LC Filter - Capacitor type

Discussion in 'General Electronics Chat' started by Skeebopstop, Dec 15, 2009.

  1. Skeebopstop

    Thread Starter Active Member

    Jan 9, 2009
    358
    3
    I've recently got my Class D audio amp hobby project up and running and now I am going back over the design wondering if there aren't ways to improve the sound quality.

    One thing that came to mind is the output stage for Class D audio is a PWM signal so an LC filter is required before sending it out to the speakers. In my design, I've used an MLCC for the capacitor purely based on the low esr. Now I come back to ask myself whether or not such an MLCC in a low pass filter might add some phase distortion?

    Any comments? I just referenced another design and they used film caps.

    Regards,

    James
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    James,
    I'm drawing a blank - what are MLCCs? Multilayer ceramic capacitors? Yep, that's it. (Pardon the mentalpause...senility is to be enjoyed, I suppose...)

    Well, ceramics may not be the absolute best for audio, but we sure used a lot of 'em in RF filters - anywhere from 5MHz up to around 500MHz; then multilayer ceramics' L parasitic really starts kicking in big time, and you have to go to SLCC's to get away from that.

    Just for example, when testing at 10MHz, a 47pF cap might measure 50pF, but at 500MHz, it's up to around 150pF. The larger the initial size of the cap, the larger the difference.

    But you'd really need to post your filter schematic to get some decent feedback.

    LC filter design and implementation is supposed to be science, but there are elements (sic) of it that seem like magic sometimes. No matter how many calculations and iterations you go through, there is always some room for improvement.

    I'd need to know your filter specifications, along with what your PWM base frequency is to start making some trial runs on it.
     
  3. lmartinez

    Active Member

    Mar 8, 2009
    224
    6
    To broad of a subject to make stipulations! Post a online diagram.......:rolleyes:
     
  4. Skeebopstop

    Thread Starter Active Member

    Jan 9, 2009
    358
    3
    Well, it is a pretty simple LC filter (ignoring the speaker which has non-linear characteristics).

    18.7uH in series with 0.47uF in parallel -------L----------------- speaker +
    |
    _ _ _
    C
    _ _ _
    |
    |
    ------------ speaker-

    The PWM carrier is > 200kHz and I set the cutoff frequency of the LC ~ 30kHz (as per L and C).

    The primary thing which made me start to ask is I used these tiny little 1206 0.47uF MLCC (yes multi layer ceramic capacitor) whereas I look at an international rectifier application note and they have some chunky monkey film caps on there.

    For your reference, my amp is happily blowing my clothes off, however I am curious if I can't improve the sound quality a bit.

    Cheerio,

    James
     
  5. Skeebopstop

    Thread Starter Active Member

    Jan 9, 2009
    358
    3
    Nooooooooo my ascii circuit drawing didn't come out very well, attached is the schematic output stage (excuse the ugly schematic, when doing hobby stuff I tend to 'smash' schematics together).
     
  6. rjenkins

    AAC Fanatic!

    Nov 6, 2005
    1,015
    69
    Purely as an experiment, I'd try a rather higher cutoff, in the 50 - 100KHz range, and cascade three filter sections (with slightly different, decending frequency values to avoid resonance peaks?).

    That should overall give a harder cutoff, but with less effect on the higher audio frequencies.

    What's the modulator section based on? That's the other really critical point for distortion in a Class D design.

    Nice project!
     
  7. Skeebopstop

    Thread Starter Active Member

    Jan 9, 2009
    358
    3
    Yeah it's a fun one. Blowing my clothes off as we speak with.

    I think the 40dB of an LC is enough. I ran the simulations with varying loads and the resonance peak isn't really an issue and inductors rated to 8.6A @ > 18uH are pricey.. Besides, I have a pot on the front end which adjusts the self oscillation frequency (which tends to vary over a 100kHz range depending on the duty cycle and other parameters) and I knocked it up to keep the range over 600kHz or so and noticed almost no difference, so I think I'm alright.

    I'm still working on understanding the control loop. It is with the IRS2092 from international rectifier and drives a 2nd order delta sigma modulator with a transconductance amp on the front end (pretty heavy but I'm surprised at how great it sounds for class D) with unfiltered PWM feedback! Craziest little chip.

    I have mostly ready that ceramic capacitors in general seem to introduce alot of THD or phase distortion, so I was curious if the LC filter would have a similar phenomenon but now that I sit here thinking about it, if ceramic caps can cause phase distortion and increase THD quite heavily (read a paper on it) I don't see why similar effects wouldn't plague the back end of an LC filter considering there are heavy dependencies on frequency for impedance so the filter would be tuned differently for every frequency! I think I'll put some film caps there instead and just see if I notice anything.
     
Loading...