[hpsdr] mercury problems: seemingly solved

[george byrkit]

Dear all,

I did get a lot of help on Teamspeak last night regarding my problems with Mercury not running properly.  The gist of the diagnosis
was that it was a 122.88 mhz oscillator problem.  There have been some oscillator failures.  There is also the possibility that
because of how the osc is mounted on the Mercury PCB that those connections could fail.

Pressing on the oscillator can got D8 on, and D9 flashing and D10 flashing twice as fast as D9 (the expected/desired condition).
Resoldering the 4 connections to the oscillator chip seemed to fix the problem, but pressing on the can seemed to sometimes
interfere with the proper operation (flashing would stop).

Thinking back, I've  had issues in the past with the Mercury 'losing sync' (or showing RED on the 'sync' button on KISS) when I was
doing some testing while building up the Hercules in that rig (one of 2).

So this morning, I resoldered again the 4 connections.  It seems much more reliable.  I'll give it a go, but if the problem
re-occurs, I'll send it to Scotty for analysis.  The vendor is interested in these failures, because they may need to change/improve
their manufacturing process.  If you listen to the Teamspeak, you'll find that originally there was a 5% failure rate when the TAPR
build of Mercury was conducted.  Scotty reported that they had 1 failure with the build that iQuadLabs did of Mercury, about to be
released for sale (look for the announce on Monday...)

If anyone sees similar problems (D8 not on, D9 and D10 not flashing), then first check the soldering of the 122.88 mhz oscillator to
the Mercury PC board.  If that doesn't cure the problem, it's likely an oscillator failure.  Phil VK6APH reports that he's removed
the cover to the osc 'chip' and found a surface mount capacitor that had a broken solder joint on one end.  The manufacturer is
keenly interested in this issue, since the part does sell for approx $50 each!

Thanks to Scotty and Phil for their 'remote diagnosis' skills (like from a half-world away in Phil's case.)  Their previous
experience seeing similar issues was a god-send!  At this point, my Mercury seems to be back in operation, and upgraded to v2.9a
(which is what the current version of 2.9 is).  The fact that it occurred after I updated from V2.7 to V2.9 was mere co-incidence.
Being a software developer, I'm keenly aware about the difference between co-incidence and causality.  Many tend to confuse the two.
Sometimes, it's hard to not leap to a conclusion of causality, when it's only co-incidence.

73,
George K9TRV