Electrical Noise help

Discussion in 'General Electronics Chat' started by Tobias, May 5, 2009.

  1. Tobias

    Thread Starter Active Member

    May 19, 2008
    I have a board built up. I forgot to have a ground plane put on it. I know I screwed up there so no need to take me to task on that one.

    I am testing this board with a racing ignition operating on my test bench. In essence what this deal does it controls a digital pot (AD 5220 50k) based on the position of the BCDs. For example when SW2 is at 5, the digital pot moves to 10k. The Wiper and one leg of the digital pot go to the ignition system. I am measuring the resistance between the wiper and leg with a Fluke Meter.

    When one of the pins on Conn3 get 12v, it grounds out a pin on the pic and the code instructs the digital pot to step to 0 ohm.

    So I have a 12v battery on my test bench with a MSD ignition, coil, spark plug off the coil and a motor turning a crank trigger wheel. I hold the trigger pickup to the crank trigger wheel and the signal triggers the ignition to fire the plug.

    My digital pot device is using the same power and ground, just like in a race car. The housing is grounded that holds the board.

    When I watch the ohms between the wiper and leg while the ignition is firing the ohms go from 6k to 60k then back to 0ohm and then climbs to up again.

    This leads me to think the PIC itself is fine but the two inputs to the digital pot (CLK and U/D) are having some problems. I also have a couple LEDS that go high when an input from CONN3 is highand they do not do anything stupid. I do have the U/D pin pulled high with a 10k resistor and the pot is always incrementing high even after starting back at zero.

    What can I do to help the CLK pin? I am not too sure if a pull-down or cap or both would be the best. Or maybe something else.

    I am attaching the schematic and board drawing.

    Thanks in advance.
    Last edited by a moderator: May 5, 2009
  2. bertus


    Apr 5, 2008

    Do you have decoupling capacitors near each chip?
    These should be about 0.1 μF each.
    It can happens that circuits start to behave strange when they are not present.

  3. nanovate

    Distinguished Member

    May 7, 2007
    You can put a small RC filter on the CLK input.

    Also why is the PIC's pin 8 (VSS) not connected? All VDD, VSS, AVDD, GND etc pins should be connected.

    How is the PIC GPIO pin configured for the CLK pin?
  4. Tobias

    Thread Starter Active Member

    May 19, 2008
    The CLK pin is an output that I use the PULSOUT command.
  5. DonQ

    Active Member

    May 6, 2009
    It may help to realize that when you tie into the spark coil, you are hooking to an incredibly noisy signal. Yes, even the "low voltage" side of the coil. When the points (or the silicon equivalent) "break", the inductive kick of the coil will put a spike on the low voltage side, maybe even hundreds of volts. And it will ring. And it will transmit throughout your board. And once it is there, it will make any circuit behave in a way that defies explanation!

    Best to isolate this signal from your board, either with off-board filtering, opto-isolation, or some combination.

    Also, automotive power is notoriously noisy. This is a case where a big capacitor is not enough. Parallel any big capacitor with a small one also. Big capacitors have internal parasitic inductance, so high frequency noise is not filtered well. The small cap will get this. You can also include an inductor in series with the incoming power. Again, big inductors include parasitic capacitance, so bigger is not necessarily better.

    A ground plane is rule #1, but is only for getting rid of small amounts of remaining noise after the steps you take to stop noise from even getting as far as your board. Its also a good idea to fill in unused areas on the trace side of your board with grounded polygons as well.

    There is no way to have too little noise on your board. It saves you the possibility of having just a little bit too much noise. Even a little bit too much noise, in the power or the signal, will burn up a lot of late nights trying to find the problem.

    This is the sort of problem that needs to be "designed" out instead of "fixed" later, because "fixing" it later hardly ever really fixes it.