AVR-Chat

When it comes to dimming leds I would doubt very much that you are at 
your limit with software. Driving/multiplexing/dimming leds is a very 
non-critical non-time sensitive low overhead (all that sort of stuff) 
operation.

What I can't figure out is why you have run your leds in this 
configuration? The transistors are doing nothing for current boost, they 
are simply acting as controlled resistors. The problem is that the cpu 
can really only sink an overall maximum per chip so you need to buffer 
the I/O pins typically with grounded emitter BJTs. I would imagine that 
your variation in brightness is due to the cpu being overworked (RTM).

As only one of many ways can I suggest that you drive the leds via one 
or more  74HC164 8-bit serial-parallel shift registers and then regulate 
the led power rail either via your preference of a hardware circuit (555 
pwm or DC voltage) or better still via software controlling a PNP. 
However, if you are simply after regulation and not necessarily variable 
dimming then you will probably find that simply buffering the drive 
solves your regulation problems, assuming that the power-supply is not 
the problem.

I have run 64 segment outdoor bargraphs in this fashion with a humble 
8-pin pic. If the pic can do it plus receive multiple channels of serial 
data in bit-bashed fashion then it does not seem likely that a more 
efficient avr can't.

*Peter*


tymebrit wrote:
> How do you go about dimming a 7 segment display (controlled by an
> AVR)? I can't do it via software (already reached the limit), so I'm
> limited to hardware. Basically I don't want the segments getting
> dimmer or brighter as more or fewer of them turn on. Right now this
> is my circuit:
> 
> http://i6.tinypic.com/1zf3mtl.jpg <http://i6.tinypic.com/1zf3mtl.jpg>
> 
> and it works well. I'm just curious if I can do it less
> complicatedly. I.e. one transistor, chip, or whatever. Need to
> reduce parts count. Thanks,
> 
> Brit