Weird issue - A Sound 'skip' Every 20 Seconds caused by too many HDDs?

[Dangleberry]

Greetings.

I'd like to start by thanking all the great reviewers on JG for your thoroughly entertaining and passionate reviews. Please, never stop.

If someone could please help figure out this weird issue, t'would be much appreciated.

So, specs wise I have a 4 year old Silverstone OP700, powering the following 24x7:
- Gigabyte GA-EP45-UD3 motherboard (30W-40W?)
- C2D E8400 OC'd to 3.74 Ghz (65W TDP + a bit more?)
- 2x 2GB DDR2 RAM
- Geforce 9800 GTX (140W Max)
- Basic 4 Port RAID PCI card
- 9x 3.5" hard drives
- 7 fans (excluding PSU and Video card) - 4x120mm (7W each) 2x140mm (4-5W each), 1x80mm.
- 48W 4 channel fan controller for the 4x120s.
- Creative 5.1 soundcard
- LG DVDRW drive (IDE)

Almost always, 3 of the 4 120mm fans are intentionally off at startup (1 is for CPU) to reduce draw, and then turned on 10 seconds later.

So, this problem has been occuring for close to 2 years now:

If there are 8 or more hard drives running, exactly every 20 seconds, any media that plays, regardless of format, has a very slight hickup/skip in the sound.
Video isn't affected and neither is mouse movement or anything else, it never causes a BSOD either, (it usually takes 2 days max before I get tired of the skipping). For 4+ years now it still gives 90+ days of uptime with no problems other than windows being a bit sluggish.

If I unplug 2 of the 9 hard drives, things are fine.

Please correct me if I'm wrong but this could possibly be caused by drawing too much power with more than 7 hard drives - but what could be occurring every 20 seconds in a PSU to cause the sound glitch?

Reading JG's review of the OP650, it's not a bad PSU, so I'd expect the 700W to hopefully be in roughly the same class:
http://www.jonnyguru.com/modules.php...=Story&reid=39

Assuming, worst case scenario that each hard drive takes 3x their operating wattage (10W) to spin up, so around 30W, 9 hard drives would draw a max of 270W at spin up and then, assuming they can't idle, would soon drop down to 90W. But a relatively instant 270W draw should just be a slap in the face for a 700W PSU and not a knockout uppercut.. Right?

Although for clarification, if someone could please advise if the startup power draw from hard drives is all drawn from the 12V rail or a 5V rail or a combination of the two, that would be helpful in understanding things.

Also, it makes no difference whether the RAID card is used or not, as there are 8 SATA ports on the motherboard - and there's no staggering feature on either of them.

Applying the same rule to all of the fans, spinning up would probably be a bit over 110W, and then drop to around 40W. (So with drives and fans we're up to around 400W?)

I read on the Backblaze blog (they build storage based racks), that the 5V rail is what hard drives typically demand the most power from, so having a high wattage on that is important when running lots of them.

Another thing I noticed was that the OP700 has quite a generous +5V max wattage, even compared to some 1000W+ PSUs:

OP700:
+3.3V +5V +12V +5VSB -12V -5V
Max.(Amps) 24A 30A 58A 4.0A 0.5A /
Min.(Amps) 0A 1A 3A 0.1A 0A /
Range (%) +/-3% +/-3% +/-3% +/-5% +/-10% /

combined +3.3, +5V 180W
combined +12V 696W/ 58A

So that brings me to wondering what the flying puck is causing the sound glitch..

Any ideas?

[Zap]

I would expect the system to shut down if there isn't enough power.

Does it matter which controller the drives that you unplug are on?

Have you tried just unplugging the data cables, so that the system thinks the drives aren't there but they are still drawing power (if they still stay spinning)?

Have you considered moving some drives out of the system? Either replace smaller capacity units with higher capacity, or using an external drive bay. This 4 bay USB 3.0 and eSATA box is $70 right now after coupon code EMCXPWT44 (exp.6/19) and a $20 rebate. It even gets bundled with a free USB 3.0 PCIe card.

I'm wondering if it is a DPC latency problem and not a power problem?

[blivit4]

It has nothing to do with the PS, and audio latency problems with the MB.

[Jon Gerow]

It's a bus width problem. Not a power problem.

Using a sound card instead of the on board sound *might* solve the "problem", but you can't be too sure.

[Dangleberry]

Thank you all for the responses.

Zap, I'm quite sure it doesn't matter as I recall at least having only 1 drive on the extra onboard controller and 2 drives off the RAID card. I'll test this properly again though to be sure.

I'm also pretty sure I've tried not having SATA cables plugged into 4 drives but left the power plugged in, and the problem still occurred but I'm definitely going to do this test again - I've had the issue for a while and every few months I'll think of another way to test, but I just don't remember them all clearly. :/

Unfortunately I'm kind of a step ahead of you on the external enclosure idea, that link seems to be of a rebranded Vantec HX4 of which I have 3, and they're full. (btw, that little 80mm fan at the top doesn't do enough to cool the drives in that thing, I'm going to do a basic mod and replace the 4 rear drive brackets with a 120mm fan and figure out how to power it via USB or something).

Just read up about DPC latency, interesting.


Blivit4 & Jonny, are your suggestions related - so can bus width issues affect DPC latency?
Also, out of 10, how bad is it for me to not stagger say, 10 drives on an OP700? (I'm planning to RAID1 (with a better RAID card) my OS drive so +1)

Jonny, I apologise, I also have a little soundblaster 5.1 sound card, I always forget about it as it's right under the video card and even with the case 2 ft away from me and open I have to duck my head down to the keyboard to see it. :S
I'll add it to the specs right now. (and the DVDRW drive, but that's all of it - I promise :S)

And, running 24x7, would an OP700 also deteriorate by that rule of about 5% per year so, it should be down to around a 550W PSU by now?

Another thing I thought of trying is using 2 of these external hard drive power adapters to supply power to 2 drives and then plugging the SATA into the motherboard and see if the problem still occurs.
I believe you guys that it really shouldn't be (isn't?) the PSU but I'd like to be 100% sure if that's ok :P

Will do some testing.

[McSteel]

Is your soundcard PCI or PCI-E?
Are you willing to try ASIO drivers and an ASIO-capable player? Personally, I love XMPlay, it's just so versatile and well-written.
Mentioning ASIO because it has latency compensation as a standard feature, and it works flawlessly.
Do your HDDs use/support power management/saving features? You could use HDDScan to experiment with that.

## EDIT ##
Didn't see the post above as I had ~30 tabs open and only came back here to respond 15 minutes ago, without refreshing...

Anyway, the "power rating deterioration with time" is complete bollocks. The only thing that does degrade in any meaningful quantity over time is electrolyte in the capacitors.
And even then they're either still working as intended (though a little less vigorously) or they fail (bulge/leak). You might see an increase in ripple/noise and a slight worsening of the voltage regulation, but ultimately, your PSU will not change it's maximum output over time (and certainly not by 5% per year!)

The problem lies elsewhere, and probably should be corrected with software.

Oh, and, are you using the last/latest BIOS available for that motherboard? Are you willing to experiment with beta/modded ones? Perhaps there are some out there that have a newer firmware for the HDD controller...

[Kougar]

That's a lot of hardware... don't suppose you have a power meter to check the draw from the outlet? Depending on OC settings and program load it should easily do over 500w draw, possibly even more.

It's a bus width problem. Not a power problem.

Using a sound card instead of the on board sound *might* solve the "problem", but you can't be too sure.

How is it a bus width problem? Do you mean between the southbridge and northbridge? That's a 2GB/s link and even running all the mechanical drives full out with transfers should not saturate that, but I'd assume most of the drives are idle at that? SATA is a serial point-to-point topology so I'm not sure what else ya might be referring to for a bottleneck.

Edit: Didn't see the above posts. Since your mention it I found the DPC Latency checker to be fairly useful in the past, have you tried it yet?

I'd also reiterate what MCSteel asked, is the card PCI or PCIe based? JonnyGuru may be on the money about a PCI bus bandwidth problem if you're using a PCI soundcard, whoops.

[mariush]

It's probably a DPC issue...

Download and run the tool on this page: http://www.thesycon.de/deu/latency_check.shtml

If it's around the yellow line or above it, that explains the stuttering.

A solution would be to stop whatever uses interrupts or other things that affect DPC... some fan/temperature monitoring softwares can cause this issue. If you use them, see if things change if you disable them.

[Jon Gerow]

If it's on board sound, which is what I assumed (sorry, nothing said otherwise), it's using the PCI bus.

[mariush]

If it's on board sound, which is what I assumed (sorry, nothing said otherwise), it's using the PCI bus.

Not always. Some of the latest Intel "chipsets" are completely PCI free.

Some motherboard manufacturers add pci to pci-e x1 bridges for onboard devices and to add 1-2 pci slots for backwards support.