Go Down

Topic: why run traces on the bottom (Read 969 times) previous topic - next topic

db2db


When making a double sided PTH board, is there a downside to running most traces on the top? Seems like it's more like SMT, where you can easily see what's happening.

Yankee

No problem with a one sided board, top or bottom, if you can manage it. Usually though you won't be able to route everything on one side and you have to go to the other to get from A to B.

James C4S


Seems like it's more like SMT, where you can easily see what's happening.

I would not get in the habit of following traces on either the top or bottom side of a board.  First time you try that trick on a 4 layer board, you'll get burned.
Capacitor Expert By Day, Enginerd by night.  ||  Personal Blog: www.baldengineer.com  || Electronics Tutorials for Beginners:  www.addohms.com

majenko

The "bottom" layer is only more used because it's "traditional".  In the early days most boards were single sided.  Components were largely through-hole.  The solder, therefore, had to go on the opposite side of the board to the components.  It became logical for the components to be the top, and the copper tracks to be the bottom, as that was how it sat on the desk easiest.

Then along came more common surface mount components.  Double-sided boards were still very expensive then, so it became a mix of through-hole components on one side, and copper tracks and surface mount components on the other.  The distinction between "top" and "bottom" became blurred, as there were now components on both sides.

As board manufacturing and design technology has advanced still further, and we now have boards with many layers (4, 8, or even 16 in some cases), the whole concept of "top" and "bottom" is really quite meaningless.  It's only when the board is used in conjunction with other components, such as enclosures, other boards, human interface devices, etc, that the orientation of the board becomes apparent.

So no, there is no need (or real desire) to keep as many of your traces on a specific layer, other than for certain circumstances.  Sometimes it is important to have ground and power on certain layers, or for traces to be on specific layers for impedance or other technical reasons.

And then there's the purely aesthetic reasons.  I believe that a PCB should look good as well as function well.  It is an art form and should be designed as such.  A well designed PCB is easier to work out what is what than a spaghetti mish-mash of components.  If the PCB contents are laid out well, you don't need to manually follow the traces.  You know what each area of the board does, and you can predict where many traces should go and trace them using a DMM in continuity test mode.

pwillard

#4
Jun 25, 2012, 01:26 pm Last Edit: Jun 25, 2012, 01:29 pm by pwillard Reason: 1
A board should use traces as needed by the design.  The bottom preference is mainly "historical" as it was needed for added mechanical strength for through hole parts.  While you CAN solder non-SMD devices to the top side (aka component side) of a board, you lose a good amount of structural support (IE; think about mounting non SMD capacitors/resistors and IC's using only TOPSIDE copper).  You do gain the bonus of not having to drill any holes if you do that.  I imagine it would be easy to break a part off the board...   pulling the copper...  (Anyone for the lazy solution?)

On a DUAL sided board, it's pretty much up to the designer to do what makes the most sense.  If you grab some spare boards from around the house you will likely find that boards have a directional preference.  IE; one side will be used primarily for North/South vias while the other side will be used for East/West vias.  To me, this approach makes the most sense.

Grumpy_Mike

Quote
one side will be used primarily for North/South vias while the other side will be used for East/West vias.


Does this mean that if you rotate the board through 90 degrees then it automatically flips over?   ]:D

MarkT

For high-frequency high-gain amplification sometimes you want maximum separation of input from output components to prevent unwanted feedback from stray capacitance (this could cause oscillation).  With 4-layer boards (groundplane internal) you can place input components and tracks on the opposite side from the output components and tracks (assuming surface-mount), thus saving space.  Similarly you can keep sensitive analog circuitry away from noisy digital signals.
[ I won't respond to messages, use the forum please ]

db2db


a board, you lose a good amount of structural support (IE; think about mounting non SMD capacitors/resistors and IC's using only TOPSIDE copper).  You do gain the bonus of not having to drill any holes if you do that. ri


If I lay out a board with traces all on the top, doesn't Eagle still make a drill hole, and that hole would be plated? (and then soldered) The solder is in that hole and also has a pad the same size on the other side of the board back?  Do we need to specify that?

MarkT

PCB fabs wouldn't bother plating holes for a one-layer board - you may have to tell them is a one-layer board though(!)

Plated holes do make stronger solder joints for the through-hole components, so if you can afford two-layers you get that advantage. 

Recently there's a fashion for one-sided boards made of aluminium - the board is coated in an insulating layer and the copper traces deposited on top of that.  The advantage is that all surface mount components get good heat-sinking (and groundplane) automatically.  For simple high-power circuits (LED lighting arrays for instance) this is a nice solution.  But through-hole components cannot be used.
[ I won't respond to messages, use the forum please ]

db2db

Yes, a two sided board, with just a few traces on the back.

florinc

I don't know if others agree or not, but I always try to have the majority of the traces on the bottom, so that in case I need to "debug", I can easily follow them, cut them, re-route them etc. This would be harder when most of the traces are covered by components.

pwillard

#11
Jun 26, 2012, 01:48 pm Last Edit: Jun 26, 2012, 01:50 pm by pwillard Reason: 1
Quote
Does this mean that if you rotate the board through 90 degrees then it automatically flips over?  


Grr   haha  funny.


I make my own boards,  one sided, using the toner process... so my opinions are going to be skewed since I don't have the advantage plated through holes or as many topside traces as I want.

So, for me... I consider topside traces to be "components" or "jumpers" and assume that none of my non-jumper "holes" require electrical continuity from bottom to top.  This means that most of my board is "bottom wired" and I actually place SMD's on the bottom side as "mirrored" objects.   If I were to migrate to a 2-sided PCB for a FAB house, I would likely continue my practice of keeping nearly everything bottom side and only use topside as needed.

To me... making a PCB is like a puzzle to be solved.  I take personal pride in laying out a board that has the least possible amount of top side "jumper" wires.

If I were to design a SMD-only PCB... I would just flip TOP and BOTTOM layers.

db2db


I don't know if others agree or not, but I always try to have the majority of the traces on the bottom, so that in case I need to "debug", I can easily follow them, cut them, re-route them etc. This would be harder when most of the traces are covered by components.


This is a good argument.  I might change my plan now.

Go Up