Pretty Sure a Noise Problem

Discussion in 'General Electronics Chat' started by Tobias, Dec 9, 2009.

  1. Tobias

    Thread Starter Active Member

    May 19, 2008
    158
    0
    I made this deal that will allow me to monitor ignition timing on an engine. The device has two inputs, Tach and TDC. The Tach signal is a square wave that goes hi when the coil fires for each spark plug. The TDC signal from a variable reluctance sensor. The TDC signal goes to a component that turns the sine wave into a square wave.

    The tach turns on the output of a flipflop and then the TDC turns off the output of the flipflop. So if the ignition timing is 45 degrees, the output will be high 50% and low 50%. I then take this PWM with varying duty cycle through a RC circuit to turn this square wave into a 0-5v signal. 45 degrees is not 2.5 volts.

    It works pretty good on the test bench however I am picking up noise along the circuit on the race car. I have access to a dyno next week where I can really test the hell out of this thing. Any suggestions based on the attached schematic to help eliminate the noise? I am going to bring my scope and solder wire to pins along the circuit to figure out the problem area.
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    A typical automotive environment is a brutal one; temperature-wise, shock/vibration-wise, and electrically.

    Race cars jack that up by an order of magnitude, at least.

    For starters, you will need to use 0.1uF bypass capacitors between Vdd and Vss on each and every IC in your design.

    You will also need to supply regulated voltage to your design, and I strongly recommend the use of a pi-type filter for the input power.

    Your R4/C2 low-pass filter will have an extremely slow response time. It will require 30 seconds of input to even approach the final value.
     
  3. mkbutan

    Senior Member

    Sep 30, 2008
    270
    16
    is there a dynamo near your ckt. for charging the car battery
    or simply put the magnetic rings in your input/output wire's
     
  4. mkbutan

    Senior Member

    Sep 30, 2008
    270
    16
    For starters, you will need to use 0.1uF bypass capacitors between Vdd and Vss on each and every IC in your design.

    yes you are right


    I strongly recommend the use of a pi-type filter for the input power.
    what is pi filter pl.
     
  5. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Here is just one way to improve the response of the filter, while keeping ripple low and providing a relatively low impedance output. See the attached.

    I chose the values to be not terribly dissimilar to what you're already using.

    This filter settles to within 1.5% of the desired output within 1 second, and is within 0.2% by 1.5 seconds. The opamp on the output is connected as a voltage follower; it presents a relatively high impedance to the filter, yet has a relatively low output impedance.

    This is not an ideal design, and the LM324 is certainly not the ideal opamp to use; it is slow, not rated for automotive temperature ranges, and has a relatively high offset voltage compared to more modern designs.

    However, it will sense input voltages down to ground, the output will go nearly to ground, it is a single-suppy quad opamp, is widely available and very cheap.

    If you do use an LM324, all unused inputs should be tied to ground.
     
  6. Tobias

    Thread Starter Active Member

    May 19, 2008
    158
    0
    This board is in an aluminum enclosure. Is it better to connect VSS to the enclosure or have the board float in the enclosure?
     
  7. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    A pi filter is a pair of capacitors to ground connected by an inductor.

    The inductor tends to keep the current flow constant.
    The capacitors tend to keep the voltage level constant.
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Grounding the enclosure will help. However, you have a high-impedance signal leaving the box, and the wiring will pick up lots of noise from various sources, particularly the secondary ignition wiring.

    Reducing the output impedance of the signal by using an opamp buffer like I illustrated will help a great deal.

    Using a coaxial cable will help a great deal more in reducing noise on your output signal.

    [eta]
    If you really want to get rid of the noise, digitize the analog signal and send it out logic level.
     
    Last edited: Dec 9, 2009
  9. mkbutan

    Senior Member

    Sep 30, 2008
    270
    16


    thanks for the Ans. :)
     
  10. Tobias

    Thread Starter Active Member

    May 19, 2008
    158
    0
    Thanks for your help. Turning the signal into digital is a no go with the data acq systems in drag racing. There are digital channels, but working with the suppliers can be a pain in the ass. Most teams have spare analogue channels.

    From the testing I have done on the car so far, the output is fine until the engine gets to higher rpms, then it goes to 0v. That is leading me to think I am getting noise on the TDC input between the VRS converter and the flip flop. Could noise on the output make it read 0v? I will definitely put the op-amp in the circuit.

    thanks again

     
  11. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Sorry, but without more complete info, I have to make guesses to fill in the gaps.

    Noise on inputs or outputs will cause big problems.

    I really can't be more specific, as you are working within a noise-rich environment. My suggestions will likely help a great deal, but can't cure all ills.
     
Loading...