Assuming I am ok with the extra cost of having parts assembled on the backside of my PCB, does it make sense to basically place all the main parts on the top side and then strategically place all the caps, resistors, inductors, etc (that need to be close to the pins they are associated with) on the backside using vias?
\$\begingroup\$
\$\endgroup\$
3
-
\$\begingroup\$ Not necessarily. It may be worse or irrelevant but it depends on the PCB design what is required from the circuit. Without such info what do these caps, resistors and inductors do, it can't be answered. If it is an RF board, you can't do that. \$\endgroup\$– JustmeCommented Aug 14, 2023 at 18:18
-
\$\begingroup\$ I'm afraid this isn't a very meaningful question -- assuming you are "ok with the extra cost", you can come up with quite outlandish assemblies that are possible, but highly impractical (not competitive with other methods in production). What kinds of components would be placed, what sizes, what packaging? Does the board have any mechanical constraints (surrounding geometry that restricts component placement)? \$\endgroup\$– Tim WilliamsCommented Aug 14, 2023 at 18:20
-
\$\begingroup\$ For example, I am placing an STM32H747XIHx MCU which is a BGA component. Many of the 3V3 pins are buried into the middle of the BGA but need bypass capacitors. If I mount them on the back side they would be much closer to the pin than routing it out to the edge of the BGA part? \$\endgroup\$– Kevin McQuownCommented Aug 14, 2023 at 18:58
Add a comment
|
4 Answers
\$\begingroup\$
\$\endgroup\$
It's pretty common to do that around BGA parts to get the bypass caps close to the pads.
I would be inclined to limit "back side" parts to a limited variety of mostly small passive parts such as bypass capacitors and maybe some resistors or networks. In particular, I would avoid parts that would require glue.
answered Aug 14, 2023 at 19:59
\$\begingroup\$
\$\endgroup\$
It will make troubleshooting quite a bit harder, assuming that's something you expect to be doing. There will be a step change in impedance associated with each via; whether this is a problem will depend on your signal. They will also nibble out whatever planes you have, possibly creating troublesome slots that could radiate or even islands which could interfere with heat dissipation. Unless there's a compelling reason (called out by manufacturer's datasheet, space constraints, etc), I don't think it's wise to make a special effort to put as many passives as possible on the bottom of the board.
\$\begingroup\$
\$\endgroup\$
I would say routing that way wouldn't be very efficient. You don't want high-speed signal to change plane for no reason. If you have a strict no resistor/Cap on top layer, that would imply, many signal will cross for no good reason. That would lead to a lot more vias and a much weaker plane. Also, decoupling capacitor are most effective when they are as close as possible to the pins. So placing a via in between is counter productive (if you are not obligated to, of course).
Unless you have a reason not to place SMD on top, I wouldn't do that!
\$\begingroup\$
\$\endgroup\$
That is one strategy. You will need to balance the higher cost/complexity with the limited board space if placing components on only one side.
You also need to consider the height of bottom side components, (in case you have limited underside space for mounting of the board).
Another issue would be weight/mass/shape of the component, (bottom side parts are often held by small glue dots prior to the soldering step, heavier or irregular shaped parts could tear off the board prior to the initial or secondary soldering steps).
Potential requirements of troubleshooting, calibration, extra vias, less space for ground/power plane copper, and connector insertion/removal are some additional factors.
.
