PS AUDIO: Not a pretty feature

PS Audio logo

Paul McGowan:

Thanks for all the great feedback on our new video series on Jitter. Click the link to our YouTube channel so you can view what you want. We’ll be adding a new video or two on the subject of digital audio and jitter each week, so hopefully we can all share a lot of what’s interesting to you. Ted Smith is a fascinating fellow and just sitting down to chat with him is a real treat. Ted’s been working with us for a couple of years now. He was involved with the Mark II upgrade to the PWD and you’ll see more of the results of this collaboration soon enough. Wink, wink.

Continuing our story of the Digital Lens RAM buffer product, introduced in the mid 1990′s, Arnie Nudell and I decided to break from loudspeakers and jump into the fray with a new product concept, the Digital Lens; a RAM based digital rebuilding instrument (kind of like a Power Plant of digital audio). Before the Lens was introduced there were jitter reduction devices aplenty, but none like what we were contemplating.

To build a “perfect” jitter reduction device we would have to accept the incoming data from a CD player, strip the clock from its data, store the data in a memory and then recombine that data with a new, low jitter clock. Sounds simple enough but when we went to mechanize it we discovered a problem: the CD players were all running relatively out of time from an accurate clock. From the CD designer’s perspective this mattered not because in a digital audio system, the CD player becomes the master timekeeper in the system. That’s right, whatever is feeding your DAC has control over the timing, the jitter, the pacing. It has control of everything and the DAC is just a slave. So CD player designers didn’t really care if they were a bit slower or faster than what’s considered accurate because the DAC didn’t know any better. But the Digital Lens did know better.

What was surprising to us was the amount of variability and to what products the timing error applied to. CD players that were in the many thousands of dollars were some of the worst, while cheap CD players and transports were close, but still off. To make matters worse, none of the players were consistently off in terms of timing. They would speed up, then slow down, all within the course of playing a single CD. There was simply no standard these players were following. Our prototype Digital Lens had a precisely timed, expensive reference clock that was accurate to many parts per million. CD players were all over the map.

So the first challenge we faced building a practical RAM buffer reclocking circuit was measuring the error of the CD players. Why was this important? Because if the player was slower than our master reference clock, we’d have to tell the RAM buffer inside the Lens to fill up before it started delivering the data to the DAC. And if the CD player was too fast, we’d want to deliver the data immediately but assign enough RAM space to we wouldn’t have an overflow problem. This was not trivial stuff and during the course of a single CD play, the players could vary from too fast to too slow. We’d have to measure that difference and assign the buffer to respond accordingly.

The first result of our designing a measurement system for speed error quickly turned into a new feature on the soon-to-be product: a PPM error readout on the front panel (we couldn’t resist). Yup, once we had to figure out how far off each player was from the reference, we wanted to immediately display the error for customers to see. This was actually our chief engineer’s idea, Bob Stadtherr, who gently suggests things like “you know, we could display the errors if we wanted to ……” That’s about all it takes to ring my bell and so it becomes a feature.

The errors were measured in PPM (Parts Per Million). This was a great idea and helped sell Digital Lenses, only there was a problem. Remember I told you that the more expensive players had worse errors than the lower cost ones? Just imagine the flack we got from owners of $20K transports who had timing errors worse than a $500 CD player. They couldn’t believe what they were seeing.

It wasn’t pretty.

Leave a Reply