Lawrence has been working on an Industrial (emissions Class A) pump product and has fixed an interesting EMC radiated emissions problem associated with the display panel.
Narrowband emissions caused by harmonics of a digital clock are nothing new when it comes to LCD panels and the signals that drive them. Inadequate filtering of these digital interfaces means that there is plenty of high frequency energy available to excite any parasitic antenna structure that we have created with these interconnected elements.
Normally the source of the emissions are the digital video signals from the CPU to the display. In this case, near field probing showed that the source of the noise was strongest around the display controller itself.
The first radiated emissions scan showed that there were emissions of a 60MHz clock signal present at around 300MHz and 660MHz (green = chamber background, red = measurement)
Here’s the display in question, a Riverdi RVT43 series touch screen for the Human Machine Interface (HMI) of the pump.
Lawrence took a closer look at the PCB with a contact probe and a spectrum analyser to find there was a Winbond QSPI flash running at around 60MHz.
Measuring each of the pins of the display cable (on the quad SMT resistor packs) showed significant levels of 60MHz noise on each trace, with the bottom most ones being the worst.
Looking at the circuit board we can clearly see where the problem lies:
We have a 60MHz clock trace (in red) running immediately adjacent to one of the signal traces leaving the PCB on an unshielded ribbon cable.
This is poor design practice on the part of the display manufacturer.
There is enough coupling between these traces (inductive and capacitive) for noise on the clock trace (source) to appear on the cable signal trace (victim). If we consider the fields around the traces in cross section through the PCB…
There is also the possibility of coupling inside the IC given that the output pins are immediately adjacent.
We can address the problem a number of ways
- Remove the noise at source by changing the PCB = not possible, this is a third party component. Changing the value of the series resistor packs could have worked in this case.
- Detune the antenna structure by changing the relationship of the two PCBs relative to each other = difficult to achieve, not guaranteed to work
- Adding common mode suppression on the cable to reduce the energy of the signal driving the parasitic antenna
Point #3 above was the only practical course of action.
A small ferrite core was placed over the cable and the results speak for themselves (green = before, red = after)
There is no compliance / EMC information / CE marking present on the display. But surely, as a piece of equipment or sub-assembly capable of causing radio interference, this falls under the scope of the EMC Directive and should be CE marked and assessed for EMC?
Yes, it should.
In this case, the host equipment manufacturer is having to deal with the poor hardware implementation by the display manufacturer.
Good work by Lawrence for getting to the bottom of this EMC problem.