

Edit: I am a little surprised as I thought that the Dragonfly would be better than it tested...

Certainly has more fanfare with respect to its design.

Archmiago's measurements showed it to be worse than SMSL iDEA too so I think the data is correct.

 

[mindbomb]

hold up... is the smsl idea actually better than the similarly priced fiio e10k?

 

[amirm]

On the title of this thread, the L is missing from SMSL. Feel free to delete this comment, mods.

Thanks. I fixed it.

 

[amirm]

hold up... is the smsl idea actually better than the similarly priced fiio e10k?

When I get a chance I will repeat the measurements against the E10K and report back.

 

[Jimster480]

When I get a chance I will repeat the measurements against the E10K and report back.

What I really want to see is the E10K.
Are you able to do linearity measurements for "ENOB" graphs?

 

[amirm]

What I really want to see is the E10K.
Are you able to do linearity measurements for "ENOB" graphs?

Not yet. I converted a couple of tests to USB which you saw in this review. The linearity one takes more work to convert. The -90 dB sine wave is a good time domain approximation of it though.