I'll preface this by stating unilaterally that these are not my own personal stories. These are stories as told by Jim Fahey, a field technician for a large minicomputer manufacturer, based in Maynard, Massachusetts. He has kindly given his blessing to republish these stories here under the provision that they are not monetised and that he is credited.

On to the story...

Tales from the Mill, Part 2: "Money and Power"

Here's a war story from field service. It's from around ~1979. There was a “Per Call” customer (no service contract) and they had a PDP 11/70 with all core memory that was intermittently (usually less than an hour of run time) crashing. Other engineers had been working on the problem for two days. A lot of stuff had been changed but the problem persisted. It was a tough problem because the system would literally just “lock up” nothing in the error log and the system had to be re-booted to get it going again. If you have ever worked on a PDP 11/70 you'll know that a typical system would have 10 plus power supply modules and nothing causes weird problems like a “flaky power supply”. So I spent a good part of the morning checking and adjusting the power supplies in the CPU, Memory and Unibus expansion boxes.

I should note that one of the perks of being the lead engineer in my group was that I had an oscilloscope with a built in DVM! (The ONLY DVM in my unit at the time!) While I checked the voltage with the DVM I still put my scope probe on the outputs. I did note that one of the memory power supplies had just a little ripple but was well within specs. I spent a fair amount of time re-seating modules, cables etc. The system hangs were not limited to the client operating system. I don't recall what it was (probably the system exerciser) but there was a diagnostic that would hang after running for a while. Other than being able to re-create the problem without using the customer's O/S it didn't provide any additional troubleshooting information.

So after hours of shortening busses, rattling boards and raking backplanes, I sat back in my chair and thought that “memory power supply” just bother's me. So I decided to run diagnostics while I was measuring the power supply with the oscilloscope. Then I had a brain storm – Boot the system with Cache Disabled! As soon as I did that I could see the dropouts on one of the five volt regulators in the memory box. I replaced the regulator saw that the new supply was smooth, then ran my diagnostics long enough to feel confident that the problem was fixed. I then informed the customer that system was repaired and while he was booting up his system I was filling out the Labor Activity Report (LARs) that included his bill for 3 days labor and the cost of the power supply.

The customer refused to sign the LARS so I tore out his copy (remember carbons!) and left it with him and told him he would have to take his complaints to management. The next day I also got a lesson in the “business of field service”. I had transferred to the field office from the Mill. I was the senior In-House Field Service Engineer supporting the PDP-11 Software Development Lab on ML5-5 so writing (what seemed like) a large bill was something I was not in the habit of doing.

When I got to the office my manager had already spoken with the client who did not want to pay for 3 days labor “to replace a bad power supply”. I then pretty much told him the story I have described here and he asked me if I thought he should reduce the bill. I said well I feel a little bad that I didn't investigate that power supply sooner so maybe we could shave a couple hours off. Then he asked me “If that was a contract customer would anything different have happened?”. I said “No, I don't think so”. So do you think that guy - who doesn't have a contract – gets the same level of effort as a contract customer - takes up 3 days of my engineers time, including a full day of my senior engineer should get a discount for a bill that is about the cost of the monthly charge for a PDP11/70 under contract?

The question answered itself.