dj_oatmeal said:
I have 100uF between pin 3 and GND. 2200uF and 104K on Vp. Pins 2 and 5 go straight to GND without any caps. Should I add a cap to these pins? If so I have 220uF available would that work or is that too much?
The heatsink is not grounded.
norazmi said:
it seems like you`re following the right philips schematic, vp pin7 connect to +vdc power source and make switch for pin8 to connect to vp and must have 2200uf and 104 or 104 100nf parallel to GND, no need to add any caps to pin 5,2. as i can see you add volume (ROTARY POTENTIOMETER) for the input, try remove this one and connect directly from you sony discman, sometimes wrong wiring on volume control can cause no audio signal.
I think you are both looking at the correct schematic.
The problem (with the caps) seems to be that they are not properly located.
The 0.1 uF (i.e. 100 nF/104) and 2200uF caps' bodies need to be as close as possible to pin 7, with the shortest-possible connections to that pin. If a compromise must be made, then the 0.1 uF caps should be closest to the pin.
I would also add a 0.1uF cap from pin 8 to power ground, similarly.
Any other capacitors that connect to the chip's pins should be located similarly, i.e. right AT the pins.
Any and all 'to/from' wire pairs should be tightly twisted together, e.g. input/input-shield 'gnd', power/gnd, switch and pot to/from, speaker hot/gnd, etc.
Wires with low-level signals should be routed away from high-current wires and components.
All wires and component leads should be as short as possible.
I would also try norazmi's suggestion and at least temporarily try it without the pot, removing the pot's wires at their far ends (i.e. don't leave them dangling from the chip etc, during any temporary test).
First, though, you "must" properly-heatsink the chipamp. Yes, its ENTIRE REAR SURFACE must be tightly-clamped to the large flat surface on the rear of the heatsink!
You should definitely use a thin layer of thermal heatsink compound "goo", between the chip and the heatsink (but can 'probably' get away with not using it, for testing, if none is on hand yet).
Since you probably don't have any chip-to-heatsink insulators on hand, make sure that the heatsink is not connected to anything else that is metal.
Maybe none of this directly explains or solves your problem. But the chip could have already been destroyed by either soldering without heatsinking between soldering iron and chip body, OR, it could have 'oscillated itself to death', because of concerns like those partially-addressed above.
High-frequency oscillation can quickly overheat and destroy a chip. Incorrect capacitor placement, and 'loose' layout problems, among other things, can cause HF oscillation.
It would probably also be very wise to add simple lowpass RF filters to the inputs. For that, you can put two resistors in series with each input, close to each pin (pins 1 and 9), with a small capacitor to ground from between each pair of resistors. Something around 470 Ohms for the four resistors, and 3300 pF polystyrene (probably best) or polypropylene or polyester or silver mica, or a C0G or NPO ceramic, for the caps, should work fine. You can scale the component values, as long as the R times C stays in the same ballpark (unless you want to raise or lower the filters' -3dB 'cutoff' frequency. A lower RxC raises it, and vice versa. f(-3dB) = 1/2/Pi/R/C in 'calculator format'.).
