Noise interference issues

[Tom]

I'm picking up a lot of noise with my sound cards and I'd like to see if I can do anything to help reduce any of the electronic interference that's causing the noise levels to be so bad.  I don't know what the normal...or acceptable noise levels should be, but when I monitor VU levels with absolutely no sound being input from any device, I'm seeing the meters registering as high as -70 db.  I see this basically in every one of my systems, no matter which sound card is installed.  My newest Turtle Beach Riviera is the worst, though.  As soon as I enable its INPUT jack, the noise levels shoot way up.   I've ruled out cabling problems, as I'm still getting these high noise levels when no cables are connected to a sound card's input jack.  I see this with my Audigy, Santa Cruz, and Riviera sound cards.  What's really strange is, as I move my mouse around, the noise levels actually DEcrease slightly, but only with my optical mouse.  I wonder if my KVM switch might be part of the cause for some interence.

Does anybody have any tips on reducing computer interference and sound card noise levels?

Thanks!

[Ari]

I get the same problem. I feel my recordings are too noisy, and since I don't like using noise reduction filters or normalizing the volume, I stay with relatively low quality recordings. Tips are definitely welcome.

[Tom]

Ari, do you get more noise from your SW60XG than other sound cards?  Either I've never noticed before, or my '60' is starting to go bad, but it's definitely the noisiest MIDI device I have -- even worse than the MT-32.  I can rid most of the noise by turning Voice Cancel 'ON', and lowering the INPUT to '0'.  But then, I lose its input.   I guess I really don't need it so I can live with that. 

I don't like using Noise Reduction techniques, either, but will on very quiet recordings.  Usually, there's enough going on within the song file that I don't hear any noise.  It's basically just those really quiet passages, and only when I really crank up the volume.  But regardless of output volume levels, my software shows there's noise present whenever the INPUT line level is more than about 3/4's up...on all of my sound cards.

[Alistair]

Few questions/comments.

Firstly, I use Noise Reduction with CEP 2.1, and it doesn't do anything unsavoury, not that I can detect, anyway. Just removes background noise.

Secondly, what do you guys use to measure your noise? Not sure what a VU level is. I'd like to test my setup and provide a comparison so maybe I can find a solution. I know having too many crossed, overlapped cables doesn't help me, but I don't really have any solution for that.

- Alistair

[Ari]

Ari, do you get more noise from your SW60XG than other sound cards?  Either I've never noticed before, or my '60' is starting to go bad, but it's definitely the noisiest MIDI device I have -- even worse than the MT-32.  I can rid most of the noise by turning Voice Cancel 'ON', and lowering the INPUT to '0'.  But then, I lose its input.   I guess I really don't need it so I can live with that. 

I don't like using Noise Reduction techniques, either, but will on very quiet recordings.  Usually, there's enough going on within the song file that I don't hear any noise.  It's basically just those really quiet passages, and only when I really crank up the volume.  But regardless of output volume levels, my software shows there's noise present whenever the INPUT line level is more than about 3/4's up...on all of my sound cards.

I don't know if it's noisier, but it's volume is considerably lower compared to my SCB-55 and LAPC-I, which means that when I try to raise the volume externally, there's a lot more background noise.

For digital recording, I use a program called GoldWave. It's got a VU meter too, which shows a lot of noise coming from the line-in even when nothing is playing.
I've thought of trying to get better audio cables, but haven't gotten around to look for any.

[Tom]

Most digital recording software (including CE, CE2K, and CEP) allows monitoring the recording level.  In CEP, press F10, or go to OPTIONS and select, "Monitor Record Level".  Set the recording range (right-click on the level meters in CEP and select 120db as the range), then watch the meters.  Ideally, they should not register very high without any input. 

If your SC-8850 is connected to your soundcard's LINE IN, turn its power on and see what happens with the meters.  In my case, I'm watching the levels with a mixer connected to the soundcard's LINE IN, and can then control various external inputs from the mixer.  Try it with nothing connected to the LINE IN, and adjust the LINE IN level to about 90% ... see what you get for noise levels.

The reason I don't like to use noise reduction filters (such as those available in CE, CE2K, and CEP) is that no matter how you slice it, NR alters the original recording to some degree; by either cutting off bottom or top end frequencies.  CEP is pretty good, but you still lose some of the original output whenever you apply a NR filter.  Back in the day, with louder soundcards, NR filters always produced a 'flanger' effect that anoyed the crap out of me.  I haven't noticed that with newer soundcards and later releases of digital recording software.

I've got some gold-tipped cables and it makes no noticable difference ... not that I can hear, and nothing changes on the VU's when these cables are used.  I hate to admit it, but my Sound Blaster Audigy seems to be the quietest card of the bunch.

[Alistair]

But, if it distorts the audio in other ways, I'd rather noise reduct

As Zemus showed me in the Digital Audio section, try using that setup he listed to reduce noise. It works, and I can't hear any difference between before and after (except, the noise is gone).

I don't think, when you use 4 sound modules/cards and a sound card, that you can really have some "zero noise" setup, or whatever.

Ari said:

I don't know if it's noisier, but it's volume is considerably lower compared to my SCB-55 and LAPC-I, which means that when I try to raise the volume externally, there's a lot more background noise.

Yeah, my MU100R is way quieter than my Roland gear. I wonder why?


I'll check out the VU readings with CEP later tonight, gotta study, for now. Maybe I'd rather live in blissful ignorance, though

- Alistair

[Tom]

Whenever I DO use noise reduction within Cool Edit 2000 or Cool Edit Pro (I prefer 2000 over Pro; nicer control, same features), I typically don't use the same NR filter settings more than once, since the noise isn't a constant; it's variable, depending on its source.  Using the same NR filter settings seems like you may be unintentionally removing data that you don't need to remove.  The noise level is different depending on what's connected to my INPUT port, which device is actually powered ON, and probably the proximity of other electronic devices around my computer at that given time.  So, I record the noise before I playback any music, and create a filter from the 'noise' recording.  That seems to be the best way to apply a noise filter.  It worked very well with several of the AL EMMO and QFG2VGA recordings.

I've been able to reduce most of the noise by turning OFF Input Line levels on unnecessesary sources.  I'm wondering if a USB sound card might eliminate all noise issues, as I'm positive most noise is caused by interference inside the PC environment.

[Marten]

Have you tried ground loop isolation cables?  Those can eliminate some noise, and shouldn't distort the audio (at least, I'm personally unable to hear any difference aside from removing undesireable hiss/hum).

[Alistair]

Firstly, Marten's point:

Have you tried ground loop isolation cables?  Those can eliminate some noise, and shouldn't distort the audio (at least, I'm personally unable to hear any difference aside from removing undesireable hiss/hum).

I once tried this, based on Marten's recommendations, and I have to say, it did more harm than good (assuming I bought the right kind of cables, mind you! ).

Basically, it distorted the output like crazy (especially with certain sounds), and put me in a situation where I'd recorded the majority of a soundtrack (SQ5, as it happens) and it was useless, as was the balancing I'd done to mix it, since the ground loop cables changed the way things sounded. Bleh.

Anyone want some ground loop isolator cables, on an unrelated note?

Whenever I DO use noise reduction within Cool Edit 2000 or Cool Edit Pro (I prefer 2000 over Pro; nicer control, same features), I typically don't use the same NR filter settings more than once, since the noise isn't a constant; it's variable, depending on its source.  Using the same NR filter settings seems like you may be unintentionally removing data that you don't need to remove.  The noise level is different depending on what's connected to my INPUT port, which device is actually powered ON, and probably the proximity of other electronic devices around my computer at that given time.  So, I record the noise before I playback any music, and create a filter from the 'noise' recording.  That seems to be the best way to apply a noise filter.  It worked very well with several of the AL EMMO and QFG2VGA recordings.

Sometimes, I get the feeling you're really patronising with your comments, Tom- probably unintentionally, but still..

Obviously, that's what I've done- I haven't selected part of the track as the basis of the filter. I typically let the recording run 10-15 seconds after the file has finished playing, and then use that as the basis for the filter (plus, Zemus' settings I put in as well). (Been doing it for years.)

Not sure was CE2K is like. CEP is perfect for my needs, also, I'm really used to it.

I've been able to reduce most of the noise by turning OFF Input Line levels on unnecessesary sources.  I'm wondering if a USB sound card might eliminate all noise issues, as I'm positive most noise is caused by interference inside the PC environment.

I ALWAYS disable ANY audio source I'm not using. To take the Santa Cruz for example, back when I used that, I muted everything bar aux and line in (and master volume). The rest, muted and also sliders lowered to 0% (trust me, it's just as important as muting them).

I use the FireWire version of the FireWire and USB card, the M-Audio Audiophile. It's simply awesome, better range than the Cruz (more highs and lows). At 130 US bucks, it isn't that expensive either, at least, not in my view when we're talking big quality differences. (It doesn't have a microphone port, or anything like that, so there's very limited scope for 'unnecessary sources'- it just has Wave, SW Synth, and CD Audio, I think.)
Especially when crappy 'top line' SB cards retail at over 200 bucks.

- Alistair

[Alistair]

Hey, I just changed to 120 dB range. My noise level is about half the meter (almost 60 dB), which I presume is bad.

The M-Audio doesn't have any way of changing line in, it's automatic. I have to tweak my module's volume settings to not max it out, if you get my drift.

Goodness knows what could be done to reduce it.

Just tried opening a memory intensive program, and that added to the noise. Maybe it's because my fan its working overdrive (90 degrees for a week now).

Either way, Zemus' method for noise reduction works, in my book. Maybe I should post some before and afters.

- Alistair

[Tom]

I used to get -60db readings using my old AWE32.  When the Riviera's LINE IN port is cranked to its max, I get about -72.  Shutting the LINE IN down (or OFF completely) on the Riviera has basically solved the noise issue.  It's now about -80 to -90db with my MT-32/SC-8820/SW60XG connected to the Santa Cruz LINE IN from my cheap Geneva mixer.  This eleminates my need to use any noise reduction filters.  I'm sure there's something within my PC that's causing the Riviera to pick up interference; the card's specs are too good to be this loud.  I might try switching the card to another PC.

[Alistair]

Like I said, I presume my PC being on for a while is causing some of it. Also, the FireWire draws its' power from the PC, which maybe contributes?

I hate to think how loud it is when 4 modules are all on. I'll have to get back to you.. :O

- Alistair

[Alistair]

To update.

I'm having some weird issues, too. The sound modules don't create nearly as much interference as my sound card and PC. With just the sound card on, I get about 72 dB, and about 60 with 4 sound modules on. Still, that's 48 decibels of noise with just the Audiophile, which is a long way from its' 90+ SNR ratio, etc.

It also seems to me that when I load stuff into the memory, the noise (monitored by CEP 2.1) suddenly drops for a second, similar to Tom's initial query.

But yeah. I'm pretty much getting the exact same noise readings with my sound card on and no sound modules on, so I wonder what the deal is. I presume our computers produce a lot of interference. I'm guessing with Tom's Riviera/et al, it's because the card is in the machine, and with my Audiophile, it's because I have to sit the device on the computer tower, still getting lots o' noise.

I'll try purchasing a USB Audiophile (which is supposed to be less noisy, and I also have long USB cables so I can sit it away from the PC tower), and seeing what happens with that.

EDIT: On ground loop isolators, the one I bought that was no good was this:
http://www.crutchfield.com/S-rZTwJP0zRBY/cgi-bin/ProdView.asp?g=890&I=127SNI1

Marten, is this the kind you were talking about? I may have not been thinking or asked enough questions when I bought this one.

- Alistair

[Marten]

It isn't the same, Alistair, but I don't see any reason why it would work differently than the ones I've bought.

Time for a tiny bit of electrical systems explanation:

The purpose behind a ground loop isolator is to eliminate 50Hz or 60Hz "hum" frequencies that can leak through the ground line when connecting two pieces of electronics.  Explanation of what a "ground loop" is, here:  http://en.wikipedia.org/wiki/Ground_loop_(electricity) .  The isolator uses a transformer - basically, two coils held close to each other, where the electrons in one coil become "excited" to mimic the source coil - to break the direct current connection between the cables.  See:  http://en.wikipedia.org/wiki/Transformer and http://en.wikipedia.org/wiki/Isolation_transformer.  For a description of how an isolation transformer is used in audio, see http://www.epanorama.net/circuits/audio_isolator.html .

Note that this equipment won't eliminate other noises - it'll only solve grounding issues.  Theoretically, the AC signal is duplicated precisely, so the only signal eliminated is the undesirable direct current signal.  But, as the epanorama article indicates, there may be a bit of phase shift between the high and low frequencies.  (Any electrical engineer will tell you that "theoretically" is like saying "it doesn't really work this way.")

The ground loop isolator I've used successfully is similar to the Crutchfield model, and is also categorized under "car a/v" at Radio Shack:  http://www.radioshack.com/product/index.jsp?productId=2062214

I've never noticed any distortion or sound loss from the cable, but that doesn't mean it isn't happening and I might be missing it.  Most people reviewing the Radio Shack product reported no sound loss, yet at least two claimed it "eliminated the vocals".  One claimed that the signal was reduced, requiring him to raise the volume... which reintroduced the hum.

All I can say to that is:  "Your mileage may vary."  It's also possible that the people using the product and reporting problems either received defective product, or hooked theirs up backwards.  It does matter which direction you pump the audio through the transformer (this is mentioned at the bottom of the epanorama article).

[Alistair]

It's almost impossible to hook it up backwards- it's just RCA connections.

Also, what are those two small wires that stick out of the main filter?

Finally: On car isolators, read this paragraph:

Automotive "Signal Isolators". This method is a "last resort" means of isolating line-level interconnections. One source for these devices is your local Radio Shack (#270-054, $14.99). WARNING: because it uses transformers for coupling, there is the possibility of signal degradation, loss of frequency range, and increase of signal distortion. If you decide to go this route, consider installing one on every device feeding your main preamp/receiver. NOTE: these units are not suitable for coupling RCA composite video feeds from video equipment!

(Source: Ken Simmons)

What do you read into that, Marten?

I wonder if a 'normal' ground loop cable or the ones Ken recommends on his site about this kind of thing (which I skimmed over because it's INTENSE):
http://smr-home-theatre.org/Ground-Loops/

Regards,
- Alistair

[Alistair]

Some advice I got from hydrogen Audio- and Tom, you should do this too (sub in whatever sound card you use where it mentions Audiophile).

I'm going to try it tomorrow.

NOTE: The following instructions are for Cool Edit, possibly Cool Edit Pro (I only have CEP 2 so I don't know).

A system noise floor recording , capturing just the contributions of computer and soundcard, is done by recording from the analogue inputs with nothing connected to the input jacks. Recording at 32 bit, 44.1kHz is a good choice. Record for 12 to 15 seconds. Select a few seconds of that recording to measure, say seconds 7 through 10.

Ideally the levels will be the same across the entire recording, but it is not unusual, while the clock is settling down in the first second or so, for the levels to vary rapidly. You can easily find out by clicking on the vertical zoom-in until the wave form is clearly visible on screen. Anyway, I always avoid getting too close to the endpoint when taking measurements.

Analyze/Statistics gives some useful numbers. If you were measuring a nice clean sine wave, you would see Peak Amplitude and RMS Average just about equal, computed relative to a sine wave, or Peak about 3dB higher than RMS, computed relative to a square wave. A noise floor recording, not being a nice sine wave, should have a much larger difference between Peak and RMS values; the RMS values are usually more indicative of what you will hear (or not hear). My PCI soundcard noise floor is about -86dB peak, -98dB RMS, relative to a 0dBfs square wave. Your soundcard should be about the same. If noise level are much higher, something is wrong.

With that same three seconds of the recording selected, run Analyze/Frequency Analysis. Click on Scan to build the graph. I like to use FFT Size 16384 on just about everything; it gives enough detail for most of my purposes. Always running the same analysis makes comparisons between different samples more meaningful.

If you have any hum problems they will show up as significant peaks on a curve that otherwise is unlikely to have any major variations. Its not unusual for little boxes like the FW Audiophile, using a wall wart supply, to show some 60 (or 50) Hz, along with the first few harmonics of same. These really stand out on the Frequency Analysis graph, but they probably do not represent enough energy to be heard, or to make any significant difference in your recordings.

If you see large hum peaks, and the Statistics figures are much higher than I indicated they should be, you can use Transform/Filters/Notch Filter to remove the hum. Leave the Frequency Analysis window open while you adjust the Attenuation and Notch Width to eliminate those peaks.

Run the Notch Filter, scan your selection to see the results, click on Undo to try again. When the hum peaks are flattened, run Analyze Statistics again to see how much you have to gain by eliminating the hum at its source. If all you have is the average dribble from the wall wart supply, the numbers, pre and post filtering, will not change much.

You can then proceed to making noise floor recordings of your audio source equipment by plugging its outputs into the soundcard inputs. Record while the sound equipment is running at idle. This will measure how much noise they add to the basic system. If you have ground loop problems, those should show up as greatly increased peaks on the Frequency Analysis graph of the new recording.

Regards,
- Alistair

[Tom]

Sounds interesting ... I'll give it a try.  Thanks.

[MichalN]

Just a few notes...

First, AFAIK gold-tipped cables do not improve sound quality so much as prevent the connection from going bad. If metal oxides build up on the connector, bad things will happen to audio.

Second, most, if not all, PCs create ungodly amounts of EMI. Sometimes the interference is so bad that you can actually hear it (no, I don't know why, I just know what I hear). All things considered, the inside of a PC is probably one of the worst places in the world to put hi-fi audio equipment. Just FYI, AC97 and HDA uses physically separate controller/codec implementation precisely because if the codec was part of the southbridge (where the controller is), the noise would be terrible. The codec is typically placed in the corner of the mobo furthest from the CPU to minimize the interference.

Third, pro audio equipment uses balanced connections to eliminate interference. I'm not aware of any PC audio gear that supports balanced connection (maybe someone knows better?); what's more, balanced cables won't do a thing if it's the sound card itself picking up noise and not the cable.

From what you guys are saying about the noise being dependent on what the PC is doing, it sounds like the interference is generated by the computer itself. Out of curiosity, are you running CRTs or flat panels? Does turning the monitor off make any difference? I assume you have already tried turning off other electrical devices in your room/house to see if they might be causing interference. I would guess (though I have no data) that everything else being equal, an external sound card in a shielded casing should do better than an internal sound card.

As a side note, if it's really the PC generating the noise, I wonder how the CPU type affects this. Did anyone ever try to make some measurements? If I ever get some spare time, I could probably try plugging a SB Live! into various boxes from 486 to Core 2 Duo and seeing if the noise is any different.