Perpetual Technologies P-1A D/D & P-3A D/A processors Manufacturer's Comments
Editor: On behalf of everyone at Perpetual Technologies, I'd like to thank Stereophile and Bob Deutsch for this incisive and positively well-done review. We realize that the P-1A (particularly) and the P-3A present several reviewing challenges, and we really feel that the products were well understood and reviewed. Of course, this is only the opening act, as it were. My comments will be mostly nontechnical in nature, but directly after my few words from the grandstand, Peter Madnick and Keith Allsop will weigh in with some important technical comments.
When we decided to forgo the early architecture we had designed originally, we knew it would lead to a fair bit of delay. The fact that remains unchanged was that, after careful analysis of the prototype we presented at HI-FI '99, we knew that we would need the additional computational horsepower (and then some) that the Analog Devices SHARC® chip provides us. This 32-bit Floating Point solution was the only answer to doing it "right" from the beginning. We're glad we decided to do this, terrible delays notwithstanding.
The important next steps are what most are waiting for, and Perpetual Technologies is ready. As I pen this reply from Taipei, Taiwan, I have just finished demonstrating 24-bit/96kHz full-band speaker correction (what we refer to as SOCS, "Speaker Only Correction Software") to rapt audiences in Hong Kong and China. The results were outstanding. All persons could clearly hear and appreciate the results of the speaker corrections. It was fun and exciting for me, as well as for everyone in attendance.
By the time this reply is published we will have delivered SOCS to hundreds of waiting audiophiles all over the world. This program will be ongoing, as we are in the process of working with some of the leading speaker companies from around the world to archive their speakers' data. No simple task...Readers should call us, or visit our ever-changing website, to find out when we expect to have SOCS for your loudspeakers.
For the audiophile using loudspeakers either out of production, of a DIY nature to begin with, or whose original maker does not have the data sets we require to build correction files from, we have an elegant solution. Enter our Mobile Measurement Kit (MMK). The MMK includes a computer, a calibrated microphone, and detailed instructions, and is a simple-to-operate data-collection system that will allow us to build really great correction sets for a particular loudspeaker/room scenario. We envision the cost to rent this system to be around $250, plus the $399 needed for the SOCS. This is an economical and elegant solution to a big problem. I want to emphasize that this is not our planned Room Correction System upgrade. That comes later, and will include measurement hardware and the use of the Internet for data transmission and filter downloads.
The P-1A represents the first step in a series of revolutionary technologies. Many have taken to calling it the Sony Playstation&Q#174; of the audio industry. You think it's one thing, when in fact it is many. With the P-1A and its USB interface we can do so many things for the audio- or videophile. Soon, we'll have plug-in software-driven enhancements that will make the P-1A a must-have for everyone. Just be patient with us.
When you couple the technological advancement(s) of the P-1A/P-3A, made affordable through our "Internet Direct" distribution system, you get one formidable combination. We are really proud of these products, as well as of the distribution architecture we are building through our sister company, AV123.com.
Again, we'd like to express our thanks for a positive "first installment." We look forward to participating with the magazine in developing the rest of the story. I'm sure it will be one that everyone will find a happy ending to.—Mark L. Schifter, President & Founder, Perpetual Technologies
Editor: We'd like to thank Stereophile and Bob Deutsch for his excellent and highly complimentary review of the Perpetual Technologies P-1A/P-3A products. As Mark Schifter mentioned in his letter from Asia, we'd like to offer a bit of explanation regarding the interesting measurements that John Atkinson got for the P-1A. Obviously, the excellent P-3A measurements speak for themselves.
The P-1A was designed to be a general-purpose audio digital signal processor. The Analog Devices SHARC DSP chip was selected for its 32-bit block floating-point math and 60-megaflop operation speed. All of this power is required because the Resolution Enhancement process, which can be described as a dynamic "adaptive-predictive" algorithm, executes a large number of multiplication/addition operations per sample. The algorithm is dynamic; that is, it is designed to operate differently on input signals that change in both amplitude and frequency with respect to time. The algorithm has been optimized, over these last five years since the introduction of the Audio Alchemy DTI•Pro 32, to work at maximum performance with real music and not steady-state test tones.
Unfortunately, there are no simple real-world measurement techniques to show the performance improvements a P-1A can give. We rely on mathematical simulations of the enhancement algorithm using complex dynamic test signals and, of course, listening tests. The attached graphs are two of the simulations we work with.
Fig.1 shows a difference of the summation of the non-fundamental components of 16 different burst sinewaves, before and after Resolution Enhancement and then assembled on one plot. From front to back, each curve represents a 6dB level reduction (therefore running from 0dB in front to 96dB in back); the X axis is frequency, with DC at the left and 22.05kHz on the right. The Y axis is the dB level, and the Z axis (height) is Resolution Enhancement in +3dB steps from 0 to +15dB. What one can discern from these graphs is that, as frequency increases and level decreases, the algorithm does the greatest amount of enhancement. Starting between 10kHz and 11kHz, depending on dynamic conditions, the real power is applied, with up to 15dB of enhancement to a 22kHz, -96dB signal.
Fig.1 Difference between the summation of the non-fundamental components of 16 different burst sinewaves, before and after Resolution Enhancement by a Perpetual P-1A. (6dB/vertical div.)
Fig.2 shows three cycles of a 5kHz sinewave graph showing the waveform before and after Resolution Enhancement. In this case, this is a special P-1A code that does not add dither, and therefore more clearly shows the operation of the enhancement. As expected from the first graph, the algorithm is most effective on higher frequencies. The Stereophile result in fig.11 from the review is actually very good in comparison to fig.5, and would have been even better if the masking effects of dither had been removed, as we have done here.
Fig.2 Waveform of a 5kHz sinewave before and after Resolution Enhancement by a Perpetual P-1A.
With respect to the jitter measurements, again we rely on mathematics, showing sub-50ps jitter figures. We don't use Mr. Miller's technique of in-band jitter measurement on the P-1A because we are not confident that the dynamic in-band algorithms in the P-1A are not contaminating the measurements. We have had, however, excellent results with his system measuring DACs.
In conclusion, we appreciate the opportunity to give you some insight into what the P-1A can do from a mathematical perspective. Our listening tests (and those of Stereophile) have shown us to have been very successful. Certainly the correlation of these different measurement systems with listening tests needs further research; we urge your readers to see for themselves how we've done.—Peter Madnick, Keith Allsop, Perpetual Technologies