From: Winfield Hill
Subject: Re: blocking signal to unpowered CMOS
Date: 20 Nov 2002 03:51:47 -0800
Organization: Rowland Institute
X-Newsreader: Direct Read News 4.11
Bob Wilson wrote...
> Jumbaliah email@example.com says...
>> Is it good practice to use an NPN (3904) transistor to block a signal
>> to an unpowered CMOS IC? I have a circuit that has a 5V signal on the
>> collector all the time. However when I switch off the circuit, the
>> base is left connected to other unpowered circuitry and the emitter of
>> the 3904 is connected to the input of the unpowered CMOS IC via a 10k
>> resistor. The ground for the circuit is always connected. When the
>> circuit is powered up the IC gets power and the base of the 3904 also
>> receives power thus allowing the signal to pass. (I know they're may
>> be timing issues as to which powers up first)
>> I've read about clamping diodes, but they only limit the signal's
>> voltage range and not block it from an unpowered IC.
>> Is this good? Bad?
Jumbaliah, I cannot follow your description of the 3904 connection,
can you please be more explicit?
> An unpowered CMOS IC will do funny things when voltage is applied to
> its input. What happens is that the voltage goes through the upper
> of the protection diodes, and onto the interal Vcc line, and the
> IC springs to life.
> There are a variety of ways to ensure the inputs are left "unpowered"
> choose one that works for you.
I agree, some method is necessary. At least a series resistor.
One that seems nice, although I haven't used it, employs JFETs.
Its purpose is to reduce the value of the series input protection
resistor to retain a fast logic connection, without increasing
the fault current into the unpowered device's protection diode.
. | |
. ____V_ |
. | | |
. ----' '--/\/\--+---
Use a small low-capacitance low-threshold-voltage JFET, such as a
2n3687 - pn3687. Hmm, I'm sure there's a more available choice.
Another method employs small MOSFETs, with their gates connected to
the destination +5 supply. When that supply is off the FET is off.
Low-threshold-voltage MOSFETs are necessary, because the destination
logic is robbed of a that voltage, say 0.7 to 1.5V for a 2n7000.
This is not a problem for HCT logic, etc., with 5V input swings and
a 1.2V logic threshold.