Suggestion Part alternatives for the MAX7219/MAX6954 7-Segment LED Driver

[mino-fm]

For reliability reasons, I do not suggest driving LEDs directly with the microcontroller outputs unless current is under 4 milliamps and the lower the better.

Reliability is no problem using 4094. The current limiting resistors are inside these chips. Older CMOS devices have weak outputs.
Over thirty years I assembled serveral thousands of digits. Avarage current is about 5 mA/segment. Nor any digit neither any segment failed. Using newer high efficiency LEDs the digits are bright enough for daylight applications (transparent colored filter is needed).

[David Hess]

There are two reasons for my warning; one of which still applies to the high voltage CMOS 4094.

1. The output current warning applies to non-driver outputs.  Part of this is the limitation on the power pins; eight 25 milliamp outputs yields 200 millamps through a ground or power pin which is a typical absolute maximum rating and not always specified.  25 milliamps might seem excessive but if the display is multiplexed, it represents a much lower average current through the LEDs but not the driver.  High voltage CMOS is so weak though especially at low voltages that it can be excluded.

2. ESD through the exposed LEDs can go directly into the ICs and insulating shielding, like the plastic housing of an LED or LED display, can make this worse by increasing the breakdown voltage.  Specially driver ICs may be designed to handle this but not general purpose logic.  Some inexpensive large junction transistors used as drivers help a lot with this unless ESD protection diodes are used.  Where this really matters, a conductive screen should be used across the display.

Update:

3. Using external driving transistors allows routing the relatively high currents away from the signal common.

[Mr. Scram]

The microcontroller obviously needs enough output pins but I would also look for a part with a UART, USART, SPI, or whatever for serial communications rather than trying to bit-bang a receiver.  Using a UART for the serial link may be advantageous because no clock or chip select is required.

I considered PICs first because I am familiar with them and they are available in easy to prototype 300mil DIP packages up to 28 pins providing plenty of I/Os but AVR could work also.  The price would be competitive with a 74HC595 solution and better if multiplexing is used.

Thanks for your input, but I was mainly wondering about limitations in regards to current. Would you use individual transistors?

[mino-fm]

There are two reasons for my warning; one of which still applies to the high voltage CMOS 4094.

Concerning 4094: You are thinking about problems that do not exist. Reading out a display is done by eyes and not by ESD-fingers ;-)

Would you use individual transistors?

For multiplexing you always need transitors for driving common anode or cathode pins.
Looking at ATmega48 the total current of VCC or GND is 200 mA. Divided by 8 segments (incl. D.P.) each segment can get about 25 mA which is delivered by a single output pin (max. 40 mA). But the common pins has to be able delivering 200 mA. These currents need a transistor anyway.
Multiplexing of 8 digits will reduce the eff. segment current to 25/8 = 3.1 mA done by current limiting resistors. That is enough for high efficiency LEDs.

Whether you like multiplexing or not depends (for me) of the hight of the digits. Displays <= 10 mm (0.4") are simple to root so multiplexing should be preferred. Displays >= 12.7 mm (0.5") offer a lot of space below them or on the other side of PCB. So it is very easy to use parallel driving (SMD 4094) and cascading many more digits as by multiplexing.
If you use larger digits >= 50 mm (2") where each segment consists of two or more LEDs you need transitors too. Here again parallel driving is the best way. Multiplexing would cause high peek currents and a lot of noise. ULN2803 is a compact form of segment drivers up to 24 VDC.

[David Hess]

Thanks for your input, but I was mainly wondering about limitations in regards to current. Would you use individual transistors?

I would absolutely use external transistors or drivers if the display is multiplexed.  If not, I might still do it unless the currents are low or for higher reliability.

[optoisolated]

I really like the idea of using MCUs as an LED Driver.. MCUs with crap loads of DACs are cheap as chips.. Definitely gonna give that a go I think. Gives me a lot of flexibility with IO expansion if I need it as well.

[floobydust]

Are you wanting bright displays? With 7 segments+DP, 20mA each = 160mA max. per digit.

You're going to need something to buffer digit-driver current you can pull from an MCU I/O pin, and see if the MCU can handle outputting 160mA total from the IC package for the segments. Example ATmega328/P rated sum for all I/O 100mA. If you are thinking about direct-drive. High voltage LED's like blue/white/green AlInGaP with Vf ~3.2V would need a 5V MCU.

Adafruit uses 'HC595 and TPIC6B595 instead of MAX7219, or HT16K33 LED driver rated 200mA/digit and 20-50mA/segment, under $1.50 on eBay.

[optoisolated]

That TPIC6B595 looks like it will do the business.. Might order some in and mess around with them and see if they will work. Its a bit pricey but there's already libraries available for it which will reduce my dev time a lot. Cheers!

[Giuss]

TM1637 can be another option

[MudAndSnow]

I recommend considering an MCU. I looked around at various options for 7 seg drivers and ended up going with ATMega328.

It is common, cheap, easy and versitile. I used that one MCU for everything in that project: 4 digit 7seg, 4x4 keypad, etc.  You'll probably want one with more outputs but it seemed like a no brainer to me once I picked a common MCU after spending way too much time digging thru the plethora of display drivers to find the right match.

Sparkfun has an example of ATMega328 as 4 digit 7 seg driver here:
https://www.sparkfun.com/products/11441