I believe a pull-down resistor at the base should be used. Even if all the inputs are LOW, the base could still carry current from outside sources (static, EM, etc and switch unreliably.

While I suspect that eight parallel reverse-biased diodes would provide sufficient leakage current to take care of that, it is certainly a good idea.

You could put the pulldown at the junction of the diodes and the base resistor, in which case it would have negligible effect on the normal operation of the circuit. If you put it at the base, then you need to allow for the amount of current that it will steer away from the base. But with a base current of 1.3mA, you only need to steer 1% of so of it away and so something like

Rpd = 100*(0.7V/1.3mA) = 53.8kΩ should do. So put a 10kΩ resistor there and call it good. You'll be steering about 5% of the current away, but given the 5% tolerance on the resistors, you should be able to accommodate this with ease.

 

It works!

It's a IR LED. It's a test circuit for this: http://forum.allaboutcircuits.com/showthread.php?t=89719&page=2

So I followed the link, and the link in there, to the LED spec sheet and it appears that you will likely have a forward voltage drop of about 1.1V at the current you are targeting. With the 220Ω resistor you will probably get about 17mA of LED current. My guess is that this won't cause you problems and it is still close enough to β=10 that you should be fine as far as a pretty constant Vcesat. But you might consider bumping it up to a 270Ω resistor.

 

So I followed the link, and the link in there, to the LED spec sheet and it appears that you will likely have a forward voltage drop of about 1.1V at the current you are targeting. With the 220Ω resistor you will probably get about 17mA of LED current. My guess is that this won't cause you problems and it is still close enough to β=10 that you should be fine as far as a pretty constant Vcesat. But you might consider bumping it up to a 270Ω resistor.

Yes, I'm aware we were not using the actual voltage drop in this very circuit. But as you can see in the other post, I'm planning to put these LEDs under 200 or 300mA. Which will get us to a 2V voltage drop according to figure 5.
No, I'm not planning to fry the LED nor the transistor. Each LED-transistor pair will be working under 6.25% of the time (duty cycle). Well, that's discussed in the other thread already.
And yes, I know that I'm going to loose a pair if the clock freezes

 

Yes, I'm aware we were not using the actual voltage drop in this very circuit. But as you can see in the other post, I'm planning to put these LEDs under 200 or 300mA. Which will get us to a 2V voltage drop according to figure 5.
No, I'm not planning to fry the LED nor the transistor. Each LED-transistor pair will be working under 6.25% of the time (duty cycle). Well, that's discussed in the other thread already.
And yes, I know that I'm going to loose a pair if the clock freezes

That all sounds reasonable and it looks like you are considering the right things. As you probably already know, when you increase the collector current by about 20x, you will need to increase the base current by the same factor and reduce the size of the base resistor by a corresponding amount. You also want to check the transistor's data sheet and make sure that it can handle peak currents of 300mA in the collector and 30mA in the base and also what the corresponding Vbe and Vce values are. The pulldown resistor can stay the same size.

 

That all sounds reasonable and it looks like you are considering the right things. As you probably already know, when you increase the collector current by about 20x, you will need to increase the base current by the same factor and reduce the size of the base resistor by a corresponding amount. You also want to check the transistor's data sheet and make sure that it can handle peak currents of 300mA in the collector and 30mA in the base and also what the corresponding Vbe and Vce values are. The pulldown resistor can stay the same size.

Yes. I didn't reach that point yet. In fact, I'm not sure if I'm going to use a 2N3904 or one with better power dissipation factor. I'm driving the base with a CD4094 so that will be another limit. I don't want to use two transistors for each LED.