I just finsihed a project that has a 6275 on it for driving LEDs.

The device was connected to a SiLabs C8051F124 processor. At first when we were integrating software on the target hardware this chip was not working well at all!! Several fixes were required and then it began to work just fine. Let me explain.....

Because there was a significant delay time to final target hardware that was ready for software integration I built a separate software development board that connected to a SiLabs 'F124 eval board via the DIN-96 expansion connector. This board used a 'HC595 shift register and some simple transistor drivers for the LEDs as opposed to the Allegro 6275 chip since it was someting I could get built over a weekend. For simplicity I just ran the HC595 off the 3.3V supply and configured the control line outputs from the 'F124 as push pull outputs. The software development on the test platform worked flawlessly with regard to getting LEDs to light, blink and flash!!

When the day of first circuit boards arrived....naturally it was the Friday before the Monday that prototypes, with software, HAD to deliver to the customer!! Myself and the hardware engineer worked at integration into the wee hours of Saturday morning. It must have been around 1 AM when we got to the point of trying to find out why the 6275 LED display driver was not making LEDs light. Bleary eyed we, me and the hardware engineer, stared at the oscilloscope for about an hour before we realized that the 6275 chip was connected to the 5 volt rail on the target board... 3.3 volt push pull outputs from the 'F125 made very nice looking waveforms on the o-scope BUT were not good enough to reliably get the 5V 6275 part to clock correctly. Finally it was clear that I had to change the I/O pins to quasi-bidirectional mode and add pullup resistors if there was a hope of getting the 6275 to work. After making the changes the LEDs started to function.

However there was a still a bit of flicker in the LEDs. Even though the LED state was generally static the software was written to update the shift register in the 6275 once each 100 milliseconds. The flicker seemed to indicate that shift register was not taking the proper data each and every time it was loaded in the serial software bit-bang routine. It turned out that we traced the final problem to using too high of value pullup resistors on the 6275 Data IN and CLK lines. Lowering the resistor values made the signal rise time faster and permitted full functionality of the 6275 chip.

In conclusion....if you are going to use the 6275 chip make sure to provide logic inputs to the chip that have FULL 5 volt level swings and to that with a reasonably speedy rise time.

Let me also add that if I were designing the hardware from scratch I would probably not use the 6275 for a couple of reasons. First the mix of 5V and 3.3V is always a pain. Second this chip can end up getting rather hot if you try to pump a lot of current through the LEDs due to the fact that the outputs drop about 0.7 volts at higher currents. The chip is JUST fine for lower current applications as a local front panel type LED driver. Thirdly I think the part is a bit more expensive than what I would like to see.

Michael Karas