[Elecraft] [K3S] intermittent RF path interruption

Fri May 1 11:17:50 EDT 2020

Excellent job, Ben.  Back in NASA Apollo days I recall the honor of “steely-eyed missle man” was given to someone who did a superb piece of work.  You, sir, are a steely-eye ham!  Great doc too.  Thanks for writing it up.

73,  Gary W2CS

> On May 1, 2020, at 1:19 AM, Ben Gelb <ben.gelb at gmail.com> wrote:
> 
> And now the exciting conclusion....
> 
> I decided the most likely spot for the problem to exist was in the T/R
> switching circuit in the KPA3A. Schematic on pg. 46 of
> https://ftp.elecraft.com/K3S/Manuals%20Downloads/E740323%20K3s%20Schematic%20Files.pdf.
> 
> I removed the KPA3A to take some pictures and figure out how to probe
> some points of interest, then re-installed it, and took following
> measurements:
> 
> Drain of Q4 - 0.25V during RX, 122V during TX
> Drain of Q6 - 13.7V during RX, 0.24V during TX
> 
> ... so far seems reasonable ...
> 
> Measured both sides of L3 and saw 7.42V during RX. Also seems reasonable.
> Then measured both sides of L1 and saw that it appeared to be floating... hmmmm.
> 
> L1 and L3 should basically see identical voltages during RX. L1 should
> not be floating. This means that there's no DC voltage being applied
> to D2 or D5 anode. That would mean no connection from antenna input to
> either PA output or RX path. Fits the problem description.
> 
> I took the KPA3A back out again and did some continuity checking.
> Shortly I was able to discover that one of the ends of R5 (facing L1)
> had a failed solder joint to the PCB. When I touched it with the
> probe, the bad end actually came completely free and lifted up off of
> the PCB. After starting at it for another minute, I realized that
> there is a mismatch between the component package size and the
> footprint on the PCB. The pad spacing on the PCB is too narrow, and
> even when perfectly centered, there is almost no overlap of the
> package pins on the resistor and the solder pads that it is supposed
> to attach to. This is also true of R6 and R12 (though they seem to be
> connected, at least for now - though notably R12 has a green wire
> attached to one end).
> 
> This seems like an assembly defect. Doesn't seem very surprising that
> with a poor mechanical connection that the solder joint would not hang
> on. Maybe it can (or already has been in a later run?) be corrected by
> selecting a resistor that has a package that matches the PCB
> footprint? Seems likely to me that more KPA3As out there would fail
> similarly.
> 
> I noted by contrast that R18 uses the same package as the R5, R6 and
> R12 components, but appears to have a PCB footprint that correctly
> aligns with the component.
> 
> Anyhow, I removed R5 and soldered it back on (turning it at a slight
> angle, and borrowing the nearby pad of C17 to ensure it had a good
> solder connection). Now all seems to be well again. Hear lots of
> signals on RX w/ amp inline an make power out on TX. Hooray!
> 
> Here are a few pix of the adventure. Hopefully they might be helpful
> to someone else.
> https://drive.google.com/drive/folders/1kP9cR6HbBqP1Xy35QOOd945Bm2KgYW7T
> 
> 73,
> Ben N1VF
> 
> On Thu, Apr 30, 2020 at 8:07 AM Ben Gelb <ben.gelb at gmail.com> wrote:
>> 
>> I realized that the "Tune" wattage was set high enough to always
>> switch the PA in, even when the power was dialed down to <12W.
>> 
>> With that corrected, I now see consistent behavior.
>> 
>> When KPA3 gets into "bad" state, I see no power out and hear no signal
>> on receive. Dialing the power down to <12W results in the "clack" of
>> the KPA3 bypass relays and immediately restores operation (albeit at
>> QRP levels). Dialing the power up and down across the 12W mark makes
>> the receive go in and out (no receive w/ KPA3 inline). Turning power
>> on and off does not clear the fault - still persists after it comes
>> back up. It does eventually resolve itself though and operate normally
>> for a while. But have yet to determine a reliable means to "fix" it on
>> demand.
>> 
>> But it is clearly a problem in the KPA3 and not the base K3S, since
>> all is cured once the KPA3 is bypassed.
>> 
>> I have now reproduced this on multiple HF bands as well, so not just a
>> six meter problem.
>> 
>> I started digging into the schematics. T/R circuit in the KPA3 seems
>> like the obvious place to start. Though I can't really see how a
>> single component failure in there would cause what I'm describing (in
>> particular the "memory" effect that causes the problem to manifest for
>> a while and then go away).
>> 
>> I probed the 7T and 7R voltages on connector P68A (easily accessible
>> w/ top lid removed from rig). Looks fine - about 6V on either when in
>> TX or RX, respectively, and ~0.3V when complemented. Also confirmed 5V
>> is present on appropriate pins of P68A and 13.8V is present on P68B.
>> 
>> I'd like to probe further on the PA board (while installed, so can see
>> what is going on), but while there are many exposed vias visible with
>> the K3S lid off, the silkscreen side is not visible when the module is
>> installed. So its hard to figure out what to probe. Does anyone have
>> access to board layout images, or otherwise a guide to what to
>> probepoints are available on the heatsink side (which faces up when
>> module installed) of the PCB? If not I suppose next step is to pull
>> the board out and try to map it out manually when I have a bit more
>> time, but hoping to save that step.
>> 
>> For reference, schematic of interest is on pg. 46, here:
>> https://ftp.elecraft.com/K3S/Manuals%20Downloads/E740323%20K3s%20Schematic%20Files.pdf
>> 
>> Thanks,
>> Ben N1VF
>> 
>> On Thu, Apr 16, 2020 at 11:30 AM Ben Gelb <ben.gelb at gmail.com> wrote:
>>> 
>>> Thanks all for the replies.
>>> 
>>> - I saw the same issue on ANT 1 as ANT 2.
>>> - Haven't dug out a dummy load yet... still TODO.
>>> 
>>> New clue: Dialing the power down to 8W fixes the RX path (can go back and forth across this boundary and hear the RX go in and out, coincident w/ a relay click), but *not* the TX path when in a bad state. So I will still see 0W out when set to low power. Still have to switch to another band and transmit to "unstick" the TX path.
>>> 
>>> So I take this to mean that:
>>> - With high power out, both RX and TX path are impacted by the issue.
>>> - With low power out (<8W, PA clicked out), the problem no longer seems to be in the RX path (but is still in the TX path).
>>> 
>>> When I have a bit more time will look at the schematics and see if I can reason about what might be happening.. feels like this should narrow it down pretty far. But posting the update in case it spurs any new thoughts.
>>> 
>>> Ben, N1VF
>>> 
>>> On Sun, Apr 12, 2020 at 1:03 PM Nr4c <nr4c at widomaker.com> wrote:
>>>> 
>>>> Well first I’d try with a “dummy” load.
>>>> 
>>>> The swap antenna to Ant 1.
>>>> 
>>>> Sent from my iPhone
>>>> ...nr4c. bill
>>>> 
>>>> 
>>>>> On Apr 12, 2020, at 3:23 PM, Ben Gelb <ben.gelb at gmail.com> wrote:
>>>>> 
>>>>> A bit of a mystery w/ my K3S and I could use help figuring out what the
>>>>> problem might be.
>>>>> 
>>>>> Background: I have had the radio parked on 6m for several weeks (HF antenna
>>>>> temporarily out of commission) and have left it running (receive) for
>>>>> extended periods.
>>>>> 
>>>>> Recently, I have noticed that when I do sit down and try to operate, the
>>>>> radio will get into a mode where during TX the rig will indicate 0W forward
>>>>> power during transmit. Following this zero power TX, the RX will also go
>>>>> quiet. Then seems to stay that way for some time (not clear to me what
>>>>> makes it clear up... or if it ever does if left alone).
>>>>> 
>>>>> Some experiments I have done to try to figure out what is wrong:
>>>>> - Hitting the tune button also shows 0 power (instead of usual 10-15W) when
>>>>> radio gets stuck in this state.
>>>>> - Dialing the power down to where I hear the 100W PA switch out of line
>>>>> (relay click... around 8.0W) seems to immediate restore receive and also
>>>>> causes the rig to make power again on transmit. But when I turn it back
>>>>> past 8.0W, it appears to go back right back into the bad state (so suggests
>>>>> whatever the issue is, is in the 100W PA module).
>>>>> - Switching to another band and transmitting momentarily (think I tried
>>>>> 15m) will sometimes knock the radio out of this "bad" mode. But going back
>>>>> to 6 meters will eventually cause it to happen again. Usually pretty
>>>>> quickly (several seconds of TX).
>>>>> 
>>>>> Some other notes:
>>>>> - Antenna always measures 1.1:1 on the K3S display (when TX working
>>>>> correctly). Also checked w/ RigExpert. I don't think the antenna is the
>>>>> issue.
>>>>> - All my observations have only really been on ANT2 port (nothing connected
>>>>> to ANT1 presently).
>>>>> - Haven't really exhaustively tried to test band dependence of the behavior
>>>>> because I don't have an antenna on the other bands at the moment and not
>>>>> wild about trying to transmit into a big mismatch. So although it appears
>>>>> 6m related based on my observations, that might not be exactly
>>>>> representative of whats going on.
>>>>> 
>>>>> Posting here in case I've given enough clues for someone to suggest what
>>>>> may be wrong or how to narrow down further.
>>>>> 
>>>>> Thanks,
>>>>> Ben N1VF
>>>>> ______________________________________________________________
>>>>> Elecraft mailing list
>>>>> Home: http://mailman.qth.net/mailman/listinfo/elecraft
>>>>> Help: http://mailman.qth.net/mmfaq.htm
>>>>> Post: mailto:Elecraft at mailman.qth.net
>>>>> 
>>>>> This list hosted by: http://www.qsl.net
>>>>> Please help support this email list: http://www.qsl.net/donate.html
>>>>> Message delivered to nr4c at widomaker.com
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:Elecraft at mailman.qth.net
> 
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
> Message delivered to gary at w2cs.org