7-Segment Reverse Bias

Many rookies (and also many professionals) are making a capital mistake by configuring the segment drivers (i.e. the port pins since normally there are no additional drivers here) as a push-pull (or totem-pole) output, meaning the pin is simply an output driving against Vcc or GND. Well, this works indeed, but not always and not always for a long time...

In this picture, CH1 is a segment and CH2 is a CC of a display (each directly the pin at the display, not at the controller!). The math-channel shows the difference i.e. the voltage over the diode. The pin is permanently configured as an output. You can see the segment driver bravously driving against V_CC and GND (since we are measuring on the display, V_CC only swings to about 3 V, the rest drops over the series resistor) and you can see the effects on the CC of the display. The cathode driver is active in the 3rd and 9th division, here we can see the saturation voltage of the darlington transistor of about 0.8 V, in the rest of the time it is hi-Z. The math-channel shows us that the diode sometimes is positively biased, other times it bears a reverse bias.

If we watch a digit currently driven, this is no problem but if you look at a digit that is 'off' you can see a reverse bias over some LEDs since the cathode driver is hi-Z and segments being driven low experiences a reverse voltage caused by other diodes that are forward biased (i.e. being lit on some other digit). This will not be full V_CC but two or three volts might occur.

This is why we should drive the digit only against GND (what the ULN2803 does perfectly) and the segment only against V_CC but not against GND! The solution is not to use the PORTx but the DDRx. We switch the PORT permanently high and only the DDR decides, wich segment gets some current.

With current 7-segment-displays you will frequently find a maximum reverse voltage of about 5 V so they may forgive this fault. But if you use high current LEDs for a large display the datasheet will state Avoid reverse bias by design i.e. reverse voltage is simply inacceptable and could cause anything from reduced brightness over reduced endurance up to immediate failure.

The datasheet of our display also specifies a maximum reverse voltage of 5 V but it also states Prolonged reverse bias should be avoided.... You cannot decide if the short times of reverse voltage in multiplex-use would accumulate but if you want to be sure, it would be good practice to avoid unnecessary reverse bias.

In the second picture, the PORT is permanently '1' and only the DDR is switched. You can see the segment pin not longer driving against GND but only buckling down until the segment draws no current. As the math-channel shows, the diode now gets no reverse bias any more.

All signals now are in permissible range and that's how it should be. Additionally you can see much lower amplitudes in the steep signals what means reduced EMI, which is of growing importance nowadays.

Indeed, it showed to be advantegous to keep the pull-ups active. Their very low current makes no visible illumination of the segments switched off but reduces the reverse bias that would otherwise build up over the junction capacitance of the LEDs.

The glitches you see on the math-channel are only resulting from these capacitances.