How to get rid of mains (?) hum?

[lethargic_man]

I've had the intention for a while of transferring all the music I have for my Hybrid Music System to epicyclic, and then burn it to CD-ROM, before the hardware it runs on (1987-vintage BBC Master and 1986-vintage Music 500) complete packs up. A year ago I had an attempt, but ran into problems with what I thought was mains hum. compilerbitch suggested plugging everything into the same mains socket, so, finally having got around to moving the pedestal housing my stereo, CD player and Beeb across the room (and getting and tacking down longer speaker wire so I could connect the first of these back to the speakers on the side they had come from), I tried this out... and found the situation unimproved. My father suggested cutting the earth contacts on the Music 500 output, as I'd probably got a mains hum loop, but that made no difference either.

The funny thing is that when I simply play the Music 500 through epicyclic, it plays fine (there's a small amount of background noise, about as much as a computer fan); it's only when I try and capture the sound to record it (using Audacity) that the bad noise turns up (71k MP3). I've tried various techniques on Audacity for getting rid of it, but they always result in a degradation of the sound quality (60k MP3); better would be to eliminate the noise from the input. Anybody got any idea why I only get it when I try to capture the audio?

(Also, when I tweak the Alsa volume controls, there's a master control for playback; there's also a PCM one, and increasing that increases the noise dramatically. Increasing the line-in control also increases the noise when listening to straight play-through, but I've a feeling Audacity bypasses this.)

Comments

[compilerbitch.livejournal.com]

The other standard approach is to use an (audio) isolation transformer in line with the signal that you're having trouble with. They are not so thick on the ground these days, unfortunately, though.

[compilerbitch.livejournal.com]

*has idea*

Ah, I think I know what it probably is. The old Yamaha DX7 keyboard synthesizer used to have a similar problem.

Basically, the Music 500 has a very primitive DAC, which doesn't cut off at 20KHz. Most likely, it is giving you a spray of high harmonics that are aliasing against the sample rate of the ADC in your sound card (which itself probably doesn't have a very good low pass filter). Playing the same sound through your sound card won't cause the problem, because the audio is mixed in analogue form so the ADC doesn't play any part.

Try stringing one or more ferrite beads on the audio cable, or using a ferrite ring and winding the cable through it a few times. It's a bit of a crude solution, but it often works wonders in these cases.

(listens to mp3 to check)

Yes, that's definitely it. What you have there is definitely NOT hum, it's aliasing noise.

[lethargic-man.livejournal.com]

Hooray, compilerbitch to the rescue! Now, where might I get a ferrite ring or beads?

[compilerbitch.livejournal.com]

You may be able to get something from Maplin, but failing that, RS Components or Farnell should definitely have something...

(looks)

http://www.maplin.co.uk/Module.aspx?ModuleNo=32792&criteria=ferrite&doy=10m7

Any of these might do it. Maybe get one or two of the smaller ones and a large ring as a just-in-case, last ditch alternative? Basically, if just threading the wire through the ring/bead isn't enough, you just loop it through several times (or lots of times!) until you get enough attenuation. It shouldn't take *that* much.

If this doesn't sort it, then heavier guns may be necessary (e.g. an LC filter network or some such), but I doubt you'll need that.

[lethargic-man.livejournal.com]

My father said he'd cannibalised a ferrite ring from an old monitor and send it down to me, along with some plug adapters I'd asked for. It's taken until now him to get around to it.

I passed the wire through the ring twice, which was as much as the ring's diameter allowed for, and, well, it's a bit better (http://www.michael-grant.me.uk/images/ljtrivia/ferrite.mp3) (compare to the previous sample (http://www.michael-grant.me.uk/images/ljtrivia/raw.mp3)), but not all that much (and that may possibly be due to my volume settings on epicyclic...?).

Reckon a larger ferrite ring would help?

One of the other adapters I asked my father for is one to allow me to output from my stereo's headphone socket (it doesn't have any line-out) into my computer. I thought if I took advantage of my stereo's amplifier, I'd have less background noise, but if the problem's aliasing, that's probably not going to help...

[compilerbitch.livejournal.com]

A larger ring and more tucns would help, yes, but see also my other reply below.

[compilerbitch.livejournal.com]

turns

[compilerbitch.livejournal.com]

What you need is an audio low pass filter. It may be that the Music 500 puts out a lot of crap at just above the audio band, so you may need a steeper filter than you'd get with the ferrite approach. Do try passing it through the stereo -- it is possible that it itself has limited bandwidth, which would actually be of benefit for this purpose.

If none of the easy answers work, I could probably come up with a small circuit that you could build on a bit of Veroboard that would fix the problem.

[lethargic-man.livejournal.com]

Hi. Thanks for all your help on this. To summarise what has happened since our last episode:

My father suggested checking I hadn't got an earth braid strand touching the centre pole of my phono sockets. Investigating, I was able to reduce the noise a bit twiddling the leads, but then managed to break both wires in both plugs. (I'm pretty useless with my hands, really, and am green with motor control envy at this (http://www.michael-grant.me.uk/images/ljtrivia/clapton2.mp3) (seen via ewtikins's blog).)

I then tried playing via the stereo, but that gave me hum (http://www.michael-grant.me.uk/images/ljtrivia/hum.mp3) when not playing from the Music 500 at all! I was also getting distortion in the sound, same as for the Music 500.

At this point, it occurred to me that the common factor was that I was still my father's phono-to-jack converter; so I unplugged that and switched to lead he sent me for going through the stereo, and, hey presto, most of the noise and distortion disappeared, though not all of it (http://www.michael-grant.me.uk/images/ljtrivia/clapton2.mp3) (sample recorded off CD via my stereo). (Part of the problem was also that the half the shielding was missing from the old DIN plug.)

There was then a pause whilst my father knocked together a DIN-to-jack converter, and another pause whilst he knocked together another one after the first one went missing in the post. *gnash*

Finally, I got to try it out today, and it's sadly still distorted (http://www.michael-grant.me.uk/images/ljtrivia/sab.mp3) (though not as bad as before). (The lead is looped twice (in opposite directions) through the ferrite ring; unfortunately with the new wire I can't at present try going through the stereo as I've only one small-to-large jack converter.)

Any suggestions leap out at you from the latest samples, from your sound engineer experience? :o)

[lethargic-man.livejournal.com]

In today's episode, I fitted a second ferrite ring I just got from Maplins, adding a further eleven loops around the ring to the two I had on the old one, and once again the quality is improved, though once again not enough. Here's the new version (http://www.michael-grant.me.uk/images/ljtrivia/two_ferrite_rings.mp3), and here (http://www.michael-grant.me.uk/images/ljtrivia/mexico_from_tape.mp3), for comparison, is a version I recorded to and then played back from tape. (I suspect the noise that sounds like a burglar alarm going off in the background of the latter is what the DAC does to the background hiss on the tape; I can't reduce this because the Music 500 outputs at a low volume, and the volume settings on epicyclic are already at maximum.)

So, where do I go from here? Would adding a third ferrite ring help, do you think, or is it take for that low pass filter you suggested above?

[compilerbitch.livejournal.com]

I did a bit of digging, and it does look like the Music 500 has a powerfully weird DAC design. I saw a spec sheet online that said it sampled at about 700KHz, which worked out at about 47KHz per channel. This is best described as bloody weird -- I suspect they have avoided the need for digitally summing the signal by basically multiplexing the channels into a faster DAC, then assuming that the frequency response of the output would effectively do the summing for them. Yes, this would work, but OMG, the unwanted high frequency harmonics at unrelated frequencies that this would kick out is really not funny. I think you may need to go the heavy weapons route for this one, and use a bit of active electronics. What you need is an analogue 'brick wall' filter, which is really aggressive outside audible frequencies. Happily, you can get chips that do this so you don't have to build a really complicated multi-stage filter from opamps these days. I will upload a circuit diagram for you shortly -- I used the Linear Technology FilterCAD software to design an elliptical filter that cuts of at 18KHz (better than FM radio, which knees at 15KHz). It falls off extremely rapidly down to at least 70db of cut, which should be enough to drop any aliasing below the noise floor. I'll post screen grabs with the circuit diagram and frequency plot shortly.

[compilerbitch.livejournal.com]

Note: the main chip is a Linear Technology 1562 (you may see it in catalogues as an LT1562) -- it seems to be possible to order them online from http://www.linear.com/ directly. The opamp on the output is any generic audio opamp you might want to use, it's not critical. Whatever Maplin has in a blister pack is probably fine.



This is the frequency response. It's typical of an elliptical filter, having a few downward spikes in the stop band, but this shouldn't worry you. It should be at -70db at the Nyquist limit, so it'll do your job I think.

[compilerbitch.livejournal.com]

PS: If you want to build this on Veroboard or some such, you'll probably want the dual in line package version (LTC1562CN).

[lethargic-man.livejournal.com]

This is waaay beyond my ability. I asked my father if he could do it for me and he said yes; after a couple of months of occasional prodding, he said:

"It's not likely to be done in the near future, I'm afraid. The project requires considerable planning and design of a circuit board, which takes a huge time. There are considerations beyond the obvious such as designing and applying a suitable power supply (even if I do use an off-the-shelf unit), designing an enclosure and assembling it, to say nothing of designing and fabricating a PCB with no more than Letraset PCB transfers and copper chloride. I don't think you quite realise what it involves. If you can find an off-the-shelf kit solution, I'd be happy to do that for you as that would take an evening or two.

After half an hour of casting around on the Internet, I began to wonder if I was tackling this the right way. So I applied Lateral Thinking—what I really should have done in the first place (hindsight is, of course, always 20/20)—and tried cranking the sample rate up as high as it could go, in the hope I might capture the high frequency spikes as sampled data rather than aliasing noise, and lo and behold, a clean signal.

I suspect there's more to it than that—I get an almost clean signal at 32 kHz, but an aliased one at 44.1 kHz—but if the hat fits^*...

* As yet, only proven for ten second clips of three tracks; but hopefully it'll extend to the lot.

Thanks for all your help anyway; it was interesting learning about what they did inside the box, and if I hadn't had you to tell me I was listening to aliasing noise rather than mains hum, I might still be at square one on this.

[lethargic-man.livejournal.com]

Scratch my last comment: my solution didn't generalise; I'm still getting distortion on the high notes. I've still got one last trick in my bag, which is to fish out my ferrite rings again, but I don't have high hopes; otherwise it'll be back to pursuing the analogue filtering approach again.

The funny thing is I found a blog (http://blog.joel.co.uk/index.php?itemid=157) whose owner talks about having converted his Music 5000 tracks to MP3s, but didn't mention any problems in so doing. Sadly, he hasn't replied to my question yet. Possibly the situation with the Music 5000 isn't as bad as with the earlier Music 500.

[lethargic-man.livejournal.com]

Yes, this would work, but OMG, the unwanted high frequency harmonics at unrelated frequencies that this would kick out is really not funny.

You'd have thought, staring down the muzzle of the twenty-first century and with digital media already on the scene, it might have occurred to someone that people were going to want to do an ADC on the output, but evidently not... or perhaps they were argued down on grounds of adding to the cost.

I think you may need to go the heavy weapons route for this one, and use a bit of active electronics.

Thanks. I'm absolutely hopeless with my hands—it took me hours to solder the phono plugs back onto their wires after removing them to thread the wires through the ferrite ring, and even then I didn't do a good job, which is why I ended up having to meddle with it again afterwards—but I'll see if I can prevail upon my dad to do it for me, in exchange for CDs of all my AMPLE albums once the job's done. (I can run off copies for you too, if you'd like.)

Thanks once again for all your work and time on this.

all-embracing

[(Anonymous)]

[url=http://www.free-casino-bonus.com/]online casino[/url] [url=http://www.casinovisa.com/how-to-play-online-roulette/]slots[/url] , [url=http://www.concordiaresearch.com/games/roulette/index.html]Casino gratuiti[/url] , [url=http://www.realcazinoz.com/jp]bingo[/url] , [url=http://www.avi.vg/category.php?a=sex4sexx&cid=30]adult games[/url]