PDA

View Full Version : I/O Pin Protection



Rumple
07-23-2009, 09:56 PM
Newbie question...so new that I haven't actually received my hardware yet...

Is inserting a 10K resistor in a data line an acceptable method of shifting from a 5V TTL output level to a 3.3V Propeller input level?

Thanx.

Mike Green
07-23-2009, 10:35 PM
10K will probably work. 2K or 2.2K is better. The 5V signal will conduct through one of the I/O pin protection diodes to the 3.3V supply "rail". The resistor has to drop 5V - 3.3V - 0.6V = 1.1V and has to limit the current to under 0.5mA. A 2K or 2.2K resistor would do this. If there's a lot of capacitance involved on the data line, the additional resistance (if you use 10K) will slow down the signal changes. Most of the time 10K should work fine. A common recommendation is to use 1K, particularly when the data line comes from a 5V IC. In that case, the 5V part usually outputs a high signal more like +4.7V at most and you have 4.7V - 3.3V - 0.6V = 0.8V. With a 1K resistor, you'll get a bit more than 0.5mA, but not much. The protection diode is designed for 0.5mA, but is much more robust than that. That's the thinking process.

Post Edited (Mike Green) : 7/23/2009 3:42:58 PM GMT

BigFoot
07-24-2009, 03:03 AM
You need to be careful with 1K resistors, especially if you are interfacing to allot of 5V signals. Remember that the other end of the protection diodes are connected to Vdd. Your 3.3 volt bus will suddenly become 3.8 volts and some of the other chips on the bus might not like this high a supply voltage. Allot of the new low voltage chips spec 3.6 volts max.

Peter Jakacki
07-24-2009, 05:42 AM
As BigFoot mentioned this is a problem not often considered but where the problem can become apparent is usually when the Prop is drawing very little current so that the Vdd rises up to around 4V or so (non-destructive substrate breakdown). This can happened during reset and I have found that this can affect program loading so one way of handling this is to use a voltage clamp across the supply. I prefer the method where I use two red leds in series with a 22R resistor as this will exhibit a sharper "knee" than low voltage zeners and they tend to clamp rather sharply once the Vdd tries to rise above 3.3V.

*Peter*