So now I was the proud owner of a pretty useless Radio Test Set, since none of the RF functions worked. First I set the unit aside and thinked about scrapping or reselling it. After a few days, I took the challenge and decided to see if I would be able to repair this thing.
As it is quite usual with such old equipment, the service manual contained full schematics and parts lists of the box. For each board there is also a fine description of how it works. Having this comprehensive information, it was easy for me to discover that I'm in the need of some RF instrument (preferably a spectrum analyzer) to do the basic diagnostics. Your average scope simply won't show anything if you try to follow a 1.7GHz signal. Luckily I was able to borrow such an instrument (an Advantest R3131 9kHz ... 3GHz spectrum analyzer) from work.
I found it easy to pull one SMB cable after another and compare the real signal (using the R3131) to the expectation. Pretty fast I discovered that the high synthesizer module didn't output the 1st GEN LO signal. There was some signal visible, but not at the expected frequenzy and at least 20dB lower level than expected. Other signals, like the LO SYNTH RF from the low synthesizer module looked fine.
Pretty much shielding, the right two SMB connectors carry the 1st GEN LO. In the first place I soldered some wires to the relevant power supply rails, re-seated the board into the card cage and checked the supply volatages. They were OK.
Next I suspected the VCO module to malfunction, since this is the source of the bad signal.
I had to remove the VCOs shield to gain access to the signals around the VCO module.
The VCO is in its own shielded box inside the shielded box inside the shielded card cage. I didn't look inside the the VCO box, maybe there's more shielding? Looking at the schematic page concerning this part of the high synthesizer module, I started to check the supply and control voltages of the VCO module, using the same technique of soldering wires and re-seating the board into its box again.
The supply voltage was OK, the control voltage was stuck at one end. At some frequency setting the range selector input of the VCO (Pin 4) changed its level and the signal at J1 changed frequency. Carefully checking not to destroy any driving circuit, I did override the control voltage input using an external volatage source, and look here: The output frequency moved around proportional to the control voltage. Exactly what one would expect of a perfectly working VCO. The VCO module was OK, phew!.
Not too many parts left to check: The MMIC gain blocks AR3 and AR5. I soldered wires to their bias voltages (collector of Q5 and Q6) and found one of them not plausible (way to high). Logical conclusion after checking the transitor: The MMIC must be defective. No MAR-8 lingering around in my workshop, but some other RF gain block samples left over from past RF experiments, having similar gain and frequency specs, but other case and bias requirements. I temporarily mounted this part and tested the R2600 functions again: Everything worked as expected. I was able to generate plausible signals, receiving a local FM radio station worked after discovering and replacing the blown fuse inside the ANT connector. So the fault was located and the unit seemed to be alive again.
I decided to order some spare MAR-8 MMIC to have an original replacement for the broken amplifier. Ordering this part turned out to be somewhat difficult, not beeing available at the usual suspects, finally I've found the german Amateurfunker Magazine web shop having them in stock. When they arrived, I replaced the part, soldered the shield, screwed the module into the card cage, did a short test: Yes everything was fine.
I did some checking to confirm my suspect: Powering the electronics from a bunch of lab power supplies revealed that everything was fine with my R2600, now I'd just have to repair the power supply.
Checking the output voltages in idle and lightly loaded state showed nothing suspicious. Each of the 5 output voltages was perfectly in its specified range. Increasing the load on the +12V rails caused the supply to shut down at some point, caused by a rising voltage on the 5V rail. First I suspected the electrolytic capacitors of beeing old and lost ESR. I did replace them with not so nicely fitting ones (see them on the photo) and checked the old ones for capacity and ESR. Since the behavior of the power supply board didn't change with the replacement caps, and the old ones still looked very fine in terms of their electrical properties, I decided to put the old ones in agail, just because it looks better.
Looking at the schematics, one can see that all of the output voltages were derived from a single transformer
using an isolated forward converter scheme with coupled filter inductors.
So basically any voltage in the output circuit is coupled to any other voltage. You just cannot easily use your
scope to find partially defective parts. Each particular waveform on any output voltage looks very like any other
voltage, just other amplitude. So I started removing rectifier diodes until the problem disappeared. Yes, at some
point (all rectifiers but the +12V rail removed) the problem disappeared :-). Didn't know much more than before.
It turned out that replacing all of them (except the 40V rails, since I didn't have suitable dioded) solved the issue for now.
Don't know which one, but one of these must be bad in a bad way. Looking at them using the diode test of your
favourite multimeter doesn't show out the bad one.
So I started to think about what action was in common to the occurence of the failure: De- and re-assembly of the power supply or CPU board and opening several ribbon cable connections. Beeing known as prone to failures, I checked all of the ribbon cables and their connectors: No failures found, not even a hint of a slightly bad contact.
Both actions include the task of removing the rear panel. For some reason, possibly simply because there wasn't
sufficient free space left anywhere else, the people who built this thing decided to put the reference oscillator
on the back panel. It is connected by a small ribbon cable and a SMB RF cable to the main unit. And finally
I discovered the reason for the misbehaviour exactly there:
That little beast went slightly loose while disassembling the rear panel and sometime got contact again while I assembled it back and sometimes it did not contact proper. There's no real need to open this connector when disassembling the rear panel, so I didn't and never noticed that it wasn't plugged in proper. Pushing this connector a little bit solved the issue in a fraction of a second.
Receiving (and listening to) a local FM radio station