LED system
LED system
(OP)
Hi there, new member ...
I want to build a low voltage system consisting of a 120vac-connected custom power supply that drives multiple series-connected strings of removable LEDs.
A 32vdc power supply would drive multiple current regulators, each consisting of an LM317 and a current-setting resistor, and driving a series-connected string of up to 8 removable LEDs. The LEDs would have a forward voltage drop of about 3v and a nominal forward current of 350ma or so.
Allowing for the 3v drop of the LM317, and some wiring drop etc, the 32vdc supply should allow 8 or 9 LEDs in each string (assuming the variation in LED forward voltages averages out). If a string only has 1 or 2 LEDs, there will be a pretty significant drop across the LM317, so they should have good heat dissipation, perhaps TO-220 packages screwed to a metal chassis.
By putting a potentiometer in series with a fixed resistor on each LM317, dimming should be achievable. The fixed resistor would be sized so that when the pot is at 0 ohms, the LEDs in the string are at maximum desired brightness (current 2-3x nominal, or up to 1 amp). With the pot rotated the other way, the string is effectively "off".
Since the LEDs in each string are removable, they can potentially be connected with reverse biasing (correct polarization of the LEDs cannot be guaranteed). So each LED will be packaged with an "anti-parallel" (anode of each connected to cathode of the other) ordinary silicon diode (such as 1N4007). When the LED/diode pair is installed one way, the LED is forward-biased and the 1N4007 reverse-biased (the diode can handle the possible 32v reverse voltage indefinitely). If installed the other way, the forward-voltage spec of the 1N4007 insures that the reverse voltage on the LED cannot exceed 1.5v or so (and the Cree LEDs I'll likely use tend to allow 5v).
The same problem as old-fashioned Christmas lights exists: if one burns out, the whole series string is killed. But LEDs have lifetimes in the 10,000s or hours, so the MTBF of a string of 8 should still be long. The culprit can quickly be identified by reversing each connection until the string lights up (because the 1N4007 paired with the dead LED will then be forward-biased).
LEDs in each string will be guaranteed to have the same current and similar brightness. Extra-bright assemblies can have 2 or more LEDs in series (protected by a "backwards" silicon diode).
I'd appreciate any comments on the above, thanks !
I want to build a low voltage system consisting of a 120vac-connected custom power supply that drives multiple series-connected strings of removable LEDs.
A 32vdc power supply would drive multiple current regulators, each consisting of an LM317 and a current-setting resistor, and driving a series-connected string of up to 8 removable LEDs. The LEDs would have a forward voltage drop of about 3v and a nominal forward current of 350ma or so.
Allowing for the 3v drop of the LM317, and some wiring drop etc, the 32vdc supply should allow 8 or 9 LEDs in each string (assuming the variation in LED forward voltages averages out). If a string only has 1 or 2 LEDs, there will be a pretty significant drop across the LM317, so they should have good heat dissipation, perhaps TO-220 packages screwed to a metal chassis.
By putting a potentiometer in series with a fixed resistor on each LM317, dimming should be achievable. The fixed resistor would be sized so that when the pot is at 0 ohms, the LEDs in the string are at maximum desired brightness (current 2-3x nominal, or up to 1 amp). With the pot rotated the other way, the string is effectively "off".
Since the LEDs in each string are removable, they can potentially be connected with reverse biasing (correct polarization of the LEDs cannot be guaranteed). So each LED will be packaged with an "anti-parallel" (anode of each connected to cathode of the other) ordinary silicon diode (such as 1N4007). When the LED/diode pair is installed one way, the LED is forward-biased and the 1N4007 reverse-biased (the diode can handle the possible 32v reverse voltage indefinitely). If installed the other way, the forward-voltage spec of the 1N4007 insures that the reverse voltage on the LED cannot exceed 1.5v or so (and the Cree LEDs I'll likely use tend to allow 5v).
The same problem as old-fashioned Christmas lights exists: if one burns out, the whole series string is killed. But LEDs have lifetimes in the 10,000s or hours, so the MTBF of a string of 8 should still be long. The culprit can quickly be identified by reversing each connection until the string lights up (because the 1N4007 paired with the dead LED will then be forward-biased).
LEDs in each string will be guaranteed to have the same current and similar brightness. Extra-bright assemblies can have 2 or more LEDs in series (protected by a "backwards" silicon diode).
I'd appreciate any comments on the above, thanks !





RE: LED system
RE: LED system
As for off-the-shelf LED drivers it's hard to beat Deal Extreme:
http://www.dx.com/s/led+driver
Keith Cress
kcress - http://www.flaminsystems.com
RE: LED system
RE: LED system
Yes, I realize LEDs want constant current; I thought I was pretty clear about that, but I guess not. All my talk of voltage was simply that I wanted to make sure that the series-connected string does not require too high of a voltage to maintain the constant current (and I cannot start with higher than 32vdc or so).
Can you provide a link (or part #) of a switch-mode current regulator ? I see the ones at Deal Extreme, but it doesn't look like they allow driving as high voltages as I need (30+v, to provide forward voltage of 8-9 series LEDs). Also, the dimming issue.
As far as cost, I'm gonna pay $30+ for an AGT 32vdc supply, and then probably less than a buck per "channel" for the LM317 plus resistor/pot.
As far as reverse voltage, most of Cree's lighting-class LEDs have a maximum reverse bias of 5v or so; go to:
http://www.cree.com/LED-Components-and-Modules/Pro...
... and look at the data sheet.
RE: LED system
In general, the eBay vendors ship almost immediately; whereas Deal Extreme seems to sit on orders for an additional week or more before shipping. I've previously bought quite a bit from Deal Extreme, but the typical eBay vendor seems to be better in terms of price, variety, shipping speed. YMMV
RE: LED system
Why one wouldn't go with a canned solution if the quantity is less than 10k units per month is beyond me...
Dan - Owner
http://www.Hi-TecDesigns.com
RE: LED system
The reason I don't go with a canned solution is because, as I said, I want to be able to drive 8 series-connected LEDs, and that requires the current source to be able to go up to 30v or so. I also want it to be dimmable. If there is an off-the-shelf solution that can provide that, please tell me.
RE: LED system
Consumer products are typically in parallel. If complications are required, there are modules that can be individually addressed. Often used as building blocks for full motion signage.
Of course we can't see all of your requirements through the Internet, so I'll assume you have good reasons for the custom configuration.
RE: LED system
RE: LED system
Want a cheap switching regulator? Get one of those LM2596 modules they sell on ebay for about a buck. Put a resistor in the lower leg of the LED stream. Calculate the resistance for the desired current, the sense voltage is about 1.2V. I just bought some inline LED PWM dimmers on ebay, 5 for $5 shipped.
RE: LED system
Check out the LPF series for example. $25 or so off the shelf, all one piece and done.
http://www.meanwell.com/search/LPF-25D/LPF-25D-spe...
I use them all the time
RE: LED system
RE: LED system
mA = 1.2/R
I have a 4 - 20 mA calibrator I made over 30 years ago based on a 317.
It adds mA in steps of 1, 2, 4, 8 and 12 giving me any whole mA value between 4 and 31. In those 30 years I have replaced several switches but never had to re-calibrate it.