Out of the blue faults in a PM6654C

[SaabFAN]

Hello, it's me again with a damaged Philips T&M Instrument

This time it is my PM6654C, which I had equipped with a OCXO (Morion MV103) just last week.
Upon startup all digits just show "0" and all the LED in the buttons stay off, but when I perform the manual reset with the reset-button on the front AND then turn the knob that controls the measurement-time, a random number of LEDs come on and I can hear reed relay click.

The knob for the measurement-time controls a 555-Chip that puts out short pulses, which directly feed into the D8741 Slave-Processor that controls the display and keyboard. So I'm guessing that the fault is in that chip. I'm also suspecting the 8741, because about a month ago I repaired my PM5190 that also had a Microcontroller out of the MCS-48 family, which failed in a rather strange fashion.

Also the rising edge of the Clock is a bit slow and there's no flat top on the clock-signals to the Processors. I'm a bit suspicious about that, but I don't know if that's normal for Low Power-Schottky-Logic or not.


If it really is the 8741 Processor though, I have a bit of a problem: The processor has an internal ROM and although I was able to download the complete Firmware for this thing, I cannot program that particular chip because my Programmer is a TL866.
Is there a way to make the TL866 program a 8741, or any µController out of the MCS-48 Series?

Is there also another possible cause for this failure / has anyone dealt with a simliar problem before?
Btw. I have measured Voltages and they are all within spec, which are quite tolerant (11,4 to 12,6V on the +12V-Rail for example).

[SaabFAN]

Well, that was a rather unproductive day...
I spend about 6 hours trying to find the problem and in doing that I accidentally connected the -5V and -12V Rail, which damaged a 4011 NAND-Gate and led to the discovery that several pins of that NAND-Gate were shorted to GND at the Pins on the board.
I don't know if my soldering caused this, or if the vias connected somehow to the GND-Plane.
I managed to get that part working again by cutting traces and putting the NE555 and NAND-Gate semi-dead bug style.

After that I replaced the 8155 RAM-Chip and reseated every chip, but no luck: Still the same symptoms, but occasionally I'm getting the error-code E4.
I'm unable to start any of the test-programs though, which, in my mind, confirms the theory about the damaged 8741 Slave-Processor that reads the keyboard and controls the display.

Which leads me to my next question: Has anyone managed to program a MCS-48 type Microcontroller with a TL-866?

btw. I have attached some pictures of the OCXO-Module I built to fit into the counter. It is based on the schematic from this website: https://www.maltepoeggel.de?site=ocxozeitbasis&lang=de
Total cost: Less than 20€.
It fits very nicely and once it has warmed up, it draws about 85mA and I'm 95% confident that it is not responsible for the problems I'm having right now

[SaabFAN]

Now that I'm pretty certain the 8742 Slave-Processor is the culprit, I've bought 3 on ebay and decided to just try and make a programmer for that damn thing.

Here it is:


It is the simplest implementation I could think of and can be assembled on a breadboard.
The +21V is provided by a lab supply.

Programming the chip is pretty straight forward:
- Set the control-pins to the correct voltage
- Apply Address to D0 - D7 and P20 - P22
- Pull !RESET HIGH
- Apply DATA to D0 - D7
- Increase VDD to 21V
- Pulse PROG with +18V for 50ms
- VDD back to +5V
- Repeat 2047 times

Edit: Just realized, there need to be resistors between the transistors and the Arduino

[SaabFAN]

I've managed to repair this thing and it is back in action inside what I call the "Philips Tower"

Once I had my Programmer for the UPI-41 Slave-Processor running and ironed out all the software-bugs, I programmed one of the replacement-chips I got from ebay (one is apparently bad, another one didn't survice a slipped multimeter-probe) and put it in the socket. Powered it on and everything is running again.
Self-Tests complete without any hickups and it is now running for about 2 hours without problems.

The self-made programmer can be found here: https://www.eevblog.com/forum/microcontrollers/simple-programmer-for-vintage-intel-mcs-48-microcontroller/. Software is very simplistic and can only program chips with 1K Memory max. If there's a chip that has more memory, some modifications are necessary.

[SaabFAN]

Rather than opening a new topic, I'm reopening this one because my PM6654C is sick Again...

Today it powered up with the E4-Error, meaning that the CEX-Board (the frequency-multiplier that generates the 500MHz clock to allow the counter to measure 2ns events) isn't producing a clock-signal.
Usually the reason is that there's no reference-clock reaching the CEX-Board, which seems to be the case because once switched to the internal clock, the counter starts with no error-message.

First check was if the reference-signal is strong enough. It is: Scope shows 500 mVrms into a 50Ohm load. Looking at the signal on the scope, I then discovered that the signal from my frequency standard really looks awful. It didn't look that way 4 weeks ago when i finished building it and put it next to the window to have the best GPS-Signal. Of particular note is a 3 Vpp (1 MOhm Input) high frequency burst that lasts a few milliseconds and has a period of almost exactly 1 second.
But that cannot be the culprit, right? A 2n5771, protected by BAW62 diodes and 1.5k resistance in series shouldn't die because the input-signal is a little bit too strong. Afterall the Manual states the input can handle 25 Vrms!
Well, I'm not so sure anymore. The reason for that: The input measures Open Circuit, but the service Manual clearly states: 51 Ohm resistor to ground connected right next to the BNC-Connector.

Once I'm back from work today, I'll be opening the thing to see where the signal stops.
If it really is the 2n5771, what would be an appropriate replacement? It is a 850 MHz PNP transistor with 15V Collector-Emitter voltage rating and 200mA current-rating. Could I use a BC557? It doesn't come close to the 850MHz rating of the 2N5771, but it's task is only to amplify a 400 mVrms 10 MHz sinusoidal signal to PECL-Logic Level with a supply-voltage of +5V. Which shouldn't be out of reach for a BC557, right?

What bugs me a bit: This is the second time that something in this counter just failed without any warning, despite everything else inside of it being in very nice condition: Basically no dust, voltages are all fine, no excessive ripple on the supply-rails, so signs of abuse (well, aside from the non-existent 50 Ohm-Termination on the REF-Input that I discovered today). Why does it still develop faults?

[SaabFAN]

Well, I found the problem: Input-Signal too weak and therefore the Clock-Signal generated by IC112 seems to have been juuuuust out of tolerance.
Interestingly enough, normal counter functions were still working, but I guess the 500 MHz clock wasn't generated properly in the CEX-Unit anymore (I wasn't able to measure that because I have no probe capable of doing so). Driving the input with 100mVrms more (I had about 430 mVrms, which is enough according to the manual) made the plateaus of the clock-signal from IC112 a bit wider, which seems to have solved the issue. Apparently the signal amplitude from my GPS-Disciplined Oscillator dropped a bit since I built the thing.

[Zucca]

Thanks SaabFAN, keep posting yours EE adventures. Even if nobody is replying a lot of people are enjoying your stuff..