Air Conditioner Plant Makes S9 Noise at Home

Klaus Illhart VK3IU  had a significant noise problem that made HF operations difficult.  Klaus has prepared a detailed description of the process of tracing and resolving the source of the interference. 

Here we have an abbreviated version to give readers a sample of Klaus’s experience.


RF interference through Brivis Evaporative Cooler

Telecommunications Engineer (Australian Professional Engineer) Clyde North, VIC 

I would like to express my sincere thanks to the Brivis support team, which certainly in early steps, went above anything that I would have expected in supporting me.  I am looking forward to solving the RFI to a level where a proper use of the radio spectrum will be possible while keeping the family cool.

Problem description

When the evaporative Cooler is operational, noise levels increase from S2 on an HF Receiver to S9 over a frequency range from 18MHz to 30MHz.  Further “control traffic” can be observed in the 140 MHz to 150MHz range.



This was heard on individual frequencies at S3, even when the evaporative cooling unit is not operational and in standby mode.  Switching OFF of the evaporative Cooler at the main ac power switch board kills all interference.


When the evaporative cooler is operating, it makes the use of the frequencies between 18 to 30MHz impossible, as receive signals cannot not be decoded/received.

Antenna installation:

left antenna vertical Butternut HF6V  HF antenna 3.5 to 30MHz,  8m height on the 1.5m mast.  Antenna to the right 50 to 450MHz vertical antenna,  2m height on 1.5m mast.   The evaporative cooling unit is approximately 2m from antennas.


RFI Measurement

Measurement of the RFI was carried out both as an observation on the Transceivers, expressed in S Units and through more detailed measurement using a HP8590A spectrum analyser, which shows noise levels expressed as dBm.  The tests were carried out over the frequency range from 2 to 52 MHz.


It should be noted that the antenna used was the Butternut HF6V trapped vertical, which is resonant in sections of the HF band but performs poorly for frequencies above 30 MHz .

Evaporative cooler on standby (not switched ON)

The adjacent image shows a reference shot of the spectrum analyser, with the evaporative cooler deactivated.  The area in the yellow rectangle shows low noise activity between 20 and 30 MHz.


Controller Replaced

The air conditioner controller was changed by Brivis technicians in February 2022.  Measurement conducted after the change showed a decrease in noise floor between 20MHz to 30MHz of approximately 5dB.  This provided a slight improvement.


Noise present from original controller Noise present from new controller

Testing resumed with the evaporative cooler after it had started up and completed its Wetting cycle. Note that the new controller changed the behaviour observed.  While the old controller the noise level was present instantly, the new controller increased the noise floor moments after the fan activated at the completion of the wetting cycle.

Portable battery operated HF transceiver

A battery-operated portable HF transceiver was used to conduct independent and portable measurements on the AC power.  This was achieved with an Elecraft K2 portable HF Receiver and a simple HF Loop antenna.  (muddy meaning)


Near field investigation within the house and on the roof indicated that the distance to the Evap Cooler is not of a consequence as the level only slightly drops when the receive antenna is moved away, several meters. 


Exclusion of RF through the AC mains supply

Interference is received through the HF Antenna.  There was no influence of RF through AC power.  This has been experimentally verified after power to the radio room had been cut off at the switchboard.


My station is sufficiently RF proofed.  It uses RF grounding of 20mm2 cable into an earth rod buried 1.5M into ground.  The 240V AC supply is passed through a mains filter to prevent any RF from reaching the test equipment.


RF AC RFI filter in 240V AC supply path, CORCOM – TE CONNECTIVITY 

Power Line Filter rated at 250VAC 10A on a single phase supply.


Ferrite installation

I added three clip on ferrite clamps (material 43) on the power supply cable (yellow circled) to the FAN.  This reduced the RFI further. It was noticeable that at frequencies above 28 MHz a decline in RFI is observed. Noise was suppressed by approximately 5dB and from 22MHz to 30MHz.  


Unfortunately, I ran out of clip-on ferrites.  I am unsure if the smaller diameter clip had any impact.  It would have been useful if Brivis could advise the function of these cables.


This is already very promising; it seems that the FAN contributed heavily to the RFI.  Further tests will be done when I am able to procure additional ferrite of different materials.


Dependency FAN speed to RFI

A test was carried out by changing the FAN speed from minimum to the maximum speed. The observed result is that a change in FAN speed is not much affecting the noise level on the transceivers, but only a slight change in the spectral measurement can be observed.  

S-Unit at  FT-1000MP
Band / freq in MHz Aircon OFF Aircon ON
FAN unit minimum
Aircon ON
FAN unit maximum
29.5 2 7 7
29.0 3 5 6
28.5 3 7 7
28.0 2 7 7
21.0 2 9 9
18.07 3 7 8
14.0 6 6 7
10.1 6 6 7
7.0 7 7 7
3.8 5 4 4
3.5 9 9 8


Insertion of RF suppression filter

Some months later.  After opening the control box of the evaporative cooler it was immediately noted that Brivis applied an unknown ferrite ring, feeding through all wires of the cable loom, going to the fan motor. 


Due to the size and number of wires only one turn was applied. This approach is highly inefficient and didn’t produce much noise reduction.


Brivis block diagram

Marked in RED, GREEN and YELLOW are the main power supply feeds to the three phase motor.  The Brivis application of their ferrite core to all wires is not necessary, as only the power feed wires are suspected to carry the motor high current requirements.  The remaining are simply control and sensor wires.  


I initially tried to use this ferrite core and apply more turns to the motor power wires only, but I was not able to add more than 2 turns, which didn’t improve the noise situation.  Further the ferrite material characteristics were unknown to me and I decided to develop my own solution.


T38 Toroid filter for 3 phase motor

I experimented with two types of ring toroids F43 and T38 material, but decided after measuring both variations for my purpose the more costly T38 material was the best approach.  One toroid for each of the three phase supply wires should be applied.      
Manufacturer: TDK  R 36.0 x 23.0 x 15.0  mm   Order code: B64290L0674X037


Thanks to QRM Guru and especially Ian VK3BUF with his industrial experience and his professional advice, I settled on a counter-turn ring ferrite, 10 turns left and 10 turns right.  This as per Ian’s advice will eliminate current inside the toroid itself, but keeps a good filter characteristic.   The toroid remains cold after 10 minutes of running the fan on maximum.  Initially I experimented with insulated wire, but this limited the number of turns I could apply over the ferrite core.  I decided to go for coated solid copper wire instead.


Installation of Toroid in Evaporative Cooler

This November I finally went back on the roof and inserted the three toroids and started to measure the effect they may have on the RF noise.  The diagram below shows the wiring layout.


I had a spare F43 toroid and spare wire length, so I decided that while on the roof, I would use it as well and wound around another 5 turns.  I believe this was not necessary.  It was more a measure to reduce the excess wires, which  prevented the box closing.


Noise with inserted Choke Filters

A series of repeated measurements were taken over the following days. 

Measured was always with two methods, the YAESU FT-1000MP in S scale and with the HP Spectrum analyser showing dBm levels.


1   AC off, 

2   AC on slow fan and 

3   AC on, fan on maximum speed.


At the start I also measured the AC start up phase, where the pre-cooling happens.  The pump is pumping water through the filter matts to pre-wet them before the FAN starts up. The water pump in pre-cooling phase does not contribute significantly to the noise.


As expected, the modifications had a big effect.  The noise stayed well below the -80dBm mark of the analyser. The noise floor fell by approx. 15dB.

Improved noise figures – 10 January 2023  app TIME 1820H to 1831H

2.4kHz SSB Filter  /   500Hz  CW Filter   pre amplifier  TUNED amplification curve



These changes have taken 18 to 30 MHz from being completely unusable, to my being able to conduct good QSOs on these bands.  The effort involved was totally worthwhile, as apart from my own comfort, turning off the air conditioning on hot days would not be popular with the family.


As a future task, I will be sourcing clip-on ferrites of different materials to further suppress RFI from 20 MHz up.  


The air conditioning company, Brivis stopped co-operating after I sent the details of this report and they have not responded.  This is rather disappointing as a co-operation would have been a mutual benefit for all involved.  Future Coolers could be built with less radiated noise.


My advice to readers with similar problems is ‘Don’t give up’.   Identify the source of the noise and work out a solution which will reduce its impact.  Of course this is easier if you are the owner of the noise source, but it is more challenging if the noise is coming from neighbours.