View Single Post
Old 4th Nov 2020, 2:08 am   #68
SiriusHardware
Dekatron
 
Join Date: Aug 2011
Location: Newcastle, Tyne and Wear, UK.
Posts: 11,483
Default Re: I found it! A very sorry looking MK14.

Quote:
Originally Posted by Mark1960 View Post
Another possible way to go about troubleshooting mickie would be...
One stunt I have used from time to time when trying to troubleshoot microprocessor based systems is to fit a much lower frequency crystal or supply a lower frequency from a generator - slows everything down and makes events much easier to see. Of course that approach only works when everything on the system is derived from one main clock.

I had an interesting experience with a microprocessor controlled functional door-entry system which I eventually tried this on. The background was that this system had been in production and working perfectly for years and then it started to become apparent that new / current examples were often declaring phones to be off-hook when they actually weren't off-hook.

The sequence was:
-User presses door panel button
-Microcontroller activates relay to connect the common circuitry to the individual phone, checks for the presence of an on-hook phone.
-If the phone is present, leave the relay energised and ring the phone.
-If the phone is absent or off-hook, de-energise the relay and display 'Off' at the door panel.

Problem was that often, a phone which was present and on-hook would fail the check, and this had only recently started happening. The energise-check-fail sequence was very fast, too fast to be able to tell what was going on so that was the point at which I dropped the crystal frequency to a quarter of the usual frequency to slow everything down.

However, far from allowing me to see what the problem was, it fixed the fault! At the same time, we also discovered that raising the system supply voltage from 12V to 14V also fixed the fault. Eventually we noticed that the relays being used in current versions of the PCB were not the same make as those in older versions, so we held a 'relay race' and discovered that the changeover speed of the new relays was considerably slower than the original parts. With the original very close timing delays in the firmware the new relays weren't being given time to get across before the phone present / absent check was carried out. Once we realised that, the delay-before-check time was increased in the firmware and that problem was finally fixed, but the breakthrough came about as a direct result of running the system at a lower speed for diagnostic purposes.
SiriusHardware is online now