Using a ULN2003 for Cathode and Anode?

[Stonent]

Another post related to my time machine project...
Originally the idea was these 7 segment displays would be configured by dip switches to flip the A through G segments on and off. Yeah it would be not great way to do it, but it would have worked.
A different part of the circuit was going to have 24 LEDs that I was going to drive from a shift register and the ULN2003.

So I had ordered parts based on that assumption.

The design has since changed and now the switches will be used to turn the 24 LEDs on and off and will be trying to drive the 8x 7 seg displays with the shift registers and ULN2003 chips.

I decided to treat this circuit as if it were an 8 x 8 LED matrix with a Column Cathode, Row Anode configuration.
So I'm going to wire all the segments together and cycle through them using 1 shift register for the segments which I'll drive in parallel, and will select the digits using the second shift register.

Can anyone think of a way to use the ULN2003 to drive the anodes and cathodes or would I just be better off using 8 individual transistors on one side?

4 x TOD-5261AG-B Green Common Cathode Dual 7 Seg Disp. (Total 8 x 7 Seg displays)
2 x ST M74HC595B1 Serial to Parallel shift registers
8 x 270R resistors (Technically a 15x270R common source resistor network)
2 x ULN2003APG (7x NPN Darlington) http://www.toshiba-components.com/docs/linear/ULN2004AFWG_en_datasheet.pdf
1 x MCU (probably a ATMega328P)

[williefleete]

probably easier to use external PNP transistors on the darlington drivers for the anodes
emitter: going to supply rail via the common source resistors
base: via a resistor into the darlington driver
collector: on the anode(s) of the LED display

low side will be the cathodes going into the output lines of the darlingtons

as a side note, there used to be these awesome IC's that were a combined shift reg and 8 darlington outputs, I used at least 24( + some in a prototype) of these in 3 displays I made, they are - I believe - not manufactured any more, the part number was UCN5841

[oPossum]

A nice substitute for HC595 + ULN2003 are the TI TPIC6C595, TPIC6B595 and TPIC6A595 power shift registers. They use DMOS outputs that typically perform better than the bipolar darlintons in ULN200x series.