Warning - potential issue with M-Cubed hFX/AES16

[MahlerFreak]

All,

 

When Chris reviewed the horizontal-case silent server built by EndPCNoise, he noted that the sound quality wasn't quite up to what he'd gotten from the Zalman-cased PC he'd built himself, but couldn't quite put his finger on it, and chalked it up to software/OS setup. I have recently purchased the same horizontal-case machine, and I can be more specific: I think the Lynx AES16 doesn't sound it's best in that case.

 

When I got the server, I first listened to it with the Weiss AFI1 firewire digital breakout box, into my AlphaDAC, through Magico V3s. Superb sound. After a couple of days, I installed my AES16, thinking it would simplify the system. But I was bothered immediately by what seemed like slightly exaggerated vocal sibilance. Thinking I might be imagining it, I spent a great deal of time switching back and forth between Lynx and AFI1. I was not imagining it: the sound through the Lynx had a slight hardness/hashiness in the treble, a bit of confusion to the imaging, and a slight reduction in soundstage, which the AFI1 did not exhibit.

 

Having found some tracks that reliably revealed these issues, I went on a mission to fix the problem. Minimizing startup and processes, Vista vs. XP, ASIO vs. WaveOut vs. WASAPI, new vs. legacy firmware and drivers, Media Monkey vs. J River - none of it made a significant difference. In desperation, I finally installed the Lynx in my son's cheap HP/Vista x64 box. Viola - most of the problem went away.

 

Although it's not possible to say I eliminated every other variable, it would seem most likely that the AES16 is mysteriously unhappy inside the hFX case. That particular physical configuration may be more subject to jitter-inducing EMI, perhaps. There is a ribbon-style breakout harness between the mobo and the PCI mounting location - that may well be acting as a nice EM noise pickup. I don't know what else to speculate at this point.

 

In any event, I'd recommend sticking with the vertical case if you're having one of these built, or at least try to audition both to see if you're bothered by the same thing before you commit.

 

Edit: changed RF to more accurate EM.

 

[WoodsDweller]

... would be heat. I assume you were using a PCI riser card that allowed you to mount the Lynx horizontally? What was it on top of, the CPU? The downside of a passively cooled case is that you may have pockets of heat build-up. Running at a higher temperature may throw some of the components out of their spec-ed range of operation, resulting in some audible distortion.

 

If possible, try running the Lynx in the hFX with the top off and an external fan blowing across it. If your problems go away, it was heat. How to solve it would be another question...

 

EDIT: I'm looking at the photos and it looks like the Lynx would not be over other components. Still, I would suspect excessive heat buildup.

 

[MahlerFreak]

Thanks, heat was a suspect for me, too. But the card was barely even slightly warm to the touch, and the problem was still evident even with a cool, freshly powered-up PC. In fact, the whole unit runs remarkably cool for a fanless design ...

 

BTW, I should amend my post above - the EM noise spectra inside a PC actually encompasses a much broader band of frequencies than just RF per se, so I really mean EMI rather than RFI.

 

 

 

 

[davidR]

So are you saying that the PCI slots aren't mounted directly on the motherboard but instead connected through a ribbon cable? If so your observation could be correct about the noise. You could wrap the ribbon cable in aluminum foil to reduce EMI pickup; just make sure you wrap it very tightly around the cable and it isn't directly touching the motherboard at all as that could cause damage to it. I know a certain company who does this as part of a mod on pro interfaces where a ribbon cable is used to connect the power supply to the circuit board. What brand of power supply is in your PC and what is in your son's?

 

[MahlerFreak]

Yes, there's a ribbon cable between the mobo PCI slot and the remote PCI slot, so the PCI card can mount parallel to the mobo. That's a very good suggestion about using foil (have to make sure it's grounded to chassis though, if it's to be an effective Faraday shield). I'll try that.

 

The power supply for the hFX case is in an external box, connected through a 3 pin XLR-style cable. Don't know the brand of PS inside the box - it may be made by M-Cubed themselves. Either the cable, or the power supply itself, may be another source of EMI. Balanced AC power comes through an Equitech 2Q conditioner connected to a dedicated line, so power itself shouldn't be an issue.

 

Maybe Chris could do a couple of audiophile academies on setting up and tweaking an AES16 and software environment? How about it, Chris?