I have built the circuit pictured below. What is confusing me is that, as I add more LED-pairs in parallel, the current through each pair decreases. So, for example, a single pair of LEDs draws 16 mA. Four pairs of LEDs draw 56 mA, or 14 mA each. Eight pairs of LEDs draw 100 mA, or 12.5 mA each. I can't figure out why the current draw isn't increasing linearly as I add more LEDs. It's problematic, because it means that the LEDs get dimmer as I add more.
Current draw is being measured directly before and after the LEDs, not from the PSU as a whole, so this isn't a case of some fixed overhead throwing off my calculations. The PSU is not being over-drawn and supply voltage is rock steady at 5.15 volts.
... BTW, the "LED series parallel array wizard" said to use 56 ohm resistors for 20 mA current, but the site I was ordering the LEDs from had a minimum order, and they didn't have 56 ohm, so I decided to accept an 18% reduction in current/brightness. By my calculation, I should still be getting 16.4 mA through each pair... which is pretty much what I get when there is a single pair.
Incidentally, I originally suspected that there was some restriction in the FET that was causing the problem, such as insufficient VGS to cause the FET to fully turn on. I hooked a 12v supply up to the gate through a potentiometer and ran VGS up to about 9 volts. Current draw peaked around 3 volts and didn't increase any further, leading me to conclude that the FET is fully on by about 3 volts.
You can get lower, but the cost starts jumping up, as does the input gate capacitance, which affects switching speed, assuming you will be want to turn these on/off down the road.
Short answer? Because I have some N-channel FETS left over from another project, so I'm trying to use them if I can. I also have some PN2222's if a BJT would help, but I'd have to re-wire the circuit to be P-channel.
majenko:
With the MOSFET and a total 100mA draw, I=V/R = (5.15 - 4 - 0.125) / 68 = 0.015073529A
As the total current increases, so does the voltage drop across that MOSFET.
Oh dear. The final expected current through the circuit, at a nominal 0.0169 amps per LED pair, is about half an amp. When I saw that IDS was 13 amps, I figured I would be well within my limits. I wasn't aware of how RDS_ON affected performance.
For what it is worth, you can try keep lowering the R value across all LED pairs and see if you can get the brightness level you want, before you take the plunge to use a different MOSFET.
majenko:
125m? is actually quite high really. That is going to have an effect on the rest of the circuit.
At 100mA it's going to be dropping 0.125V, which you would have to take into consideration in your per-branch calculations.
I apologize for the elementary nature of this question, but in the interest of trying to understand vs. just relying on others' knowledge, I tried to run through this math.
Can you use a higher voltage? Say 12V?
Then you can have fewer strings, and fewer mA will be needed.
Say 4 LEDs/string, and 16 strings for 64 LEDs total still. And 320mA.
CrossRoads:
Can you use a higher voltage? Say 12V?
I haven't built the power supply for this project yet, so that is an option. My main concern is that, since these LED glasses will be worn on people's face, I'm not sure whether it's safe to use 12 volts. I mean, I know the chances are slim, but what if, worst-case, a stray wire poked somebody in the eyeball? Nice moist eyeball... would 12 volts be a hazard?
Honestly, I would much rather run the project at 12v. I could run four LEDs in parallel instead of two, and it would use 220 ohm resistors, which I already have plenty of, because that's the nominal value for a single LED at 5v. It would allow the project to run off a car battery if needed, and the higher voltage would reduce the current drawn.
The only thing that I think might be a definite concern would be that VGS for the FET is -10 to 10v. But since IGS is negligible, I think I could use a voltage divider to bring 12v down to 9v or so for the gate drive.
Well, you're not committed to that particular FET forever, right?
Most likely going to surface mount eventually to make things smaller all around?
Then you would be free to use whatever parts you wanted, vs just what's on hand for prototyping.
CrossRoads:
Well, you're not committed to that particular FET forever, right?
Most likely going to surface mount eventually to make things smaller all around?
Then you would be free to use whatever parts you wanted, vs just what's on hand for prototyping.
I'm mostly trying to cobble this project together on the cheap, as much as I can. I will probably build it once, using the same components for prototyping and final construction. I don't know how I would go to surface-mount without designing a PCB, which seems like way more work and expense than a project like this is worth.
I'm not committed to this particular FET forever. If it's the right thing to do, I will buy other components. But I always try to rummage through the "spares" bin and make what I find work before plonking down the credit card at Digikey. Otherwise, the "spares" bin just gets bigger and bigger forever, you know?
I really appreciate your contribution to the thread. Whether I end up using this FET or not, now I know about the significance of RDS_ON, where I didn't before.
