So I finally got the system sorted out. I am running a MCH server with Lynx card straight into a stack of 3 x DirectStream DACs. Control volume with the remote - the DACs don’t miss a beat and stay in synch perfectly! The server runs Dirac. This setup beats the crap out of ANY surround sound processor. Spend a lot of time ripping blu ray concerts, SACD and DVD-A, but well worth the effort.
I have a different server for 2 channel, which runs into one of the DACs over USB.
I used to do the same but with MSB Signature plus for mains, and NAD M51 for surround / center. I found the DirectStream to be close enough to the MSB to swap out it, and with the freed up cash I threw in two more DirectStream for MCH. Because the DirectStream has processing delay, you need three identical DACs in this type of MCH setup. My next move will be to replace the three DACs with the Multi Channel version of the DirectStream (Ted, Paul are you listening!).
The all Spectral Amplification is also all spanking new (bought used), and what a revelation these amps are. Compared to the improvement wrought from replacing my Class A monos + Arc Ref 5SE with Spectral DMC 30 SS series II and DMA 360 Series II, the difference between the MSB DAC and the DirectStream is indeed splitting hairs. 2 years ago when replaced the PerfectWave with MSB this was not the case - MSB trounced it. So to my ears, the DirectStream has all but closed the gap with the five figure big boys.
Why wouldn’t it work well? You mean sonically or practically?
Only snag is because my center channel is low sensitivity and Dirac DRC reduces gain, I cannot rock the house with SACD rips (these somehow play lower volume than other digital sources). The bottleneck is the center channel. The DS plays about 3dB lower than typical DACs, so this adds to the problem. I have ordered a gain stage that will add 8dB to the center channel, allowing me to jack up all channels by 8dB, so this will hopefully address this issue.
No, I was just talking about the abstract use of something in a setup it wasn’t designed/tested for. Since I didn’t do anything to keep them locally synchronized I didn’t have any idea if there’d be a practical problem or not. That’s the only problem I’d see with all non-USB inputs. (separate USB inputs will be well sunk but may drift slowly at a few parts per million.) But no matter what doing the math tells me, I always want to test things for myself before I’ll say they would work.
I’m glad someone has the confidence/temerity/resources or whatever to try it, thanks.
Are you using a PS3 for rips? SACDs play at the same volume as PCM here (0dBFS PCM == 0dBFS DSD). Perhaps there’s some sort of PCM volume leveling, processing, or whatever going on that DSD steps around?
Since I can get a bargain on DSs (I have an inside connection) I’ll have to try doing it myself when I’m not busy with some other project.
I think the synching issue is more of a theoretical than a practical concern. Having said that, Mytek and Playback Designs have an architecture allowing you to synch up the DACs if you’re using a stack in MCH application.
My SACD rips are Ps3 based. I convert DSD to PCM and then apply Dirac, not sure where, but somewhere in the processing chain, gain is lost.
If you can work out a deal with your inside connection I would definitely give this a whirl.
One concern I has was synching the volumes. It works absolutely perfect. None of the DACs ever miss a volume up/down signal and they act is one DAC with the remote control. You do NOT need a MCH analog preamp in this setup. If you have one, take it out. I have never heard one that beats the DAC direct signal path (2 channel preamps - different story). The volume synch actually works better than the RS232 and Macro programming I used on my MSB + NAD M51s - the RS232 sometimes missed a command, and the DAC volumes got out of synch.
Ah, if you convert DSD to PCM outside of a DAC you do need at least a 4dB drop and 6dB is probably best, otherwise you’ll clip now and then which isn’t pleasant.
So you figure a 4-6 dB drop is applied by Jriver during conversion to prevent clipping? That would explain it. If so jacking up the center volume with a gain stage in the analog domain would be a good solution.
SACDs can go to 11. 0dBFS on a SACD is only 50% modulation: all 0’s or all 1’s on a SACD would be 6dB higher than 0dBFS. That also has the benefit of some “soft clipping”.
Some people/programs naively don’t use the full 6dB of attenuation when they go from DSD to PCM and that will cause distortion/clipping now and then.
edorr said
My SACD rips are Ps3 based. I convert DSD to PCM and then apply Dirac, not sure where, but somewhere in the processing chain, gain is lost.
Why are you converting your DSD to PCM?
Two reasons, for MCH, I don’t have a choice, because the Lynx card cannot output 4 x 2 channel DSD over AES/EBU. I could play DSD on two channel, but in order to apply Dirac DRC in need PCM. In my experience, the benefits of DRC far outweigh the benefits of keeping material in native DSD.
vortecjr said
Why don't you trim the volume for each output in the Lynx mixer control panel so the levels all match?
Jesus R
The issue is not trimming the volume between channels. The issue is that the channel with the lowest sensitivity (center) resticts the maximum volume at which I can play, which it too low.
vortecjr said
Why don't you trim the volume for each output in the Lynx mixer control panel so the levels all match?
Jesus R
The issue is not trimming the volume between channels. The issue is that the channel with the lowest sensitivity (center) resticts the maximum volume at which I can play, which it too low.
Understood. Something is not right. Whatever JRiver is doing to the file to prevent clipping it should be doing to all the channels equally. I wonder it Audiogate can make the multichannel conversion. It could be the conversion, Dirac, gain on the center channel amp and or efficiency of the center channel speaker…
vortecjr said
Why don't you trim the volume for each output in the Lynx mixer control panel so the levels all match?
Jesus R
The issue is not trimming the volume between channels. The issue is that the channel with the lowest sensitivity (center) resticts the maximum volume at which I can play, which it too low.
Understood. Something is not right. Whatever JRiver is doing to the file to prevent clipping it should be doing to all the channels equally. I wonder it Audiogate can make the multichannel conversion. It could be the conversion, Dirac, gain on the center channel amp and or efficiency of the center channel speaker....
Jesus R
All channels are impacted equally, so the bottleneck is simply the channel with the lowest sensitivity speaker, which is the center. I'll be adding an 8dB gain stage on the center channel and I should be OK.
(Ted, Paul are you listening!).
I still am a little surprised that it works well for you but I’m not complaining