How much supply voltage swing can a latch endure and still remember

[[email protected]]

I am planning to use a bunch of 74HC595's latches to drive some LED
displays. I anticipate the supply voltage to vary (intentionally at a
slow 1mS rep rate) between 2.5 and 5v with slew rates about 1v/uS.
Can these latches still hold their state without changing at these
kind of rates? How could I simulate this to get an idea of how much
supply voltage swing that could be tolerated before data gets
corrupted?
thanks,
al

 

[Joerg]

I am planning to use a bunch of 74HC595's latches to drive some LED
displays. I anticipate the supply voltage to vary (intentionally at a
slow 1mS rep rate) between 2.5 and 5v with slew rates about 1v/uS.
Can these latches still hold their state without changing at these
kind of rates? How could I simulate this to get an idea of how much
supply voltage swing that could be tolerated before data gets
corrupted?
thanks,

Well, this spec says 2V min:
http://www.nxp.com/acrobat_download/datasheets/74HC_HCT595_4.pdf

Keep in mind that they need good bypass caps and you'll have to muscle
those around.

 

[[email protected]]

Why on earth are you slewing the supply voltage?

JF

I'm using it to control the If / brightness of the LED's . Only 1 LED
per latch string
will ever be lit. I can save space by minimizing #resistors vs simple
control circuit
without using insanely small parts. When the LED's are off, the
supply is 5v , LED's
on and supply is about 2.5v to 4v +/- some ripple depending on LED
color.

I understand Tim's comment about input levels following Vdd and am
also aware
that latch power up state depends on device characteristics and
mismatch. I'm
concerned that mismatch will throw a curveball when there is data in
the latch.

al

 

[[email protected]]

I'm using it to control the If / brightness of the LED's .  Only 1 LED
per latch string
will ever be lit. I can save space by minimizing #resistors  vs simple
control circuit
without using insanely small parts.  When the LED's are off, the
supply is 5v , LED's
on and supply is about 2.5v to 4v +/- some ripple depending on LED
color.

I understand Tim's comment about input levels following Vdd and am
also aware
that latch power up state depends on device characteristics and
mismatch. I'm
concerned that mismatch will throw a curveball when there is data in
the latch.

al

This sounds pretty nutso to me. Capacitive loading on the output pins
could lead to latch-up should they be higher than the power supply.

In theory, for a static cmos circuit, you only need the largest Vt to
retain a logic level. I'd prefer you pwm the led itself instead.
Better yet, use the max7219.

 

[Joerg]

This sounds pretty nutso to me. Capacitive loading on the output pins
could lead to latch-up should they be higher than the power supply.

What's capacitive on the outputs here?

In theory, for a static cmos circuit, you only need the largest Vt to
retain a logic level. I'd prefer you pwm the led itself instead.
Better yet, use the max7219.

Philips/NXP makes a nice PWM mux chip for LEDs. PCF-something.

 

[[email protected]]

What's capacitive on the outputs here?


Philips/NXP makes a nice PWM mux chip for LEDs. PCF-something.

It's not likely to have capacitance across the LEDS, but given how ill
conceived this project seems, I figured I'd point out the perils of
slewing the power supply when other pins may lag the supply voltage. I
have no idea if 74HC chips use epi, but given the cost, I doubt it.

Factory latch-up testing does not move a hard supply. You don't know
how many problems have been caused by systems designers that refuse to
use shutdown pins because the datasheet might say a few uA is
consumed, so they use a pass device on the power pin.

 

[Joerg]

It's not likely to have capacitance across the LEDS, but given how ill
conceived this project seems, I figured I'd point out the perils of
slewing the power supply when other pins may lag the supply voltage. I
have no idea if 74HC chips use epi, but given the cost, I doubt it.

Factory latch-up testing does not move a hard supply. You don't know
how many problems have been caused by systems designers that refuse to
use shutdown pins because the datasheet might say a few uA is
consumed, so they use a pass device on the power pin.

You'd probably consider many of my circuits ill conceived

Sometimes when every penny counts you've got no other choice. I've seen
one design where they supplied a uC from mains via a resistor. The
rectifier was <gasp> a substrate diode. Now I wouldn't go quite that far
but it sure was clever.