Single Sided LM3875 PCB... Crit please.

[justblair]

This is the PCB, the power supply pins did not label this morning... But from left to right they are

0v -24v +24v

The Schematic

I have built this according to the schematic on the Decibel Dungeon site and is one sided for conveniance. I aim to etch this myself, though will develop it into a more complete board, with preamp etc on the same board.

The filter to safety ground is missing, as I will have 1 filter for all the amp stages.

The connection from Signal Gnd to Power Gnd (0v) is currently missing from the PCB also. I'm looking for advice on how I should execute this. My own thought is a .25mm trace from square pad on the input pins to the 0v line adjacent. Is this fair enough?

By the way the caps are orientated the way they are to allow me to mount them on their sides. I want to make the board profile as flat as possible as this will be part of a 160mm by 100mm board that fits in a hammond extruded alu case.

Comments please?

[BWRX]

Hi blair. After a quick glance it appears that there is no feedback resistor connected to the output. It's hard to tell with that peach color, but are the input signal and power supply grounds connected?

[justblair]

Originally posted by BWRX
Hi blair. After a quick glance it appears that there is no feedback resistor connected to the output. It's hard to tell with that peach color, but are the input signal and power supply grounds connected?

Ye sorry, I should have mentioned in the preamble that the feedback resistor is going straight onto the LM3875 pins...

The input signal and power supply grounds are yet to be tied together. I was referring to that in my question above, what is the best way to tie these, I had thought a very thin (.25mm) trace between the input and power supply ground?

Other than that anything obvious that can be improved?



Here is a high contrast version of the bottom of the board...

[BWRX]

Some things I did for my layout with the LM3875:

Locate both supply caps to the left of the chip. Remove pin 2 (and the through hole on the board) so you can run a thicker power trace to pin 4. Use supply rail decoupling caps with a smaller package. Mount the feedback resistor on the board instead of directly to the pins of the chip. There should be ample room for this. Connect the power and signal ground using a plane, just keep the signal ground away from the high current paths around the supply caps and output speaker ground. This shouldn't be hard to do if you locate the power ground on the left side of the board between the caps and have the signal ground on the right side of the board. I may have time tomorrow to whip up a quick layout if that didn't make sense.

[theAnonymous1]

What software did you use to make the layout?

[justblair]

Originally posted by BWRX
Some things I did for my layout with the LM3875:

Locate both supply caps to the left of the chip. Remove pin 2 (and the through hole on the board) so you can run a thicker power trace to pin 4. Use supply rail decoupling caps with a smaller package. Mount the feedback resistor on the board instead of directly to the pins of the chip. There should be ample room for this. Connect the power and signal ground using a plane, just keep the signal ground away from the high current paths around the supply caps and output speaker ground. This shouldn't be hard to do if you locate the power ground on the left side of the board between the caps and have the signal ground on the right side of the board. I may have time tomorrow to whip up a quick layout if that didn't make sense.

Thanks for that. Unfortunately its GF night tonight, so I wont get a chance to try some of those ideas till tomorrow...

Originally posted by theAnonymous1
What software did you use to make the layout?

I used Diptrace, so far my favourite designer. I found it a lot easier than Eagle and a fair bit better than PCB123. This is my first audio PCB, previously I have made sensors etc, marks out of 10?

Thanks for the comments so far any other ideas are welcome as well..

[justblair]

After pondering some of BWRX's advice, I came up with a second design.



The design is for a pair of chips. This time the power rails are thicker, and the decouling caps on the left of each chip. I have mounted the resitor on the board.

I have omited the input caps. I will filter dc out (if necessary) further up the signal chain.

The signal ground and power gound are now connected,

This is designed for use with a single power supply.

I have connected several of the NC pins to the power ground to make the routing easier.

The two chips will share a heatsink, which I will also couple to the alu case I have in mind....

Comments please, before I start the next section of my design...

[gootee]

Have you researched the actual sizes and lead-spacings of all of the components? What make/model/voltage caps are you using for the 1000uF positions? And what type are the 0.1uF polypropylenes?

- Tom Gootee

http://www.fullnet.com/~tomg/index.html

[justblair]

Originally posted by gootee
Have you researched the actual sizes and lead-spacings of all of the components? What make/model/voltage caps are you using for the 1000uF positions? And what type are the 0.1uF polypropylenes?

- Tom Gootee

http://www.fullnet.com/~tomg/index.html

The lead spacings are pretty general just now, which I think is fine for the resistors, but as you have pointed out the big electolytics are on the small side.

The design is such though that I can accomodate larger, or even multiple electrolytics with 5-10 minutes of alterations to the board.

Consider this a rough sketch. I am going to add buffers etc over the next few days. My aim is to work in a modular fashion so that I can easily produce a range of pcbs for trying out.

Right nowI am intersted in any electrical/layout mistakes such as grounding, trace widths etc

[woolly]

amateur opinion - would be better with more space between tracks or could be a tricky when soldering and spilling over tracks