Go Down

Topic: Probably simple transistor question (Read 1 time) previous topic - next topic

Shadow

I'm currently building a lighting setup using the TLC5940 to drive a bunch of 3w led's... The problem is the tlc5940 is only capable of putting out 120ma per channel and I'm hoping to bump that up to 1a.

The led's are common anode (+) and the tlc5940 switches the - which works out fine. Originally I thought I could just use a darlington array but the ones I got act as inverters where the output is +.

Is there a different type of transistor that I should be using that someone can point me in the direction of? I'm learning all this as I go so I'm not entirely clear what I should be looking into doing.

Planning on driving up to 96 channels so the package plays a factor in this.

Thanks!

Anachrocomputer

What type of darlington did you get, that switches the positive side?

Have you seen the ULN2803 chip, which has eight darlington drivers that switch the negative side of the load, to ground?  I think they're rated at 500mA per driver.  Otherwise, you could make discrete darlington driver with transistors such as the BFY51, which is rated at 1Amp (plus another small-signal transistor like the BC548).

If you've got 96 channels, at 120mA each, you'll need to be quite methodical with your wiring, and you'll also need to be sure that the wire is thick enough to handle the current.  You'll have nearly 12A running through the power supply wiring.

peter247

The ULN2803a is my do all chip for general buffering and driver relays etc.
http://peter224722.blogspot.com

Shadow

#3
Feb 06, 2009, 03:01 am Last Edit: Feb 06, 2009, 04:36 am by Shadow Reason: 1
The one I was trying to use was an uln2003an. Looking at the datasheet and comparing the uln2003an it looks functionally the same to the uln2803.

I figured it out I think. I had to put a resistor between the TLC out and +5v and that solved it.

I just did trial and error to figure out the proper resistor size, can anyone point me in the direction of the proper way to decide? :)

Grumpy_Mike

Please be aware that while the darling driver can switch up to 500mA per channel you can't have all of them switching that current. The maximum total current switched on at any one time is only about 650mA. See:-
http://www.thebox.myzen.co.uk/Tutorial/Power_Examples.html

As to the resistor, this will not matter because the TLC has a constant current drive in it which is set for all output channels by a single resistor, see data sheet for how to calculate it. Therefore you could put virtually anything as a load resistor to switch your external drivers.

big_mark_h

Just a quick note about the TLC5940. Although it says in the datasheet that you can 120ma per channel, you can't have all 16 channels at 120ma at the same time! I found this out the hard way by destroying a few chips before I found the solution.

Theres a calculation in the datasheet which you can use to work out how much power you will disspate depending on how many channels are on, what their current is limited to, etc. I worked out that 25ma per channel at 5 volts was about as much as I could safely use if all channels were on at the same time, but I limit the current to 20ma with a 2k resistor between IRef and ground. I don't think that you will be able to use the TLC to limit the current if you are using transistors to switch your LEDs on and off.

Shadow

#6
Feb 06, 2009, 06:48 pm Last Edit: Feb 06, 2009, 06:50 pm by Shadow Reason: 1
Thanks for the info grumpy_mike, I'll definitely have to take that into consideration =( Is there a good way of doing what I'm looking for? I was planning on joining 2 of the in/outs so I'd have a total of 4 channels per darlington. Hoping to push 700ma @ 5v. It will be pwm'ed, but I'd like to have the ability to run it at full if possible, otherwise I"ll have to put something in the code to prevent that.

big_mark: Yeah, I'm not using the tlc5940 current limiting, I've got my own resistors on the led's. Wasn't aware of the power limits on that chip as well though, something else I have to take into consideration now :)

Grumpy_Mike

Quote
Hoping to push 700ma @ 5v.


Is that per channel or per chip? You might push the chip to that as a total, you might like to help it by sticking a heat sink on it. But 4 channels of 700mA is way over the top. You would be better off going for individual transistors or FETs to drive that.

Shadow

#8
Feb 06, 2009, 09:51 pm Last Edit: Feb 06, 2009, 09:54 pm by Shadow Reason: 1
That's per channel....

Are FET's packaged in arrays? I'd rather not have to solder 48 transistors, but if thats what it takes then thats what I have to do :) Are you able to point me at a specific part?

There's an appnote on using the tlc5940 with mosfets:
http://focus.tij.co.jp/jp/lit/an/slva280/slva280.pdf

Is this the path I should be taking?

Grumpy_Mike

Quote
There's an appnote on using the tlc5940 with mosfets:
Is this the path I should be taking


No this is for using the TLC with a higher VOLTAGE supply. You want to make it drive higher current.
Stick with what you said before, substitute a resistor for the LED and use the voltage at the chip to drive a transistor or FET.

Quote
I'd rather not have to solder 48 transistors

Sorry but if you want to control 48 channels of 700mA each that's what you will have to do.
A 2N2222 is cheap enough and will handle buckets of current especially if you use a heat sink. Otherwise a hex fet would probably not need a heat sink, have a look at those.

Shadow

#10
Feb 06, 2009, 11:49 pm Last Edit: Feb 06, 2009, 11:50 pm by Shadow Reason: 1
Little confused by what you said about swapping the led for a resistor? Wouldn't these be wired up the same way as the darlington array?

Thanks for the help grumpy, appreciate it.

Grumpy_Mike

Quote
Little confused by what you said about swapping the led for a resistor?


In the normal TLC5940 you connect the LEDs from +5V to the pin of the chip. You want to connect the pin of the chip to a resistor and the other end to the +5V. Then connect the pin of the chip to the base of a transistor (no need for a base resistor) or FET used to drive the LED. Note this will invert your signal so full on for the chip will produce full off for the LED. I thought this was the configuration you had in mind.

Shadow

#12
Feb 07, 2009, 12:48 am Last Edit: Feb 07, 2009, 03:12 am by Shadow Reason: 1
Quote

You want to connect the pin of the chip to a resistor and the other end to the +5V. Then connect the pin of the chip to the base of a transistor (no need for a base resistor) or FET used to drive the LED. Note this will invert your signal so full on for the chip will produce full off for the LED. I thought this was the configuration you had in mind.


Ah ok, this is what I'm doing right now for the darlington array. I'd rather not have to deal with cooling so I'm looking at the hexfets/mosfets but I'm clueless as to what I'm looking for in a part.

700ma @ 5v = 3.5w so that means a pn2222a would be around 300 degrees without a heatsink if I'm understanding your power link properly :p How do I determine what I'm looking for in a mosfet to do this? Googling around has only come up with "mosfet basics" that are still way above my limited knowledge :(

Edit: Any particular reason to use a mosfet instead of just single higher rated transistors? I haven't done the math yet but at first glance I think a tip102 would do fine.

Grumpy_Mike

Quote
700ma @ 5v = 3.5w


NO.

The transistor doesn't have 700mA through it with 5 volts across it, that's the load.
Look at the data sheet for the on voltage that is the collector emitter voltage when on, this is about 0.6V it is that you multiply by 700mA to get the power in the switch. This is not the same as the power in the load.

The reason FETs can switch so much current is that when they are on they present a very low resistance of less than 0.1 ohm therefore they can take a lot of current without generating much heat.

Shadow

Understood.

Looking at the datasheet (http://www.fairchildsemi.com/ds/TI/TIP102.pdf) for the tip102a, in the charts it lists 2.8v for 'Base-Emitter On Voltage' with a test situation of 4v @ 8a. Since there's no chart I'm not sure how to figure out at 5v but I'm (maybe stupidly) assuming that if it can handle 8a @ 4v, then 1a @ 5v shouldn't be a problem. The datasheet also doesn't list the thermal resistance so I'm not sure how to do the math on this one.


Go Up