Picking Output Filter Parts

[BJosephs]

I’m going to take a swing at a class D amp board and I’m wondering what are the current recommendations for low distortion output filter parts (inductors in particular). I’m not going for huge power levels and I’ll be using either TPA3244 or 3250 using a 10uH and 1uF lowpass filter in BTL mode. I’ve heard air core are the lowest distortion but it seems the class-d specific inductors from Wurth and Coilcraft are all ferrite. I’ve also seen some dual inductors and suggestions to prevent coupling between filter parts. So what’s the state of the art approach here? Is there a benefit to oversizing the inductors to prevent saturation?

Brian

 

[JonSnell Electronic]

Inductors do not induce 'distortion'.

The inductance and current handling are critical.

Use the correct value and all is well.

 

[MorbidFractal]

Inductors do not induce 'distortion'.

The inductance and current handling are critical.

Use the correct value and all is well.

Inductors do induce distortion. Their value is non-linear (current) amplitude dependent. Circuit gain depends on their value. You have a multiplier. You have a source of distortion.


... runs away.

 

[BJosephs]

Inductors do induce distortion. Their value is non-linear (current) amplitude dependent. Circuit gain depends on their value. You have a multiplier. You have a source of distortion.


... runs away.

This is my understanding as well. Several data sheets cite the output filter as a significant source of THD, even including THD graphs for different brands and part numbers. And TI has an app note on post-filter feedback that is justified by reducing inductor distortion by 6dband allowing cheaper inductors to be used. It’s worth discussing so I’m here asking for recommendations.

 

[JonSnell Electronic]

Inductors do induce distortion. Their value is non-linear (current) amplitude dependent. Circuit gain depends on their value. You have a multiplier. You have a source of distortion.


... runs away.

That is why you must calculate the values before playing with different inductors. It is a science.

 

[chermann]

hi


for me is that what developer like doctormord- doc- is doing . it is a kind of "rule" for me: he use ferrite covered coils...he did some measurements on coils an their thd...etc:


chris

 

[doctormord]

That is why you must calculate the values before playing with different inductors. It is a science.

As long as you keep them out of saturation, all is fine. If not, they contribute to distortion. Aircoils don't saturate but radiate alot and are bulky.

Btw. The table shown by chermann is based on excessive testing at TI for they latest EVMs and to give hints, what inductors are preferable.

 

[BJosephs]

hi


for me is that what developer like doctormord- doc- is doing . it is a kind of "rule" for me: he use ferrite covered coils...he did some measurements on coils an their thd...etc:


chris

Thanks. This is the kind of insight I was hoping for. Looks like CC MA5172 or Wurth 7443631000 are the best performers in 10uH.

 

[chermann]

please ask doc/voltwide and others... i am just learning and follow their experience.
chris

 

[nigelwright7557]

I haven't had any problems with none air core inductors.
I use T106-2 core with 12 turns of enamelled wire on them.

 

[doctormord]

T106-2 is always a good choice, Coilcraft MA5173-AE are from the same core material.

 

[peroz]

I am trying to find some inductors as well, and I asked in another thread, but the anwer I got was, that the inductor should have been tested to a frequency, higher than the switching freq. of the amp. The amp is 3116 mono.
Can't find one like that...


Would any of these be okay for a 22uH one:
Obročaste dušilke | Elektronske komponente. Distributer in spletna trgovina – Transfer Multisort Elektronik

 

[chermann]

Hi

my idiea:

TPA3116 mono => PBTL = datasheet write 10µH

i use at my dual mono chip amp the bourns 10µH its small size SMD
Bourns SRP1250-100M
cost 1,8 euro at conrad and the datasheet give me a good current availibility thats fit to the max current of the tpa3116 amp.

chris

 

[doctormord]

I am trying to find some inductors as well, and I asked in another thread, but the anwer I got was, that the inductor should have been tested to a frequency, higher than the switching freq. of the amp. The amp is 3116 mono.
Can't find one like that...


Would any of these be okay for a 22uH one:
Obročaste dušilke | Elektronske komponente. Distributer in spletna trgovina – Transfer Multisort Elektronik

If it has to be 22uH:
DTMSS-20/0.022/10-V FERYSTER - Inductor: wire DTMSS-20/0.022/10V | TME – Elektronske komponente

If you're limited to TME, then better go for Ferrocore inductors like these HCI:
Ferrocore HCI2114

with 15 or 22uH.

 

[peroz]

Thanks for your replies, guys.



I want the 22uH because of these guidelines:
https://www.diyaudio.com/forums/cla...no-amplifier-stereo-system-8.html#post5608515
for 8Ohm speakers.


I like TME because my company has quite good discount and we buy from them weekly.

I don't know how to work with SMD, that's why i never even checked the Ferrocore ones.


Is external filter a good idea or is it better to just change the components on the amp pcb?

 

[doctormord]

Just change at PCB, the Ferrocore inductors are usable with wires connected to as the conductors are wide enough for easy soldering.

 

[chermann]

Hi Doc


Can you explain me why you use at PBTL a 15µ-22µH instead of 10µ as the datasheet?


thx
chris

 

[doctormord]

10uH is out of stock, and it isn't important if BTL or PBTL, speaker impedance is what counts. If you still have 8R even at PBTL, 15-22uH is perfectly fine. In the end it all belongs to the desired filter frequency.

 

[nigelwright7557]

10uH is out of stock, and it isn't important if BTL or PBTL, speaker impedance is what counts. If you still have 8R even at PBTL, 15-22uH is perfectly fine. In the end it all belongs to the desired filter frequency.

I wind my own from t106-2 cores.
About a dozen turns gives 10uH