When I read about this report from Stewart last night (as well as the responses indicating that other had experienced a similar effect) I decided to set up the bench test to see if I could emulate the problem. I've since discovered that these reports are valid, the K3 PA will generate a severe spectral "comb" if there's insufficient isolation between the RX and TX antennas when the KXV3 RX input is used.
To execute this test, I ran the K3 into a dummy load through an RF coupler that displays 26 dB of isolation. The output of this coupler was split (3 dB splitting loss) between a Tektronix 496 spectrum analyzer and a Tektronix 2701 50-ohm step attenuator. The output of the 2701 was fed into the K3's RX input. Initially, the attenuator was set to 0 dB. The K3 was keyed at full output (100W or +50 dBm), which results in a +21 dBm signal back into the K3's RX input. Under this condition, if the RX port is enabled, the K3 will generate a spectral comb with the most severe spurs only 10 to 20 dB down from the carrier. Turning the RX input off clears the problem. I was able to create this comb spectrum on 160, 80, and 40 meters, with 40 displaying the highest-power spurs. Lowering the input power to the K3 via the 2701 attenuator indicates that the "sweet spot" for generating this spurious oscillation occurs when there's an input signal of about +20 dB. This would correspond to a minimum of 30 dB of TX antenna to RX antenna isolation when operating at 100W output power or 40 dB of isolation at 1 kW. I have spectrum analyzer photos of the spurious signals generated by the K3 under these conditions if anyone is interested. I ran the same test under the same conditions with my K2/100, and this problem does not occur. The bottom line here is that a +20 dBm signal (100 mw) will create severe spurs on the K3 but not the K2 when operating in the 40M band and below. Above 40M I was unable to duplicate the problem, perhaps a higher power to the RX input is required. It's important to note that anyone who has +20 dBm or more coming back on the receive antenna is asking for trouble in the first place, and additional isolation and/or a shorting relay during TX operation is probably a good idea. >From the reports, it sounds like there have been a few folks who had rather high power coming back in through the RX input for a while and didn't notice it until this secondary symptom was noted in the K3. I haven't looked at the RX input circuit in the K2 to see how it differs from the K3 so that I can find the root-cause of this problem. Scott N7NB K2 #1285 K3 #219 ------------------------------ Message: 34 Date: Thu, 31 Jan 2008 16:14:13 GMT From: Stewart Baker <[EMAIL PROTECTED]> Subject: [Elecraft] K3/100 RF Feedback/Spurious oscillations ? To: Elecraft Reflector <elecraft@mailman.qth.net> Message-ID: <[EMAIL PROTECTED]> Content-Type: text/plain; charset="iso-8859-1" Very strange effect noticed today when tuning up on 20m. At power levels above 20 w on 14.175 my external Palstar SWR meter would suddenly leap from 1:1 to above 5:1. Also the K3 SWR display showed similar. High current was also reported. No problems with either a TS850S or K2, both OK up to 100W. Switching the RX Ant switch to the main antenna or removing the RX Antenna BNC lead cured the problem. I use the RX antenna input for 2 x Pennant loops for the LF bands, but find that they sometimes give a better SNR than the main antenna on higher bands. The main antenna system is a doublet fed with open wire feeders into a fully balanced (no toroid on output) ATU. My initial thoughts is that there might be some spurious output generated by an obscure feedback path from the RX antenna input. If this output were on a frequency well separated from the desired then it could show up as as high SWR. What think Ye ? 73 Stewart G3RXQ _______________________________________________ Elecraft mailing list Post to: Elecraft@mailman.qth.net You must be a subscriber to post to the list. Subscriber Info (Addr. Change, sub, unsub etc.): http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/subscribers.htm Elecraft web page: http://www.elecraft.com
