[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Preventing a REALLY expensive mistake.

Original poster: "Jim Lux by way of Terry Fritz <twftesla-at-uswest-dot-net>" <jimlux-at-earthlink-dot-net>

> Original poster: "Christopher Boden by way of Terry Fritz
<twftesla-at-uswest-dot-net>" > 
> >Original poster: "John Williams by way of Terry Fritz
> >       If I were in that situation I would give serious thought to
> >large industrial scale isolation transformers on the power feeds into
> >the high voltage lab.
> Been looking into this. Anyone know where I can get a 100 to 200A 240VAC
> transformer?

You, mean, of course, 25-50 kVA, as that is how transformers in this
general size are specified.
That's a pretty big isolation transformer, and a fairly big "dry
transformer", to boot.

Most isolation transformers (ones designed specifically for this kind of
use, with good common mode rejection between windings, shielded windings,
etc.) are in the under 5 kVA range.

As a practical matter, any reasonable transformer, combined with some good
line filters, would probably do as well.

Do you really NEED 50 kVA?  How many computers are you running?  You might
also want to consider running lots of littler transformers, with the
transformers closer to the loads.  If you use the one big feed, then the
feeders on the isolated side are going to be pretty long, and have great
opportunity for picking up noise.

Here at JPL, we generally put a separate iso transformer in each rack.
Then, there is a carefully designed grounding scheme to make sure we don't
get loops, etc.

You also need to be aware of grounding issues, particularly with lots of
isolation transformers.  The best isolation transformer in the world (i.e.
a DC or mechanical link) isn't going to help if the bad stuff gets through
on the grounds.

> I've been kicking around the idea of completely covering the floor with
> steel plate. If the plate were well grounded I would think you could stand
> on it while discharges struck it and you'd be fine. I'd have to account for
> creeping and such (it couldn't go wall-to-wall because the floor would have
> a different expantion rate than the steel and the steel would buckle) but if
> I left 6" on 2 walls it sould be fine.

Get any of the textbooks on HV lab design, and you'll have all the info you
need. Your problem is similar to that for substation grounding where they
want to make sure that the "step potential" (i.e. the voltage difference
between your feet when you take a step) isn't too high when a lighting or
switching impulse is being conducted through the ground.

There are also IEEE specs on this sort of thing.

> If you do have to build your own walls I would
> >think about turning the computer area into a faraday cage.  All it would
> >take is a bit of chiken wire and some serious crimping tools during
> >construction.

Building a GOOD (as in functional) faraday cage is non trivial.  You might
wind up spending a lot of time and effort for not very good performance.

You need to address several issues:
1) What are you worried about (i.e. line voltage spikes, EMI for radios, etc.)
2) What is the path for the interference (conducted over power lines,
radiated through the air, conducted through the walls?)

Then, you can attack the things that need work.  By and large, TC's don't
radiate very much RF power, so a bit of distance will probably be all you
need.  I'd worry about conducted interference, possibly coupled from the
secondary via capacitance. (i.e. if you have a power line running through
the wall next to the coil, it might act as an antenna and pick up a
signficant amount of power).

For conducted interference, the key is making sure that all of your current
paths don't overlap.  That is, make sure that the path the return current
from the sparks takes doesn't share ANY conductors with the wiring to the
computers. (i.e. bring everything back to the panel to join together).  For
the last, watch out for capacitive coupling between feeders in the same
conduit or raceway.

For instance, if you set up a ground grid under the coil and in the walls,
you want the spark current to flow back to the bottom of the secondary, not
to some other ground rod in the yard, the service panel, etc.