Hi Peter,

PeterD wrote:
On Fri, 11 Jun 2010 13:29:06 -0700, D Yuniskis
wrote:
GregS wrote:
In article , D Yuniskis wrote:
I recently had to repair a small LCD TV that was
blowing -- very SLOWLY -- it's DC mains fuse.
I.e., the set was drawing more current than it
was designed to draw. But, not a catastrophic failure
(e.g., nothing *shorted*). In fact, the set would run
for a day or more at a time "perfectly".

The fused supply fed the primary DC-DC converter for the
set. I.e., damn near all of the loads hang off the multiple
outputs of the switching transformer.

After tracking down the problem, it occurred to me just
how hard it is to do such things -- since schematics never
tell you what sorts of *currents* pass through each circuit
node (though you can often find indications of *voltages*).

So, how *should* this problem have been approached (without
risk to the set), out of belated curiosity?
I often use light bulbs on the mains to limit current.
Not an issue with this set. The set ran "normally" (except
blowing the fuse every day or two).

I also think SMPS would make attempts at limiting current
pretty useless (for such small loads)

i don't have one, but the old Textronix Hall current probe
was usefull for tracking shorts.
Again, no shorts here. Everything *worked*.

I also use the hand method of feeling warm paths and parts.
Sometimes an IR temp meter.
I think a PIR imager would have been helpful. Though not
sure if even that would have helped (though it probably would
for some types of failures)

Have you tried putting a meter on the output of the power supply to
see if the current draw is in side the monitor,

Set consists of a "brick" and the display itself.
Brick provides three (?) separate supplies to the set:
- inverter power
- audio amp power (?)
- the rest of the set

Inverters are fused separately and individually.
Fuse that was failing was in the "rest of the set" branch.

That fuse feeds the main switcher *inside* the set
(not to be confused with the switcher in the *brick*!).
The internal switcher seems to have five or six outputs
(i.e., all the various supplies needed by "the rest of
the set")



and how much more than expected the current is?

But that's the point -- you have no way of knowing what
the "expected" current is! The nameplate will tell you what
the nominal power requirements are (from which you can deduce
the nominal current). But, that applies to the entire set!
How much current should flow through each of the fuses
(two for the inverters plus two for the audio + rest_of_set)
isn't something you can know with any certainty.

Except to say that the fuses' ratings should exceed the
nominal current they are each expected to pass! :

Given that the set runs for a day or more without blowing
the fuse in question suggests the "excess" isn't that
severe (or, the fuse is significantly overrated *and*
the brick is overdesigned to be capable of supplying
that "excess").

You also may want to check to see if the
backlight system is what is drawing the excessive current.

See above.

Trying to current limit the input of a SMPS is usually not viable. As
input voltage drops, the SMPS simply tries to comphensate, until it
reaches an unstable point then (often bad) unexpected things happen
(or it simply shuts down).

Correct. All you do is change duty cycles.

My question regards how you can get this sort of information
from service documents. I.e., at best, you'll see voltages
marked at various test points in the circuit. And *maybe*
some *nominal* waveforms. But, nothing that would clearly
allow you to diagnose this type of problem.

(e.g., drawing enough extra current to blow the fuse implies
the duty cycle of the internal switcher is higher than
intended. But, nothing in the types of documentation
that I just mentioned would tell you what the duty cycle
*should* be "for a typical load")