main relay vs ignition switch while driving

well, mine have never lasted 20 years that's for sure. i'm 6 for 6 on
88-91 civics having this problem. i first encountered it when a vehicle
was only 10 years old, and its previous owner had evidently had problems
with it for some time prior to selling.

regarding solder, this is a soft alloy that operates at a highly
elevated temperature relative to its melting point. expose it to
thermal cycling [the relay runs hot you'll notice] and you have a
problem just waiting to happen. the solution is to either use a
different switching arrangement that doesn't generate as much heat
[cycle] /or/ to use a different jointing method like spot welding or
crimping. but the relay manufacturer should know all this. i still say
this relay is a cheap and cheesy design. the circuit board is low
quality and the relay internals are designed primarily to prevent
intervention, not for serviceability [either kind]. i say mitsuba knew
exactly what they were doing with this relay right from the start
[relays are old technology and their problems are well known] and that
they elected to go for what they knew would result in life limitation.

life limitation is nothing new. i once had a car clock that failed. on
disassembly, i discovered a soft solder rivet had separated breaking the
electrical supply. the interesting thing was, the rivet was held in
tension by a spring! solder [lead] tends to creep over time, especially
when kept warm. life limitation? you bet! there was no other
practical explanation for the rivet/spring combo. a fuse would have
protected against overload and the spring had no mechanical function.
anyway, i soldered a wire in place instead and the clock worked again,
just like a repaired relay.

 

Something like that. The crux of the problem is that the solder is used to
support the relays instead of the relays being cemented to the board and
solder used to stabilize the connection. Beginners are taught not to use
solder for strength or for carrying current, but I guess relay module mfrs
aren't bound by the same code. They use solder for both simultaneously.

Mike

 

Disclaimer - only my guess:
Lead alloys seem to make better contact when carrying current, as seen in
car batteries with broken internal connections. I realize battery terminals
do the opposite because of powerful corrosive influences. Anyway, I suspect
the current creates a hot spot that effectively welds a small spot in the
connection and adjusts itself in size so that the connection stays slightly
below the melting point. When the current stops and the spot cools it is
prone to cracking, and something must change when the whole thing cools off
even more.

As I said, just a guess.

Mike

 

Yeah, the solder will fail in 8-10 years. But it's the
electrical components that will last.

Spot welding or crimping would be expensive and not already in
their machine assembly line. One solution might be to replace the
relays with transistors, heat-sinks and some breathing holes.

I don't know why a spring is put into a [digital?] clock without
a purpose. You sure the spring isn't a spring resistor?

 

transistors always drop voltage - not necessarily what you want. and
30A rated devices aren't cheap.

it was an electric dial clock. no, the spring wasn't a resistor - it
would have been rated at ~10+W for a milliamp application. a "real"
resistor would have done the job better if that was what was required.
besides, it was hooked onto the chassis at one end, and onto the tab
retained by the lead rivet at the other - no reason to use a lead rivet
when it could have been brass, copper, etc.

 

They also are much more vulnerable to failure than a simple
electromechanical relay.
Lots of destructive spikes in an auto electrical system.

Maybe a bimetal thermal protection device?
If too much current is drawn,the strip bends and breaks the electrical
connection.

 

indeed. interestingly, since i fitted magnecor plug leads to my civic,
shifting has been much better [it's automatic]. i haven't scoped the
"before & after" on any of the ecu inputs, but it seems to me that if
ignition-induced signal noise is suppressed, the ecu can manage its
business a bit better. probably wouldn't be noticeable on a stick, but
on the auto, it doesn't "fluff" around shift points any more. weird,
but most welcome. i haven't touched anything else. didn't even take
the plugs out.

no, wasn't bi-metal, it was literally a lead-type rivet holding a brass
tab into which one end of the spring was hooked, with the other end of
the spring hooked into another part of the chassis. maybe it was
bismuth with a low melting point like those trick spoons. but the point
is, it served no purpose that a fuse couldn't perform, and once failed,
the clock not only didn't work but was "irreparable". bizarre why
anyone would go to such lengths to obsolete a clock.

 

It's the current you want not the voltage. Transistors are proven.

Common, solid state devices have high marks for reliability than
mechanical relays. A relay will have moving part that can produce
arcing. Also, any spikes can be quashed by a shunt diode.

 

it's watts you want to look out for. watts = volts x amps. if the
voltage drops across a p-n junction, and it does, about 0.6V, at 30A,
that's 18W you're wasting [and heat you have to dissipate]. relays, for
all their other faults, don't have that kind of problem.

http://en.wikipedia.org/wiki/Diodes

agreed, but they're not always the best solution. using silicon
transistors on a 12V system is going to step down your voltage 5% each
time it encounters a junction. that makes relays look attractive from
an efficiency viewpoint.

 

In applications where zero voltage drop is important there are already
ways to do this by selecting the appropriate transistor. A small voltage
drop of (.2V or .6V) isn't enough to slow down the fuel pump.

The relays used in the main relays can cost more and will generate
heat. So much heat that you can't put your fingers on it for more
than a second. Eventually, it'll run so hot that the contacts will
weld together.