Interesting Results from Accuphase Hi-Rez DSP/DACs


I am in the process of restoring a rare pair of Apogee Full Range speakers, when I made the decision to ditch the analog xovers and go ALL DIGITAL, using Accuphase's fascinating DF-35 digital xover. In this way, I would be totally digital up to the power amps.

Briefly, the DF-35 includes 2 completely independent digital xover processors, called Filter Amplifiers (FA), with an option for 2 more. Each FA gets a copy of the signal and has various options, including the ability to set high/low xover points with up to 96db slopes, not to mention support for both SACD and DVD-A!! The DF-35 requires a digital preamp to drive it. The only major drawback is that you need a power amp for each FA to drive. Still, eliminating both the analog preamp AND xover is a huge benefit.

Anyway, Accuphase recommends you use their digital preamp (DC-330), which is outrageously expensive. Fortunately, there are some cheaper options. I went with the Z-Systems RDP-1 digital preamp and Meridian 500 transport. Keep in mind, if you want to play hi-rez (SACD or DVD-A), you will need to go with their DC-330 and DP-100 hi-rez transport. I just wanted to see the concept work.

One of the interesting features of the Z-Systems product, is that it can output 16, 20, or 24 bit word lengths that can also be dithered to lessen the effects of quantization noise. It was the use of this feature with the DF-35 that produced some astonishing results.

Many people have complained that the Accuphase SACD players / processors are lousy with red-book CDs and have problems, in general, with upper frequency roll-off.

The RDP-1 defaults to 20 bit words, because that is the most common TRUE resolution of most commercial DACs that most of us use. When I played this through the DF-35's FAs (one drove a REL Strata III sub in bypass mode, the other a JRDG Model 6 / Martin Logan CLSIIz combo), the sound was shrilly and harsh. And, there was severe roll-off on the high frequencies that was so dramatic, you could literally hear a fizzling sound - yuck!

I was absolutly crushed! But, after consulting the manual, Z-Systems does recommend using 24 bit words when the more advanced hi-rez processors come out. Well, the DF-35 is definately one of those. So, I instructed the RDP-1 to send 24 bit word resolution, and the sound improved so dramatically that I thought either RDP-1 is defective with 20 and 16 bit words or you do not want to keep a hi-rez DAC guessing! The sound was incredibly clear to the point where I was hearing new information on recordings I have owned for a long time and knew well. I found the dither definately helps with some of the "bite" in the upper frequencies. The sound is still a little lean, but I'm not sure it's the DACs fault. The overall presentation is absolutely spectacular however.

These results are really interesting. Unfortunately, I cannot afford the more expensive DC-330 at this time; so, I cannot test out SACD or other hi-rez formats. But, it would appear that Accuphase processors are not too keen on low resolution information. The RDP-1 not only offers a high-resolution digital volume signal, it also eliminates jitter effectively.

My conclusion, so far, is that if you are using an Accuphase hi-rez (SACD) processor and red-book CD media, you might want to consider adding an independent up-sampler to improve results; assuming you're one of the many audiophiles complaining about red-book cd performance on Accuphase hi-rez dacs.

Anyone else want to comment?
fdriver
Please "ADD" the following to my previous post:

The overall effect on the speakers, in general, was the most interesting. The CLSs now played much louder, and seemed a little faster. The REL also seemed to go deeper, as it violently shook the room during recordings that previously did not have the same result!? I can only presume that removing bass from the CLS signal allows the electrostats to perform more efficiently, while removing high-freq and analog xover from the REL signal has a similar effect.
I was using a REL/CLS setup to test out the process, whilst I was restoring the Apogee speakers. The REL allows you to run it using it's own xover, or in bypass mode for situations like home theatre, where the signal has already been digitally split; hence, no analog xover is needed.

In order to keep the original message from becoming too complicated, I ommitted my setup scenarios and the resulting speaker interactions. Originally, I used an analog preamp with the REL and CLS running in parallel, as I did not want a xover-effected signal going to the CLS for purity reasons.

Here is the setup and results:

In my case I set the first FA to 56hz or less, 96db slope, and had it output via SE to the REL, which was set to bypass it's own analog xover and amplify the signal directly. The second FA was set at 56 hz and above, 96 db slope, and had it output XLR to the Rowland amps, driving the CLSs.

When I fired it up, it actually worked! But, there were some problems. The CLS and REL were completely out of sync and the sound had a hard, metalic quality that I did not like. When I read the RDP manual, I discovered the problem. It defaults to 20-bit words to work with typical processors, but recommends 24 bit words for more advanced (hi-rez) processors, like the DF-35. The RDP also has dither (noise that smoothes out the quantization noise caused by converting high-bit signals (24 or higher) to low-bit (16) during the recording process) and an excellent parametric equalizer.

I set the RDP to 24-bit word and the sound changed so dramatically for the better for reasons that I am still not sure of. I then instucted FA #2 (controlling the CLS) to delay the signal by 6 cm. It worked perfectly - the CLS and REL were now playing in sync - amazing!

Also, one of the CLS speakers plays a little louder than the other. I first instructed the RDP to compensate for this (digitally!) and it also worked - the image was centered. However, I realized that the subwoofer was also getting the same adjustment AND that each FA in the DF-35 could also do the same thing! So, I reset the balance on the RDP-1 and instructed FA #2 to decrease the louder (left) speaker and the image was still centered!

Also, each Filter Amplifier (FA) in the DF-35 has a neat feature for subwoofers; if you set the switch on the back of any FA to "sub" position, it will actually generate a mono version of the stereo digital signal.

This is especially ideal for those who use one subwoofer with only one corresponding input, or those who use one mono amp for a single sub and don't want to fork out the money for another amp/sub, or those who want to use two subs and want identical signals sent to each one. If you still want stereo sub ops, just set the switch back to normal - it's that simple!

So, the digital xover (even with a xover-less CLS) did make a big difference. With the multi-driver Apogees, these differences should be more dramatic.
Could you give us a little more info on exactly how you will use the outboard crossover with your Apogees? And how will you amp them?

I frankly find it a little strange in theory that you should report hearing an improvement based on the way you have tested this so far, just using the digital crossover to direct the signal between your full-range speakers and the subwoofer. I thought the whole point of REL subs was to leave the mains running full-range and send the same power amp signal to the sub for best integration. By hooking the sub up in bypass mode, it seems all you have done is substitute the digital crossover for the analog one that REL prefers you not use anyway. The conventional wisdom on why this would improve the sound of the M-L's is freeing them from having to reproduce the low bass, but how did you set the controls to blend the two, and do your results invalidate REL's concept in your mind (at least with the CLS's)? In any case, I don't think anything you may have heard in this preliminary experiment could be counted as a blow in favor of digital crossovers per se, since your speakers are a crossoverless design. I also kind of doubt (as you suggest) that the digital source's higher dither/bit depth setting could have caused the amount of improvement you heard unless something was actually going wrong with the interface at the lower setting.

Anyway, I find this sort of thing very interesting, and do consider it to be one the waves of the future, but as one could have predicted, it seems as if not many A'goners are interested in discussing something that may so upset the analog apple cart (and I note this as an analog guy myself, who won't be buying any high-rez digital for quite some time I feel). But then again, any way you do it, multi-amping with an external crossover is always going to be a minority approach. Best of luck with your project!
Tireguy,

Thanks for the info.

However, I can tell you that eventually everything WILL be done in the digital domain. There is simply too much to lose with an analog preamp and/or an analog xover. Analog xovers are a definate compromise in sound quality. I can hear the improvement when bypassing the xover quite clearly.

However, I do agree that the technology is new. The Accuphase DF-35 has no real competition yet, except for the new Tact 2.2X, which offers a digital xover for subs, but not for replacing the xovers in a true multi-driver speaker, like the DF-35 can.
I had an Accuphase DC-300(the prelude to the 330) and it was ok but NOT worth its retail price- I have not heard the DC-330 though I have heard the DP-85 and it was similar but more resolving then the DC-300 I think it comes down to if you like the accuphase sound then its perfect for you if not keep looking- many discerning audiophiles love there accuphase gear, but if you are questioning its performance it probably isn't for you. I don't like the idea of doing EVERYTHING in the digital domain like accuphase does(with that sort of processor) to much room for error in an area that hasn't really been perfected.