3.3v5vlevel-shiftingpower supply
In my design I am using 74HC245 buffers between a input signal of +5v and controller pins of +3.3V. What is the supply value I have to use for buffer (74HC245). I want to use +3.3V as the supply, but in data sheet it is mentioned that maximum VIH is Vcc. Please suggest me what happens if I provide +3.3V as supply voltage and feed input pins +5V?.
I am providing the link for buffer data sheet below.
http://www.nxp.com/documents/data_sheet/74HC_HCT245.pdf
Thanks in advance
@AKR: Your suspicion about the body diode is right. Since the drain was higher than the source, the transistor was acting as a diode rather than an open transistor. So, fix #1 was to invert the transistor. In the battery only case, the transistor would initially act as a diode, so Vs = Vd - 0.7, then Vgs < Vth and the transistor turns on, hence preventing the voltage drop. The other fix was to add a Schottky diode to prevent current from flowing back to the gate of the FET.
For folks who are interested, here is the final design:
Thanks for the tip! Moving on to other problems...
Your circuit cannot block forward currents because of the MOSFET body diode. Notice how it is shown in the application diagram from your TI datasheet:
Even if you pull the gate voltage to 0, the body diode will still conduct, however it will drop some tenths of a volt, and so consume power. The reason to use the TPS2412 is that when the TPS2412 applies a high voltage to the FET's gate, it creates a conducting channel through the MOSFET, reducing the power consumed by the FET.
On the other hand when the "voltage source" is not available, the TPS2412 can still prevent reverse current flows from the "common rail" by pulling the gate low.
Best Answer
If you want to translate 3.3V signals to 5V then you must power the 74HC245 with 3.3V, because at 5V it might not reliably detect a 3.3V 'high' logic level. However it will also only output 3.3V, which may not be enough to reliably drive 5V logic.
Any 5V inputs will inject current through the protection diodes and out the Vcc pin into the 3.3V supply. Depending on how 'stiff' the supply is, this may either exceed the protection diode's current rating, or raise the 3.3V supply to ~4.4V (which would be bad news for anything on the 3.3V side that can't handle the higher voltage).
You could wire resistors in series with the I/O's on the 5V side to limit diode current, but they will slow down the signal rise and fall times - and make the 5V output level problem worse.
If you power the 74HC245 with 5V then you have similar problems, but on the 3.3V side.
So the answer is:- use a proper level translating buffer such as the 74LVCC4245A, which is powered by 3.3V on one side and 5V on the other.