Highly confused about Texas Instrument TAS Class-D IC and LC output filter requirements for post-filter feedback

Hello everyone,

I put some effort into designing an output filter for a class-D IC. Here I am using Coil Craft's and Murata's SPICE models to calculate the filter with an 8 Ohm load.

Lines:
  • Red line is the capacitor impedance
  • Green line is the inductor impedance
  • Purple line is the filter response
Screenshot 2024-08-22 at 12.35.03.png


Filter:
  • -3dB cutoff is around 25kHz, just outside the audio band
  • The filter works until about 10MHz where the parasitics of the components have a resonance.
  • Sure the values can be tweaked, but this is not a bad starting point for now.
Post filter feedback

The IC that I am using TAS2780 has a post filter feedback (PFFB) feature and I thought that would be good to use to keep the THD as low as possible. So I started reading about the PFFB requirements.

1724327511815.png


It seems crazy to me to specify the filter cut-off frequency to be greater than 10MHz when this device is switching is 380kHz.

What could they mean?
  1. The rise/fall time of 1/10MHz = 100ns. So maybe if the rise/fall time of the IC is around 10ns or less, the filter could filter that out?
  2. The cut-off frequency they mean where the parasitics break the filter? This is around 10MHz for the L and C parts I am using.
They have an application report SLOA260 that was not that helpful, if focuses on using ferrites.

Questions
  1. Can anyone experienced with Class-D and filter let me know if a -3dB cut-off just outside the audio band is sensible?
  2. Has anyone used TAS ICs from Texas Instruments before, or the TAS2780 specifically? Can you suggest a sensible filter with or without the PFFB feature.
  3. What the hell are they talking about in the data sheet and application report when they specify the cut-off must be greater than 10MHz?
 
moving the filter into the feedback loop affect the stability of the amplifier.

My understanding is that the filter needs to have certain properties for the amplifier to be stable. What I don’t understand is what those properties are when using an inductor and a capacitor all of the documentation they publish is not very helpful in that respect.
 
Inserting the filter into the feedback loop gives an additional 2nd order low-pass inside the feedback loop that will degrade stability.
You could say as well that the feedback loop needs some properties to be stable with the inserted filter.
This is a common issue with feedback loops.
The implications are documented in basics of feedback loops.
 
Hello everyone,

I put some effort into designing an output filter for a class-D IC. Here I am using Coil Craft's and Murata's SPICE models to calculate the filter with an 8 Ohm load.

Lines:
  • Red line is the capacitor impedance
  • Green line is the inductor impedance
  • Purple line is the filter response
View attachment 1347671

Filter:
  • -3dB cutoff is around 25kHz, just outside the audio band
  • The filter works until about 10MHz where the parasitics of the components have a resonance.
  • Sure the values can be tweaked, but this is not a bad starting point for now.
The plot shows -3dB point at about 60kHz, not 25kHz.
 
Thanks for all the comments and useful docs. folks.

I think I have got to the bottom of it now.

It seems the post-filter feedback feature of this IC is designed for ferrite beads only. I was not expecting this.

The filter cut-off should be placed at around 10MHz to filter the harmonics of the switching frequency.
 
What is special about the TAS2780? I get the impression that it is not a great class D IC at all just by looking at the datasheet. It is not exactly competing with stellar numbers (PFFB or not) 🙂
 
Last edited: