Derf transform applied.
Fred Bloggs said:
Larry said:
message
[Apparently quoting the OP]
I've got a simple pic based controller that turns on power via a relay
to a 1000W transformer used for high energy lighting. When I turn the
power on the processor ends up reseting. I've traced this back to about
a 2 micro second low going pulse break in the power to the processor.
Adding capacitance, and even putting in diode based filter has not
helped. I'm guessign the emf field from the transformer is causing
inductance via the wires on the prototype (this design has not gone to
board yet). The power glitch happens on the output of the LM2940 5 volt
reg I'm using, and it is supplied by a 24 volt DC off line power supply
that doesn't glitch. I've tried various caps on the regulator with no
help. I'm switching the power anywhere in the AC cycle, and it doesn't
always reset. Anyone have any ideas of things to try?
I rewrote the software so that a reset is part of the normal power on
and off sequences for this device, if it happens. While this works I
find it a real cop out.
This is almost certainly being caused by the relay and its solenoid field cutting the area formed by the 5V regulator to Vdd to
GND and back to the regulator. You can verify this by supplying power to the relay through two twisted wires as well as moving it
away from the PIC circuit altogether.
Most relays have a closed magnetic path going through
iron everywhere except for small construction gaps and
the gap which varies to operate the relay. Allowing much
flux into the surrounding air is bad for the efficiency, and
would be a sign of a poorly designed relay, if it happened.
[derf] I like that "gap which varies to operate the relay" [derf]
the very principle of operation is to "vary" the gap to "operate" the relay [derf]. How the hell do you pull in the armature
otherwise
[derf]
When considering the magnetic circuit, it is useful to
separate the incidental gaps due to construction from
the gap for which the magnetic circuit exists. You need
not get all excited thinking I meant to write a primer on
magnetically induced motion. I merely reminded readers
that only certain flux does any good in such a device.
I notice that you chose to not deal with my point, which is
that flux setup outside the relay is wasted and minimized in
well designed relays.
I have used quite a few relays in close proximity to other
circuits without seeing the magnitude of effect you claim,
including circuits made with single-layer etch for cost
reasons. So I am quite skeptical that your "certainly"
is well founded in reality.
[derf] what you seem to be missing is the timescale of the interference with 2us being a bit short for anything induced by a line
frequency action-
The reliable observation is the system malfunction. With
the breadboard construction reported by the OP, that
"about a 2 micro second low going pulse break in the
power to the processor" is legitimately suspect. I think
you need to learn to weight the evidence you consider.
not to mention repeatability going against an unsynchronized activation-
The OP stated "I'm switching the power anywhere in the AC
cycle, and it doesn't always reset." This repeatability issue
works directly against your "almost certainly" hypothesis.
whereas the DC coil circuit is self-synchronizing.
Does that mean it turns on when it turns on?
That fancy language of yours is confusing.
[derf]
Since you seem so fond of your "solenoid field cutting the
area" idea, I'm going to provide more detailed reasons
for an objective person to discount that hypothesis.
The OP mentions a problem encountered upon energizing
a relay and mentions only 24 VDC or 5 VDC as sources
that might be used for that purpose. Let us assume,
(or divine, as you wish), that the highest of these is
applied to the solenoid. (This assumption has little
bearing on the end result since ampere-turns are held
nearly constant for a given coil size.) The problem, as
deduced or divined by Fred, is that a voltage has been
induced by flux emanating from the solenoid sufficient
to interfere with a 5V regulator powering a uP.
Most people familiar with the elementary theory of
induction, back EMF, magnetics, and the like can see
that the flux linking that solenoid will then change
at a maximum rate of 24/N Webers/Second, where N is
the number of turns on that solenoid. Those people
can also see that the maximum voltage this changing
flux may induce in a single circuit loop outside of
the solenoid itself is k * 24/N Volts where k is the
fraction of flux that manages to escape the closed
magnetic circuit I mentioned ealier as being the most
probable form of that solenoid.
Consdering that the only flux that acts to operate a
solenoid is that which is in the volume that changes
as the intended motion occurs, it should be obvious
that suffering the IR losses necessary to establish
similar flux levels in many times that volume would
be very wasteful. A brief examination of how most
relays are constructed reveals that their designers
are not that incompetent. They arrange that most of
the flux path is either iron, (which conveys the flux
without much stored energy or required MMF), or air
which develops useful force due to that flux.
Having never yet seen enough flux linkage to worry
about, I have not measured the k value for any relays
yet, but values in excess of 0.1 would surprise me a
lot if exceeded by any but a small fraction of relay
designs out there. (I exclude reed relays from this;
they are inefficient and irrelevant to the OP's issue.)
This estimate is based on common constructions I have
seen and the idea that, when driven at their intended
levels, the iron in the path tends to not saturate during
energization. (The magnetic circuit is gap dominated.)
I have unwound a number of solenoids wound for 12 V
and 24 V service. They invariably have many dozens
if not a hundred or more turns. A 24V coil with as
few as 25 turns would be most unusual.
Based on these estimates and observations, I expect
the ordinary magnitude of Fred's "almost certainly"
induction to be less than 0.1 * 24V / 50 turns = 48 mV,
and most likely much less.
But of course, in Bloggs World, we must make a few
concessions to the vast variability of the "real"
world. So, let's allow that the relay designer was
able to get only half the flux where it could ever do
any good, (and a much smaller fraction of the field
energy there), and make it a truly high powered coil,
having, say, only 10 turns. Under those conditions,
we might see 0.5 * 24V / 10 = 1.2 V. By gosh! That
*must* be it!. The OP has spread out the feedback
for his three-terminal regulator over enough area to
capture a significant fraction of that wasted flux!
And it is seeing the 1.25 V (or so) that it wants
when there is not a proportionate output but merely
the illusion of it induced by that humongous and hot
coil designed by one of the multitude of idiot relay
designers sadly allowed to live in Bloggs World.
Another mystery resolved by astute divination.