HC165 Shift register reading bad data

I have a shift register HC165 for upto 8 buttons and after soldering it to PCB, it now reads bad data even there is no button pressed.

Everything was working fine when tested on breadboard, but now when everything is installed in a box. It is auto raising button events itself. And when I touch the box(sheet metal), buttons' events start to raise very faster.

I then grounded the whole box but nothing happened. I measured the DC voltage from box to GND and its around 9mV and when I touch the box, it gets to 45mV.

And if I removes the wires of buttons, there is no event. Which seems buttons are doing something wrong ? I even isolated buttons from box still the same problem.

Here is the schematics

hello
have you test your circuit on 5v? your circuit is ESD sensitive make better ground of your circuit. also try with pull up resistors instead pull down

also try with pull up resistors instead pull down

Click to expand...

Won't expect any difference for CMOS logic. You forgot to mention the resistor values. Several k up to 100k seems appropriate.

Like abc_de I suspect an ESD or signal quality problem. Without knowing the hardware details it's difficult to suggest a fix. A standard debouncing algorithm, requiring e.g. 10 to 50 ms stable input state before detecting a key press will be probably able to hide the problem.

abc_de said:

hello
have you test your circuit on 5v? your circuit is ESD sensitive make better ground of your circuit. also try with pull up resistors instead pull down

Click to expand...

I can't use 5v. The microcontroller works with 3v3. How should I make better ground?

FvM said:

Won't expect any difference for CMOS logic. You forgot to mention the resistor values. Several k up to 100k seems appropriate.

Like abc_de I suspect an ESD or signal quality problem. Without knowing the hardware details it's difficult to suggest a fix. A standard debouncing algorithm, requiring e.g. 10 to 50 ms stable input state before detecting a key press will be probably able to hide the problem.

Click to expand...

I'm using 10k resistors, what hardware information should I provide?

Thank you.

try to prevent ground loops use decoupling capacitors at power supply and inputs.
second suspect is clock line of your circuit also use decoupling at CE pin of ic

abc_de said:

try to prevent ground loops use decoupling capacitors at power supply and inputs.
second suspect is clock line of your circuit also use decoupling at CE pin of ic

Click to expand...

There is a cap of 0.1uF at Vcc of hc165, should I add tantalum cap ? Also as you said use caps on inputs, does that mean 8 inputs of buttons?like 8 caps parallel to respective resistor?

Did you really leave pin 10 floating?

You should ground it or random noise will be shifted along the flip-flops.

Brian.

yes each pin must have decoupling caps parallel to respective resistors.
another way try to slow down clock of IC and distance between ic and mcu must be low as much possible

betwixt said:

Did you really leave pin 10 floating?

You should ground it or random noise will be shifted along the flip-flops.

Brian.

Click to expand...

Done, still same.

abc_de said:

yes each pin must have decoupling caps parallel to respective resistors.
another way try to slow down clock of IC and distance between ic and mcu must be low as much possible

Click to expand...

the distance is very low. I will try to slow the down the clock but that would be not good.

can you show your complete schematic and hardware pictures for better troubleshooting