DIYbanter - View Single Post

Joseph Gwinn

In article ,
wrote:

On Mon, 08 Nov 2010 09:54:06 -0500, Joseph Gwinn
wrote:

In article ,
wrote:

On Sun, 07 Nov 2010 16:26:14 -0500, Joseph Gwinn
wrote:

In article ,
Bill Noble wrote:

On 11/7/2010 8:56 AM, Joseph Gwinn wrote:
In ,
[snip]

But I need to point one thing out: A transformer feeding a resistive
load like a hot wire is not an inductive load unless the transformer
in
question has extremely large leakage inductance (transformers in
tombstone welders would qualify).

The presented load is in fact largely resistive, and I think the
point
made by another poster about DC unbalance saturating the transformer
core is a key issue. If one is in fact driving an inductor, one can
have both problems at once; the problems are independent.

In any event, Lutron et al make semiconductor dimmers intended to
drive
resistive loads through step-down transformers.

Joe Gwinn

Joe - not correct because of the way triacs work - if driven by a pure
sine wave, it is true (or nearly so). But if you turn on say 90 deg
after zero crossing you have a huge current spike - that is due to the
inductive nature of the transformer and the fact that there is no flux
in the core at turn on.

It's true that the triac turns on abruptly, in a microsecond or two. At
90 degrees, this imposes a full voltage step on the transformer input.
But, wouldn't inductance slow the spike down, whatever the state of the
core? This is a classic homework problem. The current rises linearly
from zero, and the higher the inductance the slower the rate of rise.

You would, ideally, use a reverse phase dimmer/controller. They turn
the current OFF at a particular phase angle, instead of on. Some of
the high end dimmers for magnetic transformer low voltage lighting are
reverse phase dimmers. They are NOT cheap. They do, however, reduce
the EMI/RFI emissions significantly and almost totally eliminate the
filament hum common to standard Thyrister dimmers (which Variacs also
totally eliminate , which is why they are still used in many studio
lighting systems). Darn electronic theatre lighting systems are pretty
noisy - I work next to one several hours a day.

Reduce filament noise and EMI? That certainly makes sense. There are
many patents on reverse-phase dimmers as well, with reduced noise and
EMI given as advantages.

As for variacs and EMI, I have a war story: In the 1970s, I lived in
Baltimore. A friend of mine was an electronics tech at Johns Hopkins
University Hospital. The hospital did much basic research, and the
electrophysiology department had a screened room (big walk-in box made
of copper window screen fabric and having a metal door), within which
they attempted to record nerve potentials and pulses, but were greatly
hindered by electrical noise despite the screenroom.

My friend was asked to solve this problem. It developed that the
overhead lights, which shone through to copper screen fabric, were
fluorescent. Ouch. No wonder. Replaced the fluorescent lamps with
incandescent lamps controlled by a wall-mounted variac dimmer. The
noise level inside the screenroom dropped to insignificance (relative to
the inherent noise of the instruments). One could also dim the lights,
making the instrument displays easier to see. The doctors were very
pleased.

I would guess that the EMI from the fluorescent lamps was enough to get
through the power-line EMI filters where power enters the screenroom to
supply the instruments. The good doctors probably also left the door
open, but the powerline path was probably by far the most important, as
only noise below a megahertz or so can affect electrophysiology
experiments.

Joe Gwinn

Work out the math if you don't believe me, or
try it with a transformer and a battery and a scope - measure the
current pulse at turn on with say 12V applied to the 12V winding of a
transformer - you are looking for the first 1 ms of current

This sounds like we would have significant DC current through the
transformer winding, which could saturate the core. A saturated core
will certainly cause a current spike. Maybe I don't understand the
proposed test circuit.

I recall that Lutron had patents in this area, and these patents
provided a summary of the issues to be solved, so I did some searching.

A good discussion appears in US patent 4,876,498. A later version of
the same patent is 4,954,768. Leviton reacted with their own solution,
in patent 7,482,758. And 5,477,111 speaks directly to control of
inductive loads like motors. (Go to http://www.pat2pdf.org to get
copies.)

Anyway, the big issue in these patents is DC causing saturation, and (in
two-wire circuits) inductance-caused errors in knowing when to trigger
the triacs.

Joe Gwinn
The emi from the flourescent bulbs was not just going back the power
lines. The radiation from the bulbs themselves was significant, and
anything short of a full faraday cage wasn't going to catch it all.
The copper screen was not sufficient (either not full enough coverage,
or not grounded, etc)

It was a commercial faraday cage, and I'm pretty sure it was grounded.
If closing the cage door solved the problem, my friend would not have
been given the assignment.

My reason to think that conducted EMI was the issue is that those lamps
are far too small physically to be very good at emitting noise at 1 MHz
and below. Nor would 1 MHz waves fit through the door, so leaving it
open would have little effect, as observed. At 1 MHz, the wavelength is
300 meters.

Joe Gwinn