Well, you could further optimise it by replacing each NPN transistor and associated base resistor with a zener diode as illustrated below.
However, with 7 segment displays, it is usually the segment wiring that is the problem. It is very nice to have all the corresponding segments wired in parallel. If it is scaled up to more displays, that problem multiplies. But as you said;
So what does this circuit look like now?
This is an IR remote big 2 digits 7 segments counter (option increase, decrease, set and clear)
Thanks for every one
5 series LEDs and a 12 V supply is a bit dubious, even with red LEDs.
You will need a decent heatsink for a 7805 regulator from even 12 V. A switchmode converter as a daughterboard would be better. Actually, you can get them in a small pack which mounts to the same pads as a TO-220. 
If you have a 10k pull-up on ~SRCLR, you may as well use one resistor for both. The only reason for having such a resistor is as a placeholder in case you later wish to control this pin. Or you could include the capacitor to ground but as it clears the shift register and not the latch, it actually does nothing useful.
If you wish to suppress any spurious output during boot, it is better to hold the output enable "G" HIGH.
1 Like
It seems somewhere along the line we switched from common cathode to common anode. Is there anything comparable to the TPIC6B595 for common cathode displays?
After reading the manual, I found myself did not even respect the recommended thermal via layout and making the board hotter using 7805... I need to learn a lot of thing I will move the power supply circuit to separate board, reduce the number of led to 4 per segment and use the G signal and browse old topic in this forum for better understanding...
The only components consuming current at 5V are the Nano (~20mA), the tpic chips (<5mA) and the IR receiver (5mA). I would have thought the Nano's on-board regulator would be able to cope with that, so the 7805 might not be required.
Thanks I didn't know that!
There does appear to have been some refinements to the design during the course of the thread.
post #1 shows a common anode set up
post #4 shows a common cathode setup
post #23 reverts back to a shows a common anode set up
MIC5891 appears to be a power high side shift register:
In post #1, I interpreted Q2 and Q5 as switching the common cathodes of the digits.
Anyway, thanks for the link to the high-side version of that chip.
I agree that it has been done in a particularly unusual way, but the anodes of all the segments, for a particular digit, are grouped together at the power supply positive terminal. The switching of the individual segments is done on the low side at the cathodes. So, that qualifies common anode.
In a more conventional layout, the transistors (or arduino pins) switching the cathodes would have been directly gounded. In this case, though, these are grounded through a single transistor, presumably to provide a way to blank an entire digit for multiplexing. Normally, this switching transistor would have been between the power supply and the (common) anode but the arrangement shown could be made to work.
Looked at another way, it would be relatively simple to integrate the common anode display (figure 4 below) in the OP's wiring scheme. It would be very difficult to integrate the common cathode (figure 2 below) display.
from: https://www.jameco.com/Jameco/workshop/TechTip/working-with-seven-segment-displays.html
Well I guess we'll just have to disagree about that. He has simply moved the individual segment switches to the bottom. If he had left them at the top where they usually are, you wouldn't call it common anode. But the switches to turn the digits on or off are on the cathodes, so I think that makes it a common cathode display. It functions like a common cathode display.
Build a prototype on breadboard. Test it, including running it for several hours to make sure it does not get too hot or fails. Then design your PCB.
@imla Has consistently shown common anode design. Post #4 was by @groundFungus .
Indeed. Post #4 was not from the OP.
The design in post #1, as a subsequent discussion has highlighted, is not so easy to cleanly categorise into the conventional common anode / common cathode definition because of the arrangement of the resistors in the segment chain and the position of the switches.
There is not much "common" about it because the segment resistors cannot be shared between corresponding segments on different digits.
Edit
But, having said that, the OP's designs were always "closer" to a solution with one or more low side shift registers (such as the TPIC6B595) than high side shift registers (such as the MIC5891). The switching on the low side gives the design that "common anode" type look.
After looking here and there, I found a "better" solution!?
Using 2x8 or 1x16 channel constant current led driver.
Post a link to it. The 9-12volt range rules out some of the more popular led drivers.
That's switching the segments. But switching the digits is also on the low side (even lower), which gives the design that "common cathode" type look.
Instead of struggling to attempt to classify that design as either a conventional common anode or common cathode design, let's just call it a hybrid. It has features usually associated with both of these models. Anyway, it is unlikely that we'll be seeing anything like it in the near future.
My solution for a similar project (big 4 digit score board): I used I2C and took a board from Horter & Kalb.
Edit: 1 board per digit!