Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
Thanks, folks. . .what I wanted always seemed to be just one more mouse click away. . .
For the Everlight white LED 333-2UWC/CB data and curves it seems this LED can be modeled by an ideal 3v Zener and an incremental resistance of 33Ω in the 0 to 30mA range, to give Vf = 3.0v to 4.0v.
@ If=20mA, Vf = 3.8v avg and 4.5v max (i.e., 97.5% of the time the Vf will be <4.5v at this current) which gives me some idea of the tolerance on Vf.
One curve sort of shows brightness = power dissipated, which is strongly dependent on If for an LED.
Using this model I think I can predict the max/min expected current flow through an LED given a voltage source driving the LED, or given a voltage source and a series resistor with known tolerance driving the LED.
Given a white LED modeled as a 3.5v +/-1v Zener is series with a 30.00Ω resistor.
If the Zener voltages in a batch of LEDs are normally distributed around 3.5v and two standard deviations equals 1v, then if you connect 100 randomly picked diodes to a 4.5v source, 96 will give a current of 0 to 67 mA and you won't need instruments to notice this variation. LEDs kind of reveal themselves, like an ammeter that reads out in brightness.
Now connect 30 ea. of these diodes to a 30 x 4.5v source. If the SD of one diode = 1v then the SD of this series string will be sqrt(30x[1^2]) = 5.5v.
Now the current will vary from 27 mA to 39 mA and if all the LEDs are lumped together this brightness variation probably won't be noticed.
I'd be a little careful however of any data sheets i look at when it comes
to white LEDs. Some of them are much older than the newest LEDs out there.
This translates to a lower average voltage than shown on the data sheet.
Of course also to be considered is the LED aging, which can reduce
LED forward voltage drop over time even at the same current.
I'd be a little careful however of any data sheets i look at when it comes
to white LEDs. Some of them are much older than the newest LEDs out there.
This translates to a lower average voltage than shown on the data sheet.
Of course also to be considered is the LED aging, which can reduce
LED forward voltage drop over time even at the same current.
So I have to guess at wider tolerance limits, or measure >10 LEDs picked at random at some forward current of interest to pull out each Vz and each series R. If I do one of these for each Vz and R
**broken link removed**)
I'll spot outliers, if any.
LED aging must be small for a typ. LED operating lifetime of 100Khour, no?
It's significant in LEDs like the white Luxeons, being -0.2v or greater.
A bunch of people complained about this in the past on other web sites.
After some hours (a surprisingly short time like hours) their flashlights
would get brighter and hotter.
At this point the only thing i can say for sure is if the string (or single LED)
is driven by a constant current it is sure to work for a long time, but
otherwise it's hard to say unless of course the resistance relative to
the overhead voltage is high enough to act more or less as a constant
current as the voltage of the LED(s) changes over time.
In other words, if the supply voltage is two times the LED voltage it
will probably work ok for a long time, but if the supply voltage is very
close to the LED voltage (to make it more efficienct of course) then
we can expect problems as the LED ages because the current will
change much, and it's a negative voltage changes so that means
higher current.
I also do not have any data on the smaller Nichia type white LEDs
though.
Then again if the constant current regulator has to drop more
voltage, it's going to get hotter too.
It's significant in LEDs like the white Luxeons, being -0.2v or greater.
A bunch of people complained about this in the past on other web sites.
After some hours (a surprisingly short time like hours) their flashlights
would get brighter and hotter.
it's hard to say unless of course the resistance relative to
the overhead voltage is high enough to act more or less as a constant
current as the voltage of the LED(s) changes over time.
In other words, if the supply voltage is two times the LED voltage it
will probably work ok for a long time, but if the supply voltage is very
close to the LED voltage (to make it more efficienct of course) then
we can expect problems as the LED ages because the current will
change much, and it's a negative voltage changes so that means
higher current.
Then again if the constant current regulator has to drop more
voltage, it's going to get hotter too.
Yeah, I'm looking for these voltage/resistor limits and maybe find some optimum operating points.
It turns out my spreadsheet is not very accurate at the extreme ends, probably because it uses a discrete distribution to model a continuous one. I have to work on this some more.
I've looked at the temperature response too and i have seen that the
forward voltage changes with temperature too. I would think that if
you figure +/- 0.5v the design would probably be stable.
I think you could do some tests of your own too, by running an LED
at the required current for 24/7 and see how much it's forward
voltage changes after a few weeks, months, years even if you have
the time. You'll have to make sure the temperature is the same for
each measurement though.
I've guessed how much the temperature of an LED changes by measuring
it's foward voltage with constant current as the LED self heats after
turn on. I can get a rough idea what temperature the die reaches.
I've guessed how much the temperature of an LED changes by measuring
it's foward voltage with constant current as the LED self heats after
turn on. I can get a rough idea what temperature the die reaches.
I've used that figure yes. What i was doing was trying to find out how
well my make shift thermal compound was working on a high power LED.
I tested one with more or less regular high temp epoxy and another one
with epoxy that was made just for heat sinking, and they both came out
fairly close. I was surprised how well the regular epoxy worked, but then
i made sure to keep the layer very very thin, which as im sure you know
means less distance for the heat to conduct through, and so it still keeps
the LED cool enough even at full power (with a decent heat sink area).
I've used that figure yes. What i was doing was trying to find out how
well my make shift thermal compound was working on a high power LED.
I tested one with more or less regular high temp epoxy and another one
with epoxy that was made just for heat sinking, and they both came out
fairly close. I was surprised how well the regular epoxy worked, but then
i made sure to keep the layer very very thin, which as im sure you know
means less distance for the heat to conduct through, and so it still keeps
the LED cool enough even at full power (with a decent heat sink area).
So you could figure the thermal resistance of the epoxy layer in °C/W/mm of thickness.
It might be the same per mm as those insulating mica washers or conductive grease they use.
Yes exactly, and i did calculate the thermal properties and since the heating
was very much the same as with the purchased thermal epoxy i figure that
it was almost the same as the real stuff. I was very surprised to find this
out, because the stuff made for the job has aluminum in it for better conduction
and the run of the mill stuff was just that, any ol' stuff that wasnt made for
good thermal conduction. I guess i was expecting some usable results however,
or i wouldnt have applied the run of the mill stuff to the LED and heat sink,
but i never expected it to be so close to something that was actually made
for the job.
It's too bad i could not use the exact same LED and heatsink, although i used
another LED of the exact part number and the same exact size and make of
heatsink, so i feel that the two results were comparable.
What i did was apply the epoxy, then squished it down real good and moved
it around, sliding it back and forth a little, to get the epoxy layer as thin
as absolutely possible. I thought that just about anything should work
pretty good if it is kept THAT thin Luckily, it worked out, otherwise
i would have had to chisel the LED off and use actual thermally conductive
epoxy.
On a related subject, i have read about the electrically conductive paints
or pastes (whatever they are) for fixing or even making conductors.
I havent actually tried any ever however, and they are a bit expensive.
Perhaps you've had some experience, or maybe i should make another
thread about this? A friend asks me about this too.
I can tell you that the visual area of contact is about 1% of the actual area that is touching, which accounts for high contact impedance, either electrical or thermal.
There is some contact goop you can buy to lower the electrical resistance for switch or relay contacts and thermally conductive grease lowers the contacting surfaces thermal resistance for the same reason and by the same mechanism.
Have a visit in a dollar store. They sell 3-LED flashlights with tree white LED's for a buck. Then you'll be able to experiment by yourself to plot the E-I curve.
One caution: These LED's are very fragile to excesive reverse voltage. Be sure you FORWARD bias them.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
