74HC595 floating pin

[V8TR4]

Hi everyone,

Been using the 74HC595 for a while now and it has always been noisy. Today I
was doing some routing of a new board using clips of previous circuits when
I saw that I did not have the last 74HC595 out pin connected to anything.
Should this pin be tied to a pull down resistor when not used?

Thanks,

 

[Wouter van Ooijen]

Been using the 74HC595 for a while now and it has always been noisy. Today I

was doing some routing of a new board using clips of previous circuits when
I saw that I did not have the last 74HC595 out pin connected to anything.
Should this pin be tied to a pull down resistor when not used?

As you said yourself, it is an *out* pin.


Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
PICmicro chips, programmers, consulting

 

[John Jardine]

Hi everyone,

Been using the 74HC595 for a while now and it has always been noisy. Today I
was doing some routing of a new board using clips of previous circuits when
I saw that I did not have the last 74HC595 out pin connected to anything.
Should this pin be tied to a pull down resistor when not used?

Thanks,

Pin 9 is left open.
Use a *lot* of these chips with no problems. What's this 'noise' your're
getting?
regards
john

 

[V8TR4]

Hi John,

I use the 74HC595 for relay actuation as well as 7 segment LED display
drivers. The equipment I use them on unfortunitly is also using a 20kw
Electron beam gun used for deposition. Lots of arc outs and plasma's. These
rattle the shift register causing funny displays and relays misfiring. This
has been a big problem for awhile. I have caps across the power pins, and
pull down resistors on the 3 lines running the shift register from the CPU.
The funny thing is the only chips effected are the shift registers. I use
PIC and Basic Stamps with no problem. As well as other TTL chips.

Any ideas are much appreciated,

Mark

 

[Wouter van Ooijen]

I have caps across the power pins, and

pull down resistors on the 3 lines running the shift register from the CPU.
The funny thing is the only chips effected are the shift registers. I use
PIC and Basic Stamps with no problem. As well as other TTL chips.

Any ideas are much appreciated,

small caps (10 ... 1000 pf) on the SR inputs, especially the clock?




Wouter van Ooijen

-- ------------------------------------
http://www.voti.nl
PICmicro chips, programmers, consulting

 

[Valentin Tihomirov]

If you feed garbage to an input do not wonder you get garbage at he output.
However, I had not any problems with floating inputs fo far. It does not
depend on functioon, e.g. 595, 165 or 02, the opportunity to ger an error
depends on family of device.

 

[Sir Charles W. Shults III]

Any time I build circuitry for use in devices with high voltages or
currents, I always enclose the circuitry in a shielded metal box, also known as
a Faraday cage. Ground the box and use only shielded wiring in and out of it.
You can't just hang wires around something like this. Every one that is out
there is an antenna ready to pick up the signals being inadvertently broadcast
by your arcs and discharges. And if you are driving your primary electron beam
with AC, you have already created a noise situation.
Any time you have a pair of wires at different potential and no shielding,
they form a loop antenna. The trick is to minimize the area inside the loop,
making the induced signal very small. Also, bypass capacitors are an essential
part of any logic circuit. You must have a 0.01 to 0.1 uF capacitor across
every power and ground terminal of every chip on the board, depending on the
power draw of your chips and the current capability of your supply.

Cheers!

Chip Shults
My robotics, space and CGI web page - http://home.cfl.rr.com/aichip

 

[Colin Jackson]

Watch out for slowing down the input signal slew rate with the caps. It can
actually make the problem worse!

 

[John Jardine]

What if the upset in your main controller?

In that case you're buggered.

Mark mentioned though, that the micro wasn't being upset.
regards
john

 

[V8TR4]

I am going to see about making the modifications you mentioned this
afternoon. Run a test and see how it goes.

Maybe I should consider using a different chip all together for my relays.
The PCF8574 is an I/O expander that uses teh I2C protocol. I am thinking
that it might be more immune based on it not being a shift register.
I would hate to change chips midstream. Another idea I had was to isolate
out the shiftregister signals by having a secondary relay turn off the power
to relay from register and turn on power from a power source. This is more
of a kludge but would at least stop the chance of a misfiring. This winds up
costing me 2 extra I/0 though.

Thanks for all of you advice, on the upside to all this at least I will be
an "expert" on noise immune systems by the time I get this worked out

Mark

Hi John,

I use the 74HC595 for relay actuation as well as 7 segment LED display
drivers. The equipment I use them on unfortunitly is also using a 20kw
Electron beam gun used for deposition. Lots of arc outs and plasma's. These
rattle the shift register causing funny displays and relays misfiring. This
has been a big problem for awhile. I have caps across the power pins, and
pull down resistors on the 3 lines running the shift register from the CPU.
The funny thing is the only chips effected are the shift registers. I use
PIC and Basic Stamps with no problem. As well as other TTL chips.

This is what I do:

Use a board with a groundplane. This can usually be double sided PTH,
with the top layer used as ground plane. (You can route a few *short*
tracks on this if you have to).

All wires terminating on the board must have a TVS diode, at the wire,
to the ground plane. You then need some series impedance near the
diode before the signal gets too far onto the board.

|'''''''''''

-----[Z ] ----| Logic chip I/O

| |
TVS |___________
|
---
-

Z can be a highish value resistor, say 2k2, for inputs. For low
current outputs, say 10 ohms. For higher currents, use an inductor or
ferrite bead.

I use the groundplane as my logic 0V, and connect it via metal
mounting pillars to the chassis. (Actually there seems to be some
debate on whether this is a good idea or not. The text books and
almost everyone else seem to think there should be separate grounds
for logic and chassis.)

 

[John Devereux]

I am going to see about making the modifications you mentioned this
afternoon. Run a test and see how it goes.

Maybe I should consider using a different chip all together for my relays.
The PCF8574 is an I/O expander that uses teh I2C protocol. I am thinking
that it might be more immune based on it not being a shift register.
I would hate to change chips midstream. Another idea I had was to isolate
out the shiftregister signals by having a secondary relay turn off the power
to relay from register and turn on power from a power source. This is more
of a kludge but would at least stop the chance of a misfiring. This winds up
costing me 2 extra I/0 though.

If your system has enough noise to currupt a simple chip like a
74HC595, then I'm afraid that no digital circuitry is safe! IMHO you need
to prevent the spikes getting into the circuit in the first place.