[dorkbotpdx-blabber] ARRGGHH - Please check this circuit

[montyg at bittybot.com]

Hans,

You have some pretty large currents that you are switching (I'm  
guessing) pretty fast.  That can get tricky, especially when the main  
supply is located aways away.  If I remember right, your modules were  
daisy-chained and in actually use would be strung out over quite some  
distance.  Resistance will definitely be a problem at those high  
currents.  To keep the voltage drop down to 0.5V under full load you  
would need: (100W/5V = 20A; 0.5V/20A = 0.025-ohm) less than 0.025ohm  
of resistance in all your cabling and connectors.  Without considering  
connector losses this would require ~10AWG wire for 20' of cabling  
(think heavy-duty extension cords).

Resistance isn't the only problem though.  Since you're switching  
these high currents (at what I'm guessing are relatively fast turn-on  
and turn-off times), you need to watch out for inductance too.   
Inductance resists change in current, and will cause voltage droops  
every time you switch on and voltage spikes every time you switch off.  
  Not switching all the LED's simultaneously would help, but that  
seems artistically limiting.

A typical power delivery system for high-current switched loads will  
take care of all the higher frequency switching transients at the load  
by using sufficient capacitance so that only the lower frequency  
transients need to be supplied over the cable and bulk supply.  You  
still want to minimize the inductance though.  Inductance is dependent  
on the amount of area enclosed by the current loop, so look at your  
power and ground paths and take out as much space between them as  
possible -- twisting the power/ground wires together is a good start,  
but also consider the loop area formed by leaded capacitors and  
breadboards too.  Keeping the loop area small is also important for  
reducing EMI.

The other thing commonly done to power large dynamic loads is to  
deliver the power at a high voltage and down-regulate it using  
switchers at the load.  In your case, supplying 12V and using a  
12V->5V switching regulator (DC:DC converter) at each load would cut  
the supply current in half.  Unfortunately, this solution will up the  
cost -- switchers are quite a bit more than linear regulators, but  
switchers will reduce the current draw while linears won't.

Finally, at each load, it would be a good idea to isolate the  
microcontroller supply from the switched LED supply as others have  
suggested.  A 7805 off the 12V line sounds great.  You still want  
sufficient capacitance after the 7805 though to handle the  
microcontroller's switching transients (which are much smaller, but  
still present.)

Hope that helps!

Monty


Quoting Hans Lindauer <armatronix at sbcglobal.net>:

> By now, many of you have been blinded by my MIDI LED project at the
> meetings.  Hopefully some of the electronics experts out there can help
> me get this thing working so I can get on with my life.
>
> The trouble is, I've been trying to run the whole thing off of 5V, but
> when the lights come on fully, it /appears/ (I don't have a scope to
> confirm this) that the voltage drops below the minimum for the
> microcontroller, causing the microcontroller to reset.
>
> I had hoped that putting a capacitor on the microcontroller's power
> rail would help (I even figured out to put a diode going into the cap),
> but it still isn't working.  I think the problem is that 5V doesn't
> provide enough overhead, and trying to use caps to take up the slack is
> only going to be marginal at best if I want to be able to run the LEDs
> at full brightness (which I do).
>
> So my next step is to try powering it off of 12V instead of 5V.
> /BUT.../ when I tried that before, it burned out my LED after a few
> minutes (and I can't afford to burn out too many of those, they're
> expensive).  The guy who posted the driver circuit I'm using claims
> that it works for 3V-60V, but obviously I've found an exception to that.
>
> I don't understand this stuff enough to know what's the problem here,
> but if any of you do, could you please let me know what to try next?
> And if there's a Better Way (that won't break the bank or introduce a
> lot of complexity), I'm open to suggestions.
>
>
> Here's the driver circuit; I have 36 of these in the system, hooked up
> to the Dorkboards' PWM outputs:
>
> I=1A thru the LED (metered at 5V input)
> R1=100Kohm, 1/4W
> R3=0.56ohm, 3W
> Q1=Fairchild 2N5088BU Small Signal NPN Transistor  data sheet
> <http://www.fairchildsemi.com/ds/2N/2N5088.pdf>
> Q2=Fairchild FQP50N06L 60V N-Channel QFET Logic Level MOSFET  data
> sheet <http://www.fairchildsemi.com/ds/FQ/FQP50N06L.pdf>
> LED=LedEngin LZ4-20MA10 Hi-Power LEDs on MCPCB RGBA 10 Watt Full Color
> data sheet <http://www.ledengin.com/products/10wLZ/LZ4-00MA10.pdf>
> (I'm not using the amber die on the LED)
> Link to Instuctables page for this circuit
> <http://www.instructables.com/id/ECM4Q6NH9IEWOF32MJ/>
>
>
>
> Thanks!
> -Hans