Wednesday, February 4, 2015

RME ADI-8 DS Mk iii A/D D/A Conversion - Time marches on, and gets more accurate.

Ty Ford
I’m writing this article in February 2015. If you’re still arguing that analog is better than digital, I can’t help you. I’m OK with the idea that drums and other sounds with high transients may be recorded on properly aligned and biased analog tape recorders and that the non-linearities of that system may result in a sound you like better. The rounding off by tape compression is an acknowledged, if not expensive, effect in music production today. I get it.

File that on the shelf next to “Ford’s Theorum” (something I’ve said for years) which states that, “Good analog is better than bad digital and good digital is better than bad analog.” Further, that, “Good solid state is better than bad tubes and good tubes are better than bad solid state.” You can argue the boundaries of those intersections, but you can’t argue that they don’t exist.

The transformer - no transformer debate is sort of a corollary of the above. Yes, iron changes the sound and in some cases can be said to make things sound more pleasing. Yes, the right iron and the right number of windings can help low sensitivity mics do better by providing some “free” gain. 

Once you get tired of arguing any of the above, there’s still A/D conversion to hammer out. Fortunately, we are years and many revisions past the early and ugly attempts at 16-bit, 44.1 kHz. Yes, there was a quantum leap when good 24-bit was achieved. Yes, we can and have taken it farther, but In a world where the largest market segment is happy listening to mp3 files on cell phone earbuds, how far do we really need to go?

Audio is seldom the prime mover in the Technology Circus. More typically, audio is the happy recipient of advances in the video, medical technology and scientific measurement industries. Below those impressive moments is...wait for it...Audio. 

George Massenburg GML 8204
In my own pursuit of “financially limited perfection”, I realized that there are others like myself who can hear things that others can’t. 

Likewise, there are things I’m relatively deaf to that others can hear. We are all in pursuit of satisfaction; our own individual happy place that makes us smile when we hear “the right stuff.” There are many paths.

I improved the audio on my first Pro Tools rig almost 15 years ago by carefully selecting some good preamps - GML 8204 and Millennia Media STT-1. After settling in with them, I took aim at the A/D conversion. The hardware I was using had an ADAT optical port, and because of my budget, I tried the RME ADI-8 DS shortly after it came out in 2000. This was back before AES standards had evolved and Tascam’s TASCAM Digital Interface (TDIF) and Alesis’ ADAT ruled the swamp.

Millennia Media STT-1
Almost 15 years later, my ADI-8 DS is still in the rack and apart from requiring a recent power supply replacement, has been powered up and quietly doing its job all along. I did the power supply swap out with a little help from RME US rep Jeff Petersen at the Synthax in Ft. Lauderdale office, distributors for RME/Alva Cableware and Ferrofish in the USA.

Synthax's Jeff Petersen
The thing is, A/D converters often don’t get the respect they deserve. Petersen puts it nicely, “Converters aren’t sexy, they’re just clean and effective.” He’s right. 

The RME ADI-8 DS Mk iii that's the object of this review, and an upgrade from the piece I've ben using is just a one rack-space box with a few lights and switches on the front and holes in the back. 

It’s supposed to be inaudible, invisible. If you’re an electric guitar player with a pedal board, you want each pedal to change the sound. Fine, but the job here is to allow as much original detail though as possible.

Back then, combined with the better preamps, the first RME ADI-8 DS made a noticeable improvement in my audio at 44.1kHz and 48 kHz. The Lavry and DB Technologies I tried had more detail, but were too expensive and I was satisfied by the improvement brought by the ADI-8 DS. I could go ADAT light pipe right in to the Digidesign hardware. That was a big plus.

By replacing Digi preamps and A/D conversion, I had bypassed the entire front end of the Digidesign hardware. I don’t use the D/A conversion, just the A/D. No disrespect to Digidesign. They had a piece at a price. I just built out from that with gear that I could hear made a difference and that (Including five preamps and eight channels of A/D conversion) cost nine times my original investment in Digidesign hardware and software. I expected that would pay off for my host-based system as computers got faster and storage got cheaper. It did.

I think I started with Pro Tools V 5 and am now at V 10. I use it every day and make a living with it. Pro Tools 11 is current and, for the first time, offers faster than real time Bounce To Disk, but I’ll need to upgrade my entire Mac platform and all the good and bad can be expected to go along with that. I’m just not ready to do that at the moment.

RME ADI-8 DS Mk iii
RME ADI-8 DS Mk iii
The most notable operational difference between my ADI-8 DS and the newer ADI-8 DS Mk iii is that the D-subs connections now go to AES/EBU I/Os instead of TDIF. As the AES/EBU spec has evolved, more AES enabled gear has been created; including this upgrade for the ADI-8 DS Mk iii. About how the line of RME converters plays out, Petersen says, “The ADI-8 Pro (48k max) was the original, the ADI-8 DS (96k max) would be the Mk ii essentially, and the ADI-8 DS Mk iii (192k max) is the 3rd generation of this venerable 8 channel converter. The QS is a different animal.” I took the opportunity to give it a test run during the time my “original recipe” unit was out of the rack for its power supply exchange.

The Mk iii benefits from the 14 years of advancing technology since the first ADI-8 DS was released. The most noticeable spec difference is the lower latency of the Mk iii. It’s at 12 samples, the old one was at 40 samples. I am not bothered by the old unit’s 40 samples, but if that was a deal killer for you, see how you like 12. 

Like the earlier model, The ADI-8 DS Mk iii is an 8-channel AD and DA converter in a standard 19", 1U box. This includes ICC Intelligent Clock Control (ICC), SyncCheck ™, four hardware reference levels up to +24 dBu (instead of just three), AES/EBU and ADAT I/O, with up to a 192 kHz sample rate. 

According to RME, “Our unique SyncCheck and AutoSync technology has evolved into the new Intelligent Clock Control of the Hammerfall DSP system. HDSP is the only digital I/O-system worldwide capable of measuring and displaying the frequency of all clock sources. Even word clock! Based on validity and current sample rate the system then decides which clock source should be used - fully automated and performed in hardware. With this the HDSP system offers the easiest handling of the present clocks.”

Word Clock Implementation
The signal present at the Mk iii BNC Word Clock input can be single, double or quad speed and when lock is achieved the WCK LED stops blinking and glows steadily. RME uses what they call a “Signal Adaptation Circuit” and “Automatic Signal Centering” to reshape the Word Clock signal. This reshaping addresses several possible fault problems, including poorly shaped, low or high signal at the input and sends the new signal to the Mk iii WCK BNC output. The WCK BNC output also has built-in 75 Ohm termination, should you need to slave the Mk iii at the end of a chain of digital devices.  

SteadyClock is RME’s answer to the problematic Superclock and has been added since the earlier ADI-8 DS. It uses a 22 MHz clock signal that RME believe better answers the stability problems raised by Superclock. The Dither feature of the earlier model is also absent from the Mk iii.

Analog inputs are via eight 1/4” TRS jacks or one 24-pin D-sub connector. The TRS input jacks are servo-balanced to accept either balanced TRS audio or TS unbalanced audio. You may see the term “double balanced” in the marketing information. I had no idea what that term meant, so I inquired. Turns out it means that both the TRS and D-Sub connections are each balanced, hence “double balanced.” I would hope so, but you never know these days. Another term, Full Symmetrical, caught my eye and I was told that meant that all stages are fully balanced instead of single ended. 

While you may use only one of the inputs, 1/4” or D-sub, the Mk iii outputs have separate drivers enabling both 1/4” (TS or TRS) and D-Sub outputs to be used simultaneously. The TRS/TS output is servo-balanced, but the D-sub is not. When connecting with unbalanced audio gear, pin 3 must be lifted.

As with the analog input, the analog output also has four selections for level; +24 dBu, +19 dBu, +13 dBu and +4 dBu (-10 dBv).

A/D Input
If you have analog gear with different output levels, the DS Mk iii is even more obliging than the original DS. There’s a fourth analog sensitivity setting from which to choose. So now you have +4.2 dB (-10dBv, +13 dBu, +19 dBu and +24 dBu. These changes are all made before the A/D converter. 

The analog LED input level lights on the original DS flashed at 2 dB below FS, which I initially found disturbing, but learned to ignore. The Mk iii partially lights at -2dB FS and goes to full brightness at 0 dBFS. In the heat of the moment I don’t know if I’ll notice the difference. I’m still in ignore/denial mode. I’m not always looking at the rack, so LED hold lights may be helpful if you’re the type that likes pushing input levels as high as you can get them.

Digital Patch Bay
After A/D conversion, the audio is available at both AES and ADAT outputs simultaneously, except when in Patch Mode. The Patch Mode lights indicate the choice you’ve made and which output is active. 

Maybe the key differentiation feature for is the Digital Patch Bay. If you have a variety of digital audio boxes in your studio and they need to talk to each other, you may have had to stop, get behind your racks and swap cables and/or connections to do certain things from time to time. 

The RME web site explains this simple, push-button feature quite well, “A digital patchbay with free choice of source and destination setup can be used to convert ADAT to AES, AES to ADAT, cross-convert them at the same time, pass ADAT on to ADAT while monitored analog, and many more. The ADAT outputs also feature copy mode for connection of two different ADAT devices. These powerful and easy to use modes add significant value to the already outstanding conversion quality.

During single speed A/D conversion (up to 48 kHz), both ADAT and ADAT AUX outputs carry the same audio. A/D conversion is not available during AES input to ADAT and AES output
ADAT input to ADAT and AES output
AES input to ADAT output and ADAT input to AES output
AES input to AES output and AES input to AES output. 

In normal studio operations, the DS Mk iii will be the master and any digital devices connected to it will be slaves. That’s how it is with my setup. The Digidesign 003R is the slave and that is a choice in the Pro Tools software. Within Pro Tools, the ADAT protocol and optical source (because ADAT is optical) are the way to go. the When devices are locked correctly, the Mk ii Sync Light is on and steady. A blinking light means there’s a problem.

In Conclusion
As mentioned before, decidedly unsexy, but very important to the integrity of the signal chain. I recorded voice, acoustic guitar and drums separately on both my original ADI-8 DS and the Mk iii at 24-bit, 44.12 kHz. Try as I could, I could not detect any real difference in the sound. 

So it’s primarily about AES/EBU implementation, the lower latency, the added flexibility of the patch bay, SteadyCLock and an added input sensitivity and output level increment. My ADI-8 DS has been in the rack and powered up 24/7 for about 15 years before the power supply checked out. That’s got to count for something, too.

Technique Inc. © 2015 All Rights Reserved

Here’s the bullet list from the RME web site.

8-channel AD converter, full symmetrical design, 119 dBA
8-channel DA converter, double balanced output, 120 dBA
Low latency conversion: less than 12 samples delay
4 x AES/EBU I/O per D-sub, 8 channels @ 192 kHz
2 x ADAT I/O, 8 channels @ 96 kHz, 4 channels at 192 kHz
Word clock input and output
2 x 8-channel level metering
Comprehensive digital patch mode for full interconnectivity


Input AD: 1/4” TRS jack and 25 pin D-sub, servo balanced, all symmetrical audio path
Output AD: 4 x AES/EBU, 2 x ADAT optical
Input DA: 4 x AES/EBU, 2 x ADAT optical
Output DA: 1/4” TRS jack servo balanced, to +21 dBu. 25 pin D-sub, balanced, to +24 dBu.
Dynamic Range AD: 119 dBA
THD AD: < -110 dB (< 0.00032 %)
THD+N AD: < -104 dB (< 0.00063 %)
Crosstalk AD: > 110 dB
Dynamic Range DA: 120 dBA unmuted
THD DA: < -104 dB (< 0.00063 %)
THD+N DA: < -102 dB (< 0.0008 %)
Crosstalk DA: > 110 dB
Input level for 0 dBFS: +24 dBu, +19 dBu, +13 dBu, +4.2 dBu
Output level for 0 dBFS: +24 dBu, +19 dBu, +13 dBu, +4.2 dBu
Input/Output level at 0 dBFS @ -10 dBV: +2 dBV
Sample rates: 44.1, 48, 88.2 , 96, 176.4, 192 kHz, variable (sync/word)
Frequency response AD/DA -0.1 dB: 10 Hz - 23.2 kHz (sf 48 kHz)
Frequency response AD/DA -0.5 dB: < 5 Hz - 43.0 kHz (sf 96 kHz)
Frequency response AD/DA -1 dB: < 5 Hz - 85.0 kHz (sf 192 kHz)
Power supply: Internal switching mode ps, 100 V-240 V AC