PC microphone input noise
May 26, 2022 at 11:20 PM Thread Starter Post #1 of 4

FireDragon76

New Head-Fier
Joined
Nov 1, 2017
Posts
32
Likes
2
Location
Orlando
I've had microphone problems on my PC for years. At first, I thought it was a PSU issue, then a room noise issue, now after some testing I think it may be due to something about the computer in general (perhaps the PSU). I tested the same microphones (MEMS headset mics and an electret condenser) on another PC that has a high end MB and power supply, and there's about a 9-10 dB difference in the signal to noise... all presumably electrical (I moved the PC out of the noisier room).

My current power supply is an Antec Neo Eco Gold Zen 500W bought during the early part of the pandemic (summer 2020). It replaced a ten year old Antec 450W. It didn't seem to improve the noise. My motherboard is an ASUS A320. Low end but not junk. The headphone output is decent and that's all I use now, I got rid of the Soundblaster AE-5 because the microphone input was actually worse than my motherboard. I can crank up the MB audio and the background is completely black even at very high volumes I'd never listen to. Which is impressive as MB audio used to be pure junk years ago.

I use NVidia RTX and even it was unable to clean up the audio from the Soundblaster. I love RTX but it's a beast (a significant performance hit in any kind of 3D gaming) so I am trying to use lighter solutions like RNNoise. And RNNoise seems to have particular problems with the noise of my PC (it works perfectly fine on the other machine I tested).

I also have a Rolls preamp and a larger electret microphone, as well as a high end boutique preamp and some dynamic microphones (probably my favorite, but difficult to drive). The line input on my PC works fine with all of them. The Rolls has a fair amount of noise with a dynamic microphone, because it's made of cheap stuff, but it works great with headset electret mics and so I just used the Rolls for a while connected to my headset mics and it did reduce the noise a fair amount. But I'd prefer a simpler solution.
 
Last edited:
May 27, 2022 at 9:15 PM Post #2 of 4
If you have slots on the bottom of the case, flip the power supply over to draw outside air. The put the AE-5 on the bottom pci slot right next to the shielded side of the power supply and far from the video card. The AE-5 can go in any pci slot, not just the short ones.
 
May 27, 2022 at 9:55 PM Post #3 of 4
If you have slots on the bottom of the case, flip the power supply over to draw outside air. The put the AE-5 on the bottom pci slot right next to the shielded side of the power supply and far from the video card. The AE-5 can go in any pci slot, not just the short ones.

I got an old Antec Sonata case, the PSU is at the top of the case.

My motherboard is an Asus A320M-C. There are no lower PCI-E slots that are usable. The lowest slot is a regular PCI (non-express).

I'm guessing the power supply is the weak link because I studied the motherboard and the audio layout looks well thought out and there shouldn't be alot of noise. Probably, it's due to the bias power to the microphone being dirty (and motherboards generally have no space to clean up that sort of thing, from what I gathered talking to folks on youtube).
 
Last edited:
May 29, 2022 at 4:57 AM Post #4 of 4
I got a used Sennheiser SC75 communication headset off eBay to try out, since years ago when I was big into internet PC gaming I used something very similar with my soundcard, and it worked just fine. The microphone is very good! I only need 10 dB of gain to get a very acceptable signal. Performance between my PC and the reference test PC is identical in terms of signal to noise (dynamic range is about 30 dB vs. 10-15 average). So while it's nice to know that my PC can perform better, I am still let with a mystery as to why there's a stark difference with inline mic's performance in terms of SNR. I am guessing it is due to different impedance, or perhaps it is due to a difference in power supply quality (self noise of Realtek chips are too low to be a factor).
 
Last edited:

Users who are viewing this thread

Back
Top