Win 7 64 bit and Fireface 800 - digital rubbish in Audio output

[Cirrus]

Just got a new PC to replace an old one at my studio. It took a couple of hours to get the RME Fireface working without major driver conflicts etc, and eventually I've got to the stage where if I turn the RME on after logging into windows it works fine - all ins and outs available, no dropouts in my DAW...

But there's a constant whining, beeping ticking noise in the background. The beeping noise is in time with the mouse and the whine changes pitch depending on what's being shown on the screen. And the noise comes through my speakers even when the Fireface is turned off! I've read through the thread on this forum that points the finger at ground loops and will of course try rearranging wires, plugs etc, but my old PC has no problems with noise when slotted into the same setup.

Anyone have any suggestions? Things to look at? I'm wondering if there's some interference happening within the PC between the firewire bus and the onboard graphics.

[dmills]

It's a straightforward ground loop of the traditional firewire pain in the arse sort.

A bodge that might help is to drop into the bios and set the option to prevent the thing going into C3 power saving state (This reduces the power fluctuations that you hear as boops and buzzes at the cost of some more heat), but the real answer it to break the loop somewhere.

Regards, Dan.

[Exalted Wombat]

Just got a new PC to replace an old one at my studio. It took a couple of hours to get the RME Fireface working without major driver conflicts etc, and eventually I've got to the stage where if I turn the RME on after logging into windows it works fine - all ins and outs available, no dropouts in my DAW...

My FF800 works fine on this Windows 7 PC - no need for the legacy drivers or any other fiddling about - but like you I have to be sure it's powered down when booting, else I (inconsistently) get a BSOD. I've reported this at the RME forum, but just got "no-one else has ever had this problem..." If this is what you're getting, perhaps you'd like to try telling them again?

Your problem does appear to be a ground loop. I don't imagine a studio PC would be a laptop with a problematic power supply, so it should be fixable.

[Hugh Robjohns]

Different computers have different internal grounding arrangements, different hardware and thus produce different noises! This does sound like a common or garden ground loop, so I'd follow the usual rules for removing the loop first and foremost.

That means making sure the computer, speakers and interface mains plugs are all in adjacent sockets of the mains distro board, and that youre using balanced audio cables between interface and speakers. If that doesn't cure the problem then you'll need to start breaking the audio cable screens. The easiest way is with a line transformer box, such as the ART Cleanbox2 or similar. You could alternatively snip the cable screens inside the plugs at the speaker end -- but make sure you label the cables so they dont get used for other purposes by accident.

H

[Cirrus]

Thank you both (edit - all three) for the replies!

Yes, same for me - if I turn the FW800 on before comp then it pretty consistently does a BSOD within 15 seconds of logging in, saying something like "IRQ not less than or equal" along with some incomprehensible code. I'll go and say my hellos on the RME forum.

[Pete Kaine]

"saying something like "IRQ not less than or equal" along with some incomprehensible code."

Sounds like it's still driver related then. Have you updated your chipset driver and turned off the stuff in bios you don't need / is recommended that you do.

Try steps from 7 - 11 on here : http://www.native-instruments.com/en/support/knowledge-base/show/752/windows-7-tuning-tips-for-audio-processing/ and make sure you turn off the stuff in the bios that affects CPU clocking (eist & c states as well as anything that says "downclock to improve thermal temps" or the like)

Oh aye... that "incomprehensible code". Try sticking it Google, first rule of tech support!

[Exalted Wombat]

I suppose I should mention that the computer that does it here is a Scan build, and had the problem from day one. It's an easy work-around, so I didn't fuss too much.

[ef37a]

I have a dim recollection of a Firewire interface earth loop problem that was not treatable by the usual remedies running to quite a few posts here a few years ago?

Some one, can't recall who, said that the problem was in the MOBO set up but was not prepared to give details into the public arena.

Sorry I can't be of any more specific help.

Just a thought. Might be worth checking that all the MOBO fixing screws are present and tight?

Dave.

[Martin Walker]

Ironically I once again mention the perils of digital cables that may result in ground loops that are not as easy to solve in the current SOS August 2014:

www.soundonsound.com/sos/aug14/articles/qa-0814-01.htm

And you are right about that obscure motherboard issue, but I believe there were two:

1. One involved a BIOS setting that the poster wouldn't reveal.

2. The other was a 'ground loop like' problem from a different SOS forum user that turned out to be due (IIRC) to the way the motherboard was grounded via multiple standoffs from the metal case.

Martin

[ef37a]

Ironically I once again mention the perils of digital cables that may result in ground loops that are not as easy to solve in the current SOS August 2014:

<a href="/sos/aug14/articles/qa-0814-01.htm" target="_blank">www.soundonsound.com/sos/aug14/articles ... -01.htm</a>

And you are right about that obscure motherboard issue, but I believe there were two:

1. One involved a BIOS setting that the poster wouldn't reveal.

2. The other was a 'ground loop like' problem from a different SOS forum user that turned out to be due (IIRC) to the way the motherboard was grounded via multiple standoffs from the metal case.

Martin

Indeed! I remember the earth post issue and was why I belatedly mentioned checking mounting screws.

On point 1, I seem to recall that the poster gave a "commercial" reason for non-disclosure?
I thought at the time it was rather unhelpful and smacked of not wanting to lose "a nice little earner"!

Also, do not bother taking a scalpel to a USB cable and stripping off 1cm of screen. Does bugger all to help earth loops!

Dave.

[Martin Walker]

Also, do not bother taking a scalpel to a USB cable and stripping off 1cm of screen. Does bugger all to help earth loops!Dave.

And also prevents the USB device being correctly recognised in some setups

Martin

[ef37a]

Also, do not bother taking a scalpel to a USB cable and stripping off 1cm of screen. Does bugger all to help earth loops!Dave.

And also prevents the USB device being correctly recognised in some setups

Martin

Ah! Had it fixed the earth loop I should have bridged the shield with an nF or so.

Dave.

[Pete Kaine]

My FF800 works fine on this Windows 7 PC - no need for the legacy drivers or any other fiddling about - but like you I have to be sure it's powered down when booting, else I (inconsistently) get a BSOD.

I suppose I should mention that the computer that does it here is a Scan build, and had the problem from day one. It's an easy work-around, so I didn't fuss too much.

I don't recall hearing about this one (excuse me if my memorys vague!). Could you generate or do you still have that bsod error to hand?

I'm trying to think what the variables might be with that and what you've outlined. The two that come to mind would be:

Its being powered off the system buss only and isn't getting enough juice during the boot phase to enumerate properly everytime - have you tried / are you already using a wall wart to power it?
(I can't recall if that model could be buss powered?)

Otherwise I guess when it's turned on in windows it's forcing a driver call in the OS. I would question here if it's pulling a newer/different driver in windows than during startup, but if as you say it's a day one thing then your not likely to have older driver components lodged in the OS anywhere.

I can only assume you've done the obvious firmware/driver update round for the FF800 itself. I'd certainly be keen to see that BSOD as I say, I don't recall coming across this one before.

[Exalted Wombat]

The FF800 is mains powered.

Yes, latest firmware and drivers all round.

This particular computer, a few years old now, was designed to be overclocked. We went back and forth for a while trying to find stable settings. Eventually I gave up and abandoned overclocking entirely, since when it's been acceptably stable - though it's still the only computer I've owned recently that EVER crashes.

This is all quite independent to the start-up crashes when the FF800 was attached and powered. All peripherals were removed while I was trouble-shooting the general stability.

[Exalted Wombat]

Should anyone be interested in trouble-shooting, here's the minidump and associated files.

https://dl.dropboxusercontent.com/u/259 ... nidump.zip

[Cirrus]

Just in case legions of people are following my story with baited breath;

I put the old computer back in place of the new one, same physical setup with leads etc, and it's pretty much totally silent even when I crank up the gain.

I tried the PCI firewire card in the old computer, again still silent.

I plugged the new computer back in and it was a buzzy squealing clicky mess. Tried loads of different power-related bios settings, and while they changed the character of the noise they didn't reduce it.

Started unplugging things from the Motherboard - The front of case USB, audio ports, CD drive, both hard disks, and no change. Still makes noise when the mouse moves or the processor does stuff (it's got a fancy mouse-operated bios, it's voodoo I tell ye!)

So I have to believe that the problem is a grounding issue on the motherboard or power supply since they're the only common factor at this stage. I do note the case is powder coated inside and out, so I wonder if a ground somewhere isn't making a connection. I'll have to disassemble the comp and see I suppose. If that doesn't fix it a new motherboard will have to be on the cards.

Oh well. This is why I got into music I suppose - chasing infuriating technical issues! While I'm sure transformer isolating or using balanced connections would probably help, it'd be an annoying workaround since the RME will at times have 20 cables plugged into it and move between various rooms. Really the only option is to fix the noise if it's to be usable.

[Exalted Wombat]

Yes, there's no point in messing around with power settings. We need to kill this noise, not modify it.

What computer, what motherboard? Look at the mounting holes in the motherboard. Does EVERY one of them hold a screw, going into a metal pillar? Hook a temporary cable between the computer case and the case of the FF800. Double-check the ground connection in both computer and FF mains lead, make sure they are plugged to the same outlet. Unplug everything except power and FW cable from the FF, listen on headphones.

[ef37a]

The powder coating could be an issue. The coaters sometimes do not plug the holes!

Check the simple things first. Is there solid continuity from IEC earth to case? (might have to scratch it!)

MOBO mounting holes generally have a solder pad top and bottom. Check that these are low R to chassis and low to associated screw . Also check that any obvious earth points on cards etc go to chassis.

Note that a DMM will read about 0.2-0.4 Ohms for a short. Sharp probes!

Might be worth trying another PSU?

Dave.

[DC-Choppah]

And the noise comes through my speakers even when the Fireface is turned off!

Perhaps you are running the motherboard audio device to your main mixer? It may be noisy.
Mine is. You can 'hear' the mouse and video display. I guess it is only for driving cheap PC speakers. So I always mute that channel when working in the DAW.

[Exalted Wombat]

And the noise comes through my speakers even when the Fireface is turned off!

I've just notived that statement too! Presumably the speakers are fed from the Fireface? What about when the FF is turned off AND its power cable is pulled out (not just switched off)?

[Exalted Wombat]

Cirrus - I don't see your report on the RME user forum yet? Shall we get this moving?

http://www.rme-audio.de/forum/index.php

[Cirrus]

http://www.rme-audio.de/forum/viewtopic.php?id=20348 here we are, not had much in the way of a reply...

[Pete Kaine]

Should anyone be interested in trouble-shooting, here's the minidump and associated files.

https://dl.dropboxusercontent.com/u/25954546/minidump.zip

Thanks, yes I do recall some of the back and forth now from previously.

The error code produced in that BSOD is for a out of memory address issue by a hardware driver trying to access it. The returned error is "hal.dll+23DA" (for background on the HAL see : http://en.wikipedia.org/wiki/Microsoft_Windows_library_files#Hal.dll ) which would imply something on the hardware side has a driver that doesn't wish to play nicely with the OS build. It however doesn't move us any further forward as it doesn't indicate which piece of hardware that might be causing the unlaying problem.

It's possible that this has become a known issue in the years since and which ever driver is responsible may have been fixed now. A full round of updates might be worthwhile if it isn't too disruptive at this point?

[Cirrus]

A follow up - I took the computer home and did some probing with my multimeter.

The case is entirely powder coated or painted, and it seems the various panels were powder coated before assembly. The result is that none of the metal case is grounded at all. So the motherboard is entirely grounding through one power cable to the psu, and the faceplate/firewire jacks on my pci card aren't grounded at all - I suspect they're supposed to ground through the back panel of the case.

I did a round robin of the other three computers in my house and they all have grounded cases and pci card faceplates.

I bet that's the problem - one noisy ground path for the whole system and a floating firewire card means once I plug a firewire cable between it and the grounded fireface loads of digital noise gunk results.

So I guess the options are either try to ground the case, which is going to be a pain I think since every panel is isolated from every other panel, or create my own grounding scheme with random bunches of wires, or find a different case which is bare metal on the inside. Ironically the higher priced cases that look good and promise low noise operation are all powder coated inside and out, so it's proving hard to find one that inspires much confidence.

[ef37a]

A follow up - I took the computer home and did some probing with my multimeter.

The case is entirely powder coated or painted, and it seems the various panels were powder coated before assembly. The result is that none of the metal case is grounded at all. So the motherboard is entirely grounding through one power cable to the psu, and the faceplate/firewire jacks on my pci card aren't grounded at all - I suspect they're supposed to ground through the back panel of the case.

I did a round robin of the other three computers in my house and they all have grounded cases and pci card faceplates.

I bet that's the problem - one noisy ground path for the whole system and a floating firewire card means once I plug a firewire cable between it and the grounded fireface loads of digital noise gunk results.

So I guess the options are either try to ground the case, which is going to be a pain I think since every panel is isolated from every other panel, or create my own grounding scheme with random bunches of wires, or find a different case which is bare metal on the inside. Ironically the higher priced cases that look good and promise low noise operation are all powder coated inside and out, so it's proving hard to find one that inspires much confidence.

Tying all the panels together should not be that hard? A drill and some self tapping screws.

The MOBO fixing posts are more of a problem. The MOBO will have to come out and the tops of the posts scraped to get bare metal.

This all reminds me of an immense cockup at the network factory.... One of the (lesser!) design guys came up with a cheap way to build a 2mtr rack cab using the new M5 "Racksnap" product. Then someone pointed out that every panel was now isolated from every other and network cabs MUST be earth bonded!

Guess who got the job of making up the eyelet earth straps and drilling the cab!

Dave.