Et Cumulus ex Cirrus
A bevy of graphs that compare Cirrus Logic converters
There’s no denying that many things sounds so much more compelling in languages other than English. Would we have been as impressed with Aida and Requiem if they had been written by Joe Green instead of Giuseppe Verdi? Hmmm...
As I noted that the AD/DA converters that I’m evaluating in the Behringer X32, the Echo AudioFire 12, and the Mackie BlackJack, are all made by Cirrus Logic. Perhaps it’s the effluvia from Tropical Storm Bonnie that has propelled clouds to be top-of-mind: Ergo, we have a pile of Cirrus’, equally expressed as a cumulus of cirrus. As Classical Conversations homeschooling mom agree, Omnia sonat magnus in Latin. (Everything sounds impressive in Latin.)
The X32 uses the CS5368, and the (early model) AudioFire 12 has CS4272’s as does the Mackie BlackJack. Both the CS5368 and the CS4272 have the same dynamic range specification. The one outlier in the Cirrus AD/DA group is the Echo AudioFire Pre8. Its claim to fame is the super-quiet THAT Corporation integrated preamp chips (probably the model 1510), running alongside the AKM AK4620B converters. Keep those preamp chips in mind in the upcoming article about preamp comparisons.
In general terms, all of my AD/DA converters did very well. My good friend Bernie was kind enough to create some more data points using his Apogee Quartet. A few years back, I had recommended that he buy it, and when he did, I was both pleased and nervous. One never knows if there’s some hidden disaster in a newly-announced product, and $1300 can test anyone’s ability to forgive. In the case of the Quartet, a nasty low-level D/A conversion problem did arise, but Apogee fixed it under a gracious warranty that was a bit beyond the original terms and conditions. Anyway, I had high expectations from the Apogee Quartet, and it did not disappoint.
Based on the measurements from the Apogee Quartet, it can serve as a benchmark for all the preamps and AD/DA converters pretty much anywhere in the electronic world. The noise floor was bouncing along the theoretical minimum for wide-band converters working in the analog realm, offering about 21-bit conversion. In fact, as seen here in [Figure-14] the Y-axis has been expanded to 146 dB to allow the noise floor to be appreciated. If the graphs for the Quartet’s performance had been maintained at the standard 120 dB Y-axis, there would be no difference between the source 992Hz sine wave, and the resultant loopback +4dBu line level -18dB chart. Moreover, the preamp was notably clean, with only a hint of a 3rd harmonic above the noise floor when driven at 30dB gain. C’est assez bien! Preamp behavior will be documented in a separate article.
Just as the the X32 converters were given a high level signal to display, the Quartet was driven with a -2dBFS signal on the +4dBu line level input. The resulting graph is shown in [Figure-15]. A mild set of harmonic products do pop up, implying that somehow, some of the preamp circuitry is being tickled by the strong input signal. Given that the -96dBFS 3rd harmonic in this figure is 'way down at about 2dB SPL the way my monitors are calibrated, it’s safe to say c’est pas trop mal. (Not too bad)
Among the AD/DA products that I personally own, first up is the (AF12) Echo AudioFire 12. I have the original ones with the upward-pointing firewire ports. That means the CS4272’s are installed. For perspective, note the list of a few other products that use the same CS4272 A/D converters:
Focusrite Liquid Saffire 56
Focusrite Scarlett 6i6
Digidesign Eleven Rack
The later Echo AF12 versions had the AKM AK4620B converters, the same as the Pre8. This product never got enough market share because it arrived before ‘bypassing the preamps’ became an imperative among the home studio crowd. During its heyday, the AF12 had to compete with products that sported integrated preamps, and therefore those high-function products seemed to be a better value. Only MOTU had famously kept (during that time) to the purist formula of a line-level-only product with the 1224, 24IO, and 2408, all of which I’ve owned at one time or another.
I did find 2009-vintage complaints on Gearslutz that one cannot bypass the preamps on the Pre8. Although multiple posters wisely pointed out that the AF12 exists for those who want the bypass, the arc of the discussion proved once again that if people don’t want to like something, they’ll make up a reason, and doggedly stick to it. My favorite observation in that thread was a comment about people who’ll spend $600 for a product, and then complain that it doesn’t come with a yacht and a maid.
Here’s the AF12 graph in [Figure-8]. Nothing to see here folks, just acknowledge that the noise floor is not quite as low as the X32’s, but there is no evidence of a signal hygiene problem.
Next, comparing two devices that have 114 dB Cirrus Logic converters, but experiencing them through preamps, just as we are with the X32. In the Mackie BlackJack [Figure-10], the Onyx preamps are set for 0 gain, and in the Echo Pre8 [Figure-9], the on-board THAT Corporation 1510 preamps at 0 gain are in the signal path as well.
The graphs for these devices again undergird my assertion that the big deal about preamp bypass is a purist notion masquerading as engineering insight. Furthermore, preamps that are straight wire with gain get maligned as being sterile if they’re not liked, or transparent if they are liked. Preamps with intentional levels of transformer-based hysteresis distortion and 2nd harmonic tube distortion will get criticized as 'wooly' if they’re not liked, or having ‘mojo’ and being 'thick' if they are liked. I’ll save the rest of the rant for the upcoming preamp article.
A medium-sized version of [Figure-6] is here for your convenience.
- Ted Gary of TedLand
June 2, 2016
Additional information on CS4272 converters added April 5, 2017