The last chapter in the "Jitter Saga" has yet to be written. But the penultimate chapter has just been written by Grimm Audio, who have challenged the most sacred concept -the commonly-held notion that internal clock always performs better than external. Digital clocking has been a hot topic among audio engineers in recent years. Someone started the idea that the built-in clocks found in most converters were poor. And it snowballed from there. Reputable companies started selling dedicated system clocks. Before long, the notion that most digital converters needed a better clock became accepted as gospel. Add the fact that there exists a very real need to have a master clock when connecting multiple digital devices, and the market for digital clocks was established.
The CC1 is a master clock from Grimm Audio, a company founded by four of The Netherlands' most prominent engineers: Bruno Putzeys, Guido Tent, Peter van Willenswaard, and Eelco Grimm. We tested the CC1 in my mastering setup and with our Pro Tools HD systems at Treelady Studios. At the same time, Bob Katz ran a Grimm CC1 through trials at his Digital Domain studios as well as with a Pro Tools HD rig at Phat Planet Studios in Orlando. Numerous converter brands were used, including those from Digidesign, Crane Song, Benchmark Media, Lynx, Mytek, and TC Electronic (in the System 6000), in the quest to answer the following questions. Does an external clock improve converter performance? Can an external clock degrade converter performance? Are there tools that measure clock jitter effectively? If not, why not? Given an impeccable clock signal, does our gear have well-implemented word-clock inputs that can take advantage of the better clock? And could we confirm or deny our hypothesis in blind testing?
Unlike many manufacturers, Grimm can buttress their claims with genuine high-resolution measurements and cogent explanations that don't stretch the laws of physics. Their claim is that there is a window of opportunity to improve a converter's jitter by using an excellent external clock that has very low levels of low-frequency jitter. The external clock enters the converter through its PLL (Phase Locked Loop), which is able to reduce incoming jitter to some extent, but below a certain frequency, it can only mirror the incoming jitter. So the Grimm clock, because of its extremely low levels of low-frequency jitter, can improve the performance of some converters.
So let's examine the unit. The CC1 is cosmetically remarkable, being equally suited for the recording studio or audiophile listening room. The outer housing is powder-coated with a coarse adobe-colored finish. The front panel is solid wood (abachi) with the markings artfully burned into the surface. Status is provided by flush-mounted LEDs, which are nearly invisible until the unit is powered. Small stainless steel buttons provide controls for bank and source-rate choices. The rear panel offers connectivity and configurability: XLR ins and outs for AES3 links, word-clock in, and sixteen BNC word-clock outs. The WC outs are divided into two banks: 1-10 and 11-16. By the way, the Grimm has many unique features, including one that allows you to reclock the signal going into the AES input of a DAC using the Grimm.
The CC1 is capable of driving two different sample rates simultaneously, provided they are of the same multiple. For example, I was able to pitch source mixes at 88.2 kHz from a Lynx AES16 and capture them at 44.1 kHz on an RME AES-32 while keeping my entire system slaved to the CC1. You could not, for instance, run 96k and 44.1k at the same time, as they are not multiples of the same whole number (e.g., 96k being double 48k).
It's one thing to have measurements, but are they important to normal human ears? To try to settle the controversy, both authors set up tests using the CC1 in combination with many commercially available converters. Both ADCs and DACs are potentially subject to degradation or improvement when fed external clock, so it was important to isolate which side of the stream was under test. Additionally, you must know if the manufacturer uses an integrated or independent clocking approach. What does that mean? With integrated units from Mytek and Digidesign, the master clock drives the ADC and DAC simultaneously. But in the Cranesong HEDD 192, the DAC is independent. Thus, the HEDD's DAC is not affected by its wordclock input. All the other integrated units in this review sync to a single master clock, so you can generalize that reported degradations or improvements affect both ADC and DAC. The moral is, "Never assume; know thy system architecture". Bob's blind testing focused on the DAC section of the Digidesign unit, but was also confirmed by measurements and listening that its ADC behaved similarly. Meanwhile, Garrett's testing focused on the ADC.
BK: We conducted a blind test at Phat Planet with 8 listeners, 10 trials each -a total of 80 trials. Chief engineer Aaron Gandia and I carefully chose the source material, trained the listeners, and presented the comparisons in a way that made it as easy as possible to detect the differences. Nevertheless, a rigorous blind test is extremely hard to make, especially with such subtle sonic differences; while I took the test I found that if I lost concentration for even a moment, I could easily make a mistake. While having more trials increases statistical accuracy, ironically it potentially decreases accuracy because the listeners become fatigued. It's amazing that 60% of the total trials were correct; these listeners correctly identified and preferred the sound of the Grimm clock compared to internal clock on a Pro Tools HD 192 I/O interface. A 60% score from 80 trials means that the odds of getting this result by chance alone are only 2.7%, so we should regard this listening test as very indicative that the Grimm clock makes the 192 I/O sound better. That's what I heard. Listener comments on the review sheets include: "I heard more clarity with the Grimm, much more open sound and more depth." "Internal clock: graininess in the upper mids. Low end, high end, and stereo image were better with Grimm." "Grimm did seem to have more dynamic range. Stereo image is bigger with Grimm."
GH: From our Pro Tools tests at Treelady, we focused on the AD side of things. We recorded sources through a Lynx Aurora 16-VT and a Digidesign 96 I/O on internal clock and also slaved to the CC1. Playback was done on the Crane Song Avocet to remove the DA from the equation. We agreed that the Grimm CC1 improved the sound of our stock Digidesign 96 I/O units. Comments included a greater sense of depth, cleaner midrange, and smoother highs. Most notable was a tighter bass response. We were split with the Grimm CC1 paired with the Lynx Aurora 16-VT. Most of us thought the low mids and bass were improved with the Grimm at the expense of a more aggressive, even spitty, top end. In that trade off, we felt the Lynx would sound better using its internal clock.
BK: Many converters' performance degrades when placed on external clock. In my measurements and listening tests, I found that the Mytek 8x192 definitely performs better on internal clock, as does the HEDD 192, so if you own one of these, it should always be the master clock in your studio. However, according to Grimm, the ubiquitous Digidesign 192 I/O converter performs better on external clock, if a superior external clock is used. GH: So how could an external clock degrade the performance of some units? We found some reasons. First, the design of the external clock could be faulty. It is possible to improve the frequencies in some areas while increasing distortion in others. Second, it may not be the clock's fault at all. Some converters have such noisy WC inputs that it doesn't matter if you have a space clock, atomic clock, or the Finger of God; as soon as the clean clock hits the PLL, noise and distortion are added, polluting the signal. Kind of like pouring bottled water into a toilet bowl and giving your dinner guests a ladle if they're thirsty.
GH: So what did we learn? First, there is a lot of misinformation circulating about external clocks. Whether that's intentional or accidental doesn't really matter; what's important is that audio engineers should educate themselves. (Good sources include white papers from companies like TC Electronic and Grimm; peer-reviewed journals; and books such as Mastering Audio by Bob Katz. Bad sources include web forums and advertisements.) The more we understand the tools of our trade, the better the recordings we can make. Second, using an external clock does not categorically improve every converter. In fact, performance can actually degrade in some units. Finally, if you need a house or master clock, realize that not all timepieces are created equally. In this regard, the Grimm CC1 has no peer. With dual frequency banks, ease of use, and AES reclocking abilities, the CC1 represents the most advanced state-of-the PCM clocking art.
BK: I think the last chapter in the "Jitter Saga" will only be written if and when a manufacturer manages to make a good-sounding, affordable converter that is audibly jitter-immune. There are many converters currently on the market which claim to be jitter-immune, but to my ears, exhibit slight, audible differences when playing otherwise data-identical sources -for example, comparing the sound of a converter with an AES3 input which produces different sound when fed from the DAW than from the CD player (with identical data).
Fortunately for us, the situation has greatly improved in the last 10 years, with the audible differences now so small that we can make sonic judgments without encountering any big surprises. ($2995; www.grimmaudio.com)
Tape Op is a bi-monthly magazine devoted to the art of record making.