• 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). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Teardown of Musical Fidelity V-DAC

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
990
Likes
1,367
Location
Europe
#1
I borrowed the Musical Fidelity V-DAC over the weekend to make some measurements. It has been bought by its owner some 8+ years ago for about € 200. It's no longer available and has been replaced by the V-DAC II which is also no longer available.

It comes with a 12V/500mA plug power supply which is presumably not switched (too heavy). This is how it looks:

Musical Fidelity V-DAC Top.jpg


Inputs are on the right side:

Musical Fidelity V-DAC Right.jpg

You can switch between USB and SPDIF Coax/Toslink, but you can use only one SPDIF input at any time.

Outputs are on the left side:

Musical Fidelity V-DAC Left.jpg


Teardown

I got kind permission to open the V-DAC. This is easy as you just remove 3 screws per side and the PCB slids out:

Right side open:

Musical Fidelity V-DAC Right open.jpg


Left side open:

Musical Fidelity V-DAC Left open.jpg


PCB top view:

Musical Fidelity V-DAC PCB Top.jpg


PCB bottom view:

Musical Fidelity V-DAC PCB Bottom.jpg


What a relief to see proper manufacturing at work.

Measurements

These are the specifications (excerpt):
  • Maximum output: 2.2V
  • THD: 0.004% (20 Hz - 20 kHz) (calculates to -88 dB)
  • Frequency response +0/-0.1 dB 20 Hz - 20 kHz
  • Signal / noise ratio: 116 dB A-weighted
  • Crosstalk: -104 dB (20 Hz - 20 kHz)
USB was limited to 44.1/48 kHz samplerate for whatever reason. SPDIF Toslink works until 96/24, SPDIF Coax works also for 192/24.

Dashboard
Let's start with my Dashboard (well, not such a nice one like @amirm's), measured with a Siglent SDS 1202X-E (DSO), an RME ADI-2 PRO fs and REW, using USB input (connected to a Linux-PC running current Kubuntu):

Musical Fidelity V-DAC THD 1kHz 0dBFS USB 44-24 RCA R+L FFT=32k avg=4.png

Musical Fidelity V-DAC Level 1kHz 0dBFS USB 44-24 RCA R+L.png


Left channel: THD: -98.8 dB, Noise: -97.2 dB, SINAD: 94.9 dB, Output: 1,70 Vrms
Right channel: THD: -87.8 dB, Noise: -96.1 dB, SINAD: 87.2 dB, Output: 1.67 Vrms

THD: The left channel is some 10 dB better than the right channel regarding THD, but even the worse channel is (just) in the specs. It stays on this level over the full audible frequency range and is the same for both SPDIF inputs as well (not shown here, see attachments below).
Noise: PS noise is well within the inaudible region.
Output: Voltage is 1.4 dB below the standard of 2V which may give the V-DAC a disadvantage in a test without level matching. It's also quite short of the specified 2.2V.

IMD
I measured IMD using SPDIF Toslink with the CCIF signal (19 + 20 kHz):

Musical Fidelity V-DAC IMD CCIF 0dBFS Toslink 44-24 RCA-R+L FFT=32k avg=8.png


Left channel: IMD: -97.6 dB
Right channel: IMD: -87.9 dB

Again the left channel (blue) is about 10 dB better than the right channel (red).

Multitone

The multitone signal with 32 different sinus signals of equal amplitude emulates "music". SInce it requires 192 kHz samplerate I played it with the vlc-player via SPDIF coax:

Musical Fidelity V-DAC Multitone32 Coax 192-24 RCA-R+L FFT=256k avg=8.png


Again the left channel (blue) is about 10 dB better than the right channel (red).

Jitter

Musical Fidelity V-DAC Jitter Toslink 44-24 RCA-R FFT=256k avg=8.png


This is definitely inaudible. Via USB jitter was much worse but I don't trust the VLC player not to mess up the J-Test signal, as JA's measurements at Stereophile were identical to SPDIF, so I'm not gonna show it.

Reconstruction filter

I measured the frequency response of the reconstruction filters with white noise at -10dB and heavy averaging. Since the RME cannot use different sample rates for input and output I used an EDIROL UA25 to feed the V-DAC via its Toslink input.

Musical Fidelity V-DAC White noise -10 dBFS USB 44-16 and 48-16 RCA-R.png

For both 44.1 and 48 kHz samplerate the cutoff at the Nyquist frequency is not sufficient. However a 20 kHz sinus at 0 dB is totally suppressed in both cases (images being at 24.1 and 28 kHz, respectively) so we are probably good here.

Low Level Resolution

I fed the V-DAC a dithered 1kHz sinus signal at -90 dBFS in both 16 and 24 bit sampledepth to find out how much better 24 bit is over 16 bit:

Musical Fidelity V-DAC 1kHz -90dBFS Coax 44-16 vs 44-24 RCA-R+L FFT=32k avg=32.png


The noise floor for 24 bit drops some 13 dB which results in a resolution of 18 real bits. Here both channels are identical.

Intersample Overs

Intersample overs are something which should not occur in theory, but does occur in reality due to loudness war and/or bad mixing/mastering. What it means is that the sample values are not higher than 0 dBFS (which they can't) but the reconstructed signal can be up to 3 dB higher. Here are examples of a sinus signal with and without intersample overs:

intersample-overs.png


The frequency of the sinus is 1/4th of the sample rate, so it is sampled with 4 points. Now let's assume that the signal is sampled at the crossings of the horizontal lines, which is at 45, 135, 225 and 315 degree. The pink sinus is at full amplitude (+/-1) although the samples are at +/- 0.71 (green and light blue).

Intersample overs occur when those samples are higher than +/-0.71. If they are at the maximum possible values of +/-1 (yellow and blue) than the corresponding sinus (orange) has an amplitude of +/-1.42 which is 3 dB above 0dBFS.

There are not many DACs which can reconstruct such an illegal signal. Neither can the V-DAC. Shown are intersample overs at +3 and +1 dB:

Intersample_Overs+3dB.png



Intersample_Overs+1dB.png


This is what it should look like:

Intersample_Overs+0dB.png


Crosstalk

Musical Fidelity V-DAC Crossfeed 20Hz-20kHz Coax 48-24.png


I'm not sure whether the result is correct. Left to right is definitely out of spec but not audible at all. Remember that even the best cartridges for vinyl replay have no more than around 35 dB crosstalk suppression.

Conclusion

Oldie but goodie. The results are inaudible for most use cases, and if you own the V-DAC and need no better USB input or other fancy features you are well advised to keep it.

Edits
  1. Added missing attachments: THD+N from 20Hz to 20 kHz
 

Attachments

Last edited:

graz_lag

Major Contributor
Joined
Oct 13, 2018
Messages
1,234
Likes
1,269
Location
Le Mans, France
#4
Excellent work @LTig !

I have been owning the V-DAC II for abt. 6 months back in 2014, bought and sold in the used market for the same amount of money, 150 EUR.
Upgraded (I thought it was an upgrade ...) towards the Teac UD-501.
My unit was stamped as Made in Taiwan, information that surprised me as I have always considered Musical Fidelity as an UK-based manufacturer.

Glad to see abt. the good performances you got out from the tested unit.
 
Joined
Jul 26, 2019
Messages
45
Likes
108
#5
My unit was stamped as Made in Taiwan, information that surprised me as I have always considered Musical Fidelity as an UK-based manufacturer.
Musical Fidelity makes the overwhelming majority of their products in their Taiwan factory, and I think that's an excellent choice; Taiwan really does excel at quality manufacturing and I've owned a number of Musical Fidelity products over a decade ago and their craftsmanship has always been nothing short of superb (other than their cheap remote).

I still remember when "Made in Taiwan" was synonymous with single-use throwaway junk, but those days are mostly behind. They're also in the cutting-edge of semiconductor manufacturing.

Also thank you @LTig for this review! The V-DAC II always seemed like a cute little device.
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
7,261
Likes
7,993
#6
I somehow overlooked this review. Excellent work. Very nicely done @LTig

I've a Musical Fidelity V-link for converting USB to SPDIF which looks like it uses the same exact enclosure. A friend had one of the DACs like you tested here, and it gave him fine performance for a few years.
 

Thomas savage

Power hungry desperado
Moderator
The Watchman
Patreon Donor
Joined
Feb 24, 2016
Messages
7,547
Likes
6,395
Location
uk, taunton
#7
Don't worry @amirm this ' review' dose not come close to the value your novelty panthers provide ..

Your still the champ , come on now don't go to bed feeling upset by this !


PS this guy failed to invent the iPod too so hes no better than you ... Sure he lacked the resources of one of the world's largest companies and all that expertise at his fingertips like you had but still .
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,650
Likes
35,029
Location
Seattle Area
#8
PS this guy failed to invent the iPod too so hes no better than you
I didn't invent the iPod but did invent the other three, more useful ones:

 

Thomas savage

Power hungry desperado
Moderator
The Watchman
Patreon Donor
Joined
Feb 24, 2016
Messages
7,547
Likes
6,395
Location
uk, taunton
#10
I didn't invent the iPod but did invent the other three, more useful ones:

Damn so close , now I understand why you have that picture of Steve Jobs lacquered on the bottom of your toilet.

You can start the day with a smile at least .
 

Veri

Major Contributor
Patreon Donor
Joined
Feb 6, 2018
Messages
2,973
Likes
2,911
#11
I owned a V-DAC II long ago :) It was a really nice device. I sold it to some dude casually wearing HD800s outdoors. memories :p
 

Rja4000

Addicted to Fun and Learning
Joined
May 31, 2019
Messages
534
Likes
369
Location
Liège, Belgium
#12
Thanks
That's good to see others' reviews and this one is very good.

I like the idea a DAC could rebuild overloaded signal (well, after DAC itself, I guess), but that means that a true 0dBFS peak-to-peak signal would then be reproduced with lower output level, I guess.

About crosstalk, I think the RME is quite high on that. Did you try it in loopback?

I know it makes displays more busy and harder to read, but isn't that better if we always include the loopback as reference in the plots and measurements ?
What are the ranges you were using for the RME input? I guess +13dBu @0dBFS?

Those are no critics, for sure!
But I'm interested if we could standardise somehow those alternative reviews, to be able to compare them. Not everyone is owning an AP, but if we put enough details in those alternative reviews, they could be quite useful and help increase the amount of reviewed devices.
But to achieve that, method and documentation of test have to be somehow standardized.

Another example: I'm still struggling with multi tone test signal. Could you publish the exact tone list, with frequencies, level and phase?
I guess frequencies are set to avoid spectral leakage at 192kHz, 256k FFT window size.

About tool: I use Virtins MI pro, because REW doesn't allow automation. But I now always compare with REW, for control, so I guess the tool itself doesn't make a difference.
 
Last edited:

pozz

Contributor
Patreon Donor
Editor
Joined
May 21, 2019
Messages
802
Likes
1,063
#14
Excellent and comprehensive.

I'm with @Veri on standardizing the presentation. I think the best way would be to have a separate thread showing loopbacks and describing other details, which later reviews can follow and link to.
 

digicidal

Addicted to Fun and Learning
Joined
Jul 6, 2019
Messages
867
Likes
1,120
Location
Sin City, NV
#15
This needs more jumper wires strewn about, some messy soldering, and perhaps a battery. Sorry, considering the age, cost and relative performance... just figured this should have twice as much as a competing DAC (since that one was more than twice the cost). :p

Great job on this @LTig well done!
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
7,261
Likes
7,993
#16
Excellent and comprehensive.

I'm with @Veri on standardizing the presentation. I think the best way would be to have a separate thread showing loopbacks and describing other details, which later reviews can follow and link to.
This was discussed a bit in the past on ASR. I don't mind the slightly different presentations especially when well done like this one. But I'd be in favor of standardizing. I suppose the main problem is which software to use.
 

pozz

Contributor
Patreon Donor
Editor
Joined
May 21, 2019
Messages
802
Likes
1,063
#17
This was discussed a bit in the past on ASR. I don't mind the slightly different presentations especially when well done like this one. But I'd be in favor of standardizing. I suppose the main problem is which software to use.
I don't mind the differences either. Less a template than a way to make measurements more consistent, like an outline of conditions for test signals and plot settings. I think a range of software is ok as long as the background is known. I liked the sweep spectrum plots you showed in Audacity, for example.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
22,650
Likes
35,029
Location
Seattle Area
#18
There are a few things that can be done to make measurements more useful to broader set of readers we have:

1. Use graph resolutions less than 1000 pixels horizontally. Anything beyond that will cause the forum software to resize it and cause the image to be blurry. And at any rate the fonts become too small to read.

2. Always save the graphs as PNG as this is lossless versus jpg or cutting and pasting into forum. If you can, use 8-bit format as that makes the file size much smaller, and cost me less in the long run in hosting them. :)

3. When you can, try to draw the same conclusions I do in translating SNR into bits, THD+N into SINAD, etc.

4. Put in a lay person summary as I do at the end.

5. If you can label the graphs inside them as I do, it would be great. It is a great teaching aid but of course, increases authoring time significantly.

Steps like #3 will allow @pozz to include them in our master database of graphs potentially.
 

LTig

Addicted to Fun and Learning
Joined
Feb 27, 2019
Messages
990
Likes
1,367
Location
Europe
#19
I like the idea a DAC could rebuild overloaded signal (well, after DAC itself, I guess), but that means that a true 0dBFS peak-to-peak signal would then be reproduced with lower output level, I guess.
I've heard that the Benchmark DACs were the first to handle those overshoots. I tested the RME and it handles +3dB intersample if the volume is not higher than -0.5 dB.
About crosstalk, I think the RME is quite high on that. Did you try it in loopback?
No.
I know it makes displays more busy and harder to read, but isn't that better if we always include the loopback as reference in the plots and measurements ?
Yep - either as attachment or together with the DUT in the same plot. Here is an example I've just measured:

Behringer Ultra-Di DI20 Multitone 32 CH 1+2 + RME XLR-Loopback In=0dBu att=0 Out=0dBu FFT=256k...png

What are the ranges you were using for the RME input? I guess +13dBu @0dBFS?
Ref Level +13dBu, Trim Gain between 4 and 6 dB, to come as close to 0dBFS as possible.
Those are no critics, for sure!
But I'm interested if we could standardise somehow those alternative reviews, to be able to compare them. Not everyone is owning an AP, but if we put enough details in those alternative reviews, they could be quite useful and help increase the amount of reviewed devices.
But to achieve that, method and documentation of test have to be somehow standardized.
I second that. We should standardize each type of test, to be as close to @amirm's AP as possible, so that all measurements are comparable. This should be no problem. With the loopback result one can see when the DUT reaches the limits of the soundcard.

This is what I have noted so far, studying @amirm's measurements ('?' if unsure or unknown):
  • Dashboard: THD 44.1 kHz 24 Bit FFT=32k averages=3 (I use 4 averages as REW does not allow arbitrary numbers)
  • Jitter: JTest 48 kHz 24 Bit FFT=256k? averages=8
    I use a self created wav-file.
  • Multitone 32: 192 kHz 24 Bit: FFT=256K averages same as jitter? 32 might be fine to separate noise from distortion
    I use a self created wav-file with the same settings as AP.
  • IMD vs level: 48 kHz? FFT=128k?
  • Filter Frequency response with white noise: FFT=32k, averages=32 (or more if it helps)
  • Frequency response with sweep in REW: FFT=4M, averages=1
  • dynamic range using AES-17 protocol as created by Audio Precision ???
  • Intersample Over: I use self created wav-files with +1, +2 and +3 dB overshoot.

Another example: I'm still struggling with multi tone test signal. Could you publish the exact tone list, with frequencies, level and phase?
I guess frequencies are set to avoid spectral leakage at 192kHz, 256k FFT window size.
Multitone 32:

multitone-32.png


Multitone 7:

multitone-7.png


About tool: I use Virtins MI pro, because REW doesn't allow automation. But I now always compare with REW, for control, so I guess the tool itself doesn't make a difference.
For me REW has the advantage that I can use it on my Linux PC. I haven't found yet a way to measure THD or IMD vs level though. OTOH it's a suite for room EQ and not for measuring electronics. And its free - great job!
 
Top Bottom