74HC595 creating Fake output

[Prabhakarankft]

Dear Everyone,
Greetings! I have designed a PCB for 8 channel 24V solenoids controlling Purpose with 74HC595! The Schematic is Very simple it is just one 74HC595 IC with 8 IRF540 MOSFETs! 74HC595 ICs Control lines are from another PCB ( Latch,OE,CLOCK,DATA @ 3.3V Logic). The cable is nearly 45cm in Lenth. Shift register IC works with 5V power (Input Voltage is 15V and regulated using LM7805)

Everything works fine Most of the time! When We connect any other Power supplies(Or any type of equipment) to the AC 230V power plug, 74HC595's output goes Wrong!
Initially, I thought noises are due to Load connected on the MOSFETs! Even though removing the load from the PCBs same things are happening! MOSFETs are isolated by PC817 Optocoupler So no connection between 24V and 15V supply). So I guess this is due to Power glitches in the 230V line. Problems are happening only When We connect any other appliances

So far tried:
1. I have connected Latch pin with 100E resistor(Series) and 0.1uF Capacitor Parallel - Problem remains Same
2. 100uF Capacitor and 0.1uF (Ceramic Capacitor very close to power supply pins of IC)-Problem remains Same
3. 0.1uF Capacitor on OE Pin - Problem remains Same
4. 0.1uF 640 V Polyester Cap on AC 230V Line- Problem remains Same
5. All the solutions mentioned above in a single shot- Problem remains Same
Any suggestions to solve these issues?? Schematic files are attached for everyone's refernce . Thanks in Advance

 

[KlausST]

Hi,

using a 5V powered HC595 with 3.3V inut signals is off specification. (BTW your schematic shows "LS595" -> give consistant information)
Better use a HCT595.

Read datasheets on signal voltage levels
* 3.3V side --> V_OL, V_OH
* as well as 5V side --> V_IL, V_IH

There may be other issues, like GND bounce, noise, timing...

Klaus

 

[crutschow]

As Klaus noted, that device is not specified for 3.3V logic signals which likely makes the device very sensitive to noise on the logic high signal.
You need to convert those signals to 5V.

 

[danadakk]

If you have a DSO do the following :

1) Set it to infinite persistence, and look at supply rails to see what pk-pk noise is.
2) Turn off infinite persistence, and set scope for pulse width trigger, set it to a value
< clock PW by say 10%, and see if you get any runt triggers.
3) Set scope to trigger single shot, pulse level < 0V, say -.6v, to see if you get any
triggers.

The above can be clues as to PCV layout, excessive L in traces undamped, etc..

Post pictures, well focused, of PCB and its layout, top and bottom.

All unused inputs must be at ground of Vcc.


Regards, Dana,.

--- Updated Jun 13, 2022 ---

I see this was not observed in design -

Regards, Dana.

--- Updated Jun 13, 2022 ---

I am not sure of your source supply implementation, so this could affect how you do
your measurements -

https://download.tek.com/document/3AW_19134_2_MR_Letter.pdf

Regards, Dana.

 

[Prabhakarankft]

Hi,

using a 5V powered HC595 with 3.3V inut signals is off specification. (BTW your schematic shows "LS595" -> give consistant information)
Better use a HCT595.

Read datasheets on signal voltage levels
* 3.3V side --> V_OL, V_OH
* as well as 5V side --> V_IL, V_IH

There may be other issues, like GND bounce, noise, timing...

Klaus

Hello KlausST,
Thanks for your valuable reply! For your information, I have used 74HC595D( Nexperia Make) in my PCB. As per your input, I have changed the Power Supply from 5V to 3.3VDC by using LM1117-3.3V LDO.
The problem remains the same, as stated earlier Everything works fine Most of the time! When We connect any appliances in Power plugs(Or any type of equipment) to the AC 230V power plug, 74HC595's output goes Wrong.

--- Updated Jul 5, 2022 ---

If you have a DSO do the following :

1) Set it to infinite persistence, and look at supply rails to see what pk-pk noise is.
2) Turn off infinite persistence, and set scope for pulse width trigger, set it to a value
< clock PW by say 10%, and see if you get any runt triggers.
3) Set scope to trigger single shot, pulse level < 0V, say -.6v, to see if you get any
triggers.

The above can be clues as to PCV layout, excessive L in traces undamped, etc..

Post pictures, well focused, of PCB and its layout, top and bottom.

All unused inputs must be at ground of Vcc.


Regards, Dana,.

--- Updated Jun 13, 2022 ---

I see this was not observed in design -

View attachment 176832


Regards, Dana.

--- Updated Jun 13, 2022 ---

I am not sure of your source to supply implementation, so this could affect how you do
your measurements -

https://download.tek.com/document/3AW_19134_2_MR_Letter.pdf

Regards, Dana.

Hello Dana,
Thanks for your suggestions! I will all your suggestions and I will let you know the results! I have not much experience using DSO! I will google it and hope for the best

--- Updated Jul 5, 2022 ---

As Klaus noted, that device is not specified for 3.3V logic signals which likely makes the device very sensitive to noise on the logic high signal.
You need to convert those signals to 5V.

Hello,
I have converted my power supply to 3.3V DC. I thought this could be the reason for the issue, but the problem is somewhere else

 

[betwixt]

Klaus suggested using a 74HCT595 but you say you used a 74HC595, their logic thresholds are different, please clarify which was used.

This is almost certainly a layout problem, some trace impedance is allowing enough voltage to develop that a logic level or supply level is being breached, please show a photograph of your construction, in particular the track layout.

Brian.

 

[Prabhakarankft]

Klaus suggested using a 74HCT595 but you say you used a 74HC595, their logic thresholds are different, please clarify which was used.

This is almost certainly a layout problem, some trace impedance is allowing enough voltage to develop that a logic level or supply level is being breached, please show a photograph of your construction, in particular the track layout.

Brian.

Hello Betwixt,
We are sure We have used 74HC595D( Nexperia Make). I am very eager to know the errors in the PCB board layout to rectify! I have assumed that just routing the Air wire will give the expected result! But I am not aware of trace impedance! Please give guidance!
We have attached all the necessary documents for your kind reference. R34 & JA5 are not populated/Soldered

 

[KlausST]

Hi,

In post#5 you said you used 3.3V supply. I can't see this. Please provide consistent informations.

What's the SPI master's output levels?

Your PCB layout (including schematic) just makes connections, but it does not provide all the other features a modern PCB layout should do:
It does not:
* create a reliable, low impedance GND (plane). Not for DC and low frequencies (GND bounce) nor for high frequencies.
* care about current flow. Current and signals are no "one way" paths from A to B, they alwas need a return path back from B to A.
* give useful isolation (clearance) at the optocouplers
* give a low impedance path for the protection diodes at the output. The diodes are fast, but not your PCB layout, and not your schematic, and most probably not your capacitor.
* show any ESD protection.
* show any EMI/EMC features
... there are more issues.

Thus there is a good chance that the PCB layout is the source of your problem.
Still there is a lot, that we don't see:
* the whole wiring. It has big impact on signal reliability. Length, path, type, branches...
* the timing of your signals. Maybe you just use the wrong SPI setup regarding CLK_to_DATA timing.

Klaus

 

[Prabhakarankft]

Hi,

In post#5 you said you used 3.3V supply. I can't see this. Please provide consistent informations.
>Yes, We did PCB with 5V Power Supply, Later I have converted the Power supply from 5V to 3.3V by use of the LM1117-3.3V regulator. 10uF Capacitor used on input and output of LM1117 Regulator

What's the SPI master's output levels? That is 3.3V

Your PCB layout (including schematic) just makes connections, but it does not provide all the other features a modern PCB layout should do:
It does not:
* create a reliable, low impedance GND (plane). Not for DC and low frequencies (GND bounce) nor for high frequencies.
>Other than Copper pouring with Ground Vias on both the layers What should I care extra?

* care about current flow. Current and signals are no "one-way" paths from A to B, they always need a return path back from B to A.
>Right now We are using only one Ground Path. We will add another one pin for DC ground. My idea was regulators are used for just power up the 74HC595D and optocouplers. Total Board current consumption is just <200mA( At a time only one optocoupler will ON)

* give useful isolation (clearance) at the optocouplers
> You mean Pad to copper Clearance. 0.6mm Already provided

* give a low impedance path for the protection diodes at the output. The diodes are fast, but not your PCB layout, and not your schematic, and most probably not your capacitor.
* show any ESD protection.
> No idea about ESD / EMI/EMC
* show any EMI/EMC features
... there are more issues.

Thus there is a good chance that the PCB layout is the source of your problem.
Still there is a lot, that we don't see:
* the whole wiring. It has big impact on signal reliability. Length, path, type, branches...
>From master SPI to our PCB we are using 30 cm RMC Cables
>We using 5 Meter Cable solenoids and our PCBs only at Load side- cables are good enough to carry the Current
* the timing of your signals. Maybe you just use the wrong SPI setup regarding CLK_to_DATA timing.
>We will probe

Klaus

Hello Klaus,
Thanks for your guidance.

 

[Aussie Susan]

Hello Betwixt,
We are sure We have used 74HC595D( Nexperia Make).

You seem to not understand what you are being told. The EXACT part number is important here - the part you are using IS DIFFERENT to the one you are being told to use.
Using a LM7805 with a 12V input means that it needs to drop 7 volts - even with a small current draw that can lead to it getting hot.
Finally, I can't see what you mean when you say the "...output goes wrong". In what way? What should it do that it is not? What should it NOT do that it is?
Susan