A Tale of Two Turbos

I spent most of Saturday in the lab, swapping out a turbopump that was starting to die. How could I tell? Well, for one thing, it made an awful noise, even more than usual for a noisy pump. But after it was stopped and unmounted from the chamber, there was a simple test: comparing the rotation to a second pump that I knew was in better shape:

The pump on the left (with the copper gasket stuck to it) is the old one; the pump on the right is the one that I replaced it with. As you can tell, even with a bearing starting to go, it's a pretty damn good rotor, but clearly worse than the replacement.

The blue nitrile gloves are probably overkill, but I really like the blue-handed assassins in Firefly. That, and I was hoping to cut down on the amount of nasty thread lubricant I got on my hands when putting the bolts back on the flange.

To give you an idea of just how much fun it was to do the swap, here's a picture of the pump on the system:


It's the very top unit on the vacuum chamber, with a black computer fan bolted on top (DIY air-cooling to keep it from overheating). This puts it about shoulder height for me, but it's also out toward the middle of the optical table, meaning that I needed to get on the table to do the swap:


Contrary to what the odd perspective may suggest, I am not bent double to avoid the ceiling. I can stand comfortably on the table, provided I'm careful about avoiding a few pipes. I'm bent double because that's the only way to work on the pump unit.

You can imagine just how much fun this was for my back...

In addition to swapping out the dying turbopump (which will be shipped back to the manufacturer for a rebuild, at a cost of roughly $1000), I took the opportunity to replace the dirty oil-based mechanical pump backing the two turbopumps on the system with a quiet and oil-free scroll pump. The relatively quiet pump that had been serving as the backing pump, I moved to be the backing pump for the quiet but very dirty diffusion pump that is on the source chamber. This replaced an old Welch mechanical pump, one of those rattling blue monsters you see in every physics department, which 1) leaked oil, and b) made an ungodly racket. Now everything is cleaner and quieter, and the base pressures seemed to be about the same.

Total time required for all the pump-swapping was 4-5 hours, working by myself. That's not too bad, and it's nice to see that I'm still reasonably competent with a set of wrenches.

Update: I wrote and scheduled this yesterday. This morning, I came in to find that one of the hoses on the backing line for the big turbo had split, venting the whole system to atmosphere. Fuck.

That's experimental physics for you.

More like this

How far back are you set by the overnight venting?

Do you sometime feel like a Heterodyne?

By Lauren Uroff (not verified) on 04 Jan 2010 #permalink

$1000 to rebuild a turbo? That's a bargain.

I like the spin-down test, but I'm curious how much of the damping comes from the bearing and how much comes from the electromagnetic coupling to the motor? I've heard mytical tales of certain brands of turbopump controllers that - in the event of a power failure - power themselves off the kinetic energy of the rotor while it spins it down. I don't know if those stories are actually true.

P.S. Sorry to hear about your update. I hope everything's OK.

I've got a bunch of plastic and rubber lines in my lab for both rough vacuum and for water. Everytime I put one in I wince in anticipation of some future failure, but I'm usually too cheap and/or impatient to put in a metal line.

By Anonymous Coward (not verified) on 04 Jan 2010 #permalink

And yet, all I can think of is the episode of the Big Bang Theory where Sheldon and Leonard are talking to prospective students. Leonard spills apple juice all over the laser experiment, which results in much smoke. Sheldon then tells all the prospies that they are never going to find a job in theoretical physics. Leonard responds with something along the lines of, "The lasers are looking pretty good now, huh?"

In our lab, some guy came up with the idea to use an impact wrench to speed up the bolting process. It went well until we had to unbolt a flange, whereupon we found out that the overpowered wrench had gotten most of the bolts crossthreaded to the plate nuts in horrendously inaccesible places.