Electronics > Projects, Designs, and Technical Stuff
PCB Thickness: What to Expect
EPAIII:
I am working on a design where a combination of the normal electronic concerns as well as some geometric/mechanical factors must be combined. And I want to use a couple of PCBs in such a combined role. I am trying to work my way through the various design factors that the PCB fab houses publish and I am finding that it is somewhat difficult to put them all together to arrive at the numbers that I need.
The design is only in the very early stages so I do not have full details. I will try to provide what I can. At this point I am thinking of single layer PCBs about 1" or 25mm in diameter with a number of NPTHs in it. There are no plated holes and no active components. Until I see a reason for another board thickness, I am thinking about 1/16" or 1.6mm. And the design could easily evolve into the need for a larger diameter. That is not hard and fast.
One thing that I am considering is pressing some 4mm, ball bearing balls into several of these NPTHs. They would reach almost to the opposite side which has no copper and then be soldered to the copper donuts surrounding those holes. These balls would serve as the contacts of several switches. A problem arises because at least some fab houses require a copper free ring surrounding a NPTH before the copper of the pad begins. For instance, JLCPCB says they will leave a ring of bare FR-4 between 0.2 and 0.25 wide. I worry about getting the solder to jump that gap.
Where things get hard to design is knowing the thickness of the PCB that I can expect so I can determine just how much of that copper less ring the ball would need to force itself into in order to reach the copper donut pad.
I do understand tolerances. What I do not know is what the thickness dimension of a PCB actually refers to when they say, for instance, 1.6mm +/-10%. That translates to a range of 1.44 to 1.76 mm. Fine and good. But what is included in those numbers. The FR-4? Certainly. The copper? Probably. But what about the plating? Is that included? I see no specs for the thickness of the plating. And is there anything else to consider?
And then that 10%. Is that much really encountered often? Or would 99.9% of all boards made be a lot tighter than that? And the 1.6 mm base number. Is that really at the center of the tolerance range? Or do most boards measure near one end or the other of that range? Or do they actually use the original 1/16" number for the target value? And of course, there is a tolerance on the hole sizes which also figures into this. I really need to figure the worst case situation.
In other words, what can I reasonably expect?
Any enlightenment here would be greatly appreciated.
Georgy.Moshkin:
When I order 1.0mm thick boards, thickness is always 1.0mm (with copper on both sides) and 0.8mm for no copper areas. That's what my caliper says. It's quite strange, because I doubt that copper is 3oz (0.1mm) by default for my orders, it is possible probably thinner. I'll look into this next time I order some PCBs
I like 1.6mm boards. 1.0mm boards are quite rigid even for sizes around 50x50mm.. I use 0.6mm and 1.0mm for microwave circuits, because traces become too wide for the same impedance and may behave strangely.
For your project, experimentation would probably be the best option. I think that it is better to use plated holes and add circles made of VIAs around them. It may improve mechanical strength, especially if solder fills the VIAs too. So why do you try to avoid plated holes?
CountChocula:
My suggestion would also be to try and figure this out experimentally. You could design a test board that all of the elements you need—in fact, ideally, you'd want several variations of each element, so that you can see how the fab treats different setups. Pretty much any of the popular DYI services will make a board for a nominal fee, and so this feels like an inexpensive way to make sure that you're going to get what you need before you commit to the time it takes to design a final board.
—CC
lo_volt:
I'll suggest that you contact the board house and talk to them about tolerances and what to expect on a finished board. JLCPCB, OSHPark and most Asian suppliers sell boards cheaply because they use a standard and very common board spec. This allows them to group different boards for different customers when they fab the boards. There are domestic board houses that will work with your requirements to build the boards you need. They will most assuredly cost more but if you need something out of the ordinary, that may be your best option.
ajb:
--- Quote from: EPAIII on September 27, 2023, 07:13:59 am ---One thing that I am considering is pressing some 4mm, ball bearing balls into several of these NPTHs. They would reach almost to the opposite side which has no copper and then be soldered to the copper donuts surrounding those holes. These balls would serve as the contacts of several switches.
--- End quote ---
CNC touch probe?
--- Quote ---A problem arises because at least some fab houses require a copper free ring surrounding a NPTH before the copper of the pad begins. For instance, JLCPCB says they will leave a ring of bare FR-4 between 0.2 and 0.25 wide. I worry about getting the solder to jump that gap.
--- End quote ---
Sometimes it's helpful to tape a pair of calipers and set them to a dimension and look at it -- or if you have one of those PCB rulers with sample trace/space widths, that would work too -- for an intuitive sense of how something will work. 0.2-0.25mm is pretty small, I don't think you need to worry too much about bridging the gap, especially since the side of the ball and the top surface of the board form a slightly acute angle, which should make for a fairly nice fillet situation. If you're planning to stencil paste on, then you probably need to design the stencil with apertures right up to (or even slightly crossing into) the hole to ensure the paste is in contact with the ball before reflow, and you might also have to adjust your process to account for the thermal mass of the balls.
--- Quote ---Where things get hard to design is knowing the thickness of the PCB that I can expect so I can determine just how much of that copper less ring the ball would need to force itself into in order to reach the copper donut pad.
--- End quote ---
I don't think pressing the balls into the PCB is likely to work. Normally things that are press-fit into FR4 have sharp edges that can dig into the material, while the gaps between the edges provide space for the displaced material to flow into. The balls don't have that feature. Per my quick sketch here, the wedge angle of the ball at the top surface is 14-33° for 1.44-1.76mm of PCB thickness, which I suspect is too large for the ball to actually be retained once pressed in but too small to prevent upsetting the edge of the FR4, which will mess with your pad and make things harder to solder. You could try asking the fabricator to leave copper right up to the edge of the hole--often fabricators will do things outside of their normal spec as long as you accept that the results may suck and/or pay them extra to deal with it.
--- Quote ---But what is included in those numbers. The FR-4? Certainly. The copper? Probably. But what about the plating? Is that included? I see no specs for the thickness of the plating.
--- End quote ---
If by "plating" you mean the copper buildup, that's accounted for in the copper thickness. If you mean the finish (HASL, ENIG, or whatever), that varies with the specific finish. HASL has more significant variation in thickness over the board, so presumably has a much larger overall thickness tolerance as well. ENIG is much more consistent, which is why it's preferred for leadless parts. OSP I guess would be effectively zero thickness after soldering.
--- Quote ---And then that 10%. Is that much really encountered often? Or would 99.9% of all boards made be a lot tighter than that? And the 1.6 mm base number. Is that really at the center of the tolerance range? Or do most boards measure near one end or the other of that range? Or do they actually use the original 1/16" number for the target value?
--- End quote ---
Generally, you're going to tune any process to hit the middle of your tolerance band, so that you have maximum margin for error in both directions. I would expect any decent, established PCB fabricator to do the same, and furnish a board that's pretty close to the nominal thickness pretty much every time. Generally the final deviation is assumed to have a normal statistical distribution, with more parts closer to nominal and fewer parts further away. Analyzing tolerance stackups can get fairly complicated, but here's a basic overview of the math: https://www.smlease.com/entries/tolerance/tolerance-stackup-analysis/ (Found in a cursory google search, there are many more resources out there on the subject).
Depending on your application, it may be better to allow for final adjustment of the overall assembly to compensate for PCB/assembly tolerances, rather than trying to control all of those upstream tolerances. If you are doing something like a CNC touch probe, the standard practice for those is to include centering adjustment screws in the probe base, which compensate for any deviation in the probe body/mechanism as well as for runout in the probe tip itself whenever that's replaced.
Navigation
[0] Message Index
[#] Next page
Go to full version