TPA3255: OC shutdown always at only 9A

Hi Ralf,

Can you share the schematic to us? And what about the inductor of LC filter?

Derek,

Regards

Hi Derek,

Thank you for your reply. Schematic is exactly the same as under 10.2.4. in data sheet. Only one difference: My blocking capacitors on input A/B are 100uF and not 10uF. At first I used a different filter, but now I have changed it to exactly this type (whereby I use one 1uF film capacitor like on EVM and not two 470nF like here).  As inductor I tried both types from Wuerth Electronics mentioned in the LC-Filter-Design-PDF. At first the 7uH, now it is the 10uH. The behaviour with too early OC shutdown has not changed with original filter compared with my own filter and, as said before, exactly this filter components in original TI EVM give no problem. Thus, I think the filter is not the problem

The power supply section and analog audio input not showed in 10.2.4. is completely different  from EVM, but I designed the board in this parts without schematic. All are eralier used circuits. I only could provide the layout by E-Mail or other private message. Whereby, with very first prototype I worked with a direct connection from a input transformer (properly damped) to the TPA and the OC ehaviour was the same. Thus, analog input should not cause the problem.

Regards,

Ralf

... in the meantime I got an idea. Admittedly I only just realized that my size of the input capacitors deviates from the reference design.

On the other side there are two things that I do not fully understand. Why does it make no difference whether BTL or PBTL. So maybe a hint towards the input? ...
Then I had as I said in the beginning only a transformer as input. After that, I actually thought that an amplifier stage without DC offset would work without DC capacitors. But no, it didn´t work, then I directly got fault condition. With the 100uF it worked, but maybe this value is too high to get the full output level. Unfortunately, there is no extra indication whether the DC protection responds. Could it be that I get a fault because of DC protection and it is not an OC shutdown?

Many greetings
Ralf

Okay, sorry, forget my last post. I tried it. That is not the problem. But now I have found the reason for the shutdown. The PWM signal before the filter becomes increasingly shaky as the input level increases. At the point where I can see almost no seperated rectangular edges, the board goes into error state.
Now the big question is why that is. The clock on pins 9/10 remains stable.

best,

Ralf

one addendum yet: this behavior is independent of whether with or without load.

Hi Derek,

the inductor is one of the recommeded ones in LC-Filter-PDF on page 27: Wurth 7443631000. Its rated current is 16A and 23A until saturation, thus a lot more than the Coilcraft on EVM with its 6.1A.

Best,

Ralf

Hello Derek,
Excuse me, I have to admit, I wrote some nonsense on top. At first, I actually thought that this "jitter" was just the principal behavior in PWM. I was then confused that I did not see that on the original EVM. Maybe a measurement error. But actually I checked it twice... Can not check it anymore, because the EVM is now broken.

On my own board I have measured a lot again and I am now again of the opinion that everything looks like it should. As the input level increases, the rectangles become wider and the "jitter" varies with the frequency of the input signal. ;-)

In the meantime, I have also used exactly the filter components of the EVM as a test. The problem remained exactly the same. The BTL mode, I have now again checked more precisely. If I provide both channels with the same signal and connect a similar load, the maximum possible level is reduced to exactly half. So here too everything as expected: at BTL half the current as in PBTL. Also I modified the board and component position for very short traces, again without any changes.

Interesting is the following: Now I have halved the operating voltage, from 54V to 27V. Then I get in BTL mode from both channels together about 20A. An idea about this? Are there any mechanisms where a high voltage generates a disproportionate amount of current? Or do I have a thermal problem? Where, if I read the data sheet correctly, when overtemperature, both LEDs should light, clip and fault, right?

And one other thing: in PBTL mode with 54V supply I was able to turn up the level to maximum for a very short time and I reached about 90V, but a bit longer, not more than a second and the board goes to fault. It´s all quite crazy...

Best regards

Ralf