Friday, 9 January 2015

MEASUREMENTS: Tascam UH-7000 USB Interface (Part II: As an ADC)

TEAC & Tascam combo with size differential. The E-MU 0404USB in the background to the left.
Alright, it's time to spend some time looking at using the Tascam UH-7000 as an analogue-to-digital converter. Remember in the first instalment, I showed that this device is already a very good DAC. This time around, I wanted to see whether this device did the job well as an ADC; my main desire being to use it for vinyl needle drops and to see if it can be used as a good measuring device in comparison to the Creative E-MU 0404USB I've been using for the last couple years (which I bought mainly as a DAC back in 2009).

In terms of the underlying hardware, the Tascam UH-7000 uses the Burr-Brown PCM4220 ADC compared to the E-MU's AKM AK5385A chip. On paper at least, the Burr-Brown should be significantly better with a rated SNR of 123dB compared to the AKM's 114dB. Of course, much of the final result depends on the circuitry built around the ADC such as quality of the pre-amplifiers feeding the input signal.

I. The Spectrum of Silence

To start, let's have a look at what "silence" looks like through the Tascam UH-7000, with preamps set to minimum:

Impressive. With the ADC running at 24/96, we're seeing very low noise floor essentially flat down to -160dB across the spectrum (Note: I found an issue with this later on - see below).

Compared to the E-MU (sorry about the difference in scale!):

Clearly the old Creative unit is noisier with sporadic noise spikes reaching up to -140dB and a slight tendency for the noise floor to rise from about 35kHz and up.

So far, so good... The Tascam is doing well!

Let's now run a few RightMark tests to see how some of my DACs measure using both the Tascam and E-MU. I'll be looking at the AudioEngine D3 USB DAC as an example of a very capable but "lower tier" DAC in terms of resolution and the TEAC UD-501 as an example of a higher-end desktop DAC which would likely challenge the resolution of these ADCs. This will provide an opportunity to correlate the results between different "instruments". If the Tascam results closely follow what I've been seeing with the E-MU over the years, then at least it's suggesting that I'm on the right track with these measurements :-).

Basic setup:
Windows 8.1 Surface Pro 3 --> 6' Shielded Belkin Gold USB --> DAC [AudioEngine D3 / TEAC UD-501] --> analogue cable (shielded RCA for D3 / XLR for TEAC) --> ADC device [Tascam / E-MU] --> shielded USB --> measurement Windows 7 laptop

Tascam latest Windows driver: 1.01
Tascam latest firmware: 1.07

ASIO (TEAC) or WASAPI (AudioEngine) drivers used for playback.
ASIO for all recording.

RightMark 6.3.0 used as measurement suite.

II. RightMark Comparisons


     16/44:

As usual, let us start with the most common audio resolution - good old CD-quality 16/44. Here is an overall score sheet:

A few graphs to consider:
16/44 Frequency response: Essentially flat.

16/44 Noise floor: About the same across the board.

16/44 THD: Notice a bit more "skirting" at the base of the primary signal for the D3 suggesting more jitter as compared to the TEAC. Both Tascam and E-MU consistent in picking this up.

16/44 Stereo crosstalk: Interestingly, the AudioEngine has lower crosstalk than the TEAC despite the TEAC using XLR cables. Both Tascam and E-MU consistent also in this finding.
As you can see, 16/44 is really no challenge at all for these DACs (D3, TEAC) and measure essentially identically using both ADCs (Tascam, E-MU). The results show essentially ideal measurements. Not surprising that in the 2010's, reproducing a CD-resolution signal is really a "piece of cake" for reasonable quality digital audio gear.

     24/96:

Time to delve into the world of high-resolution with 24/96 then. Here's the summary:

Okay, a little more variance this time around compared to 16/44. However we basically see that the TEAC is capable of better noise level and in turn dynamic range compared to the AudioEngine D3 DAC (about an extra 1/2 bit or 3dB better for the TEAC). Interestingly, the E-MU actually turns in slightly better numbers in both noise floor / dynamic range as well as composite distortion numbers than the Tascam. Let's have a look at the graphs then:
24/96 Frequency response: In terms of the DAC, the TEAC has a flatter more extended frequency response as demonstrated with the Tascam measurement. It's interesting that the E-MU tends to roll-off the ultrasonic spectrum >20kHz very slightly (<1dB difference at 40kHz).

24/96 Noise floor: Not much difference really... TEAC slightly lower and this is consistent for both the Tascam and E-MU measurements.

24/96 THD: More evident than with the 16/44 measurement above, there's more "skirting" with the AudioEngine suggesting perhaps more jitter. But what's that peak at 30-40kHz showing up on the Tascam measurement?

24/96 Stereo crosstalk: Again, interestingly less crosstalk measured with the AudioEngine D3 and this is again consistent with both Tascam and E-MU as measurement devices. Hmmm... There's that odd spike at 30kHz again with the Tascam measurement of the TEAC DAC.
Despite a bit more variance, we're basically seeing consistency in the measurements using the Tascam and E-MU devices. However, I noted that unusual spike in the Tascam measurement of the TEAC UD-501 at 30-40kHz. How odd... We'll look at this in greater detail in a bit.

III. Dunn Jitter Test Comparisons:

So far, RightMark results are reasonably consistent and there's agreement between the results from my E-MU and the new Tascam unit in terms of how the two DACs measure. How then do the J-Test results look?

AudioEngine D3:


TEAC UD-501:


Unfortunately I pushed the volume a little higher with the Tascam on these UD-501 screen captures. As a result you can see noise spikes like the 10kHz spike with the E-MU 24-bit J-Test show up in a more obvious fashion (refer to the image above of the frequency spectrum of silence with the E-MU).

In any case, I think it's quite evident that there's not much discrepancy between the Tascam and E-MU. Both the AudioEngine D3 and TEAC are demonstrating rather good J-Test spectra, consistent whether using the Tascam or E-MU as ADC "measurement" devices.

IV. Tascam, we have a problem...

Remember up above when I looked at RightMark results with the Tascam and saw that usual ultrasonic noise above 30kHz with the 24/96 measurements of the TEAC?

Here's what I found (these are all from the left channel). When I start the Tascam "cold", there doesn't seem to be a problem:
Tascam UH-7000 24/96 "silence" at startup. Beautiful!
Tascam UH-7000 24/96 "silence" 15-minutes turned on. Couple little oddities.
Tascam UH-7000 24/96 "silence" 30-minutes turned on - LEFT channel. Huh? What's that?
As you can see, as the machine "warms up", noise starts creeping into the ultrasonic range between 30-40kHz! Swapping USB various cables made no difference nor did changing computers (Acer laptop and Surface Pro 3 tried). Of course I also made sure the Tascam wasn't situated close to another device which could be causing this noise. I even tried it with my old MacBook Pro and saw the same problem with OS X drivers.

I also played with the various settings of the device mixer itself to make sure I didn't accidentally turn on some DSP effects but to no avail.

BTW, here's a look at what the mixer panel looks like with knobs and buttons for effects like the compressor, noise suppressor, EQ, limiter...:


Finally, I plugged it into my Belkin power conditioner which made no difference so I am left with the conclusion that the noise is originating from within the device itself.

The noise isn't as "loud" with the right channel but obviously still there:
Tascam UH-7000 24/96 "silence" 30-minutes turned on - RIGHT channel.
As best I can tell, this seems to be something that happens as it warms up. In cooler ambient temperature (like my basement), it takes longer but will eventually show this noise pattern. If you look at the 24/192 FFT, it looks like there's quite a bit of noise up at 60-80kHz as well (maybe the 30-40kHz noise is a subharmonic):

Tascam UH-7000 24/192 "silence" 30-minutes turned on - LEFT channel.

I wonder whether this is noise from the switched-mode power supply (SMPS) given the close proximity to the audio circuits and whether this could be suppressed with better line filtering or RF shielding of the power supply. Well, here is demonstration of one piece of audio equipment where the audiophile practice of "warming up" the gear actually deteriorates performance.

V. Conclusions:

Since I bought this unit from Amazon, I had the opportunity to return it hassle-free... And that's what I ended up doing after putting it through its paces over about 3 weeks. I made a 24/96 vinyl needle drop of the Paul Simon Graceland album (Side 1 - same as what I did for the LP Test a couple months back). No question, it sounded good; punchy dynamics, excellent details, sounds just like the original LP playback. The ability to fine-tune the level controls with those large knobs on the front panel and the beautiful LED indicators to monitor for clipping will certainly be missed!

Throughout the test period, I never ran into any crashes with the driver software (unlike the issues I have with the E-MU on occasion). I also like the aesthetics of the device - goes well sitting on top of my TEAC UD-501.

I don't know if this particular unit was defective but I've received feedback from a friend in Europe with the UH-7000 that he's seeing noise in his unit as well:
European Tascam UH-7000.
If this is a systemic issue, I certainly hope Tascam fixes this perhaps in future hardware revisions (hopefully it's just a firmware/driver issue)... I'm very happy with Amazon's return policy and I must say I'm quite impressed by the little E-MU 0404USB workhorse despite its age and buggy drivers that will on occasion hang with samplerate changes. The Tascam has demonstrated to me that the E-MU ADC tends to roll-off the ultrasonic frequencies slightly (only -0.75dB up around 40kHz) and the noise floor is obviously not as clean as it could be (perhaps related to the little wallwart power supply).

Ultimately, the noise I found with the Tascam ADC once it warms up is of low-level and would not be audible in regular use (certainly not audible with LP needle drops, but high-resolution multitracking studios and those using effects processors might care) due to it being ultrasonic and down at below -100dB (a subjective reviewer would not have been able to pick this up). However, it is an unexpected finding which does show up when I do high samplerate measurements. If it were not for this, the unit would have stayed as a fixture on my audio equipment rack in the sound room.

Can anyone recommend to me a good ADC that is highly accurate, reasonably priced (maybe $500-600USD), stable Windows drivers, has good level controls, and provides good visual feedback of volume levels to avoid clipping? I'm certainly happy to purchase a Tascam unit like this one again in a few months and see if this issue is gone if by then I haven't found a better ADC.

-------

I hope everyone had a wonderful Christmas and New Year time! Time to get back to a hectic work and family life... May your 2015 overflow with good music :-).

34 comments:

  1. This comment has been removed by the author.

    ReplyDelete
  2. The Roland R-05 is very good for around $200. Very easy to use since you don't need a computer. Rips on that come quite close to my Benchmark ADC1 (which I bought used for $800). You can buy the Roland from B&H Photo which also has a good return policy.

    Not sure where you are located, but if you are near in the Pacific NW, I will be happy to let you borrow both these units.

    ReplyDelete
    Replies
    1. Thanks for the tips vijay! I'll look into the Roland R-05. I see there's a music store that sells it locally so might see if I can get a loaner to play with. Would be good for vinyl rips and convenient portable recording! Although I don't do many 24/192 recordings/measurements, I see that it is limited to 96kHz.

      The Benchmark looks like an amazing unit. The AK5394 ADC chip should be significantly better than the E-MU's AK5385.

      Thanks for the generous offer, Vijay, unless you're here in Canada (I live in Vancouver), I'd hate to borrow something as expensive as the Benchmark lest anything happens to it; especially crossing borders.




      Delete
    2. Even on my Benchmark, I do vinyl rips at 96kHz even though it supports 192kHz because it keeps my files smaller and I don't see a benefit with 192 vs. 96.

      I come up to Vancouver a couple of times a year usually (I live in the Seattle area). Next trip is likely in summer and I will be bringing my Benchmark over to get some needle drops off a a couple of friends' TTs which are similar to mine but with different cartridges. I then get to see if I like a cartridge back in home setup :-)

      Great work with this blog, btw. I don't have your level of technical know-how in this field, but if I did, I would do exactly what you are doing :-) Thank you and please keep up the good work.

      Delete
    3. What I wanted to say but didn't was that we can try to meet up and test the Benchmark then if you are open to the idea (would understand if you are not).

      Delete
    4. Nevermind - http://www.stereophile.com/content/benchmark-adc1-usb-ad-converter-measurements

      :-)

      Delete
    5. Thanks for the update vijay.

      No worries! Plenty of potential ADCs to try out. I'll make a trip out to one of the pro music shops in the next couple weeks to see what "better" stuff they might carry. So long as the return policy is decent...

      Delete
  3. Thank you for the extensive review. I noticed the same ultrasonic noise on the unit I just received so I opened it up and confirmed that it is the smps. I was able to reduce it some with makeshift shielding but a linear supply would be best. I suspect the reason your european friend is seeing different interference frequencies is due to the difference in 50 and 60 Hz input to the smps. The output from the smps is a single 10.9 V DC supply so I’ll try it with a bench top supply and see if that cures things before building a linear supply for it. Other than the switcher noise the unit is great. The price in the US has dropped to $399.

    ReplyDelete
    Replies
    1. Thanks for the comment Dave. Certainly looks like what I was seeing isn't an isolated case. Let me know how the experiment with the bench power supply works out.

      Did you notice the noise immediately or did it "warm up" with the noise building as time went on with your DAC?

      Amazing how quickly the price has dropped... $399!

      Delete
    2. The bench supply worked well, no noise issues. The unit draws 500-600 mA at 10.9 V.

      With the internal switcher I noticed the noise almost immediately (with in a few seconds of power on after it had been off for several hours), I didn't observer it building with warmup it was ~20 dB over the noise floor from the start, with the inputs turned all the way up.

      The unit has good shielding from outside interference, there is copper foil in several places around openings in the chassis. The smps is made by TEAC so I’m a little surprised they didn’t address the internal noise better, a small steel enclosure on the smps would probably reduce or eliminate the issue.

      There is plenty of room inside the chassis to mount a small transformer and LM317 linear voltage regulator circuit board once the smps is removed. That’s what I’m planning to do to fix things.

      Delete
    3. Cool, Dave! Thanks for the update. Good to hear confirmation about the SMPS noise.

      Hmmm, I hope TEAC/Tascam will do the job and come up with a revision that fixes this! If anyone notices this happening, please drop a note here. I'd love to grab an improved revision :-).

      Delete
    4. This comment has been removed by the author.

      Delete
    5. Hi Archimago,

      I have contacted TASCAM (TEAC Europe GmbH) about your findings.

      You might want to know that they are currently looking at a solution but priority is not high, since - as you described - in most cases no audible impact is to be expected.

      Nevertheless, I will buy a unit since it suits my needs (unlike the 'Yellowtec PUC2', which suffers dropout and latency issues) and let us hope, that TASCAM is fixing their product in due time.

      One request, should you one day buy another (hopefully revised) UH-7000: please do perform some measurements using the device as USB -> AES3 converter. :-)

      Best regards,
      Daniel

      Delete
  4. This comment has been removed by the author.

    ReplyDelete
    Replies
    1. Wow. Nice!

      I might get in touch with you if I decide to purchase this unit again!

      Delete
    2. Hi, I'm interested in the linear power supply you have for the Tascam UH7000. Can you please let me know how I can purchase one of these units. I'm not sure how to contact you by email. Can you please provide contact information?

      Delete
    3. Hi, I have finally completed the Linear Power Supply for the UH-7000 to resolve the noise after 19Khz
      Before
      http://img.photobucket.com/albums/v159/projectile/UH7000-Old_PSU-192_zps7ttvbjbe.jpg

      After
      http://img.photobucket.com/albums/v159/projectile/UH7000-New_PSU-192_zpssxyrdrd1.jpg
      http://img.photobucket.com/albums/v159/projectile/IMG_20151011_1619033_rewind_zpsujpjdrpa.jpg
      http://img.photobucket.com/albums/v159/projectile/IMG_20151011_1625234_rewind_zpsyd0aql42.jpg

      Initially there are some issue with noise with the SMPS and Linear PSU, I discovered the cable that link from the Headphone jack to the DAC PCB is causing all the issues

      Another interested in this PCB can drop me a message or email

      Delete
  5. Thank you for the nice review. i might grab one as a monitor controller. It is the same price with UD-301

    ReplyDelete
  6. I wish you still had this to test. I would have suggested you try a different external clocking source to see if it was something going on with that crazy clock source they have that is supposed to be ultra low jitter. It has to have amazing fast playback and is temperature sensitive in its operation. That might explain why it is changing when its on for longer - because of the tempature shift. It might be better for better jitter specifications, but that feedback circuit could be creating some interference in the input channels.

    ReplyDelete
  7. The best way to get a great mark is asking a writer from globalessays to satisfy your academic needs.

    ReplyDelete
  8. I have a few questions for Archimago.

    1) You mentioned that you had problems with the analysis software, RMAA 6.4.0, on a windows 7 laptop, so went back to RMAA 6.3.0. I downloaded the current version, RMAA 6.4.1, and cannot seem to record wav test files (when I check the box) to my windows 7 laptop. Is this similar to your problem? I cannot find 6.3.0 to download now, so I guess I’ll try the only other previous version I see at the Rightmark website, 5.5.
    2) My second question is what frequencies did you use for IM testing, the default 60/7000 Hz mix, or another, such as 18/19 kHz, and have you compared the result differences?

    ReplyDelete
  9. This comment has been removed by the author.

    ReplyDelete
  10. Hi, I have finally completed the Linear Power Supply for the UH-7000 to resolve the noise after 19Khz
    Before
    http://img.photobucket.com/albums/v159/projectile/UH7000-Old_PSU-192_zps7ttvbjbe.jpg

    After
    http://img.photobucket.com/albums/v159/projectile/UH7000-New_PSU-192_zpssxyrdrd1.jpg
    http://img.photobucket.com/albums/v159/projectile/IMG_20151011_1619033_rewind_zpsujpjdrpa.jpg
    http://img.photobucket.com/albums/v159/projectile/IMG_20151011_1625234_rewind_zpsyd0aql42.jpg

    Initially there are some issue with noise with the SMPS and Linear PSU, I discovered the cable that link from the Headphone jack to the DAC PCB is causing all the issues

    anyone interested in this PCB can drop me a message or email

    ReplyDelete
  11. This comment has been removed by the author.

    ReplyDelete
  12. This comment has been removed by the author.

    ReplyDelete
  13. Hi Archimago, what is the serial number of your UH7000? I wonder if the noise affects only some units and not all -- maybe a bad batch of earlier serial numbers? I'm very interested and planning to buy this interface and would like to avoid any serial numbers that may potentially have this noise problem.

    Thanks very much!

    ReplyDelete
  14. hello there. i am very interested in this as adc dor vinyl recording. i was wondering if the problem was solved with new win drivers. any news?

    ReplyDelete
  15. I recieved the unit today, Europe though, and with latest firmware and driver, issue with the artifacts Tascam on 21khz ++ is not gone. I am trying now to unplug different things inside to check if there is one of the cables. eg. the jack plug cable goes over to the power supply.

    ReplyDelete
    Replies
    1. Thanks for the update Svein.

      That's really unfortunate! I would re-buy this unit in a heartbeat if they can fix this anomaly. I really liked how stable the drivers were and the quality otherwise seemed top-notch when I had it!

      Good luck!

      Delete
  16. Ive done some work on my Tascan UH-7000 this weekend. please read my article :)

    http://heavycorner.com/tascam-uh-7000-fixing-issue-noise/

    ReplyDelete
    Replies
    1. Great job Svein on tracking this down! I can see your text but not the images at the moment. Looking forward to more articles on this!

      Delete
  17. For various reasons, I recently ended up getting one of these. The noise splats at 20 kHz and multiples are still there, unfortunately. The good news is that replacing the power supply with an external one got rid of it. I tried a number of different 12 V power supplies I had on hand and got the best results with a Seasonic SMPS (no idea where, when, or how I obtained it). A couple of Maplin-branded ones also worked well, whereas with some there was noise similar to how it was originally. Apparently the issue is not (only) shielding but also sensitivity to power supply ripple.

    Now if only it were supported under Linux...

    ReplyDelete
  18. This comment has been removed by the author.

    ReplyDelete
  19. Hi and sorry ... i did not know how to edit the 1st comment.

    I am writing just to confirm that using a different psu (a 12VDC unit works fine) actually eliminates the bumps in the noise floor.
    The idea would be to place a dc socket on the rear panel of the unit in order to be able to use any decent 12V/2A power supply, linear or smps.
    Unfortunately I do not have the skills and the equipment needed to troubleshooting on the original psu.
    Maybe just some more caps strategically placed could solve the issue.
    Thanks again for all the kind and valuable advice, gino

    ReplyDelete