[tech] An interesting problem

[DeZmond]

Hey everyone,

This one is really just to pick the brains of some of the more technically inclined, because I have a problem that's been bugging me for two years and one week now... my laptop.

Those of you playing CS in the glory days will remember that listening to my microphone was an experience like being stabbed with a small nail several times over... and it's even worse trying to record something.

The problem seems to be that at precisely regular intervals of about ~1.5 seconds there is a small noise - analysing a waveform taken by a blank recording shows it is nearly identical every time and in a definite pattern. I've been trying to work out the cause for some time now, and I believe it to be one of the following:

1) The disturbance is caused by disk writes. At set intervals data is read/written to the HD and the movement of the drive causes a noise.
Likelihood: medium

2) There is a problem in the interaction between the CPU and somewhere on the motherboard, most likely the north bridge, and the CPU cannot process all the data, or is causing some other problem.
Likelihood: high

The reason I say 2) is likely is due to the fact that under intense system loads (eg a high-powered graphics benchmark, or even sometimes using the DivX web player to view a SD-res movie) the system performance seems to correspond to the pattern I've noticed with the sound recordings - the performance seems to pause for a split second at the same intervals. Coincidence? You tell me.

Anyway, I would really really really appreciate any thoughts whatsoever on this one - it's been doing my head in for a long time now and I want to just sort it out. I'll fetch any information that you think could be of use, download and run any programs, benchmarks or whatever. If it is of use I'll even set up a blank recording with the anomaly showing.

Naturally rep will be given to any helpers and my eternal gratitude

Let's go bug squashing.

 

[Pestcontrol]

Ah, that's a nice challenge to figure out.

Question 1: Is it analog or digital noise? If you open the recorded noise with a hex editor, do you see the same values repeating? Then it's digital. If it just looks like a similar waveform everytime but isn't an exact replication, then it's analog.

For digital noise there's a hope of fixing it, because it's likely some wierd driver/software/phase of moon problem, for analog noise it's just bad laptop design.

To test if it's disk noise, load some linux livecd suitable for recording and see if the problem persists. If you can run some burnin program like prime95 in the background and see what difference that makes.

 

[DeZmond]

I'm guessing it's analog, for checking the waveform shows slight variations, which increase when the main fan is active. As you say I'm not sure if there's much that can be done to fix it, it's just rather irritating and I wish it didn't happen .

 

[Pestcontrol]

Poor sound is a common problem in laptops, because everything is so tightly packed there is much more interference. Your best hope is an external USB soundcard or a USB headset.

 

[DeZmond]

I think I'll just end up getting a USB headset and shall see if that cures the problem.

What do you think of my second theory that there's a possible problem with CPU/motherboard bandwidth causing a similar issue with general system performance?

Another side note is that I'll soon see if the RAM is involved since I'm getting a new 1GB module when I return.

 

[Pestcontrol]

Are you experiencing short lockups under load? That's most likely some driver issue. It could be that during these lockups certain lines are loaded (an interrupt channel for example) that interferes with your sound. That may also explain why the flaw went unnoticed during the laptop's design: It takes crappy drivers to expose it.

I don't think it's a bandwidth issue, lack of bandwidth gradually reduces system performance (by increasing latency).

You can use perfmon to monitor interrupts. Since my perfmon is currently magically broken and it's been a while since i used the tool, i cannot tell you how or whether you can monitor specific interrupts and so isolate the driver that gives you trouble. Sorry.