Question about how transistors should be connected (Design)

[psycholoner]

I've been through quite a bit of basic transistor theory, and done calculations to back it up, but there are still things I'm confused about.

I'm lost at how one should use either the emitter or collector to run a load. Well, to be more specific...
http://www.armory.com/~spcecdt/electronics/LED_matrix/LED_matrix_schem.pdf

What I'm doing right now is multiplexing single RGB leds and using PWM to control the colour. In the diagram I found, the circuit uses an NPN and a PNP to run an led, and its sourcing and sinking through the emitter. What I'm wondering is, does it need to be? Can we run the led by switching the transistors and use the collector side instead? I've actually tried it and it seems to work, but I want to know some possible downfalls and design flaws that might appear. And does it even need to be 2 opposite types of transistors.

For example, using 2 NPN transistors and connecting the led to the collector and emitter of each transistor respectively. It loses its symmetry and it feels like there might be a problem with it, but I don't know enough to be able to tell if its alright. Essentially, the transistors are being used in a digital sense, and as long as you can enable the led when you put in the proper inputs, does it really matter how the transistors are arranged?

[Psi]

does it even need to be 2 opposite types of transistors.

You need the two types (pnp and npn) because you need one transistor to pull low and the other to pull high so the led has a voltage across it.
(Since a transistor can only pull in one direction. GND for NPN and  VCC for PNP)

The only way to use the same type would be if one side (row or column) had load resistors to provide the pull in the other direction.
But that introduces other issues and isn't an ideal solution.

[psycholoner]

I still can't see why you can't use 2 of the same transistor. If you took an NPN transistor and replaced the PNP by orienting the transistor with the emitter in the other direction, and used highs to turn the transistor on instead of using lows for PNP, then shouldn't it work? I mean, in that sense, you aren't really pulling the transistor high, but you are still sourcing the current right?

Since I'm just using the transistor as switches, does that kind of thing really matter? Is it not good enough to have the LED turn on only when both transistors are active?

[jhdiii]

Yes, you can use the same type for both ends of a driver.  The transistors in the LED circuit are being used as emitter followers, which provides maximum current gain and no voltage gain and requires no biasing resistors.  Current gain is exactly what's needed in that circuit, so using emitter followers for both ends gives simplicity.

But if you want to use e.g. all NPN transistors, you can use a common-emitter circuit for the pull-down side.  In an emitter follower, the voltage follows the input voltage, while in a common-emitter circuit the drive is inverted - a pull-up current into the base will produce a pull-down current through the collector.  The one extra thing you have to provide is a biasing resistor between the driving circuit and the base, to limit base-emitter current.  It might appear to work without that, but it's a bad idea.  You need to know the transistor's current gain so that you can provide the desired output current divided by that (plus some headroom) to the base.

This isn't just a digital thing - in analog electronics, emitter followers are used to provide current amplification for varying voltages, and common-emitter circuits are widely used to provide combined current and voltage gain.

[psycholoner]

Thanks. Thats what i wanted to know. I was mostly just wondering if it was a valid design. Ive done a lot of small signal amps too so i know about how analog circuits work too. Ive been doing calculations to safely push the transistors in to saturation. Calculation wise its more complicated so Im not actually doing it like that though.

My other question was whether it is alright to use a common collector configuration with the NPN and PNP transistors instead. Youd end up turning on a transistor when it cant amplify any current, but is this a safe configuration?

EDIT: sorry, I mean common emitter.

[allanw]

No, you can't flip a transistor and use its emitter as collector. For example, the 2N2222 has maximum: Vce = 40V, Vcb = 75V, Veb = 6V

[psycholoner]

what? no i didnt mean just flipping the transistor. I meant switching out the transitors with each other and connecting the load to the collector. so the NPN has its emitter connected to ground, or i should say pulling the current.

[Psi]

yeah, sorry.

I didn't fully read the design, didn't see it was emitter follower.
Ignore my previous statement