How to reduce EMI from PWM control of high power load?

Discussion in 'The Projects Forum' started by urb-nurd, Aug 25, 2015.

  1. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Hey guys
    Still my age old project, but circumstances require i focus on it again.
    You guys were invaluable while i was finishing my degree and really did save me during numerous breakdowns.
    Thanks for that AAC! - i owe my degree classification to the sanity instilled here.
    Genuinely - Thank those that helped, the common culprits and those less so.


    Now down to business - i am using PWM to control a 100W load at 24V, so the RMS load current would be around 4.2A.
    I am aiming to produce a design that will minimise EMI emission to make CE certification more achievable in the future (beyond me at the moment).


    Currently i am PWM'ing 24V AC with no filter stage, and designed for AC to this point as i know AC can be filtered to recover the sinusoid and effect variable amplitude sine-wave dimming. Thus converting my PWM output into a nice smooth AC with little EMI to consider.

    Although it seems that it may not be so simple to filter at these power levels (or so i am lead to believe), please tell me i am wrong - i hope so haha.
    My other option is to use DC, but again i still have high pulsated load currents producing unacceptable EMI, and .... unless i'm missing something - i cant think of a way to filter DC.

    So before i can progress, i need to determine how i can reduce the EMI output of a 100W+ load controlled with PWM.
    • I could use 240V to reduce the load currents if this would ease filter design, though it imposes other issues with my load choice.
    • A user of another forum suggested DC however i cannot see how this can be filtered to negate the load current pulsations.
    • Another option is to investigate other control methods such as phase-angle control and cutting conduction on the trailing edge, though i am hoping this isn't the go to answer of the best educated here as i know the least about this option - but if its superior then i will embrace it.
    The end intention is to create something that can then be produced and sold legally in Europe abiding by CE certification.
    If i was in the US i could get away with a-lot more, but being where i am - i believe i need to over-engineer things. Which is fine by me.

    As always AAC - any help is greatly appreciated.
     
  2. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,795
    951
    Have you tried shielding methods such as metal enclosures and outer conductor shielded cabling? Unless your PWM frequency is in the radio band you should have a very low emitted power level from voltage transition\switching transients to contend with.
     
  3. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Your post addresses two very important points i failed to mention - I haven't considered shielding at the moment as i was attempting to design for minimal EMI, then use shielding if required.
    The justification for this is the nature of product i am developing cant necessarily be shielded as easy as other things, there are thermal considerations - though if i required internal shielding (or external for the prospectively protruding load) i could use a mesh and reduce this issue massively - simple faraday cage?
    I will incorporate shielding just to be safe (especially where it isn't an aesthetic eyesore), but as i actually cant quantify the EMI emission from my experience - a-lot of this is conjecture.

    The second point is i actually have NO experience or perception on the nature and magnitude of my EMI emissions.
    I am basing my current view on the fact that i had a circuit that was phase-angle control of a 25W load at line voltage (leading edge conduction), which failed CE certification due to the EMI emissions. This was not my circuit through i was aware of it and involved in discussion.
    It is from this that i got my perception that 100Hz switching at 25W with 120-240 volts produces enough current pulsations to fail the CE cert.

    So i figured that 25Khz switching at 100W+ with a load rated for 24V would have the same issue to a much greater degree.
    Though this is based on experience and rudimentary logic, i may be missing a-whole-lot of the picture.
     
  4. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,795
    951
    The experience with phase angle control may be quite different from your current project. It sounds like one used AC current directly and this one uses rectified power from a filtered supply source.
    A simple AM radio tuned to a fairly empty radio band, when held very near your circuit should reveal any possible problem areas.
    Another good detector for AM and FM emission is a TV tuner on an empty channel. I mean the old picture tube sets, not flat screen stuff.
     
  5. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,293
    1,262
    Schematic?
     
  6. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    I am using two MOSFETs to control an AC load and could show a schematic though the extra drive circuitry would detract from the questions at hand.

    I have a toroidal 25V transformer for my AC supply with two MOSFETs to control the load for each respective sinewave excursion.

    I am not rectifying for pulsed DC, though i did do this prior and also understand the EMI/RFI considerations of a toriodal transformer of course.

    As far as the circuit i have already, it is working well - however for further development i need to ensure i have made the correct choices with regards to:
    • Rated load voltage - high voltage to reduce EMI? Though lower voltage heaters are more mechanically resilient thus better for PWM
    • AC/DC supply - Either way, PWM induces high load current pulses and noise emission
    • Switching device choice (currently MOSFETs) - could be improved with IGBT or other, could also change is PWM control is scrapped.
    • Noise emission - AC permits filtering to recover PWM'ed sinusoid but i am unsure, shielding will minimize also.

      Thanks for the help folks - if required i can post a Schematic but it would be no more than a bulb connected to two MOSFETS and an AC supply with gate drive circuitry.
     
  7. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,293
    1,262
    Buy a big bottle of pepto, cause you are probably in for some heartburn.:D
    I'd put it into a metal box. Then the next step is to keep the noise inside the box.
    To do that I think you could make the output more like a buck regulator with a big inductor and a filter cap to keep the high frequency from getting on the output.
    Then treat the input like a switching power supply with the input filter.
     
    RamaD likes this.
  8. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    By a buck regulator, do you mean to use the PWM with active and reactive components to produce a variable DC (AC?) output?

    Similar to the 1,2 and 4 quadrant motor controllers that use PWM to produce a variable pseudo-sine-wave?

    Another thing mentioned to me was to try use DC and then filter this PWM'ed dc to produce a variable DC output, using an RC filter with R as my load and C to match (this is where it may go awry).

    Don't worry @ronv , i have lived this life long enough to have my PPIs on hand haha. Ant-acids are for babies.

    Thanks for the response.
     
  9. Kermit2

    AAC Fanatic!

    Feb 5, 2010
    3,795
    951
    You need to measure the radiated spectrum and determine what power and freq spread you have.
    Again, I cannot believe you are generating anything significant in the MHz range and nothing with enough power in the 100s of KHz range to interfere with those radio bands.
    Where exactly are you putting your little PWM light bulb anyway? International Space Station? NSA listening post?
    Just slap it in a cheap aluminum box and use shielded cable, or do the spread spectrum analysis and compare to the Certification Agency regulations to see where you stand if that is really an actual concern.
     
  10. crutschow

    Expert

    Mar 14, 2008
    13,056
    3,245
    Often such a circuit is filtered with both a common-mode choke plus inductor/capacitor filters at both the input and output to meet radiated and conducted EMI requirements.
    How elaborate a filter you will need can likely only be determined by trial and error.
    Goggle EMI filter for some ideas.
     
  11. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Thanks for all the info folks, i am not being overly anal with this.
    Rather trying to design for CE certification which is very much more strict than the rules in the US.

    I have experience with a design that failed CE cert and it was of much lower frequency and power than this device. So i am simply being aware that it WILL fail CE cert if not designed appropriately - i wish i could just throw it in a metal box and call it done.

    I have done some reading on EMI filters however i didn't really get anywhere as nothing i could find dealt with the power i am trying to - rather most were for audio purposes.
     
  12. crutschow

    Expert

    Mar 14, 2008
    13,056
    3,245
    Here's info on EMI filters for power.
     
    urb-nurd likes this.
  13. Ancel UnfetteredOne

    Member

    Jul 3, 2015
    36
    14
     
    urb-nurd likes this.
  14. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Just been through the posts above.

    That homemade probe looks good, though i would need to get a spectrum analyser i believe - ill look into the prices.

    @crutschow I checked that PDF, it seems like a high current feed through filter could be used to change PWM'ed DC into variable DC.
    Screen Shot 2015-09-05 at 16.47.16.png

    I have just glanced, but it looks like the attenuation is ok with 1.68Uf
    Screen Shot 2015-09-05 at 16.58.28.png

    Now in understand this still wont work for my set-up, but it seems as though i can use one of these off the shelf feed through filter to reduce the EMI from high frequency PWM switching of high load current.

    I will be switching at high frequencies (25Khz + to try reduce component size of the filters) and may even go beyond that if it helps.

    I understand the inductor creates reactive power and slows the current rise - almost buffering and smoothing the PWM - and i guess the capacitors are for high frequency coupling to ground.
    Screen Shot 2015-09-05 at 16.56.51.png

    Though if i am wrong with any of this, feel free to slap.

    I understand that i will still need filters on the line to prevent line distortion too, but is this a start with changing my AC load to a filtered PWMed DC for reduced EMI?

    I hope so, if not. i'll drink.

    Once i get my load specs nailed down, i can realy turn up the speed to get the power supply and embedded system in place.
    Its exiting!
     
Loading...