2

I have the following circuit:

schematic

simulate this circuit – Schematic created using CircuitLab

The SN74HC00 (from TI) has input clamps, so I am expecting the 5V logic input to be clamped to about 3.3V at pins A and B, and (5-3.3)/1000 = 1.7 mA to flow through the resistor.

In stead, I measure 4.5V at pins A and B, and even VCC now measures 4.5V! (The resulator is probably not happy with this, and is sinking current).

If I remove the SN74HC00 from its socket, the voltage at the regulator's output returns to 3.3V .

Later, I put the SN74HC00 back, disconnected the 3.3V supply from VCC, but kept the 5V input connected. VCC still measured 4.5V .

I also tested this with an M74HC00 (from ST micro) (whose datasheet does not mention clamping), expecting it to break. Same results, and the chip still works with 5V supply.

What is wrong with this circuit? Why does it not clamp the input?

rudolfbyker
  • 163
  • 7

1 Answers1

3

Chips' ESD protection circuitry typically works by directing any current that would pull an input above VDD, out the VDD line. With its input tied to +5 volts via 1K resistor as shown, and VDD rigidly fixed at 3.3 volts, the 74HC00 would probably clamp its input to about 3.5 volts, and would take the 1.5mA flowing into the input and direct it out to VDD. If other circuitry powered by VDD took 1.5mA or more, the redirected current would reduce the supply current draw by 1.5mA. Otherwise, the current will have to go somewhere else.

Some regulators will perfectly happily sink current if their output voltage exceeds the proper level; others may be destroyed, and some simply won't care (when the output voltage is below the regulation point, they'll supply current, but if it's above they'll simply sit idle). If you don't have something in your circuit which can absorb 1.5mA when an input is high, you'll have to add something. The simplest such thing may be a 2.2K resistor from the input pin to ground; that will absorb about 1.5mA when the uC is outputting +5, while wasting nothing when it isn't.

supercat
  • 46,736
  • 3
  • 87
  • 148
  • 1
    Or use a logic family with 5V-tolerant inputs such as 74LVC. – Ignacio Vazquez-Abrams Apr 03 '14 at 16:10
  • Or increase the 1k to perhaps to 10k, up to 100k or more, if speed and icapacitive coupling with fast nearby lines is not a concern. The input current of 74HC is so low. – fgrieu Apr 03 '14 at 16:36
  • Speed is of concern. I'm using this as part of a 5-0 to 3.3-0 logic level converter for SPI clock and data out lines. (The data in line uses 3.3-0 to 5-0 conversion, using 74HCT00 with V_CC=5V and inputs=3.3V) The regulator is LP2950N. I think is must be sitting idle, then. I'll try adding the (load? pull-down?) resistor tonight... :) – rudolfbyker Apr 03 '14 at 18:05
  • It works beautifully! – rudolfbyker Apr 03 '14 at 22:53