South Pole Diaries 1997/98



Tuesday 2nd December 1997 - The superposition principle

From John Storey.....

The day got off to a bad start with the return of the bad weather: 20 knot winds, blowing snow, low visibility etc. Once we got out to the MAPO building we wished we hadn't: the nitrogen liquifier had decided to dump a few litres of ethylene glycol onto the floor, making an unwelcome mess and soaking the boxes that Abu was sitting on. We wasted a while cleaning it up, then called the environmental folk over to complete the job.

The nitrogen liquifier is in any case a useless piece of junk and should be thrown away. There is a large box labelled "construction debris" just outside the MAPO building into which it would fit quite nicely.

In the foolish belief that everything was more or less working ok in the AASTO, I decided to move the "super" so that it and its monitor were no longer occupying the only flat bit of bench space in the entire AASTO. This required turning it on its side, disconnecting the Dallases, and hooking it up to the lurid piece of ethernet cable it had taken me three days to extract fron the station comms. people. As will be seen, each of these actions was to have unfortunate side effects.

1. Turning the AASTO on its side results in dreadful grinding and scraping sounds every time it reboots. It's completely unlike the "mating wasps" noise of the late lamented PC104 super - more of a mating Galapagos tortoise noise.

2. The lurid pink ethernet cable didn't work. Naturally we didn't discover this until it had been carefully threaded into position and all the cable ties neatly snipped off.

3. When the super is devoid of Dallases, it hangs for about 30 seconds while it mourns their absence and considers how to cope with a future life of loneliness. Each time it did this we thought it had crashed and tried to restart it, which meant we got nowhere fast.

It's hard to believe that just moving the super to a new postion would have so many ramifications. It's just another example of that important piece of physics: the super position principle.

Michael logged on from UNSW and was able to download new software and spend a very consrtuctive few hours debugging things. The MISM chopper timing problem is apparently fixed by changing the priority of the interrupt that services the chopper pulse. Michael also tested the elevation drive motors and found them to be essentially perfect.

More worrying, though, was the discovery that the nism cooler would no longer turn on. This was particularly worrying becausue it was working just fine yesterday. Once the satellite went down I turned my attention to fixing this.

It was easy to establish that everything was fine, but that the 5V control signal was not being generated by the PC104 computer. In fact, none of the functions controlled by the Diamond Systems card were working. So, hardware or software? Careful measurement showed that in fact the outputs were changing by about 5 mV in response to computer commands, which seemed to exonerate the software. It was therefore time to disassemble the PC104 stack.

For anyone unfamiliar with PC104, it is a compact system of stacking computer cards that resulted from an April Fools Day competition run by the IEEE for the design of the most irritating and fiddly little computer cards possible. By the time you're decked out with the necessary wrist straps, nipple clamps etc to control static electricity, and have waded through the brown slime to get at the connectors, it's a very uninviting task.

Anyway, the replacement A/D card was installed, found to generate the right voltages, and... still didn't work.

Finally, it turned out that the highside switch, although working, had decided to convert its input from "CMOS compatible" to "40 ohms". Presumably this is what blew the A/D card, since the clown that designed the power supply card hadn't put any resistors in series with the CMOS outputs to protect them (4.7k resistors have duly been added). When a device with a peak current rating of 74 amps decides to lose it, it's best for other components to keep their distance.

I washed the old Diamond Systems A/D card in bicarb soda (it looked pretty bad) and tried it again, but it's clearly taken a mortal blow.

Later in the evening I found that the new Diamond systems card was also dead. As it turned out, we'd been doing all our previous tests with the original Nism card, the one with the big hole blown in it by the departing highside switch. Late last night I put in the "spare" power supply card that had been sitting on the bench all winter. I'd checked it to the extent that it was making the right voltages, but no more. It turned out all three highside switches had their inputs shorted to ground. Highside switches may have to be added to that short list of things that don't survive freezing. Andre: let's grab some highside switches and some cockroaches and do a back-to-back comparison in our big freezer.

So two A/D cards were destroyed, and we had no more spares. Then, Ant had a brainwave and remembered that we'd packed the AFOS into a crate for shipping back to Sydney but that it was still sitting in the snow (thank heavens this year's cargoids aren't too efficient!), and that there was an A/D card sitting in it.

Ant grabbed a Sprite, but it wouldn't start. Time out to repair the Sprite (battery terminals). Then off for a late night raid on the cargo berm before carrying off the A/D card triumphantly. The AFOS electronics box was rather crudely repacked into the crate - removing PC104 cards at -30C is no fun.

Everything worked fine after that. However the new versions of the software have a few problems: rotmis no longer works (I nearly tore the stepper driver apart looking for that one), worse still cvfmis doesn't work so there's no way to move the CVF.

When restarting the "unsafe" version it often crashes during the countdown. It always stops on the number 14.

The "start nism" and "start mism" commands appear not to work the first time the super has been powered up, until the nism and mism have been manually started and telnetted into. This is a serious problem.

Tomorrow some Distinguished Visitors are arriving for a station tour, including Vera Rubin. We will probably have to make the AASTO look spick and span, and vacuum the carpet one last time.

Now, the problem of trying to talk to the DCU via port 4. Recall that it works fine on port 1 - ie pretend the DCU is an AFOS, set the baud rate to 9600, and no worries. An attempt to use port 4 results in no response, the telnet session with the super hanging (permanently), and the keyboard hanging for about a minute. Once this has occurred, you cannot communicate with any instrument until the super has been restarted.

Now, according to Chinese mythology, four is an unlucky number - the symbol is the same as the symbol for death. Could this be at the root of this problem? Well, it may well be, because port 5 works just fine. Rather than solve the problem, I propose that we use port 5 for the DCU from now on.

I was so pleased by this that I completely overlooked it later in the evening. The nism suddenly developed an inability to communicate, and no amount of resetting it and the super would give any improvement. It would respond to a telnet by giving an "AASTO" prompt, then respond no more. It seemed clear that the CPU card in the nism had died, and I spent about two hours exploring all other possibilities before tearing down the PC104 stack and removing the card. It was then that the light dawned - unplugging the DCU from the super restored communication immediately.

Be warned, however, that if you're sitting with a window open to the DCU it will beep at you every 95 seconds as it sends off another ARGOS transmission, and scare the willy out of you.

Andre - could you please grab a 5m mains extension cord to bring down.

The Mism seems to work fine but has a large signal at all elevations. This is very puzzling. If you block the window completely the signal goes to zero, regardless of the temperature of the thing you're blocking the beam with. It's got me beat....

The AASTO is running with the super, nism and mism powered up. The DCU is not connected to the super until the communication problem can be solved. Ther are a couple of dallases only connected, to stop the super from complaining.

I hope to leave it that way now...





Further Information