The protection circuit is cycling it on and off.
No, it wouldn't be a filter at all, just a near short at higher frequencies, I strongly suggest you do not try this with a capacitor larger than about 1nF so that it's only filtering harmonic RF noise far above the switching frequency.
Without the capacitor however it would be a 1st order filter.
The tpa3116 is the same. The data sheet even points out that under normal circumstances a ferrite bead filter would be optimal. Only if the amplifier because of less than optimal circuit layout causes excessive EMI noise, very long speaker cables are needed, or highest possible efficiency is desired should a LC output filter be used.
Please note that a ferrite bead filter isn't an output filter in the traditional sense as it's basically just a wire at the switching frequency. It's basically just to get higher dampening of RF noise.
Just like the tpa3110, the tpa3116 can also be used completely without filter but in both cases speaker wires must be very short and the load should be inductive in nature. The main difference is the tpa3116 has 4 times higher output potential so it also can generate 4 times as much EMI.
Last edited:
Since the chip has it's protections, I think it's not fried. Should I connect FAULTZ and SDZ to (possibly) get it back to work?
Would it be possible to physically shield the amp and thus the EMI, like magnetic shielded speakers?
Thanks for sharing your knowledge all along, you've made this thread much more relevant.
Yes. And some electronics and computer parts come with similar shielding, however the wire itself may irradiate noise. Perhaps shield the wire as well, like coaxial cables?
One of the latest Wi-Fi technolgies I've seen is like that, installing a in-wall cable in a building, and this cable has a few holes exposing the metal, and acts as as antenna.
I was inspecting the board, and most likely (the thermal grease did not allow me to check the marking dot precisely) pins 2 and 3 are already connected together. So, dead?
Welcome to the world of switching amplifiers.
Google D-amp interference and read all the problems of guys fitting switching amps into their cars. No one seem to be able to offer a solution.
Actually long speaker leads makes things much worse. TI suggests leads of no more than 15cm, if I remember correctly, to avoid EMI.
I tried to use an industrial SMPS to power some stepper drives but the EMI caused stepping errors that I could not fix.
The main source would be the speaker cables. The tpa3116 data sheet suggests 125cm cable as the longest without LC output filter. I recommend shorter than that tough. Think active speakers. It's far better under any circumstances, class D amp or not, to run power and signal cables to the speakers than it is to run speaker cables from the amp to the speakers.
Last edited:
Yes, they have to be. That would cause the cycling click noise you hear. That means the chip most likely works. Probably the output inductors are fried if they were hot enough to burn you.
I also favor active speakers. For this system I'll be using MiniDSP. I could install the amp inside the box, but since I've already ordered bi-amp banana plug plates, I guess it will be right outside. Summing the wire inside and outside it will be around 50 or 60cm.
I just checked the smd caps after the chip, both channels are good. Then I proceeded to remove the inductors, but my extensive manual habilities did not permit to remove them without damaging solder pads and peeling off some of the traces. But it still can work.
I am just intrigued about what caused the overheating. It wasn't overvoltage (the lytics are not even stuffed) and it did not play at abusive levels.
I know class D has some peculiarities, but there are several working class-D comercial car amplifiers. Also a friend used a Sure TK2050 4 channel without problems.
Wonders of wonders, I have just fixed the EMI problem I was getting when trying to receive FM.
I clamped 4 TDK ZCAT1325 0530 ferrite clamps on the speaker leads right up by the board.
The ferrites were purchased on line.
I did not think I would ever get the problem sorted because of the long speaker leads I have.
I clamped 4 TDK ZCAT1325 0530 ferrite clamps on the speaker leads right up by the board.
The ferrites were purchased on line.
I did not think I would ever get the problem sorted because of the long speaker leads I have.
Last edited:
Exact same effect that a ferrite bead filter on the board would have except it would be cents instead of dollars in cost.
In both cases the ferrite clamp/bead has a resonant frequency around 100MHz +-10% where they will have high impedance thereby stopping FM radio inference which as we know is in the 87.5 to 108MHz band. Outside this range, and particularly in the audio range the ferrite clamp/bead will have almost no effect. Typical impedance will be less than 10mOhm.
That's also why I say that a ferrite bead filter can't really be called a filter by normal standards. It only works in the FM band far above both the audio range and the switching frequency of the amp.
Last edited:
Well, more of a hypothesis. In the PCB filter the current goes through the inductors themselves, but with the ferrite around the speaker cable there is no direct contact, just filtering the irradiated noise.
Electromagnetic contact is direct contact. Besides when you make the tight turn (or turns) of the speaker cable on the ferrite core inside the clamp, you introduce all sorts of parasitic capacitances, and quite possibly higher DC resistance than from a ferrite bead as well.
The beads were 10 for about $8, shippiing incl. For me it was quicker to just to clamp them on.