how to solve amplifier's oscilation?

[justin66]

i will try to be as detailed as possible in short text.
first i have made a pcb on a predrilled board also called breadboard using circuit found http://andy-progetti.com/small-and-high-performance-audio-power-amplifier-lm3886/#comment-98
it worked good no heat problems etc. than i think non isolated lm3886t touched one of the power rails with its heat sink and a turn on pop was present after that. later i have build the proper etched pcb also found on the same website. and OFC i have bought a new chip. it is working but when 4ohm load is connected it is heating rather quickly to where the same heat sink for previous board was too hot to touch. im using the same power supply-25 0 25+. im not sure if that was the case with first board but now i can hear a tinny 50hz tone on the speaker. i decided to take out the chip in the first board to fit it to an etched pcb and it didnt didnt work. it just overheated with no output.i have put it back to breadboard. with a working chip when i disconnect the 4ohm speaker and take off heat sink with power on chip gets warm to the touch after few seconds and heat is rising slowly. i will include pictures for the first pcb, second and my psu.
psu is diode bridge connected to negative and positive. output is two 2200uf electrolitics and ground is connected to a center tap of electrolitics.

[ebclr]

If you are sure power supply is ok
Try the following ,If you have a metallic Box, wire a ground of your circuit to the box and put everything inside the box close the box and check if noise disapear

[justin66]

psu is a mess but its ok. i think chip is oscilating and causing heat witch is a problem

[ebclr]

50 HZ normally comes from power supply or bad ground, or ambient. It's not probable that a 50Hz comes from the chip

[justin66]

i wrapped it in aluminum foil the 50hz humm is so tiny that its not noticable if you dont put your head near the speaker. its close to background noise. heat isssue is still tresent after wrapping it and tiny tiny 50hz is still there if i disconnect one of the center lead of the transformer than its loud but now its so tinny almost not worth adressing i just thought it might help

[3db]

The PCB doesn't look the same as the design in the article.
Where's has all the copper gone ?
It was there for a reason ie. to make the thing more stable.

Edit  Oops seems as though that isn't copper.

[justin66]

i didnt really know how to use eagle xD so i just printed provided eagle project and changed all blue and green traces to black now i discovered one thing. my amp pulls 0.05A acording to my multimeter when there is no audio cable connected. when audio cable ground is connected it pulls 0.05A too. when there is positive audio signal with no ground there is that loud humm(other end connected to pc). if i disconnect audio cable(from pc) it sounds like i was touching the end of positive... you know that humm. and finally if i short out audio input ground to positive. it pulls that 0.5A dropping current as it heats up

this is the one i used modified a bit i didnt know how to remove those dotted lines in eagle it should work the same tho

[justin66]

i removed this cap and problem seems like dissapeared but i dont understand why?

[ebclr]

Regarding the heat did you use thermal paste between the heat-sink and the amplifier ?

[Kleinstein]

The capacitor at the input should not have much influence on the 50/100 Hu hum. It is more about high frequency (e.g. AM radio) coupling from, the input. The cap is anyway better from the non inverting input to GND, not between the inputs.

Hum is more caused by shielding or not so good ground layout. The GND layout of the board does not look that good and might cause some hum at 100 Hz from ripple currents to the supply caps. It depends on the supply how bad it gets.   

[Siwastaja]

I'm no expert in feedback loop stability issues, but it seems to me that C3 that you removed is drawn wrong, it should be from amp + input to ground, i.e., to form a low-pass RC filter with R2 to limit the audio bandwidth getting in. With unlimited bandwidth and a lot of gain, any noise can start the oscillation.

Secondly, I think that integration action in the feedback loop, i.e., reducing voltage gain on high frequencies, would increase stability. C6 does just that, but R5 is too high for it to work properly.

I found out the hard way with such an integrated amp that even the exact application schematic and even the PCB layout supplied in the datasheet was completely unstable and oscillating in the megahertz range, causing the integrated protection to pop. Adding these two small caps, one at input and one in the feedback (without a series resistor like R5 at all) solved the problem.

[Audioguru]

The author of the project you copied did not use the schematic recommended by the LM3886 manufacture that did not have the 220pF capacitor.
The main filter capacitors are far too small at 47uF. They should each be 4700uF or more.
Your 25V-0V-25V transformer voltage will produce plus and minus 34.4VDC which the datasheet for the LM3886 shows for an 8 ohm load. The datasheet shows a supply of plus and minus 28V when it has a 4 ohm load.
Your homemade transformer looks too small to supply a stereo amplifier with 136W of output power and another 136W of heat for a total of 272VA.