BTL output filter

[JohnW]

Hi Ted,

Using electrolytics as mention already is an extremely bad idea - I guess your using them as your under some cost constraints - but if costs is so critical then I would suggested (very reluctantly) to use ceramic caps - But Pls. don't tell anybody that I said that.

Unless the electrolytics have a low ESR at the switching frequency, internal heating will evaporate the electrolyte over a short time, and these caps will fail very quickly – seen it in many a cheap SMPS’s from China.

John

[JohnW]

Hi Charles,

Do you have any experience / recommended values for Zoble components? I was thinking of adding an external switchable Zoble for speakers which have a steeply rising inductance at HF.

I don't have the issue as I use ESL's (the opposite in fact), however most budget box speakers have a rising impedance at HF - good for Class A/ AB amps, but not good for Class D.

John

[phase_accurate]

Personally I would use values that keep the amp stable in a no-load situation. I.e. I would regard flatness of frequency response secondary to safety.
One must never forget that an LC lowpass without load is seen as an LC series tank circuit (i.e. almost a short at resonance). So the snubber should be dimensioned for minimizing this effect IMO.

If the snubber is just there to achive response flatness (with load connected) however then I would dimension it like the ones used for driver impedance linearisation. If you try to compensate for Rdc = 6 =hms and Lvc = 100 uH (as average values) this would give a capacitor of 2.7 uF approx which isn't what I would call small.

Regards

Charles

[ted12]

John
I just left the lab today but I will use ---MKT 270Vac cap--- for C2,C3 in the coming Sunday.
Ted

[JohnW]

Ted,

The Caps. will help – but I suspect that the load is very inductive – so more important to find a non inductive load.

Have a good weekend,

John

[ted12]

Finally, I use the 4" speaker as the load.
The peak to peak voltage of output filter is corrected.
Audio input (sin) adjusts from few hurdard freq to 20kHz, the output (maX peak volatge)only slightly decreases about 3 or 4V.

[JohnW]

Ted,

Good to hear - sounds like your load was too inductive,

Good luck with your design,

John