Monday, 25 February 2013

MEASUREMENTS: Logitech Squeezebox Transporter [Updated June 26, 2013]


It's Squeezebox Transporter time!


In order to do the measurements, I brought the gear downstairs to the basement which is an electrically quieter environment. There's a Belkin PureAV PF60 power center there for all the equipment. Also, the measuring computer is now the AMD Phenom X4 laptop with Win 8 usually used by my kids :-). The laptop was running on battery - I could detect a 0.5dB difference with the AC adaptor plugged in down at the -110dB noise level.

One note about the XLR measurements you'll notice - the THD levels are a bit higher than RCA. I believe this is a result of the fact that the E-Mu 0404USB could not handle the XLR voltage from the Transporter and I had to use the analogue attenuators to (just barely!) avoid clipping.

Lets start with the 44kHz signal:
Name:  44_Results.png
Views: 153
Size:  29.6 KB

Undoubtedly, the XLR output is significantly better than RCA. One observation is that through the RCA's, the stereo crosstalk remained around -90dB whether the signal was 16 or 24-bits. I'm not sure if this is the limit of the Transporter itself, or has to do with the cables I used - a pair of AudioQuest 6' interconnects. I don't remember which model of AudioQuests these were (bought a few years ago in a moment of weakness :-) but they are longer than the 3' I had been using to measure upstairs.

Now for 24/96:
Name:  96_Results.png
Views: 152
Size:  29.3 KB

Again, we see the -90dB stereo crosstalk limit with the RCA output. XLR's measure is fantastic! Likely hitting the performance limits of the E-MU 0404USB and ~3dB better than the Essence One (of course the E1 did not have the benefits of a power filter or low noise environment of the laptop running on battery).

The WiFi router was in the same room as the Transporter hence the 90+% wireless strength. No difference in the measurements whether WiFi or ethernet.

To show in graph form the difference in noise floor between the RCA and XLR:

Notice the noise spikes like at 60Hz using the RCA cable (the AQ construction seems to be shielded but can't confirm unless I cut it open!).

Conclusion:
1. Overall the Transporter measures well! Phenomenal XLR performance - best I have been able to measure so far; >19-bit dynamic range.
2. A bit unclear about that stereo crosstalk measurement with RCA however. A bit higher than I'd have expected. Might need to try a different cable and see... (In case anyone wondering, I have not opened up the Transporter to change the unbalanced volume attenuation.) --- SEE ADDENDUM: I believe it's the cable!

ADDENDUM - (Cables do make a difference :-)

Further exploration of the stereo crosstalk issue shows that it likely was the cable afterall! I switched the analogue out from the AudioQuest (can't confirm but the cable looks like the current "Yosemite" model on their web page) to a pair of old 6' Tributaries interconnects (probably 50% the cost). As seen below the Tributaries improved stereo crosstalk by ~2.5dB:
Name:  AQ_vs._Tributaries.jpg
Views: 144
Size:  26.3 KB

Here's how the graph looks (AQ [white] vs. Tributaries):


Furthermore we're starting to see that 24-bit audio is starting to show better separation compared to the 16-bit data with the Tributaries suggesting that the cable was the limiting factor! I expect a good 3' cable would improvement the measurement even further. At this point then, I believe the Transporter's RCA stereo crosstalk is just fine.

Interesting that the more expensive AudioQuest measured worse than the Tributaries in this setup. As a result of this, I'm going to demote the AQ's to my CD player - something I would never know to do unless I ran this objective test.

------

Before I leave the Transporter alone... I wanted to see if turning on the TosLink effects loop affected measurements. Normally, I have the Transporter --> TosLink --> Behringer DEQ2496 (room EQ) --> TosLink --> Transporter as DAC, so it'd be nice to know that the DEQ in digital mode doesn't affect the final sound (tested at 24/96).

With the DEQ2496 on bypass mode (ie. no room EQ processing):
Name:  Tributaries_RCA_-_TosLink_Effects_Off_and_On.png
Views: 251
Size:  25.2 KB

Using the Tributaries RCA cable as output, no difference whether the DEQ2496 was digitally in line or not. Note that I used rather generic TosLinks - well constructed relatively thick plastic optical cables bought on sale for $10 each. Alas, by this time I had disconnected the XLR's and I didn't care to disrupt the Transporter setup again :-)

------

Transporter Jitter measurements (Dunn J-Test 24/48, XLR):



Direct from the analogue output, this looks good!  The usual beautiful Transporter jitter plot. The 6 tiny spikes/sidebands are estimated as <300ps total (I think Stereophile measured 230-270ps) - nice corroboration!

Now what happens with the DEQ2496 in line (remember, the digital data is now going through 2 TosLink cables as described above)?
Obviously quite a bit more jitter has been injected by the Behringer! My estimate looking at the top 8 sidebands suggests that the jitter level now is 2ns using some measurements in WaveSpectra!

Since I can just turn the digital loop on and off, I can do instantaneous A-B'ing of the sound. Even in this condition, I cannot say the clearly increased jitter is at all noticeable. What can I say, even knowing this I'm just fine with keeping the DEQ2496 in line and use the room EQ function since *that* is audible! At least now I can say I've done an A-B test examining the effect of 2ns jitter for myself.

Addendum: Feb 27, 2013
Thanks to slimdevices forum member "tpaxadpom" who measured the digital output with the AP2722 unit:
AES/EBU 377.3-424.5 ps
SPDIF RCA 566 ps
SPDIF BNC 283-330.2 ps (rca cable with 2 bnc adapters yielded the same results)
Toslink 1.462 - 4.103 ns depending on the cable used. I've tested 4 or 5 different toslink cables.

Looks like the Transporter's BNC connection is the winner followed by AES/EBU. TosLink worst for jitter not surprisingly.

Addendum: March 6, 2013
Got some AES/EBU digital cables - here are the measurements.

Addendum: June 26, 2013
As part of the transport measurements, I decided to have a closer look at the Transporter. Here's some more data to consider:
Lovely 24/44 square wave at 0dBFS off the RCA output. Peak voltage of 2.95V. Very nice channel balance.

Impulse response (16/44):
This is the standard "sharp" filter. Linear phase. Absolute phase maintained.

When you set to "slow" roll-off, look what happens:
Wow... Barely any pre and post-ringing! However, clearly to achieve this, roll-off is expected to be very significant.

The "big board" RightMark summary - all based on RCA analogue output:
Look at the 16/44 frequency response; clearly roll-off is quite significant. Here's the graph:
You see the response deviating significantly by about 8kHz and more than -1dB by 15kHz. For good 'youngish' ears, that's significant.

This is what 24-bit buys you in terms of noise floor - remember this is with RCA output, expected to be even better with the XLR:

At 24/96, the filters are still quite different, but inaudible difference for anyone but cats, dogs and machines :-):

RCA analogue output Dunn J-Test:
16-bit (16/44):

24-bit (24/48):

Summary: I remain very impressed with the Transporter. Kudos to Sean Adams and the Slimdevices team back in the day. Technically a beautifully designed machine and it remains my primary digital front end in the listen room. I was a bit surprised by the slope of the slow roll-off - much more than I thought!

3 comments:

  1. Hey Archimago,

    ...wanted to let you know I sure appreciated your analysis and commentary. I used a similar setup to yours for years: Transporter -> Behringer DEQ2496 -> Transporter. AES for the loop. I recently modified my setup, feeding s/pdif directly to a NAD M2. The Behringer is now in loop using the M2 (using, ironically, optical).

    ReplyDelete
    Replies
    1. Nice setup Thomas!

      Wonder if you've done an A-B comparison of the M2-Behringer loop comparing the M2 direct vs. M2-Behringer with Behringer on bypass. As noted in my later Transporter-Behringer AES/EBU post, there was a measurable decrease in jitter compared to TosLink. I never could hear the difference however.

      Delete