|
Home
/ FAQ
/ News Classifieds / Events |
Audio Asylum Thread Printer |
Get a view of an entire thread on one page |
41.183.0.21
In Reply to: RE: Forever destined to listen to 96khz or less posted by theob on May 29, 2009 at 13:58:43
The first test involves increasing Juli@'s latency to a maximum 128 samples (that's the maximum I would recommend for 192k output). I suspect you have tested this and got metallics.
Second test is to never jump directly to a track, instead hit Next till desired track.
I want to use Juli@'s I2S output to ESS Sabre's 9012 Stereo DAC (133db SNR) but this problem bugs me.
Did not like 128 latency but I have settled on 64. Sounds good plus I seem to be having fewer metallics as my linear ps on p4 settles in. Also I'm running sox at 192 with alias box checked. This means no filtering of aliases right? If right that means lower cpu load.
I have tried no mouse navigation within a file and for 2 days or so I've gotten 1-2 metallics a day--not bad. Now I will try 128 latency with the juli@ control panel (in past I've preferred 48 sonically) but I assume this reduces cpu load.btw sox upsampling within cplay is better (by a small margin sonically) and infinitely better for parameter setting and disk file usage. I've just saved 15 gigs of hd!
so thanks again for a great cplay 2.0vs 26 release!
Edits: 06/04/09
I ended up setting cpu voltage higher (.896 on cpuz) and host clock control to 153 and it seems to be better (soundwise and metallics wise). Thank you ray konkle for this suggestion.
I got metallics a couple of times while trying to change tracks with the mouse (sorry I just forgot). And although I had to reboot it seems not as bad while running sox at 192.
Also I implemented a Power One power supply (12 volts. 1.7 amps thank you sondale, mark, robert, ray and gstew) for the p4 and it sounds really good (I'm sure it has more breaking in to do) but it also seems more metallic resistant. I have to figure out where to put it now.
But all in all I may be getting a Lynx 2b. Hate to abandon the juli@ but it is really a distraction to reboot all the time. If I can get less than one a day or so metallics I could live with the juli@. We'll see.
I tried only the 'next' or 'previous' buttons or keyboard for navigating within a file all day yesterday and no got metallics (more below in response to Bertel). When I was actively working as an engineer doing problem solving we used the technique called 'Is/Is not' map. It simply was the listing of conditions that are associated with problem occurence or non problem occurence. With cplay version 2.0v18 I never got one experience of metallics ever-over maybe hundreds of hours of use. I went back to it several times just to listen to music during 'heavy metal' days and it never failed in this way. Yes cplay versions higher than 18 are way way better sonically but all were subject to metallics. Not trying to blame cplay or cmp this is just a statement of fact that may shed some light on what is happenning.In addition to only using keyboard for within file navigation I am currently @ 157 hcc (was 155 the day of the big metallics)--so far after 1.5 days ok.
Edits: 05/31/09
Cics,
I can already comment on the second test:
My version of the metallics occurs even when cplay just continues playing and moves to the next track - then (at random) metallics can occur or not. I restart the track, sometimes once more, sometimes again, rarely again, and sooner or later the track starts without metallics. I could discover no pattern whatsover yet to give me a hint on possible causes (other than the combination of latencies maybe).
However, since I am currently focussed on getting the system optimized for Buffalo32s first, it isn't an issue for me for the time being, i feed it with 16/44.1 since it does upsampling anyway. Once that is running fine I'll try 24/192 and will report.
Best,
Robert
you said '...My version of the metallics occurs even when cplay just continues playing and moves to the next track - then (at random) metallics can occur or not. I restart the track, sometimes once more, sometimes again, rarely again, and sooner or later the track starts without metallics. I could discover no pattern whatsover ...' --------- was very close to my experience before I went 2.5" hd, fanless cooling except ----when metallics were bad (real bad) only a reboot would make them go away. Now that I am into 2.5" hdd, fanless cpu cooling and using sox I get metallics much less frequently and I get the kind that are less than 'real bad' (except for other day). yesterday, with using only keyboard to navigate within a file I got none. Also with sox I have not yet gotten any at auto track changes or loads.
Robert I found out that cpu voltage and speed settings affect my metallics. I am right now at .85 volts (as set in bios) and 157 host clock control. I'm sure your equivalent but very low settings may impact your metallics.
what's very frustrating to me is I can hear more purity @ 150 hcc but I get the problem more frequently plus sometimes I get instabilities (system shuts down). So finding the sweet spot is perhaps different for all of us. One thing though I believe impacts metallics, is power supply purity. I am using the P4 Ryelands cap mod which is great sonically but I'm not sure it makes my sys more 'metallic' resistant. Others with linear external ps P4 feeds may want to comment how it may effect metallics. As I recall you use a linear ps for P4 so maybe it just depends on voltage/speed levels plus whether you use juli@ or not.
one last thing I have a benchmark too but I like juli@ analogue out better. When using Lynx did I understand correctly that you can only go digital out with an AES cable? Is there no i2s out with Lynx?
Theo,
I am at 0.76785 (not sure about the last three digits, can't look it up since I am not at home today) and host clock at 150MHz, that's stable but lower values don't work (have to restart from BIOS). At that values everything was fine with Lynx AES-16.
IMHO power supply purity does not help to cure the metallics - I am on pure battery power with well-sized caps (so I think) for P4 and P24 and got the metallics just the same as when I power everything with the Antec (I have tested that already). Maybe things are different when I power Juli@ externally as suggested by Alfred/sonics, will test in the next few days.
However, my task for the next few weeks is to go from Juli@ via i2s to Buffalo32S and optimize power supply (and possibly also caps and resistors?) in that chain. Since Buffalo32S does upsampling I output 16/44.1. Once that is running fine, I will switch to 24/192 and 24/178 (thanks Jean) and see if the metallics are back and how stable and effortless Juli@ can perform. Will report.
And yes, as far as I know you can only go out digital from Lynx with their breakout cable that provides you with an AES/XLR plug which I ran into the Benchmark DAC1, worked absolutely stable and reliable. I tried both to find any possible way to (i) go out i2s and (ii) power the AES16 externally, but failed with both (probably could find a way for (ii) but (i) is impossible AFAIK, Lynx techs have confirmed this). Otherwise I would happily stay with Lynx without any doubt.
Cheers
Robert
thanks for info.
Ok I am trying not to ever use my mouse again--using keyboard to navigate. I was planning on getting a buffalo32 also but I guess you're saying metallics might interupt that too. I will try juli@ latency at higher values than 48 and see what that does (I know 48 sounds best). does that help alleviate cpu load?
Post a Followup:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2009, Audio Asylum, all rights reserved ]
[ Go Green ]
[ Take5 ]