le monde du bruit

April 8, 2007 6:58 PM

RF Filter (literally): I'm using my trusty Sony ICF-2010 shortwave radio (1990 vintage) with a brand-new Edirol R-09 digital recorder. Unfortunately, once I hook up the R-09, it spits up digital noise all over the bands.

In this case, the predominant noise is like a kind of fast chuffing, a trilling, outboard motor-like sound. The sound changes its rhythmic qualities depending on whether the unit is in standby (the chuffing) or in record mode (noise bursts seemingly whenever "packets" of data are being written to the flash card). It gets softer or louder as one touches or handles the unit - at its loudest when the unit is being handled, as the body acts as a conductor. At its absolute quietest, it simply acts as a noise floor which obliterates quieter signals and makes a significant sonic fingerprint on all but the most powerful of broadcasters.

This also happened with my Sony TCD-D7 and my Sony MZR70 Mini-Disc recorder, and not just on this radio: my tank-like Kenwood R-5000 also suffered from the ill effects of these digital recorders. Whether they use moving parts and motors (DAT, MD) or flash cards (MP3), each creates a distinctive brand of hashy noise. What can one implement to stop this egregious leakage?

I've tried all of the following:

- standard, off-the-shelf in-line and snap-on RF chokes;
- making the line extraordinarily long (up to 50') and moving the recording device far away from the radio;
- using a third device (such as a mixing board, analog cassette deck, etc.) as a "middle man" to offset any direct line-in interference;
- using both devices with their respective "wall warts" and with batteries, in combinations;
- calling both Universal Radio and Grove for expert advice.

In all cases, the noise just plows on through. I'm a bit vexed that nobody in the shortwave universe has figured this out by now. I'm not terribly solder-savvy, but I'm willing to learn to find a permanently workable solution. Any bright ideas, hive mind? Specifics, if you can...

posted by mykescipark to technology (6 comments total)

You, my friend, need an optoisolator. Or, perhaps an isolation transformer. Get one now. It will fix your problem and you will want to kiss me. It is understandable, I am quite kissable.
posted by Netzapper at 7:41 PM on April 8, 2007

As a first cut I would try an isolation transformer between the recorder and the radio. I am not sure of one which uses small radio jacks, but you could wire a female and male jack up to something like this Speco unit and experiment with the different output leads. If that fails, you might try adding a capacitor on the radio side on the hot lead (the one that goes to the tip of the plug), something on the order of 0.1uF. I confess that I am guessing a bit here, but the cost to try is super low.

on preview, the opto-isolator would be even better.
posted by caddis at 7:46 PM on April 8, 2007

Yeah, I was going to suggest an opto-isolator also but I can't seem to find any hobby circuits on the net. Here is a magazine with an article about building your own.

There are a lot of ground-loop isolatorproducts for cars though.
posted by robofunk at 8:36 PM on April 8, 2007

Used to be that you could get at least one model of optoisolator at Radioshack (at exorbitant per-unit cost, but you're paying for the convenience). Not sure if they still do, but if they don't then Digikey or Newark ought to have lots. I don't think you really need to DIY one with LEDs and IR receivers; there are tons of commercially manufactured optoisolators in DIP packaging that you can wire up on a breadboard to try.

Just as a quick check -- searching Newark (who I prefer, but that doesn't mean you should), I searched for optocouplers and then refined to those with analog outputs in DIP packages (no SOIC stuff). The Avago HCNR200 (if that link doesn't work just search) looks promising; cost you a few bucks a piece (and I'd buy several in case you fry one), but it has a 1.5MHz bandwidth, more than you'll need, and looks pretty simple to use.

In picking a part, you just want to make sure that it uses a photocell as the "receiver" on the output side. There are a lot of optoisolators made for digital use, where the receiver is a optically-triggered SCR or other switch-like circuit, that aren't designed to handle analog signals.

If you read the part's spec sheet (one for the aforementioned available here, or by just Googling 'HCNR200') it'll show some sample circuits in the application notes. You're probably going to need an opamp to boost the signal on the receiving end back up to something strong enough to be useful for your recorder. A 741 series ought to be fine.

Anyway, it's not a trivial project if you haven't done many electronics projects before, but it's not that hard. Really it's going to mostly be an exercise in reading spec sheets and following their examples on a breadboard, and testing it to make sure it works (do you have access to an oscilloscope?).

Have fun and good luck.
posted by Kadin2048 at 10:40 PM on April 8, 2007

I love you people! Thanks so much... I'll get back to you with clarifications as I try this one out.
posted by mykescipark at 2:37 AM on April 9, 2007

It sounds to me like your recorder is generating induced noise on the audio signal line. Your recorder is generating some RF noise, which is transferred down the cable to the radio recorder. HF radios are particularly vulnerable to the sorts of noise generated by consumer electronics. The noise gets louder when you handle the recorder because you are providing greater coupling to the signal cable. Does the interference happen only when the recorder is connected?
Are you using a shielded cable between the devices? And if so, is the shield really grounded? The cable shield should be grounded to the same ground used by the radio at both ends. If the ends are grounded to different locations, you'll get ground loop interference as well. From my experience, ground loop is almost always 60-cycle hum, but sometimes there is a hiss as well. An isolator, transformer or opto- will reduce ground loop but won't do much for induced noise. Also, keep that audio cable short; long cables start to look like antennas in the HF range.
Noise can also be induced on the power lines, and directly from the recorder to the antenna. If you can, move the antenna farther away from both or use a magnetic loop antenna. I find that if I use the wall-wart that came with my shortwave radio it is absolutely unusable due to the noise coming down the cable. Best reception comes only when I'm using batteries.
The experience I have with EMI/RFI issues tells me that there is almost never a "magic bullet" answer. Trying random solutions without fully understanding the problem first can get very expensive, and it can just end up with frustration. There are many situations where "just add an optoisolator" can insert more noise than you had in the first place. No one in the shortwave universe has solved the problem because each EMI/RFI problem is unique. Everyone gets his own particular set of problems. Look for three parts in the problem: A) the source, B) the path, C) the victim. Every time you have interference, all three of these are present. Figure out each, and then work on how to A) reduce noise at the source, B) remove/minimize the noise traveling down the path and C) prevent noise from affecting the victim. In your case, it's clear that the recorder is the source. The victim might be the radio, or it might be the recorder. The path looks like the cable, but part of the path may be radiated as well.
You might find this book helpful. Good luck.
posted by leapfrog at 5:41 AM on April 9, 2007