Designing my amps around TPA3118 since the beginning I am always curious to discover these in commercial applications. This one was inside the Bose Soundlink. It can be seen that the bean-counters replaced output inductors by ferrite beads here. Considering the short speaker cables and the metal housing this might comply with EMC regulations anyway.
Attachments
I just found this thread when searching "TPA3118 Ferrite Bead". I stumbled across the Bose board a few weeks ago watching a repair video. Spotted the 3118 and was surprised to see that in there. I was however shocked to see no inductors and wondered how they did it. Then I now saw in the datasheet the allowance for ferrite beads.
We've owned the Soundlink Revolve (stolen from my son's old E300D unfortunately) and it sounds pretty good to my ears. Surprisingly good for the tiny package it comes in and the price.
The only thing I could find on Ferrite Beads was this reference:
https://e2e.ti.com/support/audio-group/audio/f/audio-forum/243928/tpa3118d2evm---lc-filter
I'd like to compare the beads vs good inductors. I know little to nothing about the sound quality of either. Time to do some reading.
In general ferrite beads are cheapest solution and only work fine with very short cabling like in boomboxes. They can not be compared to output filtering networks meant for EMI free HiFi. The loudspeaker cables may be perfect antennas and this both ways.
You can put a lot of time in ferrite beads but having good output filters is more worthwhile usually. That is why some projects used overdimensioned coils, film caps etc. Not cheap but good.
Some manufacturers have papers with detailed explanation and usually the "lowest cost" section is spent on beads.
You can put a lot of time in ferrite beads but having good output filters is more worthwhile usually. That is why some projects used overdimensioned coils, film caps etc. Not cheap but good.
Some manufacturers have papers with detailed explanation and usually the "lowest cost" section is spent on beads.
Last edited:
Smaller inductors are coming closer to core saturation and add distortion.
In fact the inductors on al cheapo boards are the main contributors to THD.
A rule of thump is the bigger the inductor, the better.
In fact the inductors on al cheapo boards are the main contributors to THD.
A rule of thump is the bigger the inductor, the better.
Interesting and I guess that makes sense then for the Bose application (cable length is maybe 2" in that speaker). In my case, it may also make sense since I will likely use these boards to make powered speakers.
I've read (not in that TI thread but another) that if power gets too high, ferrite beads will not work. The TI commentator I linked to mentioned 8A rated beads. So unless my calculations are off, that means max output allowed (picking a 4 ohm load) would be , Pmax=I^2*R so:
Pmax=(8A max through inductor)^2*(4ohm load) = 256 watts
Which is way more than will be needed for this chip. Of course I may well be overlooking something.
I agree with the idea on picking good film caps. I just replaced a bunch of 47uF electrolytics with transformers and film caps on the signal path of a mic preamp and the sound is much more pleasing. I've only done a small amount of comparison on inductors on outputs of a Class D amp. I'd bet it's much the same thing.
Very good to know. I noticed the cheapo boards with 100 marked (10uH) inductors do not sound as good (I was using a 4 ohm speaker (guitar speaker) for testing a powered speaker I built in 2020). Even on the guitar speaker (lofi by all standards), the board with 330 marked (33uH) sounded better. I'm guessing in part because they were bigger sized inductors). I need to buy some real boards where through hole swapping is easier now that I know I like the 3118. It's not the best amp I've ever heard, but it's very good especially for the price. And I'd say I've still not heard it at its best (good board and components).
EDIT: I now know of the amp board called weiner. Is there another good board option? Anything with through hole and less SMD (beside the chip of course which can only be SMD). I can solder SMD parts by hand, but I don't enjoy it much with my skill level and patience. I actually prefer soldering chips to SMD resistors and caps since bigger chips are easier to hang onto.
The only (non stock to the PCB) inductors I tested were 33uH and looked like this:
No idea if that's appropriate for Class D output filtering. It worked and didn't melt... ; ). The one I used was rated at 6A. Not sure how good or bad they sounded other than that I wasn't sold on them being an improvement.
Last edited: