PCB Layout Process and Guidelines?

[jkuetemann]

It seems to be the hallmark of a good PCB layout to not have jumpers, or at least very few of them. For the DIY crowd just how frowned upon are they? IMHO a few jumpers should be fine, especially if it avoids really loony trace routing. I ask as my current rip-up and re-route looks like it may need a few links on the board.

[counter culture]

Originally Posted by jkuetemann

I'm not entirely sold on the cancellation of the fields produced by the power supply traces, they occur at different times and therefore are not in exact opposition.

It's worth reading Doug Self's Audio Power Amplifier Design Handbook where he presents some example traces showing various distortion measurements, including this. This particular source of distortion (number 6), he calls induction distortion and says it 'is probably the most widespread cause of unnecessary distortion.'

With regard to its minimisation he says 'Minimise radiation from the supply rails by running the V+ and V- rails as close together as possible. Keep them away from the input stages of the amplifier, and the output connections; the best method is to bring the rails up to the output stage from one side, with the rest of the amplifier on the other side. Then run tracks from the output to power the rest of the amp; these carry no halfwave currents and should cause no problems'

Originally Posted by jkuetemann

It seems to be the hallmark of a good PCB layout to not have jumpers, or at least very few of them. For the DIY crowd just how frowned upon are they? IMHO a few jumpers should be fine, especially if it avoids really loony trace routing. I ask as my current rip-up and re-route looks like it may need a few links on the board.

Use jumpers as necessary, the important consideration is the performance, not what people think about them or frown upon. A large number of jumpers is sometimes indicative of a low effort having been made, but if you are satisfied that they result in a higher-performing layout, then use them as you see fit.

[AndrewT]

use as many jumpers as needed.

[jkuetemann]

So, here's another iteration.

Does this represent progress? I think it does, though any layout is a compromise. Rail fusing will be on the PSU board and remainder of the output network on a remote board at the speaker output terminals. The speaker ground will be returned directly to the PSU. Any constructive comments welcome with respect to layout.

Thanks for the input thus far.

[AndrewT]

Originally Posted by jkuetemann

.............. Rail fusing will be on the PSU board ..............

Perfect, just where it should be.

I love your local HF decoupling !

Any room to fit in some MF decoupling? Try 100uF on each rail somewhere.

Add a pair of inverse parallel diodes across R3.

[alex mm]

Hi Jason ,
Layout based on your PCB , sorry if I'm not welcome in your thread.
Regards Alex.

[jkuetemann]

Originally Posted by alex mm

Hi Jason ,
Layout based on your PCB , sorry if I'm not welcome in your thread.
Regards Alex.

Good to see you! You are very welcome to participate. I started this thread to help me learn about the finer points of PCB layout through community feedback. I only wish I was half as fast at a board iteration as you. I spend a great deal of time staring at the screen, visually routing and reorganizing before really doing anything.

Do you have any specific comments about my layout, good or bad?

[jkuetemann]

Alex,

A couple observations regarding your layout, perhaps you could provide some feedback.

You provided two spades for ground connections, but they are physically connected via traces. What's the reason or rationale?

The outputs, drivers and VBE multiplier are intended to be attached to an L-bracket, the mounting holes in the upper corners probably don't need to be there, unless it is recommended to physically bolt the bracket to the PCB with dedicated hardware. Should there be dedicated fixing points or can the mounting screws for the devices be used exclusively?

Some of my earlier iterations included the output inductor since the space seemed to be available for it. Community feedback suggested it is best located at the output terminals so I moved mine off-board. Many of your layouts include the inductor in the PCB. Any specific thoughts on wether it should be on / off and why? I like the 'all-in-one' idea personally, but others seem to prefer a more modular approach with physical separation of certain blocks.

Thanks for providing some food for thought and an example from a more experienced perspective.

[jkuetemann]

Originally Posted by AndrewT

Perfect, just where it should be.

I love your local HF decoupling !

Any room to fit in some MF decoupling? Try 100uF on each rail somewhere.

Add a pair of inverse parallel diodes across R3.

Thanks Andrew. After seeing Alex's example I have a few ideas to incorporate into the next iteration.

Any thoughts on 'rail catching' diodes, reverse biased from output rail to power rails? Came to mind when you mentioned putting diodes across the 10 ohm ground isolation resistor for the audio grounds.

[AndrewT]

Output rail to supply rail diodes.
Supply rail to ground rail diodes.
If the output or the supply rail are fused then repeat the diodes to both sides of the fuse/s