Power Supplies and RFI

This is a big subject, but after years of working on power supplies, both switching and linear I concur that many switching supplies can, through conduction, radiation and common mode coupling cause an increase in the noise floor of a modern sensitive rig.  This was EXACTLY the impetus for designing the PAE-Kx33!  If you look at our website you will see some ugly examples of these other supplies which were all recommended by other hams at one time or another as being ‘quiet’ and suitable for portable rig operation.  The test results are posted here on our site, click on the “Test Results” tab:
How could hams think these supplies were quiet, when in my experience they were not?  The answer is as complex as, and related to antenna systems, namely ENVIRONMENT, RFI MODES and ANTENNA SYSTEMS.

I live in a rural area where the only real RFI to be received is coming from my own home, and I undertook a plan to reduce it as much as possible, with ferrite choking and discarding of CFL and LED lights where necessary, and ignoring great pricing on plasma TVs.  At this point in time on bands above 40M (using either the Astron or Kx33) my noise level is often S1-S3 with a 265′ OCF dipole 60′ directly above my house, indeed most of this is atmospheric noise, not locally generated.  In this environment I can detect radiated noise from devices with greater sensitivity than someone in an urban setting who has a noise level of S5-7 or higher, so those hams could possibly not even notice the contribution of a noisy switching supply.

Any power supply can create EMI by one or more of three ways: Radiated, Conducted, or by modulating Common-Mode (CM) currents through them.

Conducted, Radiated RFI
If you study the test results on our site at:
you will see that many switching supplies have very high conducted differential noise, that is noise modulating the DC current drawn by the rig.  This noise can be heard when the antenna is disconnected as the RFI enters the rig through the power port.  This will usually cause radiated noise as well, and poor design causing this type of noise usually means little attention was paid to filtering, either at the input or output, increasing the radiated noise.  Indeed most of these other switchers output conducted RFI on the order of 40 dB more than that seen operating from a battery.  The Kx33’s conducted noise is 10 dB or less above the battery’s noise.  Indeed the KX3’s internal circuitry conducts more RF to the power jack than does the Kx33.  That 30 dB difference between most switching supplies and the Kx33 represents a noise power ratio of 1000:1 in favor of the Kx33.  Considering that a properly designed rig like the Elecraft KX2 or KX3 has excellent PSRR (Power Supply Rejection Ratio), the noise floor level in the rig from conducted RFI is identical between using a battery or a Kx33.

Common-Mode RFI
To get back to basics: in a coax feedline, the reason the shield is a ‘shield’ is because the EM fields of the center conductor and shield are balanced internally in the dielectric of the coax.  This means they should not be effected by external fields, nor should they radiate external fields.  If the two currents at the antenna feedpoint are not balanced and exactly 180° out of phase, an imbalance is formed, and thanks to the laws of physics this current will flow…somewhere.  In most antenna systems the only other connection to the feedline is the chassis of the rig and everything hooked to it.  This forces the imbalance current (CM current) to flow from the feedpoint, over the outside of the coax shield to the rig chassis, and capacitively couple through the power supply to the AC line, flow through your body, etc.  If you think you do not have CM currents, there is a simple test: insert a good CM choke at the antenna feedpoint and see if the antenna matches and performs exactly the same…if the currents in the antenna system are balanced it should, but with almost all (especially portable) antenna systems, the impedance will change along with the SWR.  This is proof positive that the outside of the feedline was part of the antenna system.

Please consider that ALL ham setups need CM chokes.  Period.  Here’s Why:

These common-mode currents effectively turn the outside of the shield of your coax feedline into part of the active antenna system.  In receive they modify the antenna pattern with their own pickup pattern as well as picking up noise from wherever they are routed.   Many hams notice a serious decrease in receive noise when installing good common-mode chokes.  In transmit they cause the feedline itself to radiate and that contribution to the antenna pattern is often destructive.  When running high powers this can also induce RFI to radios, TVs and other household devices, and even develop shock or burn potentials on the rig or other attached devices.  Most antenna systems benefit from a common-mode choke at the feedpoint, and another where it enters the structure to choke currents which the feedline may have received radiated by the antenna.

Just as importantly: In all of our experience, 99% of any reported noise issues from users of our Kx33 are related to common-mode conduction to AC main currents.  Why is this important?  For two reasons:  the AC lines in almost all modern homes is extremely dirty due to the contribution of all the devices in the home, so common-mode current conducting to the AC lines will often introduce this noise into the receiver.  Also, this type of RFI means the RFI is the same on both the plus and minus leads of the supply and capacitively couples through the supply to the AC line.  Interestingly enough this common-mode coupling also applies to linear supplies as well, although in the case of switching supplies, the spectrum of noise superimposed by the supply itself on the common-mode currents is broader.  Realizing this fact, much of the Kx33 development effort went into decreasing input-to-output capacitive coupling as much as possible.  The end result of that work is a design which offers less than 70 pF of coupling input-to-output as opposed to a typical 1000 pF for other supplies.  In practical common-mode RFI terms this means that the Kx33 looks like a 650 ohm impedance at 3.5 MHz from the rig to the AC line, and a typical supply looks like 45 ohms.  This means the common-mode current is 14 times less with a Kx33 and far easier to eliminate if needed with a common-mode choke.

Bottom line:  ALL ham setups need CM chokes.  Period.

Need more proof?  You should read this paper by Jim Brown, K9YC, a widely respected expert in antenna systems and common-mode choke studies:

This is the reason why Pro Audio Engineering offers several types of Fair-Rite ferrite cores at low cost which are designed to be used as common-mode choke cores: