Studio Mains Supplies (Part 4)
In the fourth part of the series, Ben Duncan deals with interference and possible remedies for it.
Amongst other things, Ben Duncan here tells you how to cope with interference from secret military establishments.
In the July issue, we looked briefly at the parts inside standard, mass produced filters (Figure 7a) and noted that the attenuation on symmetrical (differential) interference wasn't too hot (Figure 2, July). The reason for this is that the inductors on affordable filters are invariably bi-filar wound, ie. the live and neutral turns are connected back to back. This is done to substantially cancel the inductance seen by the legitimate live and neutral currents, as consumed by the equipment. This makes much smaller and relatively cheaper inductors feasible, albeit at the expense of the symmetric interference defences. It's not quite so satisfactory, because most of the work is left to the three capacitors alone. One manufacturer's justification for this short cut is that symmetric interference should (in theory) be heavily attenuated by the power supply's reservoir capacitors and regulator unless there's any re-radiation, ie. the symmetrical RFI can spread by radiation once it's past the filter, and inside, close up to PCBs. At the same time, this facet is ironically the real reason for not worrying unduly about the relatively poor symmetrical rejection of bog-standard filters. Figure 1 shows why.
The classic source of symmetrical RFI is a bad contact, close by. In this instance, it's a heater with a naff thermostat, on the other side of the control room wall, that's in the room next door. Although the mains filter in the console's power unit only attenuates the line interference by 30 to 40dB, the capacitors and associated regulator circuitry might contribute around another 20dB - in the symmetrical mode only, you understand. But let's forget the sums. This could be much improved, but it's rather pointless unless your studio's interconnects are 100% balanced and RFI protected. That's because the strongest symmetric mains interference (ie. from next door) correlates with the strongest RF field, as received in the studio. This in turn means the contribution from down the mains will probably be swamped out by widespread RF breakthrough in the audio leads, especially those unbalanced interconnects, and speaker connections - more on this one anon.
This leaves us with two immediate remedies. One is to spend the next two weeks slipping little ferrite beads over every wire inside every audio plug, to hastily invest in a rank of balancing boxes, and converting to twisted pair cables. That sorts out the audio; then the symmetrical line noise assumes prominence, so it's time to consider more exotic filters - or link two cheap ones in series.
If this sounds tedious (let alone expensive), then harken to the alternative remedy, called prevention. This is to say that symmetrical attenuation on line filters should be enough to mop up any residual perturbations, provided you've taken steps to suppress your own, local noise sources. To refresh, local in this context means anything on the same ring main or supply, or in the same room, or any of the possible six adjacent rooms including those above and below.
An oddly effective tool for locating nearby sources of RFI (Radio Frequency Interference) is mum's tranny. It's not a vehicle - it's a portable radio. Switch to medium (AM) or long wave, and tune about to an empty channel. Swing it about, listen for the tell-tale crackling/whining, and you'll soon be using the set as a direction finder. The source will be received most strongly when it's broadside to the ferrite aerial rod (Figure 2). To find out which direction this component faces, you may need to take the back off. The condensed aerial is usually recognisable as a dark grey bar or rod, with two or three coils of silky (litz) wire, often garnished with a dollop of wax. As you turn the set around, in a 360° sphere, the level of interference should vary, to represent the relative direction (Figure 2b). One tip: strong RFI sends the receiver's front-end into limiting (not unlike an audio channel), but setting the volume low will emphasise any changes, as our ears are more sensitive to this at low SPLs. Ghetto blasters may also be used. On these, the aerial rod may be external, and encased in plastic, but the direction for a broadside will be quite apparent. Of course, any set used for tracking down RFI should be run on batteries, and wholly disconnected from the mains supply. I hope the reason is obvious.
Voltage spikes, up to 10kV, occur randomly on all mains supplies at some time or other. The highest voltages are the result of lightning, and do not occur very often and last only a very short time, typically 1µS. For this reason, although transients are an extreme (!) form of RF interference, you won't notice them directly; speakers and ears are incapable of registering a 1µS click. But you may hear the side effects, the sharp bangs and clicks as the equipment is mortally wounded... Back to transient voltages: smaller spikes of around 1kV are caused by the switching of power lines and large capacitative and inductive loads. These generally have longer durations than lightning strikes - the pulse is wider (say 1 mS) and may repeat several times (Figure 5). They're also much more common, especially when you're parked next to industrial buildings, full of machine tools and electrically operated processing. Older washing machines are a particularly notorious source of voltage transients in the home, but not more recent models, a by-product of their better RFI suppression. In other words, suppression with filters and snubbers at the source can mop up RFI and spikes all in one go.
Back in the studio, the cheapest and most effective weapon against transients is the voltage dependent resistor (VDR), primarily because of its simplicity of construction selection and application. VDRs made for transient voltage protection duties have an extremely limited VI characteristics (Figure 6). You can think of it as a fast attacking limiter - on the mains! When an excessive voltage is applied, the normally very high resistance (relative to 240 volts) drops swiftly to a few ohms or less. Provided the source impedance of the transient is high (looking back up the mains, towards the source of the voltage spike), the VDR then acts like the bottom arm of a voltage dividing network and, the magnitude of the transient is drastically reduced.
Figure 7 shows how the effectiveness of the limiting action varies with the source impedance of the transient. Essentially, this correlates with the distance and gauge of wire, between the audio equipment and the cause of all the disruption. It follows that decrepit washing machines should be treated with caution, particularly if just round the corner.
As implied by the arrows in Figure 2, once the strongest orientation has been fixed, you're left with two opposite directions to explore, for instance, the rooms below and above as the radio is limited to giving you an axis to concentrate on. For more probing investigations, in a large building say, pinpointing the interference source is quicker with two people on the job, two portable radios, and a comms link, eg: Maxon 'Speakeasy'.
Summing up on line filtering, here are three levels of attack, according to your level of operation, your piggy-bank, and how much grief you've previously experienced.
1) Bare minimum: fit internal filters to all digital gear (synths, micros, FX, digital recorders etc). An undedicated plug-in filter with integral 13 Amp plug and socket is cost effective at this stage: it can be plugged in where and whenever it's needed most, like when you borrow a CMC24 console for a weekend. Typical parts cost £20 to £40.
2) Extensive: add internal line filters to analogue tape machines, the console, and FX rack. Typical parts cost £30 to £50.
3) Comprehensive: set up a clean technical supply, a ring or spur type, with overall line filtering at entry to studio/control room. Typical rewiring parts cost £100 to £700.
And just to prove that interference isn't a new phenomena, the list Figure 3 comes from a 1939 radio handbook, and should act to get you checking out some possible interference sources you might not have thought about before, like discharge lamps (street lights), lifts and neon signs.
What happens if you simply can't solve you RFI problem because it's someone else's mess? Over the years, a variety of parliamentary acts have been passed to prohibit the operation of equipment which produces anti-social RFI. Alas, because Government deals universally in archaic technology from a previous era, when in this instance, wireless and television were the sole example of sensitive electronics in the home, the mechanisms of this instrument are blunt: no provision is made for interference to a host of modern equipment, including (you guessed) all types of audio system.
Today, the Government department responsible for regulating the radio spectrum is the Department of Trade and Industry. However, actual investigation is carried out by British Telecom, who operate a thankfully very limited number of detector vans (that are, of course, occasionally used for rather nefarious purposes)! But assuming your TV licence is up to date, they can be of help to your studio.
To apply for investigation, you should first make absolutely sure that the interference isn't on your own premises (see Studio Mains Supplies 1, HSR June 85). You can then initiate a major archaeological retrieval: ask your local post office for the form 'Good radio and TV reception' (Figure 4). They don't publicize it much, and once you've brushed the dust off it, you'll find that the form has no reference to interference experienced in recording studios; but it's 99% likely that the interference you're suffering will be manifest on one or other of the TV and radio channels listed. It follows that an application may be made, above board, on this basis.
Figure 4 shows the form filled in for a dire case of motor hash, picked up in a West country home studio despite stringent measures. It came from a cycloconvertor (which changes 50Hz mains up to 60Hz, for US machine tools) in an adjacent woodwork shop. In this instance, the RFI was quite audible on a long wave radio, brought into the studio. Short wave (SW) is another valuable wavelength for picking up and noting down as much hash as possible, whilst complying with the spirit of the form, ie. broadcast interference. Naturally, it's technically illegal to listen in to short wave radio, according to the draconian powers of the 1911 Wireless Telegraphy Act, but if asked, just tell them you're into the BBC World Service.
It's not such a good idea to skip the two week recording period advised on the form (look closely at Figure 4). For a start, it's good discipline. Going back a paragraph, the studio owner discovered the woodwork factory was responsible by careful recording and detective work. He began by noting down the times when the noise occurred over several days. At first, it seemed quite random, but by and large, the effect was inside normal working hours, therefore pointing to a commercial establishment.
When the detector van shows up, two, four, six or even 27 weeks later, you can base all field investigation on the blemishment of your impassioned long/medium/short wave radio listening, a la Holgar Czukay. What happens next may or may not solve your difficulty. For example, the source may be a 'secret' military or government installation: there's a lot of these about. The engineer in the BT van won't say very much, if this is the case. Again BT may trace the miscreant, but be unable to threaten them into getting it suppressed. Of course, if it's serious, and affects a number of other people, the DTI may take steps themselves. Failing this, you may choose to approach the owner, and ask whether he or she will allow you to undertake suppression, or pay for this to be done: £25 and a little diplomacy is far cheaper and gets results quicker than litigation.
One part of the cosmic jigsaw puzzle which actually fits is that the largest transients tend to have the highest frequency, so their source impedance is high, and are therefore attenuated most (Figure 8). As a result, spikes of 10kV and 1kV might end up limited to around the same level, say 600 volts. This sounds high, but remember we're discussing electrical peak levels. Relative to this, the maximum allowable peak voltage of the GEGB juice is (240 volts x 1.414) + 6% = 360 volts. For a period of a few microseconds or tens of microseconds, many power supplies will absorb the difference. For example, all BS rated mains transformers are flash tested to withstand a much higher voltage, and a short lived 600 volt spike won't harm them, though an 1800 volt one (say) may well do.
As we've seen, the effectiveness of a VDR is very dependent on the source impedance of the transient, which is something of an unknown factor. Therefore the degree of protection afforded by a VDR is unquantifiable, however, like an insurance policy, you'll be rewarded by skipping potential grief at a later stage. Besides, we can make the degree of attenuation more certain by adding some line inductance which sets a minimum value on the source impedance. This is already to hand, of course, if the gear has been fitted up with a filter. All we need do is place the VDR at the transformer side of the filter, so the incoming transient encounters the filter's inductors first. VDRs are also most effective when soldered close up to the protected device, in this instance the primary of the equipment's mains transformer. Even lead lengths of 1" can waste potential attenuation action relative to the fastest spikes, because 1" of wire can have significant impedance relative to 0.1 ohms (the VDR's surge resistance) and a 1 mS pulse. Some pundits disagree: they claim that certain transient voltages travel down the wiring at the speed of light, therefore the VDR should be displaced back upstream by a distance commensurate with its reaction time, say a few hundred nanoseconds. This corresponds with the transient's wavefront travelling a few feet, so to be doubly sure, you may want to mount a second VDR inside the mains plug end of your best digital gadgetry. At around 75p each, this won't wreck the bank.
Next month we'll put VDRs into practice, and take the spade into the garden to dig for a good earth.
Feature by Ben Duncan
Previous article in this issue:
Next article in this issue:
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!