So this is the active part of the Solar PV system. The Solis inverter is not too bad, HF switching around 160KHz, and it appears there is also some switching done at 29KHz: NOTE 160KHz that's s not what I call HF technology switching, I suspect multi phase switching of some sort. There are two sets of 8 panels, one set on the main roof which faces west and the other on the pitched garage/utility room roof which faces east. That arrangement does help to maximise the hours of the day when power is generated, Each set of panels is separately fed to the black filter box and then on to the two inputs on the inverter. Both feeds are metered , more to make it look pretty than anything else. WARNING: this type of inverter floats the DC feed from the panels at half mains voltage, so connecting the body of the Schaffner filters to earth cannot be done, otherwise the RCDs will trip as it causes an inbalance between the current flow in live and neutral from the distribution board.
Above is the internal view of the filter box, with the two Schaffner FN 2200-25-33 filters. Each side positive and negative from the two banks of panels has additional Ferrite chokes fitted. This works well . Below is a screen shot of the emissions at the highest panel bank,(West facing nearest the antennas). The spike on the left is 29KHz. These plots were taken with a ferrite clamp on the +ve cable in the loft just under the panel bank .
This is a device that is attached to the grid side of a Solar PV inverter and when the level generated is high enough switches, PWM, power to the immersion heater, hence the free hot water from the sun. AND FREE RFI , lots of it !!!!. As you can see from the plot below
That was taken on an overcast day, when there is bright sunlight the level at 1.8MHz , Top Band,< rises to +10 dBm. Had I been aware how dirty this was it would never have been installed , The company, the manufacturer of this thing know its bad, it was self certified and its only in the installation info for installers that customers will never see , does it state "RFI (EMC) emissions: below -42dBW ". I've worked for UKAS approved EMC test houses , and been a client when working on equipment with many others, I've never seen anything as bad as this. That's in 40 years of working , and nothing that bad since I retired.
I would appear the Solic 200 does not like driving in to a filter, as on a winters sunny day when the system is generation around 1kW it is only showing a mean output of 25v AC when connected to a 3KW immersion heater load, whereas without the filter it would normally indicate an mean 150~200V AC. This indicates the design is not capable of driving a reactive load, maybe when the generated level is above 2KW it might be able to produce a bit more. However, at this level 6mtrs is usable, and I'm not seeing levels greater than -60dBm at 1.8MHz. I have been advised by Ian GM3SEK to try reversing the FN 2060 so the Solic 200 is driving an inductive load, Unfortunately it cant drive any sort of reactive load; It is not capable of functioning with a type 43 Ferrite core with 2 turns, it struggles to function with 1 turn on the same core. This unit can only be described with multiples of mono and duel syllabic Anglo Saxon expletives ! Here is a plot when the system is generating 1.3KW, albeit there is only 25V mean at the immersion heater
Maybe I have been spoilt with the engineers I've worked with during my career in the RF/RF Test equipment industry, I have never worked with anyone who did not take pride and effort in to designing equipment , that would exceed EMC requirements and knowing that designs put forward for type approval would easily pass testing , whether to CE,FCC or NTT standards. Moreover, should there be any doubt that something would be close to only just meeting a standard it would be worked on until there was no doubt it would easily meet the required standards. Just because in the UK the HF spectrum has no real commercial value, all the long haul communications are now done by satellite or fibre cable, does not mean get it close, don't bother with Monte Carlo analysis to cope with component variation, its only for the domestic consumer market. In other countries this level of emissions would not be tolerated. Like most things in the UK lessons will be learnt after the event
I've been totally thrown by this 29KHz being there at night, the inverter must be taking power from the mains at low level light when there is not enough power from the PV panels. This is verified by disconnecting the inverter from the mains, the 29KHz disappears . AND I found that putting ferrite clamps on the cables in the loft makes the interference worse. I had put Wurth 74271211 ferrite clamps on the inverter output , two clamps with one turn of the cable on each, these do attenuate the switching frequencies from the inverter substantially, another 4 have been ordered to be put on the inverter input, either side of the main filter box.
Strangely when the light level drops so does the 29KHz level by about 40dB, and the signal plus harmonics seem far less frequency stable When generating the inverter synchronises its frequency to that of the mains, perhaps this part of the locking system, or an effect of it. The 29KHz does eventually disappear at night, which would suggest the capacitor bank on the inverter is large. Well below shows the expensive Ferrite clamps on the input to the filter box and the output, and below those how much difference it made
I don't think the top ferrite clamps are actually doing anything, so I'll be removing them, when I figure out how to use the Wurth key which does not release the clamps. So I'll probably have to disconnect the incoming feeds in to the filter box and unwind the cable from the clamps, At night time job as there is 300V DC on them , low impedance and a lot of current in sun light. Below with the sun out and generating 2.8KW The noise on 50MHz is high what I thought was progress was just the British weather being cruel. Not as bad as it was , but a lot of money spent to gain a few dB.
So where to go from here, the Solic 200 will be put on remote control , or left turned off. 50MHz: either try to blank it or only work anything to the east at night or when the light levels are low and the generated level is below 900W That is an improvement from not usable at all to the east. Maybe try to reduce the pickup from the back of the antenna.