Electronics > Projects, Designs, and Technical Stuff
9VDC + 5VDC SMPS for driving LEDS
<< < (5/6) > >>
Blade2Raiden:
Thanks a lot mariush! I'll try putting most of that into design and post what it'll look like to see what you think about it!
T3sl4co1l:
Indeed R2 can be driven from a logic pin instead of +30V; adjust its value to maintain similar current.

Tim
TerminalJack505:

--- Quote from: T3sl4co1l on February 10, 2020, 08:29:36 pm ---Indeed R2 can be driven from a logic pin instead of +30V; adjust its value to maintain similar current.

Tim

--- End quote ---

Good point!  You wouldn't need a high-power resistor in that case.  The MCU would need to be able to source ~20mA, however.  At least according to the simulation.
T3sl4co1l:
Should only need a mA per transistor base, tops (more like 0.5-0.3mA average case), then a little extra taken up by the series diodes, so I was guessing 10mA total.  But that's fine too.

And if it turns out even more bias is desirable, the logic pin can be buffered by an emitter follower, and an even lower value used (like 100 ohms).

Doesn't really matter either way; the excess current should just be wasted in the diode.  More diode bias makes the circuit more stable against crosstalk, but only in a brute force sort of way, not really enough to keep things going if one channel drops out.

Which, on that note: if an LED string goes higher voltage or open circuit, that channel will saturate.  And without much C-E current to keep the emitter voltage up, the B-E junction pulls down the bias supply.  Which dims all the others.

This can be mitigated by driving more bias, but as I said it's a rather brute-force way.  If we, say, use a modest series base resistor on each transistor (100-470 ohms say?), then turn the double-diode into an emitter-follower-boosted source (in the same way you would boost a current mirror against the same saturation crosstalk condition; not a Wilson current mirror, but another one of those 3-transistor mirror variations, I forget if it's named?), we can source the extra current on an as-needed basis.

Finally, note that a string going short will drop a lot more voltage across the transistor, cooking it pretty quickly (quickly, in terms of surviving one, maybe two failed LEDs in that string).  Expect the cooked transistor to fail as a 3-way short, so, bringing the base voltage up for everyone and making the whole thing cook.  Possibly, the supply will brown out through the offending string, for which it's nice to have a current limiting SMPS powering everything. :)

But that's not at all a game-changing revelation -- indeed, it's very nice that this has room to sustain single LED failures.  If a single LED drops 3V, then shorting one causes the transistor's dissipation to rise from, say, 100mW to 400mW, which is sweaty for a TO-92 or SOT-23, but quite survivable!  (Put big thermal pours on the collector pad.  Can also opt for a SOT-89 or SOT-223 for a higher dissipation rating.)

If you really wanted, you could even add some diodes and a common fault monitor / detector circuit to get an instant checkup on its health, as it were.  But being that this is visual, I'm guessing it'll be easier to peek at it and spot missing LEDs or strings.  No further engineering required. ;D

Tim
TerminalJack505:
Hmm...  I didn't give too much thought about the different failure modes of that circuit.  I did check to see what would happen if one of the strings went open circuit and the the current in each of the others dropped to about 50mA.

That's probably not the best circuit for either home brew or production.

The original circuit I posted has independent current control for each of the strings so it would be more robust--at the cost of using more components.
Navigation
Message Index
Next page
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod