In article , Rick Hughes
writes
On 31/01/2014 17:24, fred wrote:
Delve a bit more into the data for the LEDs to find their peak rating, I
suspect the values given are the nominal and maximum _continuous_ values.
If you can't find it on the sheet then give us a link to the data sheet.
There are other issues as well but one step at a time.
I did try that, and contacted seller ... all he has is what he provided
... he has no other data, he bought a job lot and has no data sheet.
All he could give me was:
100mcd ... forward Voltage 1.8-2.2V
20mA nominal current, 40mA maximum.
I'm happy if we set 20mA as the design criteria, need indication only
not light the room :-)
Without full data for the LEDs it's difficult to say what the pulse
handling capability of the LEDs you have is. Some are better than others
and the different materials used to generate the light make a huge
difference, red generally being more pulse capable than other colours.
After looking at some typical device data and with finger in the air,
I'll say that your nondescript leds will take 75-100mA as an occasional
short duration peak at switching time.
240V peak is 338V which would need 3k9 or 4k7 to limit switch on current
to 87 or 75 mA.
Now consider dissipation in the res with led on:
You have gone for a 470nF dropper which gives a hefty 35mA rms current.
This will be pushing 5W in each resistor which is really too much (even
with a 10W res, it would be too hot to touch). In addition, you will be
running at a peak led operating current of 50mA which again is too much.
Best to see if you can get sufficient illumination with a lower
operating current, even if it means scrapping those leds (Experiment
with a dc supply and a few resistors).
Lets try a 100nF dropper (Xc of 32k).
That gives 7.5mA rms or 10mA pk which I'm guessing will give you
reasonable brightness indoors.
That will give just over a quarter watt in a 4k7 pulse limiting res
which will be fine for a half watt resistor (always best to derate).
Next, peak power dis in the res at switching, 75mA in 4k7 gives 26W peak
meaning that we really need this to be a pulse capable resistor such as
carbon composition (yes they do still make them).
We also need the res to withstand at least the peak volts of the mains
(338V) plus a bit for transients so a 350-500V resistor.
These parts would do the job at a cost of 45p per chain:
http://cpc.farnell.com/CA07116 100nF 300Vac X2
http://cpc.farnell.com/RE06293 4k7 0.5W 350V Carbon composition
Bear in mind that the above is entirely theoretical and should be used
only as a starting point for experimentation.
In your place I would set up one of these on a switching rig (one red,
one green or a few of each) and absolutely hammer them to make sure the
leds and other components survive. At its most simplistic, this could be
running them connected to an old fashioned mechanical plug in timer with
alternating 15min on-off periods.
Starting from scratch I would probably search for some really low
current (say 2mA) but bright and pulse capable leds which would result
in reduced component stress, values and hence cost.
Finally the safety stuff:
1. Safe distances between L & N (and any series components carrying
same) on a board or through air (creepage and clearance) is 4mm. That
will mean removing intervening tracks if using veroboard.
2. Same for between L or N and Earth on an earthed (Class 1) setup is
6mm. 8mm to exposed unearthed metal parts on Class 2 I think.
3. Only use X2 rated caps and 350V or above pulse capable resistors.
4. IMO, it would not be safe to mount the LEDs in a metal panel (even if
earthed), the insulation will not have been specified for mains use. Use
a plastic panel and mount so that even if damaged, the LED junction
would be some way behind the front face of the panel.
Are you sure you don't want to scrap it all and use some skinny neons
;-)
Otherwise, good luck.
--
fred
it's a ba-na-na . . . .