Date: Sun, 27 Jul 2014 17:21:52 -0400
Reply-To: David Beierl <dbeierl@ATTGLOBAL.NET>
Sender: Vanagon Mailing List <vanagon@gerry.vanagon.com>
From: David Beierl <dbeierl@ATTGLOBAL.NET>
Subject: Re: Fridge led glowing on its own again.
In-Reply-To: <EE907CDB-B1F1-49EF-A427-1AE9EF3C85A7@shaw.ca>
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 12:36 PM 7/27/2014, Alistair Bell wrote:
>I hope Mark chimes in here, I bet he knows the answer. I mentioned a
>couple of weeks again that my blue led on indicator panel was
>starting to glow when fridge not running. This started gradually
>over the last year and culminated with it being quite bright. This
>is a led I installed in panel about 13 years ago. It's a nice little
>mod, really helps to tell at a glance if fridge is running. That is
>if led working properly.
Did you simply plug in a blue LED in place of the green one (if not, what)?
>Until about a week ago, and now again the led is starting to glow. I
>can't believe that it might be heat from the sun on van body
>transferring to combustion chamber that is the cause, but I keep
>forgetting to check at night or on cool days.
That would be very^3 unlikely.
>What could be the reason for this? Fridge stock except that 110 v
>system not connect, nor is the high amperage 12v feed to resistive heater.
So the only connections to the fridge are thermocouple and ground,
yes? That eliminates a lot of questions right away. By any chance
are the battery LEDs acting odd?
My first guess based on field experience with this board is bad LM324
(the right-hand one IIRC - at any rate the one that handles the
battery LEDs). Unplug the black separate wire to the LED panel, which
is the thermocouple output. If the light stays on that points pretty
convincingly at the chip, unless something has gotten spilled on the
board. If it goes out you have to start chasing things or bet the
investment in changing the chip against hassle with
measurements. This is a high-gain circuit where the difference
between no LED and full brightness is in the small tens of millivolts.
Here's a schematic of the flame detector with a few added goodies:
http://pws.prserv.net/synergy/Vanagon/Pilotmod_schematic.gif The
stock circuit is the part above the line, minus the 5-megohm
pot. This uses the pin-one-end amplifier (out of the four on the
chip). It's wired to have a gain of negative 370,** so when the
input (the left end of the 2.7K resistor) goes negative 10 millivolts
from the reference the output goes positive 3.7 volts which in this
circuit would put about three milliamps through a typical green LED
(and might or might not light a blue one at all). The reference is
the LED-panel B- input, which is where the problems come in, because
we're measuring millivolt differences and the thermocouple itself is
referenced to the fridge gas valve and then by whatever path the
fridge is grounded by. Now I have to stop talking, because either I
made a mistake with the circuit all those years ago or I'm
remembering wrong how it behaved when running the fridge on DC or I'm
not understanding something now. My recollection is that running the
DC would light the flame detector, and I justified that because the
7.5 amps would raise the fridge ground slightly compared to the panel
ground. The problem is that if I'm understanding things correctly
now, that should make the light go out rather than light it
more. That would imply that the thermocouple output is positive to
ground rather than negative, and that the op amp is wired with the
inverting pin wired as reference and the noninverting as input. Or
else I'm just confused. I've got no panel to look at.
**This is an op-amp whose purpose in life is to keep its two input
terminals at the same voltage. It does this by adjusting its output
so that any current flowing through the 2.7K resistor is balanced by
an equal current flowing through the 1M resistor, allowing the input
terminal to stay at (almost) exactly the same potential as the
reference terminal. The - and + on the diagram show that the
non-inverting input is being used for reference and the inverting one
for input, hence the negative gain. The battery and water tank
circuits are wired without any feedback from output to input, so the
effect is to slam the output to maximum with even a tiny difference
between the inputs -- the LM324 has a DC gain of 100,000 when wired open loop.
Yrs,
d