Noise and Emissions

Discussion in 'General Electronics Chat' started by mlkcampion, Dec 16, 2006.

  1. mlkcampion

    Thread Starter Active Member

    Nov 16, 2006
    59
    0
    Hey forum
    I have a low-flow turbine water flow meter, which consists of a Hall Senor,
    amplifer and NPN pulsed output. The meter basically give you pulses in proportion to the flow rate. Its a three wire device, +24Vdc, 0Vdc and Pulse Output. The meter is installed in a cabinet with other electronics equipment and works fine in most of the products. However in one particular product the meter was giving more pulse then it should have been.
    So began the search as to why. Firstly i installed a new meter and even though it wasn't affected as badly its readings did fluctuate more then i can accept. I went back to the first meter and decided to use this to find the source of the problem. I installed a 470uf capacitor from the meters +V and GND assuming it was some form of conducted emmissions and the cap did the job. So taking that it was conducted noise i removed the capcitor and started shutting of the other electronic devices in the unit and could'nt find any culprit. I then powered the meter from an external benchtop power supply and the meter worked great. At this stage iam thinking its the 2 power supplies that are installed in the product, This is where this product differs from others in that it contains 2 power supplies.
    What i dont know is if the power supplies (each Containing AC-DC brick and DC-DC converter) are generating the noise or is it just passing from the mains through the power supplies. The final thing i did was tried a few other capacitors ranging from 1000uf to 4.7nf to see if i could find one that did'nt work thus giving me an idea of the frequency of the noise (not sure if my logic here is correct???). All the capcitors worked!!!!!

    Another point of interest is that if i connected my scope leads to the pulse output the meter works great (I know the leeds probably have some capcitance associated with them) , and for the strangest thing after all my messing with capacitors it turns out that all i have to do is touch one leg of a capacitor (any value) to the meters +V and leave the other hanging in mid-air and guess what the meter woks!????? how is this possible???

    Also i discovered that re-routing the wire a different way around the cabinet away from the outputs of the power supplies greatly improves the reading!!

    My final conclusion is that it is radiated emissions inside our product????

    Can someone please enlighten me as to what this is, or give me an idea as to how i should really find the source of this problem??? iam not the kinda person that will just put the capacitor permanently on the board without really understanding why, i need to know!!!! :)
    Thanks
     
  2. JoeJester

    AAC Fanatic!

    Apr 26, 2005
    3,270
    1,064
    You have an EMI problem and the best method is to eliminate the source.

    Use an oscilloscope in the AC position to monitor the +24VDC so you can identify a periodic signal above the millivolt level.

    Calculate the frequency of the EMI, as viewed on the oscilloscope, and replace the module based on your calculation.

    The AC brick would be a lower frequency [line or a multiple thereof]. The DC-DC converter will be a higher frequency ... well above 30 kHz.

    It should be obvious when you replace the defective one, since the oscilloscope will reduce whatever it shows as the interference.

    If the one you chose first didn't repair the problem, try the other one.
     
  3. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    Hi,

    Check your power supply ground lead. Either it has a poor connection, or it may need to be jumped to AC ground to get rid of the noise. It sounds like connecting the scope lead made a better path to ground and eliminated the noise while it was connected.
     
  4. JoeJester

    AAC Fanatic!

    Apr 26, 2005
    3,270
    1,064
    Beenthere is correct. Check the integrity of your connections first.
     
  5. mrmeval

    Distinguished Member

    Jun 30, 2006
    833
    2
    Can you bench test the supplies and verify they are within spec? Checking and cleaning the connections can't hurt. Use an isolation transformer with the scope if you use it to test the live circuit or in fact anything that's connected to AC.
     
  6. mlkcampion

    Thread Starter Active Member

    Nov 16, 2006
    59
    0
    Hey
    I've check the 24Vdc with a 10Mhz oscilloscope, and i did find a
    perodic waveform @ 333.33kHz and 200mV pk-pk. When i saw this i had
    a look at the datasheet/evaluation sheet for the DC-DC converter and for the
    output ripple and noise waveform iam getting pretty much the same periodic waveform, unfortunately i can't make out the pk-pk or the frequency from the evaluation sheet, the grid lines are blacked out. I think what iam getting
    on the scope is a larger pk-pk but really hard to tell.(Bearing in mind that the manufactures tests are done at 100% load, i'll test soon with a larger load can't at the moment)

    Would it be true to say that even if i go replacing the power supply, iam going to effectively get the same thing either way?

    Does anyone have any idea why it is effecting the Hall sensor so badly?

    I will worked on the other suggestions to try and get rid of the noise, anymore suggestion really appreciated.

    Thanks
    mICHAEL
     
  7. mlkcampion

    Thread Starter Active Member

    Nov 16, 2006
    59
    0
    I've tried grounding the power supply to earth, however it doesn't improve the situation, seems to make a tiny bit worse!! Note also that when i connect the scope leads they dont have to be connected to the oscilloscope to improve the meter reading, in fact if i just leave the scope lead GND clamp connected to the gnd of the power supply the meter reads good!!!
     
  8. mlkcampion

    Thread Starter Active Member

    Nov 16, 2006
    59
    0
    oK
    Iam now at the stage where i've just rewired the meter. Originally the power (+24vdc and gnd) comes of the PCB on to connectors on din rail, from here it is sent to 2 more connectors, the meter is then wired to these connectors. Now i've bypassed all these connectors and went straight to the PCB and the meter works perfect. I still have the output ripple and noise on the +24Vdc going to the meter but it looks more attenuated and is obviously having less of an effect. Even if there was a bad connection on the din rail connectors (they all actually look fined) how would it have translated into higher number of pulses?

    Thanks
     
Loading...