[Elecraft] K3 RF Feedback Problem

[Ian White GM3SEK]

Dr. James C. Garland wrote:
>
>Gang,
>
> 
>
>I?m pulling my hair out over an RF feedback problem with my K3/100,
>and wonder if anybody has any ideas? I use an external station
>controller that switches key, microphone, CAT, line in/out, etc. between
>different rigs.  The microphone audio connects to the K3?s rear panel
>microphone jack and when the K3?s power exceeds about 15W (in
>SSB), the audio becomes greatly istorted. It does this when all the other
>cables to the K3 (except the microphone and power cord) are
>unplugged. It also does it when the audio is fed to the K3 via the rear
>Line In jack. Everything in my station is well-grounded. Here are some
>things I?ve noticed:
>
> 
>
>1.  The feedback still occurs when a military-type completely shielded
>100W dummy load is screwed directly to the K3?s antenna port. This
>suggests the RF from the K3 is getting into the audio line from some
>other source than radiation from the coax cables in the station, most
>likely the 12V power cord. To support this conjecture, I also notice that
>the feedback threshold (15W) is not changed when the K3 drives a
>linear amplifier.
>
> 
>
>2.  I normally use a 30A Kepco 12V power supply to run the K3 and my
>other station accessories. The RF feedback problem is improved but
>not entirely eliminated if I use a separate 12V supply for the K3. A
>common mode choke on the 12V line doesn?t appear to make any
>difference.

I just measured the common-mode RF current on the K3's 12V power cord at 
about 2mA at 7MHz, 100W into a dummy load. The background level was 1mA, 
most of which was coming *up* the line from the SMPS.

The meter consists of an RF current transformer made from a clamp-on 
ferrite core, feeding an AD8307 wide-range log detector. It responds to 
the total common-mode current from all sources combined, with a 
bandwidth from LF to several hundred MHz.

For perspective, about 1mA of assorted RF crud is a typical background 
level on computer and monitor cables, including the filtered mains input 
or DC output of a good-quality SMPS.  Common-mode currents are also 
installation-specific, because they depend on the impedance into which 
the current is flowing, so in real-life installations it's only 
justifiable to compare orders of magnitude (the location of the decimal 
point) rather than obsess over exact values.

 From that perspective, 2mA on the DC cord of the K3 is not a seriously 
high level of RF leakage from a box that is generating 100W of RF. It is 
also notable that the output cable leading to the dummy load has only a 
few mA of common-mode current (contrast that with 1.4A of 
differential-mode RF current flowing *inside* the shield).  My IC-746 
gave very similar results to the K3.

Therefore it seems unlikely that K3s in general would have enough RF 
leakage on the power cord to cause the problems you're seeing, Jim. That 
doesn't rule out the possibility of a fault on your particular rig... 
but maybe you should be looking for a broken shield connection or 
something else that has been disturbed by puling out the previous rig 
and installing the K3.

Most likely of all is a broken shield connection on a PL-259, which can 
let out high levels of RF current, which then crawls over the whole 
station.

>3. I can see some modulation-induced fluctuationss on the 12V line with
>an oscilloscope that increase with the K3?s power setting. I haven?t
>looked at these fluctuations closely enough yet to know whether it?s RF
>noise or just audio?frequency  fluctuations caused by the K3?s
>modulating current draw from the power supply.
>
> 
>
>To summarize, my conclusions so far are that (1) the 12V power cord is
>a source of RF leakage from the K3,  even though the K3 has a filter at
>its 12V power connector to minimize this leakage; and (2) The
>audio/DSP circuitry in the K3 is very sensitive to RF ? much more so
>than other transceivers in my station.  (I?ve not had this problem with
>other rigs.)
>
> 
>
>Because of this RF sensitivity, one evidently has to be very careful 
>when hooking accessories to the K3. The front panel microphone jack
>is (to me) wired in a curious way, with the Mic ground and PTT ground
>(shield) connections floating above the K3?s chassis with a 100uH rf
>choke. Similarly, both the ?hot? and ?shield? side of the rear panel mic
>input jack has series 100uH rf chokes, which isolate the input from
>chassis ground.
>
> 
>
>The problem with this arrangement is that most accessories that would
>connect to the rear panel audio input jacks have single-ended outputs,
>with a shielded cable that is directly tied to chassis ground at the
>accessory end of the cable. Thus, unless one is very careful, it is easy
>to have a situtation where the signal grounds in the audio circuitry of
>the K3 can fluctuate with respect to the K3?s chassis ground, and this
>may be the source of the RF feedback sensitivity.
>
>
>Unfortunately, I can?t figure out the detailed mechanism for the 
>feedback closely enough to figure out a solution. Foir example, I don?t 
>know whether it?s better to leave the minus side of my 12V power supply 
>floating, or to tie it to the chassis gound.   And even if I left it 
>floating, it would be tied to the chassis anyway by the other 
>accessories hooked to it, and this might cause more problems than 
>grounding it at the power supply terminal.

Think very carefully about that treacherous word "ground"!  In reality, 
"ground" connections at physically different locations are *never* at 
the same potential. Often it doesn't matter... but if the common 
negative return for all the accessories is at the PSU, the voltage drop 
along the negative return for the K3 will be injecting DC, RF and SMPS 
leakage into the entire audio system.

If the PS is allowed to float and the common negative return for all the 
accessories is moved very close to the K3, that might solve the 
problems.

>Somehow, it seems like a wiring change in
>the ground configuration of the K3?s audio circutis to improve RF
>isolation may be needed to lick the problem completely.

The unwanted impedances in the internal "ground" paths of the K3 are a 
weakness that might make a problem appear worse, but I don't think they 
are the actual source of the problems.



-- 

73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek