Wednesday, 24 December 2014

MEASUREMENTS: Tascam UH-7000 USB Interface (Part I: As a DAC)

0. Preamble

Well, look what Santa brought me early this month:

So apparently I had been "nice" in his books so he decided to see about getting me an upgrade on an ADC device for vinyl recording and measurement accuracy :-).

This is the Tascam UH-7000, part of the "pro" line of audio interfaces made by Tascam/Teac. It's a simple 2-channel model. On the front we have 2 large knobs to control the preamp input level for each channel - rotation feels smooth - good enough for reasonably precise volume adjustments. There's the "Phones Level" knob for headphone volume and when the "Link Line" LED is lit, this also serves as a master volume control for the microphone/XLR inputs. Holding down one of the two buttons for a few seconds above the "Phones Level" knob allows you to turn on the +48V mic phantom power. There are the corresponding multicolored 20-segment LED indicators; very useful to check for clipping.

Here's the rear of the unit:

To the left are the analogue inputs. This unit is meant for professional audio use so there are no single-ended RCA connectors to be found. Instead to the upper left we have 1/4" TRS balanced inputs, and XLR balanced below for analogue input. Analogue output is via balanced XLR connectors, and to the right we have XLR digital (S/PDIF or AES/EBU) in and outs, a USB 2.0 port below, and of course IEC connector for power cable.

You should not have much difficulty finding decent TRS to unbalanced RCA adaptors for line level input (should only cost <$5 at the local electronics parts store for decent gold-plated models):

The unit feels well made. It's a metal enclosure with the characteristic TEAC look, similar to the TEAC UD-501 DAC but not as wide. XLR connectors are nice with a good snug fit and reassuring click when locked. Latching mechanism feels secure and robust.

Currently, this model is on sale (I see $100 off in the US around Christmas time), I got it locally off Amazon in Canada for ~$600CAD. The device is compatible with Windows (32-bit and 64-bit) and Mac OS X. Not Linux compatible as far as I can tell.

I. The Tascam as a DAC

First, I want to explore analogue output quality when functioning as a DAC. Inside this unit is the Burr-Brown PCM1795 DAC which is the same as the Teac UD-501. The difference here is that the UD-501 is a "dual mono" DAC using 2 PCM1795s for lower noise operation versus this unit with a single DAC chip.

Hooked up to the oscilloscope, here's what a 1kHz 0dBFS square wave looks like with all volume controls at maximum:

Nice and clean. No clipping as confirmed by a 21kHz sine wave at 0dBFS. Beautiful channel balance as well.

Peak output from the XLR at 8.72V or about 6.2Vrms. (+18dBu referenced to 0.775Vrms. I believe this level can be maxed out to +24dBu according to the specs sheet if I use the mixer/control panel.)

Impulse response of 16/44 signal:

Typical linear phase filter with the usual pre- and post-ringing characteristic of fast roll-off filters. Absolute phase maintained.

As usual, I'm going to measure the output with the trusty E-MU 0404USB I have been using for the last couple years for consistency and comparison. The signal path looks like this:

Windows 8.1 laptop (Surface 3 Pro) --> shielded USB cable (Belkin Gold) --> Tascam UH-7000 --> 3' XLR analogue cable --> E-MU 0404USB --> shielded USB cable --> Windows 7 laptop (Acer)

Measurement software: RightMark Audio Analyzer 6.3.0 (I found some issues with the newer 6.4.0 version on the Windows 7 laptop).

Playback software is the newest foobar2000 1.3.6 with ASIO plugin.
Latest Tascam driver (1.01 for Windows) and firmware (1.07).

1. 16-bit Results:

Starting with 16-bit audio, here's the "big table" with results from a number of other DACs I've measured over the years:

As you can see, I've included a number of different models - ranging from the USB stick DACs like the AudioEngine and Dragonfly, to the iPhone 6, and of course better "desktop" units like the Oppo BDP-105 (oops forgot the P in the table), Logitech Transporter, and Teac UD-501.

Apart from the Dragonfly which has a bit more distortion and higher stereo crosstalk, 16-bit audio really isn't much of a challenge for modern DACs capable of high-resolution 24-bit audio these days.

A couple graphs:
16/44 Frequency Response
Notice a little more bass roll-off with the Tascam compared to the others, about 0.5dB down at 20Hz; otherwise not much to see here...

16/44 Noise Level
Good all around low noise performance.

Bottom line is that 16-bit audio is just great with any decent modern DAC...

2. 24/96 Results:

I believe 24/96 is the "sweet spot" for DACs these days. It's ubiquitous for any DAC worthy of being called "high resolution" including the simple USB DAC/headphone amps like the AudioEngine D3 and Dragonfly. Most high resolution album releases seem to target this resolution and "vinyl drops" tend to be 24/96 as well so it's important to function well at this resolution. A good DAC in 2014/2015 should have no excuses for not measuring well at 24/96 IMO.

Again, let's start with the "big table". This time, iPhone 6 not on the list as it's incapable of 96kHz.

Not bad! The Tascam UH-7000 squeezes out slightly >110dB noise level and dynamic range. On this chart it looks really close to the Teac UD-501 but in fact those results I have for the Teac are with the RCA output - the XLR results would be better and likely beyond the capability of the E-MU 0404USB measurement device.

Distortion values for the Tascam is low and it maintains very respectable overall stereo crosstalk result. However, if we look at the stereo crosstalk graph, it looks interesting:

There's a notable rise in the crosstalk starting around 500Hz and gets worse with higher frequencies. I have not seen this before with the other DACs. At 20kHz with -66dB crosstalk, it's arguable if this is ever audible of course despite the unusual finding.

24/96 frequency response: slightly more roll off with the Tascam at frequency extremes.

24/96 noise floor: Dragonfly worst performer here.

3. 24/192 Results:

Honestly, I feel that 24/192 is excessive already for consumer playback and I prefer not to have such large audio files eating up hard drive space with questionable benefit! Others disagree of course. (BTW: Consider also whether many 24/192 recordings actually have content in the high frequencies; for example, take a look at the recent 2014 release of Neil Young Time Fades Away.)

In any case, here are some numbers for the Tascam DAC and a couple others that support this samplerate:

The Tascam isn't as good as the Teac and Oppo in terms of noise level. Remember, the Tascam is about 3/4 the price of the Teac and 1/2 the price of the Oppo, plus it's also an ADC device which I have not talked about yet!

A couple graphs to review:
24/192 frequency response: Again the Tascam seems to roll-off more at the frequency extremes (not necessarily a bad thing to attenuate the ultrasonic content to reduce intermodulation distortion).

24/192 noise floor: Not too shabby!

4. Jitter

Using the standard J-Test as in my other posted reviews...
16-bit J-Test: Primary signal at 11kHz, spectrum from 5kHz to 18kHz. Jitter modulation pattern easily seen (good demonstration of the low noise floor).
24-bit J-Test: Primary signal at 12kHz. No sidebands around the primary signal. Minimal "skirting" around the primary signal.
What can I say, it looks nice. As expected for a modern asynchronous USB 2.0 interface, jitter should not be an issue. Tascam boasts that this device uses a high precision TCXO (temperature controlled crystal oscillator) down to +/-1ppm accuracy.

5. Subjective Sound Quality

The reality is that although I've measured the DAC function here, it's unlikely I will be using this device as a DAC in my listening room... My primary interest with this is for ADC recording like vinyl needle drop archiving or as a measurement device like what I'm doing with the E-MU 0404USB over the last couple years. I connected the XLR analogue output to my main audio system in the soundroom for a couple of evenings to make sure the sound was good; indeed it sounded great - very similar to the Teac UD-501 from last year. Soundstage was nicely presented when played back through the Emotiva XSP-1 preamp and Paradigm Signature S8 speakers with SUB 1 subwoofer. I've also spent more time listening to the headphone output switching between the Sony MDR-V6, Sennheiser HD800 and AKG Q701 over a few evenings while doing other work.

I noted that the headphone out seemed more powerful than the Teac UD-501 especially for more demanding headphones like the AKG Q701. The sound is full and clear at maximum volume with no evidence of strain. Channel balance is very good throughout. Even though the frequency response measurements suggest relative bass roll-off, I can't tell a difference as music still sounds nice and full on the low end (not that one would expect to hear any anomaly unless one had truly golden ears since we're looking at <0.5dB down at 30Hz difference between the Teac and Tascam DACs!).

Whether through the headphones or speakers, sonic detail like Muddy Waters' guitar sounded precise and startlingly dynamic on "Country Boy" (off Folk Singer, Classic Records 24/96 HDAD rip). Vocal separation; that sense of "air" between singers on the new Pentatonix Christmas album (That's Christmas To Me) sounded excellent (check out the recent release "Mary Did You Know?"). Another holiday treat was Andrew Gant & Vox Turturis's Christmas Carols from Village Green to Church Choir also available as 24/96. A joy to experience these tracks through the Tascam with my Sennheiser HD800 for personal listening and also through the Signature S8 speakers with the family! Fantastic ambiance in the performance venue captured in high-resolution with excellent dynamic range throughout this "classical" recording (I always enjoy a nice performance of Veni, Veni, Emmanuel this time of the year).

As I have expressed over the years, decent DACs these days sound fantastic. The main impediment to good sound has more to do with whether the amp is adequate to drive the transducer and things like impedance matching for headphones to make sure the frequency response stays reasonably balanced (unfortunately the Tascam site doesn't list the headphone output impedance and thus far I have not tried low impedance IEM headphones).

6. Summary (as a DAC)

So far, so good. Plug-and-play DAC functionality with no issues around drivers so far. No crashes or any incompatibilities with foobar2000 or JRiver. I have not tried it with the Mac and assume it should be rather straight forward as well with the OS X drivers.

From a sound quality perspective, I have no qualms with the Tascam UH-7000 so far. The sound is subjectively excellent and objectively, distortion is low with good dynamic range and low noise floor. It has unusual stereo crosstalk results with increasing crosstalk as frequency rises but I expect this to be of academic interest only and inaudible. No evidence of jitter anomaly. So while not measuring in the "top tier" of results I have seen, the Tascam UH-7000 does perform admirably and would be commensurate with the relative cost of this device given the functionality.

Next week, we move on to exploring the ADC capabilities of this machine (what I'm most interested in)! Let's see how it stacks up in that department...


Happy holidays to you and yours! Hopefully you're all enjoying the sights and sounds of the season - and hopefully you have been on Santa's "nice" list as well... :-)


  1. I am glad DACs have become almost commodity items these days. As long as a DAC buffers and reclocks with good accuracy from the buffer, spending more than $1000 USD on a DAC seems rather pointless these days. With apps like jRiver, one can tweak the sound to their liking in the digital domain and outside of the DAC. Given this and good recording/mastering, I would gladly give up my vinyl setup. Unfortunately, too many digital albums have been butchered to achieve this goal.

  2. What does a 1kHz squarewave look like at say -3 dBFS ?

    P.S. I must have been a bad boy this year.
    Santa gave me nothing of this sorts !
    Not even the $6.- eForCity - USB 3d Audio Sound Card

  3. Good argument Solderdude. I was thinking the same. How do I see, if this 1 kHz 0 dBFS Squarewave is already clipped or not, because I do not see the otherwise visible Gibbs ringing. Sure, the pre and post ringing of the Dirac impulse are much more visible as the ringing at 1 kHz (because this ringing is at FS/2 and is proportional to the proximity of the signal frequency to FS/2), but at least, there must be some sort of correlation visible at 1 kHz 0 dBFS Squarewave.

    And as a second point: Archimago, it would be really of benefit for you, if you would measure and display the jitter results at least with the HPW-Works software. We have already discussed this point some months ago, because the above shown results is of ne real value.

    PS: Will you be a the upcoming CES in Las Vegas? If yes, we could meet for a conversation.

    Best Regards

    1. Hi guys. Good points. Certainly possible - maybe even likely there's some clipping at the top of the square waves.

      Alas I won't be able to test again at -3dB... Because I ended up returning the unit... The reason will be discussed in the ADC part.

      Thanks for the reminder about HPW-Works. On my list of things to play with... Including DSD Master on the Mac, 24/96 J-Test, trying some headphone load measurements, an update of my listening room (I want to measure the freq response with the album collection and some acoustic panels in place).

      Out of town for the New Years. Have a great NY everyone... All the best for 2015!

    2. Hey Juergen. Alas not at CES. In fact, I don't work in the electronics industry although I have buddies that could get me in :-).

      Hope you enjoy the show!

    3. This comment has been removed by the author.

  4. Hi Archimago. Thank you for your reply. If you need some 24 Bit 96 kHz J-Test Signals, I can send you some (made with AP27 system). I have some experience (and contact) to DSD Master program and use it regularly for converting DSD into PCM and it works great (and have also some measurement results). The CES would have been a good chance to talk. I am heading to Las Vegas in two days, but I must tell you, that I do prefer the Munich High End show or also the Asian shows like the ones in Hong Kong or Taipei. Happy New Year and Best Regards, Juergen.

  5. Intersample Overload Margin: If you go plain / straight with the digital input signal into a TI / BB or ESS DAC chip with standard settings, then they have 0 dB Intersample Overload margin, meaning, that they can reproduce any 0 dBFS sine wave signal without any distortion of clipping, but if you have a “complex” input signal (as music), then you can get clipped samples at the digital oversampling filter, due to 0 dB Intersample Overlaod margin. An easy way to test this would be a 0 dBFS rectangluar signal with 1 kHz or higher (would be better). Best Regards, Juergen

  6. amazing, a very complete audio device as to refer to. You can visit this page, here you will find similar things. elektronik

  7. Hello, very interesting post. I'm a fan of Teac/Tascam products and was considering this unit given its potential use both as an audio interface for recording and a DAC for simple music playing. The unit has XLR analog and digital outputs but my home sound system has consumer type inputs, i.e. RCA and s/pdif. I suppose a converter box in either case would be necessary? Is one option better than the other, and which of the two would give the best sonic performance?

  8. Hello, I read with great interest the test of teac uh 7000, I am unsure whether to buy tascam uh 7000 or RME Fireface UC.
    I would use the card, especially to listen to audio from my PC and also to occasionally record my children who study at the conservatory.
    You could give me some advice?
    tank you
    Luciano Leo