[Elecraft] WSJT K2 Unreliable Output with WSJT

[w2bvh]

I have a K2 that I built in 1999. It has KSB2 firmware 1.04, IOC 
firmware 1.09 and main processor firmware 2.04P (no easy way to see what 
version of the KIO2 or KAT2 firmware is without disassembling the module).

Here is the problem: When running WSJT (FT8) the K2 keys up unreliably  
Sometimes  WSJT keys the K2's PTT and the K2 makes rf and sometimes it 
Keys the K2's PTT but there is no rf output.

The PTT from the computer (via a usb to rs-232 converter and 
RigBlaster)_ goes into the mic connector. PTT goes to ground every time 
WSJT tries to transmit -- even when the K2 fails to make rf. Similarly, 
there's audio into the K2 mic connector from WSJT every time  WSJT 
transmits-- even when the K2 fails to make rf.  This has me convinced 
that the K2 has a problem (or maybe there is a timing issue).

I measured the audio on the KSB2 and the the audio makes it out of the 
speech compressor chip on the KSB2 just fine whether or not the K2 is 
correctly making RF or is not making RF.

I also did some careful measurements on the KSB2 board and found 2 bad 
values when the K2 is not making rf when it should. The measurements I 
made were  on the dual comparator (U4) chip and the DAC (U2) chip.

1. First, the "PTT DET" signal out of comparator U4-7 is 0V when the K2 
is making rf and is 6.0V when it is not making RF. The inputs to that 
comaparator ARE THE SAME whether the K2 is making rf or is not! The 
values are 2V offset + low level audio on U4-5 and 0V on U4-6. I thought 
this indicated that the comparator was no good, so I replaced it, but 
have the same result with the new comparator chip. An  VERY ODD and 
unexplainable result!

2. Second the "PWR CTRL" signal out of the DAC  (U2-6 ) is 5.5V when the 
K2 is correctly making RF and is 0V when it is not making rf . I don't 
have a logic analyzer so there's no way for me to measure what values 
are being sent to the DAC. And The logic the 16C72 MCU implements when 
feeding the DAC is not explained in the user manual so I can't pursue 
that any further.

  Spending ~$60 on a new MCU just to see if it's part of the problem or 
not seems like a waste, at least at this point. My MCU is v1.04 and I 
don't even know if the newer firmware would work on this older board. 
Maybe someone is willing to risk lending  me another just to try out. 
(Or maybe a whole board ;-)

If you have some good ideas, please let me know. This old K2 has made 
many 1000's of q's over the years and I'd really not want to scrap it. 
If you need some of  other measurements , the K2 is still wide open and 
the scope, DMM and computer are sitting there set to go).

Any advice would be greatly appreciated.

73,
Lenny W2BVH (K2 1520)