In http://www.toyolite.com.tw/b1.htm we have
ST2221A --> Constant Current Output: 5~90mA
ST2221C --> Constant Current Output: 5~120mA
But in page 1 of both datasheets of ST2221A and ST2221C we have this phrase:
Constant Current Output : Current with one resistor for 5mA to 90mA
Again in page 5 of both the pdfs we have:
Output Current IOUT 90 mA
This makes me doubt maybe both of them deliver the same amount of current, even despite what the first link says. Right? Wrong?
Look at max the package dissapation and you will see that the 'C' is not a good choice for figh current apps.

http://www.allegromicro.com/techpub2/...295031.pdf says:
"Multiplexed displays must typically be operated at a current ãn times the nominal operating current (where n is the number of digits)".
Now in my project, I multiplex 32 columns, so the needed current will be something like 5x[at least]50mA=250mA. High flux LEDs are used:
Typical Total Flux (ƒ³v mlm) 3500 @ If = 70mA
Here we go again DO NOT pay any attention to "typicals" in a data sheet, only max and min are of interest.
If you drive your LEDs at more than than the datasheet specify as max curent you will age then real fast. A competitor did that and is not an business any more.

but ST2221 series can't offer more than 90 or 120mA. I curretly use ULN2803, I'm in doubt you know...replacing ULN's generous 500mA pin output current with 120 or maybe 90mA of ST2221's. What is that I'm not aware of? Actually I still don't understand how not to multiplex. Is it not better to use HC595+ULN2803 instead of ST2221? Although replacing four HC595s + four ULNs with just two ST2221Cs seduces me. Is it really a trade-off between current and chip count?
Yes, you need more chips when you do not multiplex BUT you can get full intensity without exceeding Imax for the LEDs (= cheaper LEDs), you do not have any synchronization issues. The added cost of higher intensity LEDs and drivers capable of delivering more current is more than the cost of the chip in non-mpx.

For each one of the 32x32 LED matrixes I'm gonna use a couple of ST2221Cs as row driver plus 32 PNP xistors driven by a 5x32 decoder made up of two HC154s as column driver. The serial data to the ST2221s on each 32x32 board comes from one of the pins of Pz and Px.y clocks them all. Now considering the need for interdigit blanking I've got two choices: 1) I may use /ENABLE pin of ST2221 to turn off the display or 2) I may use /G pin of HC154s to turn off the display. They both seem to be similar, but I read something from Allegro:
This phenomena(ghosting) results from the source driver requiring more time to turn off than the sink driver takes to turn on. Blanking will delay the sink driver turn on and will allow the source driver to turn off completely. This is performed with the OUTPUT ENABLE function.
Now, is it correct to conclude that it's better to use /ENABLE of ST2221 instead of /G of HC154?
If you do not multiplex, you do not have any highside drivers.
I have no idea what "interdigit blanking" and "ghosting" is, it must be multiplexing phenomena. It may be the names of some of the things that made me abandon all mpx for more than 16*40 - and that mpx will be abandoned in the next generation.

You truely insist on separating VCC and VLED and the GND and V0 justlike what Intel says in "Designing Microcontroller Systems for Electrically Noisy Environments" (AP-125,page 11):
Currents in multiplexed LED displays tend to put a lot of noise on the ground and supply lines because of the constant switching and changing involved in the scanning process. The segment driver ground is relatively quiet, since it doesn't conduct the LED currents. The digit driver ground is noisier, and should be provided with a separate path to the PCB ground terminal, even if the PCB ground layout is gridded. The LED feed and return current paths should be laid out on opposite sides of the board like parallel flat conductors.
But ST2221 uses only one VDD pin. Do they "internally" separate VCC of the shift register and LED driver inside the pakage or what? You see...I use a ferrite bead plus a handfull of multilayer and tantaluim capacitors to "clean" the power lines to TTL ICs an uC, while LEDs are driven with a rather "raw" VLED. I don't know...maybe I'm just griping instead of finding my "way".
The driver is connected to Vdd, the LED anodes are connected to Vdd directly, (argh.. through your high side multiplexer - why do you HAVE to multiplex, it will give you sooo much trouble on a sign this size) the cathodes connect to the ST221

Once you talked about rumors going around about 2221C heat dissipation. Also in one of your posts you named ST2221A to be the best choice, Was the emphasis on 2221 or on A?
as said above: calculate the package dissapation for the "C" and you will see.

Erik

PS we use several milliuon LEDs every month and do not have aging or fading issues, which "overdriven" multiplexed signs do.