07 November 2008

Kenwood TS-790 revisited (2)

Well, my Kenwood TS-790E has given me some more problems... read on to see how I got through them!

After last repair, I noted there were some trouble with the ON AIR indicator, as it sometimes didn't follow the PTT: from time to time it kept lit even in RX and also the opposite (off on TX). I found also some other problems.

Summing up this is what I had found:
  • Variable delay to power up display and keypad input from POWER ON: from 2 to about 5 seconds... and getting worse!
  • Inconsistent ON AIR indicator (sometimes it does not follow PTT!)
  • LOCK LED lights from time to time while changing frequency (but LOCK is NOT activated)
  • DISPLAY refresh fails: frequency changes in the unit but change is not displayed.
All of this pointed to the CONTROL UNIT (X53-2120-00). As I had previously found a leaking backup battery and removing it and cleaning the conductive residue had helped to fix other things, it seemed there could be still some problem related with it.

Thanks to my friend Mario EB5HRZ, I got on the bench another TS-790 to be able of doing some side-by-side checks... and, yes, if I swapped CONTROL UNITs from one 790 to the other, the problem followed the CONTROL UNIT. So at least I was sure the fault was on it!

Then I checked signals at the CPU. RESET signal worked exactly the same in both. Oscillators were also the same, starting immediately after power on. But all the I/O activity seemed frozen in the bad unit, until about 5 seconds passed. Checking control lines at the CPU, there was one long delay since first WR signal was generated until the next ones, in the bad CPU, but none in the good one (I mean WR signal was switching fastly on good CPU, since first moment)

I checked voltage rails with oscilloscope, and were perfect.

It seemed there was some problem in the data bus as, what happened when LEDs did weird things is that the command from the main CPU to the display CPU had some error on it so it was interpreted as another commmand. In fact, when for example LOCK LED was lit during tuning, frequency on display remained the same but there was a change in frequency. So I guessed display refresh command was corrupted and the damaged transmission was interpreted as another command.

Next day, I continued with the CONTROL BOARD resoldering I started the day before. It took about 2 hours to get it done completely.

But then, after reassembling it in the rig, it did not power up, even waiting for the 5 seconds it used to take :-(. Well, I was sure there were no soldering bridges in my work (it was done with lots of care and I am soldering since I was 9... now I am 40 ;-)!) so I thought what happened is that something was now worse than before, reinforcing the idea of a soldering or PCB trace problem.

While I was measuring with the oscilloscope, finding that CPU clock oscillated and there was address and data I/O from the CPU, it suddently came back to life, but with the same old problems (LOCK LED lighted from time to time when rotating dial and such). I powered it off again and then I got a silly idea... my guessing was that the problem could reside in some input to the CPU which was waiting for something to be ready... so I carefully touched with my finger around the CPU pins... and rig came back to life again!

I was able to do this several times and found a zone of about 4-5 pins which consistently started the rig when touched... suspicious, right ;-)?

Looking at the schematics they were all signals to and from (!) the fluorescent display (FIP). I fastly reduced the focus to the FIP BY signal (pin 4 of the CPU). It was low on start up and, as soon as it got high, rig started to work. And, yes, placing a 1k pull-up was enough for the rig to power up always immediately :-)!!!

Of course, even with the pull-up, rig continued to produce random problems... but I knew I was close!. So I traced that control line to the connector which go to the DISPLAY BOARD and, yes, there was no continuity :-)!!!. I tried to find the faulty via but as it goes below the CPU itself and the service manual PCB is not clear enough on that zone to follow the trace, I decided to run a tiny wire (wire wrapping kind) from the CPU pin to the connector, fixing it as it was made in the old PC mainboards, with small drops of adhesive spread along the wiring route (tape is only until adhesive cures)

And, YES!!!, rig started AND behaved perfectly on next power-ups... so it was FIXED!!!

Wow, how great is to find an obscure fault :-)!

Thanks you all for the patience to read all these ramblings. I hope they could help in the future to some other troubled TS-790 users.

Best regards from a very happy camper,