"Tough Guy no. 1265" wrote in message
news
On Sat, 31 Oct 2015 19:10:36 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news
On Sat, 31 Oct 2015 18:14:15 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 22:10:47 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 21:53:16 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 20:52:55 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 20:30:40 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 20:06:02 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 19:04:07 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 18:48:26 -0000, Ian Field
wrote:



"Tough Guy no. 1265" wrote in message
news On Fri, 30 Oct 2015 17:10:19 -0000, Tough Guy no. 1265

wrote:

https://www.dropbox.com/s/fy5xm8lmd3....jpg?dl=0&s=sl

This supply came with a 2 foot 9W strip light of LEDs.
After
1
week
it
began flickering badly. 240V AC in on the right. The
circuit
outputs
75V DC on the left, with a 3.5V AC ripple on it (with
the
100uF
capacitor
included, which I verified is ok). With a 680uF
capacitor
in
its
place,
there is no visible flicker.

What's likely to have broken in this circuit?

What's the chances of it lasting if I run it with the
larger
capacitor?

Forgot to add, I checked the voltage and current output
with
the
larger
capacitor, and it was virtually identical.

When you first switch on, the fully discharged capacitor
looks
like
a
dead
short for the split second it takes to charge up. The
bigger
capacitor
will
cause the turn on surge to last nearly 7x longer.

The LEDs light up very slowly with the bigger capacitor
(for
half
to
one
second perhaps), so I think it's being limited. Also,
when
I
turn
it
off,
won't the capacitor stay charged to the voltage just below
what's
required
to make the LEDs conduct, so it's not going to be empty at
the
next
startup?

If the circuit has a NTC thermistor inrush limiter,
you're
probably
OK.
I usually look for them in any equipment I scrap, when I
still
used
filament
bulbs - one of those added behind the switch plate makes
the
bulb
last
years.

They look similar to a disc ceramic capacitor about 10 -
15mm
diameter,
they
tend to be dull green or black and have a matt finish as
compared
to
capacitors that are usually shiny finish.

I can't see one. There's a resistor connected directly to
neutral
on
the
bottom right of the photo. In series with that resistor
are
the
two
capacitors (blue and yellow), going across to live.

Post a photo.

The link is at the top of this post, you must have missed
it.

Can't see anything that looks like an NTC, or a MOV, adding
them
wouldn't
hurt.

The NTC can be spliced into the mains live feed without doing
anything
to
the board, the MOV should be added across the AC input to the
bridge
rectifier.

If the electrolytic capacitor on the left was too small, the
SMPSU
circuit
would try to even out the voltage regardless - if it has
overcurrent
trip;
too big a capacitor would keep tripping it every time it
started
up
again.
But it might eventually get going.

It appears to be current limited. It takes some time to
charge
the
larger
capacitor. The LEDs gradually brighten.

A constant voltage supply driving LEDs would let the magic
smoke
out.

Well most of the LED lamps I've got change brightness depending
on
the
mains voltage. So they can't be that constant current. Maybe
they
just
have a voltage as a fraction of the mains, then rely on a
resistor
to
limit the current?

I bought various LED lamps from a discount store just to crack
open
and
see
what's in them.

One had a 4 terminal component close to the pads for the LED
wires,
I
couldn't identify the part but it probably has to do with current
regulation.

Among hobbyists, there is a circuit for powering regular LEDs
from
a
single
1.5V cell called a "Joule thief" its basically a flyback
converter
with
no
regulation, but the LED only conducts on the flyback peaks so
there
is
a
sort of limiting.

Another trick is the so called "wattless dropper" - a capacitor
in
series
with the mains so capacitive reactance is large compared to the
dynamic
resistance of the LED load, with large Xc, current doesn't change
much
with
varying mains - some commercial LED bulbs use it.

I was going to make something using a wattless dropper once. I
think
it
was just to power an indicator LED from 240VAC instead of fitting
a
neon.
Something made me decide it wasn't as good as it was made out to
be.

You have to include a resistor in series, the current peaks during
the
highest rate of change of the voltage waveform. Any spikes on the
mains
need
to be attenuated too.

I probably gave up on the idea because of how much limiting resistor
I'd
still need, and decided a neon would use no more power, although
would
eventually expire.

The only neons I've ever seen expire, was ones I was experimenting
on.

You know those big old cooker switches with a socket on them, both
with
a
neon? My parents' left the cooker switch on all the time. The neon
got
dimmer and flickerier over time.

It just occurred to me; that electrolytic capacitor on the end of the
board
may not be such a good idea!

Depending on how it regulates current and exactly where the capacitor
is
connected, if you get an intermittent connection to the LEDs, the
capacitor
ceases to be clamped and the voltage rises - discharge that into the
LEDs
and you've got a problem.

I don't foresee an intermittent connection to the LEDs. The original
capacitor is connected on the circuit board across the wires which lead
to
the LEDs. My bigger one will be external to the unit as it's physically
larger, and most easily connected at the other end of the LED strip
(which
has soldering points at both ends for + and -_.

The SMPSU section should be running at least 15kHz, so flicker would be
due
to something on the primary side.

The flicker is about 25Hz at a guess. Or maybe 50. Perhaps the full
wave
rectification is broken and it's getting half wave?

If you needed any smoothing for the LEDs,
about 0.22uF or 0.47uF would be sufficient.

Well they used 100uF. And I need 680uF to smooth it with the unknown
fault.

With a typical 20W CFL, the mains rectifier reservoir capacitor can be
as
small as 4.7uF - its not that difficult to find non-electrolytic types
with
adequate voltage rating, a few manufacturers do that because
electrolytics
are a reliability black-spot.

I rarely find broken electrolytics unless there's a design flaw and they
were too hot.

About a decade ago there was a plague of dodgy electrolytics - mostly on
PC
motherboards.

You mean the Dell motherboards? Yes, I replaced 150 of those. Some I got
a new board under warranty, some I bought 2nd hand on Ebay, then I started
replacing the caps and ended up with loads of free boards.

Industrial espionage gone wrong - one firm broke into a competitor's
office
and nicked their electrolyte formula, and then found out the formula was
incomplete.

Its the single biggest cause of ATX PSU failures - which may or may not
destroy the MOBO.

The only failures I've seen with ATX PSUs is when they're overloaded,
because cheap ones simply cannot give out more than half the current
they're rated at. And it's never the cap, usually several things melt.

I've spent a fair bit of money on capacitor test instruments because

My multimeter tests up to 20uF.

capacitors are responsible for such a large proportion of faults.

Aren't broken caps usually vented so obvious to look at?

A lot of them simply dry out, so there's no electrolyte to produce steam
when they get hot.

Some leak through the plug seal and corrode one of the leads.

Some PSU applications require such low ESR capacitors that brand new
"regular" types would fail immediately.