Inductive choke for PWM to average DC

Discussion in 'The Projects Forum' started by urb-nurd, Sep 12, 2015.

  1. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    I am trying to filter out my PWM frequency to produce a variable DC supply to my load.
    R is a halogen bulb, at rated power and voltage R is 3.84 ohms.
    Pwm freq = 25Khz
    upload_2015-9-12_15-14-10.png

    I stuck in a coil and fly-back diode and after fiddling with the inductor value just a few times i seem to have happened upon a suitable solution.
    Though in my experience, things don't just happen - hence my post here for validation.

    With R = 3.84 and L = 2 mH the cut-off frequency is around 300hz and i can vary the duty in my simulation to get a varied average dc output.

    upload_2015-9-12_15-18-38.png
    THe above image shows the load current with the lowest permissible PWM duty and the largest ripple current.

    The following image shows the duty around its max with the ripple noticeably reduced.
    upload_2015-9-12_15-20-21.png


    So i am just looking to clarify that a simple coil in series with my load used with a fly-back diode is all i need to feed variable DC to my bulb from a PWMed DC supply.

    I have looked briefly online for a part that meets my required specifications (assuming they are correct) and seem to have found some viable options, though all the coils seem rather large.

    Any guidance is appreciated.

    I am trying to create a small halogen dimmer essentially, and attempting to reduce the noise emissions by filtering and shielding.

    Thanks AAC!
     
  2. GopherT

    AAC Fanatic!

    Nov 23, 2012
    5,991
    3,736
    Yes, a choke will help. As would a capacitor. The Resistance is your load so low-pass RC, RL or RCL combination will filter your supply. It all depends on board space, budget and performance you need. Generally, high value capacitors are cheaper than high value chokes (especially chokes that do not saturate at high current). Your 25kHz frequency helps vs a 60Hz mains system (helps in terms of inductor or capacitor values and physical size, keeping them smaller and cheaper).
     
  3. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    cheers for that!

    I have really been needing to read up properly on filter design rather than darting in and out like i tend to do.

    I will look into RCL filters too. I am aiming to make the filter as physically small as possible - are RCL filters generally smaller than RL filters?
     
  4. crutschow

    Expert

    Mar 14, 2008
    12,990
    3,226
    That circuit looks basically OK.
    You want to use a Schottky diode for best efficiency.

    But why you talking about 300Hz when you said the PWM frequency is 25kHz?

    The higher the PWM frequency, the smaller the inductor you can use for a given current and ripple.
    The trade-off is possibly higher EMI and higher switching losses in the MOSFET.

    In real life you will need a high-current driver for the MOSFET gate, of course, and a faster rise-time than 100ns for the drive signal..

    Edit:
    A capacitor could be used to reduce the ripple current, but that may not be needed with a halogen bulb load since the thermal mass of the bulb prevents any significant change in the light output at such a high frequency (and your eye couldn't see it anyway).
    The thermal lag, in effect, acts like a large capacitor.
     
    Last edited: Sep 12, 2015
  5. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Schottky for efficiency : Noted.

    I got the frequency figure from my fiddling with the simulation. I found that at around 2mH my simulated results were as above, and from the fixed value of R and the chosen value of L calculated the cut-off frequency to be around 300hz.

    As i approached it from simulation, i ended up with a value that APPEARS to work but wasn't calculated using the PWM freq.

    If i were to design using the pwm frequency, where would you suggest placing the corner frequency?

    I have configured my gate drive already using a high current push-pull gate drive ic (FOD3180) and heatsinked my mosfet.
     
  6. ericgibbs

    Senior Member

    Jan 29, 2010
    2,499
    380
  7. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Already looked well into that, thanks eric.

    I am using dimmable halogen bulbs which are designed for lower operating voltages (actually temperatures).
    I also will be insulating my bulb in a metallic chamber which will help keep the envelope temperature up while operating at lower voltages.
     
  8. GopherT

    AAC Fanatic!

    Nov 23, 2012
    5,991
    3,736
    I agree but he said something about "Reducing noise emissions through filtering and shielding" in the last line of the first post.

    Urb, what is the problem with 25kHz noise? What problems is it causing? I can't imagine it travels far in the air and if it is moving back up power lines and causing issues, id like to hear about that.
     
  9. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    Yeah the filtering is for EMI/RFI considerations rather than bulb life, as you said - the thermal time constant of the filament is larger then the PWM period i believe.
    Its more about switching load currents of 6A+ at this frequency i believe, i already have issues with my monitor messing up my tv reception when i open a window on my desktop (i have videos, strange things are happening to me...).
    Plus i am wanting to look into filtering and employ it so that if i ever want to try sell what i am creating, i could try ensure ce certification.

    Oh, and the EMI from the load currents works to foul up the accuracy of my thermocouple circuit.

    Thats long in the future, but its something i would like to learn about and design for now.
     
  10. crutschow

    Expert

    Mar 14, 2008
    12,990
    3,226
    I'm not familiar with the term "corner frequency" with regard to PWM circuits.
    How did you measure that and how is it related to the design of the circuit?
     
  11. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    I mean the -3db or half-power point of the filter.

    If my pwm frequency is 25Khz, what cut-off frequency would i choose to filter out the high frequency PWM with minimal ripple current?
     
  12. crutschow

    Expert

    Mar 14, 2008
    12,990
    3,226
    That's a matter of defining (or deciding) what value of ripple is "minimal". :rolleyes:
     
  13. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    haha, i have no idea to be fair.

    I am kind of working blind with no experience.

    Someone else is probably better authorised to tell me what is best, rather than i suggest.
     
  14. GopherT

    AAC Fanatic!

    Nov 23, 2012
    5,991
    3,736
    You said, essentially, that the reason you are doing this project is to reduce the EMI. You must have an idea of how much reduction you want or need. Wether it is a numerical value of the device or a numerical value of the interference it is causing or even a qualitative result if the interference it is causing. But, if this is all based on the amount of qualitative interference that it 'might' cause, then I cannot help you. There is no way to make a design with no design criteria and no opportunity to measure success. At this point you seem to be entering the psudo-audiophile zone. Find some oxygen-free copper cables and start from there.
     
  15. urb-nurd

    Thread Starter Member

    Jul 9, 2014
    269
    3
    i see, but for that suggestion i was referring to the cut-off frequency for a single pole low pass filter.
    Though i do not know - as you stated - what it is i am aiming for as i have never been there before.
    I am basically asking for other members suggestions for where to place the cut-off frequency based on their experience, knowing that i am trying to reduce EMI but have no definite target to steer towards.
     
Loading...