Vias vs Single sided trace restriction

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

boji

Well-known member
Joined
Jan 6, 2010
Messages
2,375
Location
Maryland, USA
Just curious your opinions of a lesser evil:

I have the option of increasing via count on a 2 layer board so as to, where possible, keep long traces on the inside of a PCB. The outside would have a chassis pour, sheltering the long runs across the PCB from potential adjacent channel card crosstalk.

Should I be more concerned about the capacitive aspects of adding say, an extra 25-75 vias, or the potential EMF/crosstalk of allowing long traces on the outside plane?

I can appreciate this is more art than science and that I've left out many specifics of the design/circuits. Even so, what's your instinct tell you ought to be the greater concern? Vias or signal on the outside plane next to other pcb's?
 
I hate vias. As all my designs are though hole I make all side changes at component legs. At least then you know there is a nice chunk of solder and a wire connecting the traces on opposite sides. I have a pack of zero ohm resistors and I use those to  make links where it is impossible to route between two points without a via. I recently had a case where via was unavoidable no matter how hard I tried and I am still not happy about it. I will probably enhance it with solder.

My intense dislike of vias stems from the late 70s when I was working on a digital tape machine (computer not audio). We had some prototype PCBs in and it turned out a lot of the vias were not plated through properly so I had to go round each board adding a wire link at every single via.

Cheers

Ian
 
ruffrecords said:
My intense dislike of vias stems from the late 70s when I was working on a digital tape machine (computer not audio). We had some prototype PCBs in and it turned out a lot of the vias were not plated through properly so I had to go round each board adding a wire link at every single via.

Hello

We are in 2018 now :)

As you say with TH it's easy to change layer at component legs, but when you use SMD you have no choices except via, so we have to trust them  :)

Best
Zam
 
boji said:
Just curious your opinions of a lesser evil:

I have the option of increasing via count on a 2 layer board so as to, where possible, keep long traces on the inside of a PCB. The outside would have a chassis pour, sheltering the long runs across the PCB from potential adjacent channel card crosstalk.

Should I be more concerned about the capacitive aspects of adding say, an extra 25-75 vias, or the potential EMF/crosstalk of allowing long traces on the outside plane?

Sorry about the earlier rant. To answer your question a lot depends on how far apart the cards are. If it is an inch or so then you really do not need to worry too much about crosstalk. The chassis or 0V pour acts as a screen and a large area equipotential.  Any potential crosstalk will be coupled far better to the pour than to a small number of small area tracks on that side of the board.

Cheers

Ian
 
In general every extra via is another drilled hole (more machine time), and another plated through connection opportunity for failure.  Worse than having to manually connect vias, blind vias inside multi-layer boards can be a real biotch to troubleshoot and/or repair.

Modern PCB probably don't even count and charge for drill time any more and PCB process reliability is quite good... not like the bad old days some of us dealt with.  :eek:

I wouldn't get too carried away with trick layouts, but I did have one clever engineer working for me (last century) who used a trick where he would pair sets of high current runs (with send on one side, return on the other side) to reduce the loop area. I've seen a similar trick where a single send trace was bracketed by two adjacent return traces on single sided PCB to effectively define a zero loop area.

Unless you are mixing high current with low noise stuff, probably not a concern, but you are thinking like an engineer where everything matters and effort at the design stage that results in zero incremental manufacturing cost is time well spent.  8)

JR
 
boji said:
Just curious your opinions of a lesser evil:

I have the option of increasing via count on a 2 layer board so as to, where possible, keep long traces on the inside of a PCB. The outside would have a chassis pour, sheltering the long runs across the PCB from potential adjacent channel card crosstalk.

Should I be more concerned about the capacitive aspects of adding say, an extra 25-75 vias, or the potential EMF/crosstalk of allowing long traces on the outside plane?

I can appreciate this is more art than science and that I've left out many specifics of the design/circuits. Even so, what's your instinct tell you ought to be the greater concern? Vias or signal on the outside plane next to other pcb's?
I've found that in most cases, with analog audio and TH components, it's generally feasible to make the number of vias extremely small, perhaps one or two on a 100 sq. inch PCB. Then it's easy to make sure they are in a position where they are accessible. I have copper pour on both sides, so x-talk is a non-issue.
 
First off, thank you gentlemen for your vibes on vias.  :)

"It's generally feasible to make the number of vias extremely small, perhaps one or two on a 100 sq. inch PCB"
100sq in? That's a nice parcel of land you got there. 

My boards (left and right sandwich) is wonky in shape and short on space. It doesn't help I made the buckets of the console metalwork too shallow.  Hell if I'm redoing it tho. If I never have to tap another screw It will be too soon.

Anywho, I prolly don't understand EMF.  I figured signal would be safest behind a screen of copper.  Let's instead say I have a trace on the 'noisy' side of the board.  If this signal is surrounded (but not covered from above) in a 0V pour, is this 0V pour going to persuade EMF away from the z-axis of the trace as well?

Edit: I ask because to be honest, I am getting a little obsessive about signal only running on the inside at the cost of quite a few additional via bridges.    RR, this is despite me using component holes wherever it makes sense.
 
boji said:
"It's generally feasible to make the number of vias extremely small, perhaps one or two on a 100 sq. inch PCB"
100sq in? That's a nice parcel of land you got there. 
Typical of a 19 inch piece.

Anywho, I prolly don't understand EMF.  I figured signal would be safest behind a screen of copper.  Let's instead say I have a trace on the 'noisy' side of the board.  If this signal is surrounded (but not covered from above) in a 0V pour, is this 0V pour going to persuade EMF away from the z-axis of the trace as well?
In a mixer, you don't have to worry much about electrostatic EMI/RFI. The dominant issue is longitudinal noise, created by the resistance of the reference (0V/ground) and the voltage that develops across it due to circulating currents.

Edit: I ask because to be honest, I am getting a little obsessive about signal only running on the inside at the cost of quite a few additional via bridges.    RR, this is despite me using component holes wherever it makes sense.
You have to worry about sensitive nodes, those that carry low level at relatively high impedance; having two low-z nodes close to each other is not a serious issue.
 
I've measured / designed vias for signal integrity and it's not really an issue until you get very high in frequency.  Make the holes and pad a reasonable size and just about any modern board fab house should be able to make them reliably.

Have you considered using some SMT parts to help with routing? Thin film resistors and C0G caps in a 1206 package will be easy to solder and just as good / better performance wise.
 
Ha. Gremlins. Thanks Abby, for introducing me to Mogwai, after that pool party.

OCK3M2D.png


http://www.aes.org/aeshc/pdf/how.the.aes.began/augustadt_longitudinal-noise.pdf
 
John12, I have avoided SMD's for irrational reasons, to be sure.

I keep picturing them falling off vertical boards during a hot day for some reason. Well, that and I don't want to go back and redo the layout. ::)
 
boji said:
OCK3M2D.png


http://www.aes.org/aeshc/pdf/how.the.aes.began/augustadt_longitudinal-noise.pdf
I think the first quote is contemptuous; the second quote shows clearly that there are some people who actually understand how noise builds up in a large scale piece of electronics and don't have to call Marie Laveau to fix the problem.
 
abbey road d enfer said:
I think the first quote is contemptuous; the second quote shows clearly that there are some people who actually understand how noise builds up in a large scale piece of electronics and don't have to call Marie Laveau to fix the problem.
voodoo reference.....  ;D

JR
 
Back
Top