Triple Bypass Part-1
An orientation to the environment and process
Well folks, we're back with another treatise on preamp bypass in the epic journey to the land of pristine sound.
As a matter of establishing context and mindset, I’ve provided a series of images that show my studio topology, the equipment testing topology, my control room/mix position, the internals of a Behringer X32, the internals of a Midas DL151, and an action-packed photo taken during the firmware upgrade of that selfsame DL151.
In the last set of seven (ugh!) articles, the essentials of that measurement suite focused on the behavior or various external preamps feeding through the embedded native Behringer X32 preamps in a full-sized X32 mixer. I compared this signal path with other audio interfaces that have embedded preamps, and a couple of interfaces that were blessedly born without preamps.
In this 3-part series, I endeavor to answer some nagging questions that have up to now been answered with speculation, and the time‑honored Web Hysterical Waving of Arms (WHWOA). Perhaps you haven’t directly seen WHWOA in action, but you’ve experienced it. People make up ratings and characterizations of these products without ever having directly compared them in any way … but that’s the norm. In that sense, the pronunciation of WHWOA is essentially an onomatopoeia of the thought processes that it describes.
Some respondents have a friend with the equipment, or they've been to a venue that had Product-A, and were able to visit another venue with Product-B. Trusting long-term audio memory, and comparing different music in a different space on a different day is hardly a reliable approach to developing consultative guidance. Despite the thin ice, you'll find hardened policies formed in the middle 1990's, relationship-based anecdotes, personal affronts, and blatant guesswork, all used as ingredients of the product evaluation sausage served to one another on the Web. I hope to do better; with real measurements under controlled conditions, pictures, party balloons, and ice cream!
Just so you know what's coming, here's the lowdown on the results:
The Midas DL151 preamp signal path is as clean as the ASP800 ADAT signal path.
The Behringer X32 preamp-free TRS inputs are not as clean as the X32 XLR inputs through preamps.
The Behringer X32 XLR inputs are not as clean as the Midas XLR inputs.
Although there are no giant surprises in those bulleted summaries, the data values and proportions are significant. There are no "night and day" differences worthy of this common hyperbole chosen to emphasize little distinctions. I characterize the differences as worthy of the price difference, but only occasionally essential to getting a good recording.
One of the measurements that didn't make the original set of articles is focused on the group of TRS inputs in the X32. Those inputs don't have preamp circuitry in front of the A/D converters. There's just a digital trim/scaling in front of the CS42438 converters, not to be confused with the CS5368 converters that are bound to the X32 XLR/preamps. I will faithfully provide a few measurements and pictures of these TRS inputs.
Also let it be known that I have inspected the Midas preamp boards, and they have CS5368 converters on them too, so the measured differences are readily attributable to the analog circuitry surrounding the converters.
The two stars of this show sharing the spotlight are the Midas DL151, and the Audient ASP800. To save you some surfing, the Midas DL151 is an AES50-connected 2U box that provides 24-XLR inputs to microphone preamps. It's fan-cooled with two fans so it's not perfectly suitable for studio usage if placed close to the microphones. [Hmmmm, he mused… perhaps a pair of 80mm BeQuiet Pure Wings V2 cooling fans would be beneficial...hmmm.] Anyway, this box gets tested alongside a physically silent Audient ASP800. The ASP800 has 8 XLR/combo jacks as inputs, and two kinds of outputs. There's an analog DB25 connector and a pair of ADAT-out TOSLINK (optical) connectors.
Of course, it takes a bit of doing to get an ADAT signal into an X32 mixer. The big leap is possible with an X-ADAT card to replace the X-USB (or X-UF) card in an X32. The trouble is that the X-ADAT does not have a way of connecting to a DAW when it's in place, so the only answer is to have a dedicated X32 to house the X-ADAT, which in turn connects via AES50 to the main X32. I happen to own both an X32‑full and an X32-Core (primarily for redundant remote recording at venues that have a house X32 or M32) so it was no big deal to press the X32‑Core into service for managing the ADAT‑to‑AES50 conversion duties. You can see it as the ADAT‑to‑AES50 box in the Studio Topology diagram.
For a while I fiddled around with three different clocking topologies, attempting to visually discern any changes in the noise floor when the ASP800 got its clock from the X32-Core word clock, the X32 ADAT-in link, or the propagated X32-full's internal clock. Nothing particularly spellbinding was showing on the Voxengo SPAN displays, and it wasn't worth it to me to spend $30 for a PC utility that purports to directly measure jitter. As the noise floor is in the -130dBFS range, there was little hope of hearing anything related to clocking changes in the absence of a signal. That left me with essentially running a noise floor beauty contest. It also forced a change in the Voxengo SPAN settings.
Whereas the previous series of articles used a 120dB y-axis, this new set of tests must use a 140dB y‑axis so there’d be something to see on the graphs. If I didn’t change the axis range, six of my graphs could be construed as a picture of a lump of coal in a cave during a lunar eclipse.
Ok, I think we all know where things are now. Thanks for hanging in with me. Now on to Part-2 where the measurements begin.
- Ted Gary of TedLand
April 26, 2017