Ahhh... Gas & Fuel... I remember visiting them in Melbourne once...
(I stumbled into this thread serendipitously.)
In late June 1973 (I think) Gas & Fuel started having a lot of problems with their B6500. (Or was it called a B6700 then?) The customer was very upset. Things were coming to a head. The Burroughs wizards in the USA were not immediately available due to the 4th July holiday. At that time I was a software person in a large systems support group based in Ruislip, England. So, myself and an engineer, Dave Knight (Knights?), were sent off to the other side of the planet to have a look at the problem.
When we got there, about 24 hours after departing, there was a pile of hexadecimal dumps at least a meter high in the corner of the support room. I grabbed the top one and sat down. [Back in the UK my wife departed on our summer holiday with our two very young children. I was not very popular.]
Anyway, the dumps revealed that the machine was sometimes getting sequence error interrupts. This is a rare error condition, caused, in this case, by the different stack bit (bit 47) being on in an MSCW when it should have been off. To cut a long story short this was caused by a recent MCP change which meant that the MCP was now using stack numbers higher than it had ever used before. The comparison circuit for the two stack numbers being compared broke the circuit rules for the B6500 (there were too many loads) causing it to give the wrong answer sometimes. Eventually someone arrived from Mission Viejo and declared that, yes, we had found a design fault. Hardware updates were applied to all B6500s worldwide.