[Test-Equipment] Troubleshooting hint: Trak Systems 8820

[Bruce Lane]

Fellow techies,

	I'm writing to pass on a troubleshooting hint that may help save a few gray hairs. Just this week, I found (and, thankfully, was able to correct) one of the most baffling problems that has ever come across my bench. Considering I've been doing this for 28 years, that's saying something.

	Put simply: A microcontroller-based device I had was not intializing, and the problem turned out to be that one of its firmware EPROMs, despite having a perfect factory-preprinted label over its window, was as blank as a newly-washed whiteboard.

	Read on for further details.

	HOW IT STARTED...

	I was fortunate to get a good deal on a "tech special" GPS-referenced time/frequency combo clock at the Mike & Key ARC swap meet earlier in March. Specifically, a Trak Systems model 8820. My total investment, including funds to buy parts, to try and fix it, was around $125.

	As a side note: If you ever happen to spot one of these beauties, no matter if it's at a swap meet, government or commercial auction, or on Greed-bay, take a good hard look at it. They're very versatile. HOWEVER -- Before buying or bidding, make absolutely sure you pop the cover (it's all Dzus fasteners, quarter-turn and captivated to the top cover) and CHECK THE CODE written on the shield of the GPS receiver! It needs to be '3L' or later (mine reads '3P').

	If it is NOT 3L or later, it means that the GPS receiver itself is NOT Y2K-compliant, and it will need to be replaced. This is a tricky proposition, as Trak seems to have used some proprietary logic to interface to said receiver and they are most unwilling to release schematics. Trak will do it for you -- if you're willing to shell out nearly $2,000.

	THE TROUBLESHOOTING ADVENTURE...

	Basic testing revealed why my unit was a 'tech special,' and why I only had to shell out $75 for it. The device would power up, but the two-line LCD display panel on its front, used for status and setup, would not display anything other than one solid line of black dot-matrix squares.

	Further troubleshooting, including power supply checks and some consultation with the manufacturer, revealed that at least one of the microprocessors in the unit was not initializing. More specifically, the main CPU. Probing with an O-scope revealed a dead Motorola MC34064. This ingenious little chip, in the form of a TO92 transistor package, exists solely to provide a clean reset signal to microprocessors, and to hold that reset signal active for as long as the power input to the device is below 4.8 or so volts. 

	In my case, the chip had failed such that its output leg was shorted to ground. This had the effect of not letting any of the unit's three microprocessors (an 80C188-16, a 68HC11 series, and a 68HC705 series) start up. Thinking initially that this was the entire problem, I simply desoldered and removed the IC. Well, the reset lines went high (which is normal), but the unit still would not initialize.

	My next step, after ceasing to be upset at Trak for their silly paranoia about schematics, was to note that every single chip on the unit's main logic board was socketed for easy replacement. Since I lacked schematics, I figured that shotgun troubleshooting might help, so I spent a few $$ on chips and tried replacing all the support logic chips on the board.

	Still nothing. No initialize, no run. In desperation, I spent much time and effort (and just under $40) hunting down drop-in replacements for the primary and secondary microprocessor chips, and swapped them: Still nothing!

	It was at this point that I heaved a large sigh, prepared to give up on the unit itself, and decided to create and archive image files of the memory and PLD chips in the unit in case I ran into another that needed an update. With that in mind, I set up my UniSite programmer and my laptop, and got to work.

	Everything went well until I got to the unit's two main firmware EPROMs. One of them, a standard 27C512 device, came up with a checksum of all zeros when I tried to read it. Thinking that my programmer had picked up a glitch, I tried reading a couple of other similar EPROMs: No problem! They worked just fine.

	I tried reading the problem chip again. Still all zeros!

	Friends and neighbors, this was a true first for me. In 28 years of screwing with all kinds of electronics, from avionics to telephones, I had NEVER come across (nor expected to see) a supposedly pre-programmed EPROM that had failed in such a way that it appeared completely BLANK! 

	Small wonder the poor clock couldn't initialize -- half its microcode was missing!

	Greatly relieved that I was not simply going nucking futz, I dropped an E-mail to one of my contacts at Trak again, explaining the problem and asking if there was any way they could send the image file for the defective EPROM. Fortune smiled on me the next day when, in response, I received not one but TWO binary image files, one for each EPROM.

	I promptly burned a new set of EPROMs for the thing, and plugged 'em into the board. Next, muttering a quick prayer, I plugged in power and -- IT ALL WORKED! The display came up, the status and setup menus appeared at the press of the appropriate keys, and the thing even picked up GPS birds when I connected it to my master antenna system for such.

	EPILOGUE...

	The moral of the story is twofold: First, do NOT assume that a properly labeled and installed EPROM is good simply because it has a factory label over its window. Read it (electronically speaking) first!

	Second: Be persistent! The chances of a microcontroller chip developing a true hardware failure are, apparently, pretty low, so save looking there for an absolute last-ditch measure.

	I'm happy to report that the unit is quietly humming away as I type this. I'm going to let it burn in, in its newly-repaired state, for at least a day before I put it into service. 

	The only other adjustment I found needed was to fine-tune the master oscillator. Once I had that done, the unit phase-locked in under ten minutes, and it is now putting out Stratum-1 qualified IRIG-B timecode, a nice 1PPS pulse, and (most importantly for many other devices) a GPS-disciplined 10MHz reference signal that can, with a few keypresses, be reconfigured for 5, 2.5, or 1MHz.

	I hope this little tale saves someone else some frustration, no matter what device is involved.

	Keep the peace(es).


-=-=-=-=-=-=-=-=-=-=-=-
Bruce Lane, Owner & Head Hardware Heavy,
Blue Feather Technologies -- http://www.bluefeathertech.com
kyrrin (at) bluefeathertech do/t c=o=m
"If Salvador Dali had owned a computer, would it have been equipped with surreal ports?"