Mixed Systems and Specialization > Amplifiers, Mixers and Equalizers

150W+150W Class D Amplifier Design Secrets: How PCB Layout affect performance

<< < (4/9) > >>


--- Quote from: Red Sonja on August 08, 2011, 12:01:44 PM ---sir tanong ko lang kapag ba IC ang ginamit nde transistorize diba required sa circuit yung spider/star grounding para iwas sa humming?

--- End quote ---

narito ang sagot:


More Reminders

I will not delve into the details of class D circuitry to keep this article as short as possible. We will focus our study to the PCB layout effects. Baka kasi humaba ng humaba and discussion tuloy di natin matapos. ;)

Schematics, initial measurements, and other technical info, including PCB layout, will be published later when we start tackling the technical details of this amplifier.

Motivations for building the project

-   I was inspired by recent surge of interest to class D amplifiers and audio amplifiers in general by members of this forum.
-   I thought it is a good excuse for me to indulge myself once again into the exhilarating and magical world of analog electronics.
-   It will make a good case study for the benefit of our young and uninitiated designers.
-   I am jealous and green with envy to the attention other members get with their accomplished amplifier projects.  ;D

No questions and reactions will be entertained in this thread. Please post your concerns in the Q & A thread using the following link:


Salamat sa correction ate rdp ;D

The TDA8950J has elaborate protection built-in. It will protect itself from damage to just about anything you throw at it, except my PCB layout.  ;D Here is the story behind it.

The Killer PCB Layout

I made a PCB layout and  built on it the 150W+150W class D amplifier essentially using the application circuit recommended in the datasheet. After a thorough visual inspection of the finished prototype, it was time to power up the circuit, and that I did.

Whenever I power up an amplifier the first time, I do it as a matter of routine with the load (speaker or dummy load) disconnected. This initial step will (usually) save some of my precious components from total annihilation in case I went remiss in parts of the circuit. If no smoke or explosion occurred, I measure the idle current and amplifier output DC voltage as a matter of procedure.  If something is seriously wrong with the amplifier, in will usually reveal itself within these first few steps.

Round 1: First blood

And so as a matter of fact it did. There was no explosion, but one channel has its output stucked on one of the supply rail. The TDA8950J is equipped with protection circuitry that will automatically shut down everything in case it detects this kind of fault. And so the other channel is out too. I looked for the source of the short, it turned out, the output of TDA8950J IC itself shorted out, and has to be replaced. What caused the failure? No definite idea at this point, but ESD came to mind. I was not wearing any ESD prevention when I was handling the TDA8950J. Maybe, thatís what killed it.

Round 2: The face of Opportunity

I replaced the TDA8950J with a new one. I was concerned that my PCB, which is handmade, may not survive another IC replacement cycle. I fashioned a socket for the TDA8950J so that no soldering will be necessary in case the IC needed to be replaced again. This extra step proved its worth later.

This time, I checked the IC output and the rest of the pins for any short or low resistance path using a DMM. When everything checked well to my satisfaction, I hooked up an oscilloscope to monitor one amplifier output and then applied power once again.

The exact same thing happened! One output shorted out as soon as power is applied! I got excited by the development (believe it or not!), it is clear to me now that this class D amplifier is not as boring as I initially expected. I will learn something new on this one. After all, this is what project building is all about- learn new things, donít you agree?

to be continued...

No questions and reactions will be entertained in this thread. Please post your concerns in the Q & A thread using the following link:


nice tita +1 for you

subscribing  :)

Round 3: Insulation Breakdown?

With my third TDA8950J nervously seated on its socket, I went more slowly and more careful this time. I want to catch what causes this problem. If the IC keeps shorting out as quickly as it did in the last two sorties, I may not get the chance to observe and determine the cause.

Post mortem analysis of the zapped ICs held some clues. The output did not completely shorted out, as in 0-ohm dead short. Instead, resistances in the order of 6 to 20 ohms were measured. The TDA8950 employs MOSFET at its output stage. Whenever a MOSFET fails and shows relatively high resistance short across its drain to source terminals, it is tell tale sign of insulation breakdown. Insulation breakdown can occur if the voltage across a MOSFET rises very rapidly, and overshoot past the maximum voltage the MOSFET can handle.

Armed with this suspicion, I applied power once again using power supply as voltage as low as the IC will allow, at +/- 12VDC. In a quick instant, a waveform showed up at the oscilloscope display! This is the expected waveform right at the IC outputs before the low pass filter, hence, it displayed the expected 380kHz PWM rectangular waveform. I touched the amplifier input with a bare finger as a quick test, and the output responded with a pulse modulated display. The amplifier is now working!  I zoomed the output PWM waveform, zeroing on the pulse edges where overshoots usually occurs. There are overshoots, put they seem low enough to cause any damage.

Slowly, I raised the power supply voltage. The ďoh f**kĒ level came at around +/-18VDC.

Round 4:  It ainít over yet

With only two TDA8950J left, I began to wonder whether I can make this work at all with the remaining pieces. I decided its time to connect the load, just to see what happens next.

Again, I powered up the circuit with +/-12VDC. There were signs of life. I fed a sine signal on one of the amplifier input. Sure enough, there is an output! The output waveform shows deformations indicative of a high distortion level, nevertheless, it is a significant progress over the three previous tries that resulted only in blowing up the ICs. Now I have something to work on. I slowly raised  the supply voltage. Parasitic oscillations showed up, but the amplifier kept working. I reached +/-20VDC and stopped there, deciding to investigate what causes the oscillations.

This is where things get even more interesting. During my probing and experimentation, which lasted for hours, the IC kept working. I figured out the distortion and parasitic oscillations that showed up is caused by problems with the PCB layout. I did not made any attempt to make and try any pcb layout modifications as yet, as my attention is fixed in finding out why the IC fails.  I can now examine the PWM output waveform in greater detail, but still, could not find the probable killer wave. And then, something happened. The wire connecting the load went loose and disconnected the load. This signaled the start of Round 5.

Round 5: A shocking realization

The IC amplifier self destroys when it is operated without load! This puzzling behavior is completely unexpected and not like anything I encountered working with audio amplifier. Lolo Danny was on line at that time, and I was able to open a short discussion with him. He did not believe me at first, even suspecting I am pulling a prank on him. Apparently, lolo Danny never encountered this problem either. In the end, he too would want to know.

With my last TDA8950 now in the death row, I proceeded with an all out effort to come up with a fix that will solve this problem once and for all. I redesigned the output filter, modify the output snubber network. Needless to say, these too failed. But I am not giving up. This just got so interesting that I canít quit now. I booked my order for another 5 pieces as I say my goodbye to the last TDA8950 I just put to death.

to be continued...

No questions and reactions will be entertained in this thread. Please post your concerns in the Q & A thread using the following link:



[0] Message Index

[#] Next page

[*] Previous page

Go to full version