2

First of all thanks for even taking your time to read this question. We are hoping someone can help assist us with an issue we have been having.

We are currently working on an automotive tail light design. The issue we seem to be facing is a simple transient voltage spike (jitter?) on the regulated voltage line - specifically the +5VDC power line from our LDO to our microcontroller. Below is a diagram of our power circuitry.

Microcontroller power setup

We have a 4 +12VDC input lines and 1 ground line coming from the vehicle (see J1 connector). Each of the 4 +12VDC input lines have a TVS diode connected to stop unwanted transient voltages. The 4 +12VDC inputs are connected via a diode to the buck regulator which is configured to output 9V.After the +12VDC signal is stepped down to a +9VDC signal, it is sent to a +5VDC LDO which then provides power to the microcontroller.

I believe our issue is associated with switching on the power and the jitter we see on the 5V regulated line at the microcontroller VDD pin (pin 1). This is causing our microcontroller to turn on sections of the taillight unexpectedly and cause very strange behavior due to exceeding the upper bound of the maximum voltage allowed (+6VDC).

Here is a video of us simulating a TURN sequence. As you can see, the other sections of the LED are briefly lighting up, reverse and brake.

https://youtu.be/V7eahFkDDPY

Here is the scope output of the +5VDC pin on the microcontroller when this happen. The jitter/transient is what we believe is causing our issues.

Transient voltage spike

When we hooked up the microcontroller to a 3rd party "clean" +5VDC power supply (Arduino), we do not experience these issues.

What is the best way to make absolutely sure our microcontroller does not see these transient voltage spikes? Should we install a TVS diode at the microcontroller power pin? We added a zener diode as you can see in our schematic, but perhaps a TVS diode is better suited here such that it will react quicker.

Schematic

Bill of Materials HERE

jacob80
  • 31
  • 2
  • 3
    There are no part numbers for any of the parts. Which microcontroller is that, which 5V regulator, which 9V regulator, which diodes, zeners and FETs are those? At least one thing sticks out, there is no high frequency bypass caps anywhere. Also the PCB wiring is important, as there are high currents running through the board, and high frequency switch mode regulator. – Justme Nov 12 '21 at 16:55
  • 1
    My apologies - I have since added a link to the BOM (Bill of Materials) at the bottom of the post. Your point regarding the high frequency bypass capacitors is definitely an important one. We have installed a 1uF capacitor at the microcontroller VDD pin to mitigate high frequency noise. Is this not sufficient? – jacob80 Nov 12 '21 at 17:23
  • 1
    Where is the decoupling for the MCU? – winny Nov 12 '21 at 18:05
  • 1
    C5 is the capacitor I am referring to – jacob80 Nov 12 '21 at 18:06
  • 1
    Electrolytic? Where is it placed? – winny Nov 12 '21 at 22:42
  • 2
    Try triggering the scope on the biggest spike (eg trigger at 6V or so, based on what is visible) and zoom in the horizontal div, to get an idea of what time scale this is. Then use one of the other channels to see if any of the power mosfets are switching at that time... – Pete W Nov 13 '21 at 00:05
  • Please include the BOM in your post instead of linking to it. Posts must include all relevant information in them, not just linked. – Null Nov 17 '21 at 03:47

2 Answers2

2

I think the main cause is the zener diode on MCU supply. If for example a voltage of switcher rises above the zener threshold, the later will start conducting and cause the switcher to further rise the voltage to compensate the drop.

IMO, you over complicated things where you shouldn't but you forgot some essentials, like a unipolar TVS protection on the input:

schematic

simulate this circuit – Schematic created using CircuitLab

Next, the inputs don't need optos, since the PSU isn't isolated type, a resistor and zener would suffice.

schematic

simulate this circuit

Marko Buršič
  • 23,562
  • 2
  • 20
  • 33
  • 1
    We do have unidirectional TVS diodes on the input lines (D10-D13). I'm not sure I follow your 2nd point regarding the signal lines coming in from the connector. We added the optocoupler (IC4) to isolate the noisy inbound vehicle signals from the microcontroller pins. Are you saying that an optocoupler is overkill for this use case and just using a resistor and zener would suffice? For the record, there is no constant power supply in this design. The 4 switching +12VDC lines coming from the vehicle must both act as signals and provide power. – jacob80 Nov 12 '21 at 17:56
1

You have the headroom to use an active filter aka capacitance multiplier between the 12V supply (or even 9V supply) and the 5V regulator.

schematic

simulate this circuit – Schematic created using CircuitLab

R1, C1 form a low pass filter, and Q1 is configured as a voltage (emitter) follower. The three components combined are called a "capacitance multiplier" and will clean up the voltage supplied to the 5V regulator. (Doesn't have to be 7805, that was just what was available in CircuitLab).

The choice of Q1 will depend upon the current you need, and R1, C1 values will follow.

Math Keeps Me Busy
  • 18,947
  • 3
  • 19
  • 65