* Posts by Colintd

8 posts • joined 2 Dec 2016

Crash, bang, wallop: What a power-down. But what hit the kill switch?

Colintd

Re: A typically British response..

I remember that halon system going off in the taperoom for the Cambridge university mainframe during the late 80s. Minor fire went out, but the level of dust raised meant all the tapes had to go through a special wash/cleaning machine (actual liquid bath with sponge pads). Extremely disruptive.

Colintd

Kill switches that go bang

Rather than silence I can speak from personal experience about what happens if you hit the kill/field dump switch on an NMR/MRI machine. All the stored energy in the coil vaporises the liquid helium coolant, followed.by a very loud bang/woosh as several thousand pounds of now gaseous helium, with about x1000 volume goes up the vent pipe. Had to be done as some idiot had come in with a steel gas cylinder which was literally dragged off the trolley into the coil. Not an experience to be repeated. Try leaning near the kill switch if you go for an MRI and see the techs.reaction...

Boffins build blazing battery bonfire

Colintd

Re: I'm not worried

You're mixing capex with opex. Just because it's much cheaper to build, doesn't mean it will be cheaper to run.

Haunted disk-drive? This story will give you the chills...

Colintd

Re: Sticky disc

I'm sure you're right, and it was a Quantum drive. I'd forgotten all about them.

Colintd

Sticky disc

In the early 90's we had a Compaq 33MHz 386 as a main build server, with a hard disc (can't remember the vendor or size) where they had made a bad choice of spindle lubricant. It would run indefinitely if left on, but if you turned it off overnight, the next day the disc wouldn't spin up when you turned it back on. The "trick" was to turn on the power, then lift up the whole case and give it a sharp twist. The inertia of the disc meant you overcame the static friction and the disc would then startup. Over time the twist required became larger/more sudden, but it lasted until we upgraded to a shiny new 486.

Phased out: IT architect plugs hole in clean-freak admin's wiring design

Colintd

Re: Plot twist? What plot twist?

Having worked on these, the chassis was normally hot 100% of the time, because the oncoming mains went through a bridge rectifier with no isolating transformer. The chassis was connected to the -ve side of the rectifier, so you have a nice half-wave 240V potential on the chassis.

I was taught to work on these beasts with one hand in your pocket, as it minimized the chance of through body contact. More exciting was using a scope on these units, as connecting the shield (grounded) to the chassis (with 240V half wave) was a big no-no. The exciting approach was to let your scope float (plastic knobs only), the more sensible approach to use an isolating transformer for the TV, and then ground the chassis.

Google turns on free public NTP servers that SMEAR TIME

Colintd

Re: We have also a Google time now?

As a 1st approximation, 1m of positional error is equivalent to 3ns time error (based on speed of light). So if you GPS location is good to 10m, the receiver knows UTC to better than +/- 30ns

Colintd

Re: We have also a Google time now?

The atmospheric drift leads to 10s of nanosecond level errors, which are way less than the impact of variable internet delays. That's what many stratum 1 NTP servers use GPS references, as do many mobile base stations.

Biting the hand that feeds IT © 1998–2019