An EMC Problem Investigated

 A customer for whom I repair bits & pieces from his ferry boats asked me to look into interference to his ship-to-shore radio. The engineer had discovered the interference originated in an air conditioning unit. This being a Daikin FVXS50FV1B. I later was presented with the power supply board from their ship's restaurant dishwasher with a similar problem, again resulting from cutting costs and buying a commercial unit... see later.

I looked around the Daikin website and their various published documents and the only information I found regarding EMC was a note in a document that merely said, "Conforming to Safety and EMC rules and regulations".

I e-mailed the company asking for more information and I'm presently waiting for a response. See later.

In the meantime I decided to carry out some basic tests to check on the interference, if any, produced by the power supply board from the air conditioner. The customer had said that the interference was produced as soon as the equipment was plugged in.

 Below is a picture of the power supply circuit board. At the top left corner you'll see the components used for suppression of conducted interference from the switching power supply to the mains. There are a couple of blue capacitors (C11 & C12) grounding the live and neutral connections to mains and safety earth, a common mode choke (L301) a dark blue capacitor across the mains input (C6), and a choke in the live feed (L307). There's also a few other parts which are in the mains circuit, a small neon lamp and one or two thermistors/protection devices.

You'll note that the colour coding is basically that of the USA, white being neutral, black live and green protective earth/ground. Red looks like it's a second earth connection.

Connected to the mains filter is a bridge rectifier (DD301) then a high voltage reservoir capacitor, a chopper chip and an HF transformer (T301).

The chopper chip is an STR-L472 which is what's called a Quasi-Resonant device. The circuit will oscillate at a frequency determined by the characteristics of the HF transformer; primarily the inductance and self-capacitance of the primary winding. Because the chopper device is generating a 400 volt square wave there will be loads of odd harmonics of the basic oscillation frequency. Although the QR circuits provide better efficiency than those used in earlier chopper power supplies, the higher basic frequency is capable of a much greater range of harmonics, hence extra potential interference; in this case to marine radio.

 First, I'll explain the monitoring set-up. As the customer had already indicated the interfering frequencies it was a case of finding these so I could easily repeat testing if I decided to carry out improvements to the circuit board. I connected the board to a mains supply with no load on the DC outputs other than what's on the board. I wound a coil of about a dozen turns around the mains lead and to this connected my spectrum analyser. I fitted a rod aerial of about 40 cm to the IC-R7000 receiver and loosely wound the mains cable along its length. For final measurements I set the analyser to scan over a small band 164.5MHz to 165MHz with its bandwidth set to 100Hz. The interfering signal, selected from many, was about 164.92Mz and this was about -109dBm with the noise floor at around -132dBm.

 Below is shown pickup from the coil wound around the mains lead with mains power to the circuit board switched off. There's lots of peaks and, as most of these are precisely 25KHz, apart it looks like interference from another chopper power supply is present.

 Below is shown pickup from the coil around the mains lead with the power switched on.

The selected signal is under marker "1" and the vertical graduations are 50KHz apart. To the left there's another signal from the power supply board. Count three vertical lines left and it's the peak sitting at around -115dBm, another three vertical lines and you'll see another at about -118dBm. This makes the repetition rate at around 145KHz.

 Below are similar sections of the above two traces side by side (left with power supply off then right, on). The frequency in question is about 164.92MHz and the VHF receiver below is tuned to this signal. As there's lots of other noises in the band one has to check that what you're seeing originates from the test board. I was able to tune the receiver to dozens of signals confirmed to be emanting from the Daikin power supply board by the simple expedient of lightly pressing my finger on the HF transformer. This caused the frequency to vary whereas other signals were stable and showed no sign of moving.

 The VHF receiver is set to SSB because this provides a better indication for signals generated by the power supply board (ie. you can hear the beat note change when touching the chopper transformer). The signals appear to be relatively stable CW carriers and appear at regular intervals across the bands being monitored. For example I was able to detect the following ... 180.926MHz, 180.784MHz, 180.641MHz, 180.498MHz etc etc with strong examples at 171.1MHz, 176.069MHz, 177.64MHz and 179.356MHz. They're spaced at approximately 140KhZ.

In addition to the discrete interfering CW signals; when the circuit board is powered up the background noise level across the marine band on the receiver doubles in strength.

 What next? Well I'm awaiting the manufacturer's comments. My customer has already fitted a spare board with exactly the same results so it's not a faulty board. Maybe it's not all connected up as its supposed to be? I can't say, however the power supply board is fitted in a plastic lined metal box which can be split into two. The two halves have gaps around them so will not be good enough for screening higher frequencies so there may be some radiated interference and there are no ferrite chokes fitted to any of the wiring.

I looked at the chopper chip specification and this suggests in a roundabout way to add a small ferrite bead to leg 6 of the chip. This isn't easily done in the example I have because the chip is soldered too close the the printed circuit board.

Looking at the filter components, I guess the common mode choke is a bit on the small side.. maybe a larger one would attenuate higher frequencies better? Maybe ferrite rings fitted to the wiring might help? There's a control circuit daughter board fitted to the outside of the case and this will probably carry interfering signals so maybe a pair of ferrite rings can be fitted to its 9-way and 3-way cables?

Another thing I've seen before is the use of conductive tape stuck over slots in metal screens. This practice is often to be found in laptop computers and computer monitors.

 Before tackling the problem of providing extra suppression I heard back from Daikin. They supplied a certificate of conformance referring to EN60335-2-40 and 2004/108/EC. I read these documents and found that they excluded certain environments, in particular aircraft and ships. I'd already found the EMC requirements for the latter (EN301 843) and it specifically covered the marine VHF band. This is treated with priority, requiring manufacturers to meet a more stringent interference level. So, the air conditioner circuit board is not as suitable for marine use as other similar equipments whose manufacturers design to meet marine requirements.

Next, I decided to see what the fundamental interference signal looked like. I already knew it was around 145KHz so I connected up my half finished R1155 and tuned in the signal from the Daikin circuit board. I found a very powerful carrier wave at 124.8KHz together with a powerful sideband a few KHz higher. The sideband consisted of a rough modulation buzz based on 100Hz.

Many years ago I worked for a company, Plessey, that made cryptographic equipment. This sort of stuff takes plain language information, generally teletype data, and messes it around before sending it to the outside world. The idea is that anyone picking up the data cannot understand it unless they have a key to decode it. All very well but it's not that easy to build a black box that contains electronic signals without leaking some of these to the output wires. In fact using special amplifying equipment an interceptor can see enough of the original unencrypted information to read it. This means that cryptographic equipment needs to have filters so that what gets out of the box isn't helpful to a interceptor. There are specifications to which manufacturers need to work when building their black boxes known generally as TEMPEST. Click to see a true story.

One can't easily modify a piece of commercial gear to meet TEMPEST but one can use some of the techniques employed.

Read the following: STR-L472 Application Note....2004/108/EC....EN301-843

Next, attempting to reduce the interference levels

Return to Reception