Daily/Weather Graphs

There are two types of intermittent faults - those that are truly intermittent i.e. that come and go, and then there are those which are affected by external influences. In tracing intermittent faults a lot of time can be saved drawing up graphs of the failures to determine if it is possibly a time of day or weather related.

One graph can do both showing the time of day and weather conditions. Two lines are drawn on this graph being weather and system performance both on a scale of 0 to 10.

Weather: 10 is the hottest possible day known in the area with 0 being absolutely putrid conditions.

System Index: 10 being all working 100% with 0 being total failure.

It should take no more than a week to determine whether or not a daily occurrence or weather pattern is affecting the system, or not. If you have a truly intermittent fault then, well, there is no other way to say this but I genuinely feel sorry for you. The most that can be expected is being there at the right time when it all goes wrong.

Daily failures: Don't discount weekly failures either. I cannot forget a situation where a system was failing every Friday afternoon in a food plant but every time a technician was called to attend to the problem both the psychological impact of going out to site at this time as well as entry could not be gained as they were fumigating (to the glee of the technician) the problem was only attended to on Monday morning. Upon arrival the system merely needed a reset and all was well again. It was permanently blamed that something was being done to the system during the fumigating and an immense amount of time was spent on air-tight housing etc. Only after a new technician went to site, ahead of the time the system usually fell over, was the fault discovered. The system was unplugged, as was all other office equipment, in preparation for the weekend. But this was not all. This was a daily occurrence but the back-up batteries would keep the system running till the next morning but on Fridays the plant shut early and the batteries had not been charged fully and the system then failed. With the power wired in permanently the fault disappeared.

Many examples of failures like these can be recalled by many engineers. The moral of the story is don't take anything for granted.

Weather Failures: These are more easily determined as during weather changes signal strengths can be monitored for fading. Fading, however, is not the only effect weather can have on a system. Loose connections are usually very temperature sensitive and many an engineer has been chased to site just on the very day the temperature is such the fault does not appear. Again, using a temperature/weather graph will quickly show if the fault is something that could be trapped at a certain time of day.