74HC595 behaves randomly, need help with my circuit

Question

I'm trying to make a Nixie clock based on an ESP32. I've made a working prototype on a breadboard. Only the 74HC595 part, which is where the problem is at:

I then made a PCB:

After soldering everything to it (I've verified every solder connection, it's all good,) the 74HC595 behaves randomly. Each time I start it, the output values are all random until it settles on a number. I have tried everything I saw on the internet but nothing works. I also tried adding a bunch of capacitors but it doesn't work either.

Here are the results after soldering:

How to troubleshoot this?

Here is the test code U'm using for both the working prototype and the PCB:

//uint8_t latchPin = 19;
//uint8_t clockPin = 21;
//uint8_t dataPin = 18;

uint8_t latchPin = 33;
uint8_t clockPin = 25;
uint8_t dataPin = 32;

const byte segments[10] = { // Lookup table for digit patterns (common cathode)
  0b11110000,  // 0
  0b00001111,  // 1
  0b11110000,  // 2
  0b00001111,  // 3
  0b11110000,  // 4
  0b00001111,  // 5
  0b11110000,  // 6
  0b00001111,  // 7
  0b11110000,  // 8
  0b00001111   // 9
};

void setup() {
  // Set control pins as outputs
  pinMode(latchPin, OUTPUT);
  pinMode(clockPin, OUTPUT);
  pinMode(dataPin, OUTPUT);

  Serial.begin(115200); // Start serial communication for debugging
  Serial.println("Starting...");

  displaySegments(0x00000000); // reset on startup, not needed just makes it look better ig
  delay(1000);
}



void loop() {
  // Loop through each digit pattern (modify for specific display)
  for (int i = 0; i < 10; i++) {
    shiftOutt(dataPin, clockPin, MSBFIRST, segments[i]);
    digitalWrite(latchPin, LOW); // Latch the data
    delay(2000);     // Short delay for visibility (adjust as needed)
    digitalWrite(latchPin, HIGH); // Unlatch for next data

    Serial.print("Displaying: ");
    Serial.println(segments[i]);
  }
}

void displaySegments(byte pattern) {
  shiftOutt(dataPin, clockPin, MSBFIRST, pattern);
  digitalWrite(latchPin, LOW);
  delayMicroseconds(1000); // no need to edit this
  digitalWrite(latchPin, HIGH);
}

void shiftOutt(uint8_t dataPin, uint8_t clockPin, uint8_t bitOrder, uint8_t val) // just in case
{
      uint8_t i;

      for (i = 0; i < 8; i++)  {
            if (bitOrder == LSBFIRST)
                  digitalWrite(dataPin, !!(val & (1 << i)));
            else
                  digitalWrite(dataPin, !!(val & (1 << (7 - i))));

            digitalWrite(clockPin, HIGH);
            digitalWrite(clockPin, LOW);
      }
}

EDIT: Thanks for all your answers ! I read them all and tried all of them, but it still behaves randomly. What i tried:

• Added 104 (0.1 µF) ceramic capacitors between GND and VCC for both the 74HC595 and K155NA1
• Added some delay between each clock pulse: delayMicroseconds(2000); digitalWrite(clockPin, HIGH); delayMicroseconds(2000); digitalWrite(clockPin, LOW);
• Disconnected the power supplies of the K155NA1 and connected them to the 3V3 of the ESP32

Nothing has worked so far

Answer 1

Actually it's the other way around, I would not expect the breadboard prototype to work well either. The breadboard even connects LEDs directly to HC595 output pins which may lead to burnt-up LEDs or chips as there should be current limiting resistors.

There are no supply bypass caps on the IC power supply pins, and both implementations of the circuit uses long wiring between chips.

The only difference is, that when comparing the two, the PCB uses even thinner and longer wiring than the breadboard, so you have more problems with wiring inductance and less help from breadboard having lots of stray capacitance everywhere.

So basically the PCB has now most of the copper etched away on both sides and long narrow tracks are left behind while it should ideally be the opposite, most copper left present to carry a proper ground plane, preferably a large plane or wide tracks for supply current, and the data wiring on top of a ground plane to shield them.

The code itself should work, however with the actual PCB, the clock pulses for sending data may be too short and get distorted on the PCB and there may be reflections due to the PCB design not being very compatible with high speed design. Any series resistance on the clock pin and putting some delay to make clock pulse longer and letting the data pin stabilize before rising clock edge would be a nice addition.

Answer 2

I would like to thank all of those who tried helping me troubleshoot this problem.

Unfortunately, none of the answers worked. I therefore used the remaining pins of the ESP32 to drive the K155NA1 instead of the 74HC595.

Answer 3

Each time I start it, the output values are all random until it settles on a number

It sounds like you are seeing the register powering up in an undefined state, which is normal. There's no guarantee that any of the register bits will be 1 or 0 after power is applied.

To ensure that the register is clear (all zeros), it's common practice to apply a short reset pulse at power-on, by holding \$\overline{MR}\$ low (pin 10, which your schematic has labelled SRCLR) using an RC delay:

^{simulate this circuit – Schematic created using CircuitLab}

The values shown hold \$\overline{MR}\$ low for a few milliseconds, approximately \$\tau=R_1C_1\$, during which time all register bits are set to zero while the power supply settles.

Of course, any data operations on the register (by the microcontroller) that occur during this reset period will be ignored, so you must wait before you start shifting in data.

Answer 4

I see two possible issues here:

1. You should have at least 100nF (0.1uF) capacitors between the positive supply pins of 74HC595 and K155NA1, and their GND pins. They shoud be as close as possible to those pins and their leads/connections should be as short as possible.

2. Datasheet for 74HC595 mentions that its outputs should NOT be pulled up above its supply voltage, which is 3.3V, while K155NA1 is supplied by 5V and its inputs seem to leak currents from its positive, 5V supply.

K155 IC family datasheet found at https://www.digchip.com/datasheets/parts/datasheet/922/K155NA1-pdf.php :

Page 14 of 74HC595 datasheet from Texas Instruments (https://www.ti.com/lit/ds/symlink/sn74hc595.pdf) :