P3A layout

[IntArt]

i would like to receive comments about the layout of the amplifier.

anything that might prevent future problems.

i have added two electrolytic capacitors in both rails that are not in the original design because I have seen it here and the worst that could happen is not use them.

[IntArt]

here is the image

hxxp://farm4.static.flickr.com/3080/2864619412_8c76bb8564_b_d.jpg

change hxxp for http in the url

[CBS240]

Hi

Just a few points.... I would at least fill the empty space with copper. I would also take the negative feedback from a point in the center of where the two output resistors connect, that being node labeled "OUT", where the Zobel now resides. Having the DC blocking cap grounded to the same trace as the decoupling cap might create issues as well.(Buzzzzzzz) It needs it's own ground trace, a clean ground trace, without the current pulses from the decoupling cap corrupting it, such as the signal GND. Does Q9 need to be mounted to sense the temperature of the drivers, Q's 5 & 6? (reffering to the scematic here ) Typically with a feedback pair OS the drivers have to be thermally compensated, but maybe not here....I haven't built this one.




Anyway, good luck

[IntArt]

layout updated
what do you think?

[AndrewT]

I think the decoupling caps are too far apart.
Moving the power rails and power ground to the middle of the PCB and having the decoupling adjacent to each other fed from very short tracks should perform better.

Q9 MUST monitor the temperature of Q5 AND Q6.
Three transistors inline and facing the same direction allows a flat strip of aluminium/copper to thermally couple these together and perform as a small heatsink if you decide to increase the driver bias current. 220r limit total driver bias dissipation to just 190mW for the pair.

[tcpip]

Originally posted by AndrewT
Q9 MUST monitor the temperature of Q5 AND Q6.

In general, yes, but Rod's PCB layout does not have any provision for this, and he's said that all those hundreds of his builder-customers have never reported a problem. I suspect it's partly the stability and simplicity of the circuit, partly the (relatively) low power dissipation.

When I designed a PCB for this amp, I tried putting in thermal coupling between Vbias and driver transistors, but the Vbias Rod uses is a TO92 type thing which does not even fit into any heatsink, so I just put them physically close to each other and left them at that.

[AndrewT]

To92 can be clamped or glued to any sink. Alternatively drill a tight fitting hole and insert into the thermal paste filled recess.

[metal]

But the author of this amplifier did not put the Q9 TO-92 transistor on the heatsink. instead its on the PCB.

[AndrewT]

Originally posted by tcpip
Rod's PCB layout does not have any provision for this,

Originally posted by metal
But the author of this amplifier did not put the Q9 TO-92 transistor on the heatsink. instead its on the PCB.

then extend the To92 with three flexible wires/cords back to the PCB pad locations.

[jaycee]

I have had a board of my own design where the VBE transistor isnt near either driver, and it actually works quite well, while being thermally stable. I guess it is because the circuit is so simple.

The biggest problem I had was oscillation with the power devices, but fitting the larger decoupling capacitors on the board, as IntArt has done, solved the problem. I've suggested this a few times on Rod's own message board and it has indeed helped a few others - maybe this is where IntArt got the idea from