transistor selection help

Hi,

I’m currently trying to build a light panel that uses 288 leds.(96 red, 96 blue, 96 green)

It will be powered by a 9volt source and controlled by an arduino.

I know I will need transistors to control the leds but I am stuck on how to chose which one.

As an example if one transistor controls half of the panel and one colour such as blue, it will draw 480mA from the source.

So I know I need a transistor that can handle over 480mA and 9V but what I’m confused about is the power draw.
I know from using the led series parallel wizard that the leds and resistors will dissipate 4.6 Watts.

Does this mean that I will need a transistor that has a larger Ptotal than 4.6W?

Any help would be great thanks.

No. The transistor should be able to handle the voltage. 9V is a low voltage, that is no problem. The transistor should be able to handle the current. So you need a transistor that can handle 480mA.

Some small transistors, like the BC337 can handle that. Or you could use a larger transistor, perhaps a darlington transistor. Or you could use a mosfet.

Ok thanks, that's what I thought.

So as long as it can handle the current and voltage it's fine?

Thanks

hi, not necessarily, u need to calculate the watts, then look at the datasheet °C/°K per Watt, calculate the heating above ambient 25°C, then look at the case of ur transistor eg to220, To92, and see how good it dissipates the head and then decide wether u need a different case or what kind of heatsink

ive done this only one time so i maybe get it wrong, so it would be nice if someone with more knowledge looks over this:

Forward voltage of bc337 1,2V x 0,48A = 0,576 Watts (0,625 W max Rating - thats way too close) thermal resistance from junction to ambient 0,2 K/mW -> 0,2 K/mW x 576mW = 115,2°C +25°C ambi = 140°C way too hot

id recommend u use mosfets

Your circuit has 24 parallel circuits containing 2 LEDs and a 150 ohm current-limiting resistor.

A typical voltage drop of a red LED is approximately 1.7V, thus the two LEDs drop 3.4V and the resistor must drop the remaining 5.6V.

5.6V across 150 ohms gives us 37.33mA of current and consumes 209mW.

First of all, 37.33mA is running the LEDs a little hot. Running more current through a LED generally increases the brightness -- to point. The difference in brightness between 20mA and 30mA is marginal, methinks. So I would redesign the circuit to have 20mA or 25mA, unless you know that pumping up the current is going give you some advantage. Perhaps you could make a small test circuit and run various currents through the LEDs to see the minimum current for the brightness you desire.

Second, the purpose of the circuit is to generate light, yes? The LEDs in the circuit are producing the light and any power consumed by the resistor just produces heat. That being said, I would try to reduce the voltage (and thus the power consumption) of the resistor.

My recollection is that red LEDs consume about 180mW @ 20mA. My calcs are that each parallel leg in your drawing consumes 0.18 + 0.18 + .209 = 0.569mW. This gives the total circuit 0.569W x 24 = 14.6W.

So here is my suggestion: Instead of 2 LEDs per leg, use 4. Thus more of the voltage that is applied to LEDs, the less is consumed by the resistor. In this configuration you drop 1.7 + 1.7 + 1.7 + 1.7 = 6.8V across the LEDs and only 2.2V across the resistor. If we reduce the current to the nominal 20mA then the resistor value would be 2.2V / 0.02A = 110 ohms. 2.2V across 110 ohms gives use a power consumption 44mW, down from your original 209mW.

While that might sound like too much, because you now have 12 parallel legs instead of the original 24 you get a nice power savings: 0.18 + 0.18 + 0.18 + 0.18 + 0.044 = 0.764mW per leg, making the whole circuit 9.168W which is 2/3 of the original circuit -- and reduces the part count by 12 resistors.

I have used generic 5mm red LEDs in the calcs. You would need to read the spec sheets of your LEDs and alter the resistor values for each color.

My caveat: it is 2:21AM here and I am running a fever, so double check anything I say!

Good luck

lax123: hi, not necessarily, u need to calculate the watts, then look at the datasheet °C/°K per Watt, calculate the heating above ambient 25°C, then look at the case of ur transistor eg to220, To92, and see how good it dissipates the head and then decide wether u need a different case or what kind of heatsink

ive done this only one time so i maybe get it wrong, so it would be nice if someone with more knowledge looks over this:

Forward voltage of bc337 1,2V x 0,48A = 0,576 Watts (0,625 W max Rating - thats way too close) thermal resistance from junction to ambient 0,2 K/mW -> 0,2 K/mW x 576mW = 115,2°C +25°C ambi = 140°C way too hot

id recommend u use mosfets

The math looks about right to me. The fairchild datasheet lists Vce(sat) of .7V MAX at 500mA collector current (Ic) and a whopping 50mA base current (Ib). That would lower the dissipation to .7V * .48A = ~350mW. With their heat factor of 200C/W for junction to ambient, I get 70C rise over ambient. This is going to be hot, but it might work out in a cool environment with some circulation, so I wouldn't squeeze it to see if it's hot. I'd still use a MOSFET though since I don't like heat, not to mention that 50mA base current becomes a problem in and of itself.

chopperkid: Ok thanks, that's what I thought.

So as long as it can handle the current and voltage it's fine?

No, you have to calculate the wattage the current will produce.

Not all the 9V will be converted to heat in the transistor. Look at the datasheet for VCE at saturation, it tells you the voltage drop across the transistor when it's fully on. For a BC337 it's 0.7V. 0.7Vx480mA gives 336mW - which the transistor can handle.

Yeh all the math is right don't worry haha. The datasheet for the red led says it only drops 2.1V which seems low but thats what it says. The picture is for the blue leds if I remember with a voltage drop of 3.3V.

Sorry for being a noob but would anyone be able to recomend a suitable transistor or mosfet? I'm just getting confused with working out what will work and what won't.

Would something like a TIP31C work? I don't mind if the solution is slightly overkill as long as it won't run extremely hot and cause problems.

Sparkfun sells some nice MOSFETs for 95 cents each that will do a good job and you won't have to worry about anything including base current. They are a nice part to have on hand since they can handle far more power than you need here, but that will come in handy when you want to switch relays and motors.

The TIP31c will easily handle the current, but you will need to supply a bunch of base current since the gain (beta or hFE) is very low (like 20 or so). Something like a darlington TIP120 would be a much better choice I think. They have a gigantic amount of gain (hFE = 1000 MIN) so base current is no problem. Vce(sat) is only going to be .5V at .5A of collector current. This is only 250mW of power dissipation from a TO220 case, so heat will not be an issue. Beware that the TIP120 Vce(sat) can go as high as 4V as the collector current increases towards the maximum allowed.

You might want to read this before continuing: http://www.mcmanis.com/chuck/robotics/tutorial/h-bridge/bjt_theory.html

Ok I think I understand the calculations for the heat generated in the transistor now. If I was to use a mosfet as everyone is suggesting do I just need to find a logic level mosfet with a low on resistance that can handle the current and voltage?

like this one aairlml2502 http://www.irf.com/product-info/datasheets/data/irlml2502.pdf its super small, super cheap and can handle a lot more then u need

this is german but might help anyway: -u need one of those where it says logic level, then look for the needed package http://www.mikrocontroller.net/articles/MOSFET-%C3%9Cbersicht#N-Kanal_MOSFET

Pretty much the same power calculation only using Rds(on) resistance instead of the Vce(sat) voltage drop. You just multiply Rds(on) MAX by the collector current (Ic) to calculate the voltage drop of the MOSFET and then continue on like the other problems.

Like I said, the TIP 120 will work fine (you can get those at radio shack) and so will the logic-level MOSFETS that Sparkfun sells. I would just get a collection of what Sparkfun sells since you can use them over a very large range. Once you have to move to a TO220 case anyway, what difference does it make as long as it can handle the power. So just get some TO-220 MOSFETs that can handle 20 or 30A (or more) and use them at about 1/4 (or less) of the power dissipation claim you see stated and you will likely be fine. If it gets hot, then look for something bigger.

I don't like to see parts that are too hot to touch. Unfortunately in the real world, you'll find allot of designs that would have been better off with a slightly larger parts or slightly more cooling. My uverse dvr box is a prime example of what I'm talking about. The outside of the box is over 50C. :(

lax123: like this one aairlml2502 http://www.irf.com/product-info/datasheets/data/irlml2502.pdf its super small, super cheap and can handle a lot more then u need

That looks like a good one, especially for 3V systems it's turned on pretty decently at a Vgs of 2.5V. It's going to be hard for the OP to stuff that into a breadboard though. ;) Do they have something equivalent in a thru hole part?

lax123: Forward voltage of bc337 1,2V x 0,48A = 0,576 Watts (0,625 W max Rating - thats way too close) thermal resistance from junction to ambient 0,2 K/mW -> 0,2 K/mW x 576mW = 115,2°C +25°C ambi = 140°C way too hot

You picked Vbe(sat) from the datasheet (1.2V) when what matters is Vce(sat) - which is quoted at 0.7V max @ Ic=500mA, Ib=50mA. Vce(sat) will be a little higher at a lower Ib more suited to the Ardiono (say 25mA), but not much. So Pd will be around 0.4W, which I consider just about OK.

afremont: Something like a darlington TIP120 would be a much better choice I think. They have a gigantic amount of gain (hFE = 1000 MIN) so base current is no problem. Vce(sat) is only going to be .5V at .5A of collector current.

I think the Vce(sat) of a TIP120 will be very much greater than 0.5V @ 500mA, because it is a darlington.

For a breadboard or stripboard design I would choose a BC337 or ZTX851, and for an SMD PCB design I would choose a mosfet in a SO23 package.

Took that Vce(sat) number right from the datasheet graph (graph 2) of a Fairchild sheet. It doesn't start to get bad until you get over 1A. It stays below 1V until you get to 3A. You're right though, I misread it. It will be about .75V at 500mA. The baseline started at .5V, my bad.