Noise Problem. Both Comms Breaking Squelch

MikeMl writes:


[I've seen car owners yank the battery cable off while the alternator is
going full tilt; "The battery only starts the car; the alternator runs
it..." The result was a 65V+ "load dump" hundreds of ms long into the
car...fried computers, dead stereos, you name it.]


I dont have an internal schematic of the LR3C to know exactly what the
crowbar does. Two possibilities:

One is that the crowbar is upstream of the regulator so it just blows
the Field Breaker, effectively just removing power from the
regulator/field, leaving whatever current is already flowing in the
field inductance to dissipate itself in its own coil resistance. There
is almost certainly a catch (snubber) diode across the field winding
oriented such that the field current decays with a time constant of RL,
where R is the coil resistance, and L is the coil inductance.

I'd hope.. but is there one??

Two is that the crowbar is downstream of the regulator (directly across
the field winding).

Doesn't this assume the regulator is high-side; with the bottom of the
field grounded? I thought the consensus was they were low-side, al-la GM;
i.e. the top end of the field gets fed with +13V from the aux diodes or
battery; and the regulator is in the leg to ground.

Neither method (unless the LR3C is a lot more complicated than I think
it is) prevents a "load dump" as David Lesher describes it! This is why
I think it is a poor design. It is no better than a more conventional
OVP module such as used in Cessnas and Pipers which simply breaks the
connection between the Field Breaker and the input to the regulator.

The issue there is opening that will need a fast, high voltage device
that you can turn off. The SCR crowbar gets turned on to protect,
an easier task.

The issue with load dumps is not just the spike peak voltage; it's
how long they last...i.e. total energy. The no-battery stunt was so
harmful because the Xl of the field ensured it could not halt alternator
quickly. So it wasn't a few ms of spike, it was many hundreds of ms
worth...



Actually, I have been thinking about Steve's problem some more, and the
transient which is tripping the crowbar could well be coming from a
"load dump lite".

I don't recall the OP's remarks; did he indicate the pump cycling
was tripping the breaker? Or was it an unknown?

I agree the pump cycling could be an issue; your savior is often the
low impedance of the battery eating those spikes. If it's not, the
spikes may have gotten bigger, or the Xc of the battery lower...

One spike solution is large MOV's across the spike creator. The gotcha
there is: MOV's actually are sacrificial, as small spikes erode them
over time; large ones will cause them to explode... [If you ever shake
a "surge protected outlet strip" and it rattles; that's the ex-MOV
inside...]


I called B&C today to ask about this, and the only suggestion was to
connect pin 3 (sense input) of the LR3C as close to the battery as
possible. In other words, the battery is the spike filter of last
resort, so if there is a lot of wire (resistance) between pin 3 and the
battery, then the impedance along the wire (and in the master relay)
could allow pin 3 to see a higher voltage during the transient.

The other reason to do so is because the regulator is there to protect &
serve the battery. You don't care all that much about the bus being a
half volt too high or low; you DO care about the battery being that, and
the bus does have drop. So you want to sense near/at the battery.

But that brings a dilemma. You must have the sense line draw some
current. Why? So the regulator is sure its sense line has not become
disconnected from the battery. {If it does come loose, the regulator
says "0 volts! More power to the field, Scotty" to compensate, and
the smell of burnt silicon soon fills the area. There were some early
transistorized auto regulators that did just that; long gone I hope!} But
then, even with but a few mils of draw, if it does that for 2 months in
the hangar....your battery is dead. So it must be switched, and that....


The other possibility is to open the box, and put an appropriate filter
between pin 3 and the OVP circuit such that it doesn't effect what the
VR sees.

I wonder if an appropriate ferrite toroid on the sense lead could stop the
noise spikes from tripping the OVP without degrading regulation. I can't
recall what such toroids look like [i.e SPICE] when saturated with DC as
it would be...