TDA2050 bridged AMP output level lower than input level.

[RAZZITEAK]

i've just ordered a custom designed pcb for a AVR controlled subwoofer amp that i have designed.
the pcb is designed in two halves with two separate power supplies, with the first half being the micro-controller part and the second being the amplifier.

everything on the  microcontroller side (low voltage) works just fine, all the relays click and ll the led's work. 

however the amplifier side is almost completely dead 

the design comes from a schematic that i found online (see attachment no.1) it is a bridged amplifier based on the TDA2050. datasheet link: http://www.alldatasheet.com/datasheet-pdf/pdf/25046/STMICROELECTRONICS/TDA2050.html

and from my understanding it seemed as tho it would work, and many forums also backed this up.
So i simply copied the design to my schematic in eagle. I used  very similar if not exact values to what was specified and i used regular electrolytic caps in the layout, 1/4 watt resistors and a single rail power-supply with a full wave bridge rectifier and with 2x 4700µF of filtering (overkill i know).

however when i power it up with 20V from my bench power-supply and plug in my iphone on the input, it seams as tho i get a lower peak to peak output than what my iphone is giving out.


i've checked the power input of both of the AMP chips and they both get +20V and GND on the correct pins.
So i taught it might have been the input cap that was giving me problems, so i bypassed that by hooking my iphone directly to the input of the amp with no further success.
So then i tought it might have been the feedback resistor that was the issue, so i tried changing the value of that to both 20K
and 4,7K with no noticeable difference.
i then in desperation tried to remove some resistors from the board in order to see if that changed anything (i've tried to remove r19, r24 and r15, r29) with yet again no success rather than the output going completely silent once r15, r29 was removed.

so lastly i tried to bypass one of the chips completely and basically make a copy of the circuit that was in the datasheet for a single TDA2050 on a single rail power supply. and this gave me no further progress.


So now i'm out of ideas and don't know what to do next, the circuit draws rougly 150mAh when plugged in to 20V.

I've tried to desolder the chips in order to put them on a breadboard, but doing so with a double sided board and thruhole plating with just a soldering iron and a spring loaded desolder pump is next to impossible.


So is there anything i can do to "bodge" this circuit in any way to make it work?

thanks for the help! 
/Simon

P.S i am only powering up the amp side and have soldered a jumper past both of the relay's so i don't have to worry about them. And the amp makes a pop sound (quite loud) when it is turned on and off ( i know this is normal for amplifiers but i just think it was interesting how load the pops are i contrast to the input source)

[tautech]

Did you get only 1 PCB? If not check the spare for shorts, bad via's and open circuits. Good idea to  not have too small holes for component removal.

[RAZZITEAK]

i have five boards witch i have tested against the schematic for any shorts on the board, but since it takes basically no time to test, i did it again as you said, and i found nothing out of the ordinary

[lewis]

Is C21 really 10pF?

[RAZZITEAK]

The input cap (c21) is 10pF correct

[Hideki]

Well, there's your problem

[madires]

The input cap (c21) is 10pF correct

With the input impedance of about 20kOhms you got a low frequency roll-of (-3dB) of about 1.768MHz ;-) Please try 2.2µF.

[lewis]

Booya! 

[RAZZITEAK]

YES!!!  thank you! and i tought it was the input filter cap because i bypassed it, i guess you can't do that 

[RAZZITEAK]

*tought it wasn't

[peter.mitchell]

also, use polyester film (PET) rather than ceramic for that.

[RAZZITEAK]

Thanks for the tip! got a lot of noise before i switched to a 2,2µF PET cap

[madires]

And a polypropylene film cap would be even a notch better ;-)

[Neganur]

don't forget those 100 nF decoupling caps for the 2050, preferably as close to the amp as possible (bodge them in directly under the board for example)