2
\$\begingroup\$

I've run into a problem with one of the designs recently.

I need to drive several GPIO pins over a long-ish ribbon cable. I have used a buffer chip between my micro and the ribbon (SN74Abt162244) which is also a level shifter from 3.3V to 5V. Everything worked well until I connected the ribbon. One by one the outputs of the buffer chip started to give up.

The buffer should be able to give 12 mA so I am wondering if maybe ribbon capacitance was a bit too much and overloaded the outputs.

Maybe there is an IC capable of driving such a long ribbon?

Any suggestions are very welcome.

\$\endgroup\$
3
  • \$\begingroup\$ I've used RS-422 drivers and receivers for long runs of cable without any problems. \$\endgroup\$ Commented Jun 19, 2015 at 19:40
  • \$\begingroup\$ Thanks Leon. I have about 26 lines to drive over those ribbons. Not sure if there is enough space... \$\endgroup\$
    – radaudio
    Commented Jun 19, 2015 at 19:45
  • \$\begingroup\$ In what way do the outputs "give up"? Have driven signals many feet (10-15) at speeds up to 20MHz with no problem. The only PATA spec used ribbon cable for up to 33MHz over 18". \$\endgroup\$ Commented Jun 19, 2015 at 21:35

1 Answer 1

2
\$\begingroup\$

You should get a scope on the lines to see what is going on when the cable is connected. If the drivers are "giving up" as you say there is a very good possibility that they are being fried by under voltage or over voltage spikes due to reflections coming back down the cable. The scope can help to identify if this is happening.

Voltage spikes that go over the driver VCC by much more than 0.3V or lower the the driver GND by -0.3V can cause latchup or permanent damage to the driver. If the driver is experiencing latchup you may be able to turn off all power and let the board sit over night. It may work again after the trapped charge in the chip that kept the stray PNPN junction in the chip biased has a chance to bleed away.

One of the simplest solutions to this problem is to place small value resistors in series with the output pins of the driver. Select the resistor size such that the resistor value plus the output impedance of the driver begins to match the cable characteristic impedance.

You may want to also look at selecting a driver/level translator chip built with a slower process technology to slow down the rise and fall times of the driver outputs. This will go a long way toward helping to reduce the amount of potential +/- overshoot seen in the cable.

\$\endgroup\$
1
  • \$\begingroup\$ Thanks Michael. Fantastic answer. Looking at the CRO output pins got some serious overvoltage (went up to 6V on the rising edges). I will experiment with the resistors and change the chip to something slower in next iteration. \$\endgroup\$
    – radaudio
    Commented Jun 21, 2015 at 7:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.