pcb layout opinions

Wendy

Joined Mar 24, 2008
23,473
Since I make my own boards I don't bother with multi layer. I use the top layer to show where the jumpers will be, small short runs of wire to allow wires to go past each other.

Here is an example of my latest PCB. I used ExpressPCB to make the basic layout, then save the image as a .pdf. I use Gimp to translate a .pdf file into .gif (600DPI), then use M/S Paint to bring the pads up to my personal specs, then Gimp again to print to my laser printer 600DPI.

I use this process to make PCBs, you may want to go a different path.

How I make PCBs

I have to transfer the files via .zip, the forum brings the images down under the size requirements otherwise, as it has done here. Keep track of hole sizes, one size does not fit all. BG Micro has a decent assortment of small drill bits for sale.

As with anything you get better with practice. Building a PCB is much easier than wiring a perf board.

 

Attachments

Last edited:

steveb

Joined Jul 3, 2008
2,436
Thanks for the real information in the midst of the unsupported opinions that dominate this thread.

There is no question that you can make more attractive-looking boards when using 45° angles in traces. That's great. Everyone agrees that it's nice to make pretty boards. But don't throw around opinions and assumptions about electrical performance as if they were fact.

Unless you're one of the relatively few engineers designing circuits with multi-gigahertz signals, you don't have to worry about signal integrity issues with 90° corners. If you are one of those few designers, you will know what you need to do.

I'll say farewell to this thread with a quote from Mr. Montrose:
CONCLUSION


I have to say that this comes across to me as a very ungrateful response to people sincerely trying to respond to your post. You asked your question in a general and unspecific way, and asked for "opinions". I think you received a very good and widely varied set of responses from a group of people that have considerable experience.

I also don't think that the technical responses were "unsupported", even if they are opinions, especially since you provided absolutely no constrains on circuit/trace size, signal speeds, voltage levels or current level etc. The effects mentioned, which relate to high-speed, high-voltage, high-current which then relate to impedance, field intensity and trace resistance/inductance, are real effects. It would seem that these extreme cases are the ones where the issues could come into play, and so it seemed advisable to mention them in a general way. Just because you and most other people do not enter these domains of concern does not mean they might not matter to some people, now or in the future. Since, you did not provide details, how are we to know what you are really asking for?

In the future, if you are looking for detailed technical responses with supporting evidence, then you should ask specifically for well-supported facts with proof (not opinions), and provide a detailed set of constraints to define the domain of applicability and the specific problem to be addressed. Then we would know to restrict our responses and site proper accepted references, or real measurement data that we would obtain ourselves.​
 

colinb

Joined Jun 15, 2011
351
I have to say that this comes across to me as a very ungrateful response to people sincerely trying to respond to your post. You asked your question in a general and unspecific way, and asked for "opinions". I think you received a very good and widely varied set of responses from a group of people that have considerable experience.
...
Since, you did not provide details, how are we to know what you are really asking for?
I'm not the original poster. I did not ask for opinions, I just questioned the correctness of assertions by some that 90° trace angles are likely to cause signal integrity problems.

Unless someone specifically says so, I will assume he or she is not working with multi-gigahertz signals.
 

steveb

Joined Jul 3, 2008
2,436
I'm not the original poster. I did not ask for opinions, I just questioned the correctness of assertions by some that 90° trace angles are likely to cause signal integrity problems.

Unless someone specifically says so, I will assume he or she is not working with multi-gigahertz signals.
OK, sorry for the misidentification. Still, I stand by the comments. There are no constraints given, so your assumptions are not necessarily relevant.
 

THE_RB

Joined Feb 11, 2008
5,438
Thanks for the real information in the midst of the unsupported opinions that dominate this thread.
...
I gave 2 specific reasons that 90' bends are inferior (in post #15);
1; that track length is increased causing greater resistance, and
2; PCB density is significantly reduced.

Both points are "real information" and neither refers to the HF argument.
 

Wendy

Joined Mar 24, 2008
23,473
It is also why I do it. When I must have a 90° connection I "fill" in the corners. It is not obvious, and invisible to the scale I have shown.

If you ever try using a copying machine it becomes unusable because of this. The toner is acceptable, but the subtle distortions ruin it.
 

t06afre

Joined May 11, 2009
5,934
It is some books on this topic. In my book shelf. I have these two
The circuit designer's companion by Tim Williams
Printed circuits handbook by Clyde F. Coombs
And of course the IPC2221. I think the IPC2221 is free so find it with google
From The circuit designer's companion I quote
2.2.3 Track routing
The first rule of good routing is to minimise track length. Short tracks are less
susceptible to interference and crosstalk, have lower parasitic reactances and radiate less. Routing should proceed interactively with package placement to achieve this. It is often possible to improve routing prospects when using multi-function packages
(typically gates or op-amps) by swapping pins, so unless there are overriding circuit considerations you shouldn’t fix pin-outs on these packages until the layout has been finalised. A good CAD package with an extensive and intelligent component library will do this automatically. Similarly, you may have decided on grounds of package economy to use up all the functions in a package, but this could be at the expense of forcing long tracks to one or other function. An optimum board layout may require a few more packages than the minimum.
Many CAD auto-routing software packages will lay all tracks of one orientation on one layer of the board and all of the other orientation on another layer (X-Y routing). This works and is fast, especially for low-performance digital boards, but hardly ever produces an optimum layout in terms of minimum track length and number of via holes, and can be disastrous on analogue boards. Normally you should anticipate expending some skilled design effort at the layout stage in cleaning up the CAD output to produce a cheaper board with better electrical performance.
45° angled bends are preferable to right-angles, as they allow a slight increase in tracking density. Right angles and acute angles in tracks are best avoided as they provide opportunities for etchant traps and the subsequent risk of track corrosion. When two tracks meet at an acute angle the join should be filleted to prevent this. Tracks should not be run closer than 0.5mm from the edge of the board. From a mechanical point of view, the aim should be to balance the total coverage of copper on both sides of a double-sided board, or on all layers of a multilayer. This guards against the risk of board warping due to differential strains, either because of thermal expansion in use or because of stress relief when the board is etched, and also assists plating.
 
Top