Mixing anti-hum into hum. Doable?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

gnd

Well-known member
Joined
Jan 24, 2006
Messages
285
Hi.
24bit is cool, but is a curse too, because hum that was before hidden in noise, now sticks out.

I'm thinking about making a black-box, which would output 50Hz and 150Hz (50Hz rectified) from mains, but at low level, and mix that into mixer, to cancel out low level hum that is present.

I noticed that hum changes depending on units inserted, settings of units, and even changes with time in a day. So why not just mix into master some ANTI-HUM, to cancel out hum of equipment. This could be adjusted as needed, just before hitting REC on final mix record.

There would be outputs for 50Hz, both in phase and out of phase. And rectified 50Hz with filtered harmonics above 150Hz, again in phase and out of phase. There would be just two pots on front panel, one for each frequency. Pot at middle would give zero contribution, turn to left for out of phase, and to right to in phase. Turn more for more level. It would be fed into one channel in mixer. Or maybe stereo version, with additional pan pots for each freqency, going into stereo input on mixer.

What do you think of this idea? Was it ever implemented? Maybe I should patent it? :wink: Or is it just stupid idea, which would not work?

gnd
 
You can't do it once it's recorded...

Also 50Hz rectified is 100Hz, not 150.

Plus the spectral content will be different compared to 50Hz rectified and then ripple-remaindered through a PSU for example...

Not so simple as you might think.

Keith
 
Hi gnd,

as Keith says this will be a lot trickier than it seems. Why not use the energy you would expend on this to get to the bottom of your hum problems? Lots of us record in 24 bit day in day out without a trace of hiss, hum or buzz. Of course I could pretend I did all sorts of special stuff to achieve that but maybe I just got lucky :green:

cheers,
Ruairi
 
[quote author="ruairioflaherty"]Hi gnd,

as Keith says this will be a lot trickier than it seems. Why not use the energy you would expend on this to get to the bottom of your hum problems? Lots of us record in 24 bit day in day out without a trace of hiss, hum or buzz. Of course I could pretend I did all sorts of special stuff to achieve that but maybe I just got lucky :green:

cheers,
Ruairi[/quote]

Yes, I also cannot hear any hum in my mixes, of course. But that damn 160+dB range spectrum analyzer is driving me crazy. Before I had it set up, my life was so much simpler. What you hear is what you get. But these digital gizmos are hell. Now I can see even what I cannot hear. :?

What drives me crazy is that all is fine, and then I just plug another unit in, and here we go again. There is no end to it! With every new unit I can observe another variation of those nice two ripples at 50Hz and 150Hz. And if I bypass something, again something is there, just a bit different. I just wish I'd have a dial to simply dial them out. Ehh.... Maybe best to just forget about the idea, probably waste of energy, like you say.

Best that I just reduce range on spectrum analyzer to 120dB.... :?

gnd
 
Here is the thread where causes and solutions were discussed:

http://www.groupdiy.com/index.php?topic=22066
 
[quote author="gnd"]
Yes, I also cannot hear any hum in my mixes, of course. But that damn 160+dB range spectrum analyzer is driving me crazy. Before I had it set up, my life was so much simpler. What you hear is what you get. But these digital gizmos are hell. Now I can see even what I cannot hear. :? [/quote]


Sounds like it's time to turn off all of your visual "crutches" and start mixing with your ears. Unless, of course, you're going to include a free spectrum analyzer to everybody who buys every record that you sell.
 
> Was it ever implemented? Maybe I should patent it?

It is about as old as lamp-socket powered radios.

Nearly half the 5-tube radios had a hum-bucker winding on OT.

Much more elaborate schemes have been used. Radiotron touches on a few.

The core problem is that in-phase and out-phase is not enough unless you have a very specific and simple hum source (cheap filtering in the same box). Everything is phase-shifted. You need continuously variable phase. It must be adjusted for EACH harmonic... they all have different phase shift. And if you mix two tracks with different hum sources, the cancellation needed will change every time you move a slider.

Hum should be cured at the source. That's not so very difficult at the 90dB level. It is doable near 120dB, though you may have to get ALL the wall-power supplies out of the room (the Chicago classical FM station did that once). For 160dB, I think you need to move to the north pole, live on batteries and fishsticks. And you might just pick up a mix of Canadian 60Hz and Russian 50Hz. Maybe Mars?

Put masking tape across the bottom of your display.
 
[quote author="PRR"]> Was it ever implemented? Maybe I should patent it?

It is about as old as lamp-socket powered radios.

Nearly half the 5-tube radios had a hum-bucker winding on OT.

Much more elaborate schemes have been used. Radiotron touches on a few.

The core problem is that in-phase and out-phase is not enough unless you have a very specific and simple hum source (cheap filtering in the same box). Everything is phase-shifted. You need continuously variable phase. It must be adjusted for EACH harmonic... they all have different phase shift. And if you mix two tracks with different hum sources, the cancellation needed will change every time you move a slider.

Hum should be cured at the source. That's not so very difficult at the 90dB level. It is doable near 120dB, though you may have to get ALL the wall-power supplies out of the room (the Chicago classical FM station did that once). For 160dB, I think you need to move to the north pole, live on batteries and fishsticks. And you might just pick up a mix of Canadian 60Hz and Russian 50Hz. Maybe Mars?

Put masking tape across the bottom of your display.[/quote]

:grin: :grin: :grin: :grin: :grin: :grin: :grin:

Thnx PRR. It is good to see that I was reaching to far.
Some careful planning and distribution of wall-power supplies and maybe separate power distribution for switching-power supplies and linear-power supplies may be actually enough.

And masking tape is not even needed, I can simply adjust sensitivity range of analyzer. :grin:

Thnx
gnd
 
I have seen some tracking filters in the past that would lock onto the power frequency and trackit and filter out up to several harmonics, but they were not 100% effective.
Several years ago, Roland brought out a realtime digital rack mount box made for professional audio which really did work and eliminated all powerline artifacts without affecting the desired sound. I don't recall what it was called, and it went for several thousand dollars. I did see one on ebay last year, and it went for a couple hundred, and I thought about bidding just to have it on hand. They are rare, but you might want to keep an eye out for one.

Jim Zuehsow
 
Right, in the real world it doesn't work out very well. The hum comes at you in various and sundry combinations of harmonics, which change all day long. If you do get lucky and manage to null it out, you may well find that five minutes later you are adding hum instead of cancelling it out.

As suggested above, better bet to use the energy and time to finding and fixing your problem(s) at the source.
 
Old-old CoolEdit 2000 had a "noise reduction" tool. You showed it several seconds of "silence" contaminated with noise. It measured that and did a "profile". Then you ran that profile against your actual track. It was easy to over-do, end up with a warbly watery queer sound. It was nearly useless against impulse noise, even the foot-clomps as the choir sings while climbing the stair. But used with taste, I have done well on blower rumble, truck rumble, and buzz. I even made a significant improvement on a camcorder recording of solo marimba. The camcorder and internal mike were too far from stage, the camcorder motor was as loud as the decays. It was never intended to be used, just go in the scrapbook. But the High Quality Master Audio tape got lost in the mail, and this was the only fallback.

CE2K was bought by Adobe, I think they call it Audition, and I don't know if that feature was carried-over. (Adobe changed the program so much that I refuse to look.)

And it really is a 15-bit tool. CE2K does higher bitdepths, and maybe the NR tool works down around the 20th bit. But clearly it was tuned for reducing "argh" to "ehh", not to get inaudible hum invisible on a pointlessly hyper analysis.

What are you recording that you can even SEE 140dB down? In live recording of what they like to call good music, my low frequency noise floor is rarely 70dB down. All that blower rumble. Even with room-size mufflers under the stage, the random noise makes any fix-pitch noise invisible (unless I correlate for an hour). I do see 110+dB in the 2K-4K octave, and more on one mad percussion piece, but most good music captures well on bad old analog tape. I find that digital does not give me better recording, it just makes me lazy.

And I dare you to find three non-studio non-fanatic listening spaces with >100dB S/N. An awful lot of listening is done at less than 40dB S/N, between modest speakers and high room noise. When I translate a mix for PC-Speaker listening, it's all about the top 30dB.
 
Well, the problem with a noisy recording is that the noise gets very audible after some processing. If you send your vocals through a chain like LA2A-1176-Pultec and strive for an in-your-face-sound you're bound to get the -80dB noise/hum into an area where it can be heard easily even with listening equipment as problematic as a low-resolution mp3 through cheap headphones. I'd rather keep the noise than use some artefact-producing noise reduction tool, but hum really is another matter. I think I'll have to get back to my Pultec and 1176 and try to lower the hum, since it is still slightly above the noisefloor...
 
Old-old CoolEdit 2000 had a "noise reduction" tool. You showed it several seconds

Yes, I know Cool, and used it extensively. Now I use Audition. Same NR (or better) is in audition. But there are realtime options on the market, like Waves. Cool, and now Audition, are both 32bit applications. Actually, Cool was one of the first to be able to work with 32bit files, even before there was IEEE standard for 32bit files, so Cool had support for something like 5 different 32bit formats (diferent bit orderings).

Such thing surely works, but I'd preffer to not use it no master, because it removes ambience too.

What are you recording that you can even SEE 140dB down?

I think we have some misunderstanding here. For example, white noise measuring 0dB (full range) on peak meter, is measuring at some -33dB on spectrum analyzer in Cool Edit Pro. Peak meter measures power of entire signal, not amplitude of its components. It is different for pink of brown noise. Single sinusoidal signal, say 440Hz, measures the same on peak meter and spectral analyzer. But as soon more frequency components are added, peak meter reads more than level of separate components on analyzer.

It also depends on ballistics of peak meter and analyzer, and possibly depends on more things like windowind and such....

Like on my mix analyzer, full range sin 440Hz signal shows at -8dB of analyzer, and full range noise shows at -50dB.

So, while noise range shows 0dB at peak meter, it is at -50dB on analyzer.

And 140dB range on analyzer means with white noise 90dB on peak meter. While S/N ratio is 90dB, noise components are at -140dB on analyzer. It is actually 90dB SNR. So we are not talking batteries and north pole here, but actually just normal real life figures of 90dB SNR.

But I agree, 90dB peak meter S/N ratio is still just fine. But I'd like to improve that a bit, since my sound card has over 100dB SNR.

gnd
 
[quote author="David Kulka"]Right, in the real world it doesn't work out very well. The hum comes at you in various and sundry combinations of harmonics, which change all day long. If you do get lucky and manage to null it out, you may well find that five minutes later you are adding hum instead of cancelling it out.

As suggested above, better bet to use the energy and time to finding and fixing your problem(s) at the source.[/quote]

You are right, I agree. Slight mistake, and I will be actually adding more hum into mix than taking it out. And thats really likely to happen, especially when I get into it, I may easily forget about tweaking that anti-hum button. It will have to be cured at source.

thnx

gnd
 
[quote author="living sounds"]Well, the problem with a noisy recording is that the noise gets very audible after some processing. If you send your vocals through a chain like LA2A-1176-Pultec and strive for an in-your-face-sound you're bound to get the -80dB noise/hum into an area where it can be heard easily even with listening equipment as problematic as a low-resolution mp3 through cheap headphones. I'd rather keep the noise than use some artefact-producing noise reduction tool, but hum really is another matter. I think I'll have to get back to my Pultec and 1176 and try to lower the hum, since it is still slightly above the noisefloor...[/quote]

Yes, same with full mix. Having noise at -90dB in final mix, without post-processing, and possibly some 3-6 dB headroom.... It is easily boosted 10-12dB at mastering, if not more :? , and there you are with 75dB SNR, or even 70dB SNR final product, which is quite audible, I'd say.

...
 
With most softwares
for short segemnet, and and if you have some hum only recording at the beginning of the track, you can cut and paste phase reverse segment's in sync with the hum wich is clearly visible on large zoom scale.

Of course it's a pita for long and multiple tracks, but i guess thats this is the basic operation of software plugs.
 
If you need such low noise level for 80 dB compression it is logical to use a noise gate as well. :cool:
 

Latest posts

Back
Top