Tiny PCBs - score, route, or ?

All good tips, thanks.

It sounds like tabbed routing might be the best approach for the smallest scale (which is important here), allowing us to have parts closer to the edge without the stress. Maybe with a throwaway strip between the parts for the indent technique you described.

Or, I'll tryout your technique of v-scoring with a router before assembly, if we can't find a shop that'll do it for cheap enough.

Thanks, Richard

Olimex

will panelize and separate boards at no extra cost. They start with standard 160mm x 100mm boards (single- or double-sided) and make as many of your design as will fit, for a single fixed price. For instance, I recently had them build a 1" x 1.5" design; that works out to 25 boards, for a total of about $43 including shipping. (The price varies slightly depending on the presence of nonstandard drill sizes.) I'm pretty sure I've had them do boards as small as yours, in the past.

The shipping is a bit slow because they come from Bulgaria, and the quality is not quite as nice as what I've had done locally, but it's well within spec, and the price is very hard to beat.

I have not yet found anyone else who will panelize for free.

On the original topic, though: you can use tin snips to cut fiberglas board. It helps if you warm the board up with a heat gun first.

I recently had some small boards made that used a combination of V scoring (done with the router, with a pointed bit) for straight edges between the boards and side rails (to hold the boards together for parts placement and soldering), and routing for the corners and curves and between boards in the row. This left rough edges only in the middle of 2 straight sides per board, that I clean up by sliding the broken board along a sheet of silicon carbide paper. Where the routed slot meets a V groove, the V groove aligns with the board side of the routed slot. So I sand till I can't see where the broken edge meets the routed edge and I have good dimensions.

On another board that was designed by someone else, for us, they used the routed with tabs, but curved the routed line inward about a half slot width on each side of the tab so the tab broke just below flush with the board edge. Of course, this means that it broke sticking way out on the other side, but this was not another copy of the board, but a temporary support rail that was discarded.

I think you need about a diameter of board between the holes.

It takes a carbide tipped blade and makes a lot of dangerous dust.

For small runs, you can use a carbide tipped hook knife to pull a V groove out of the board along a straight edge guide.

If you break boards apart, either along V scoring or drilled holes, it is a good idea to keep surface mount parts a few board thicknesses away from the break line, to reduce the chance of pulling the metalization off the parts or cracking them in half. It is also a good idea to have the resistors and capacitors along the edge run with both pads equally far from the edge, instead of with one pad along the edge and one away from it, for the same reason.

Sacrificial connectionstrips are a good way to work with small boards. You can have the board house route most of the board outlines, with a web of disposible strips connected just enough places to hold the little borads well enough for assembly.

Our board house does that as part of the outline routing option. It is just an alternate router bit.

Walter, thanks.

Actually Olimex is the most promising so far. I'd mailed them for details, but they are closed for the holiday season. (PCB Pool also offers this, but their prices are outrageous.)

Olimex will depanelize, but I really don't want a bag of tiny PCBs in the mail - they'll be very hard to assemble. I'm hoping that if I provide the routing template as a layer file, they'll do tabbed routing. How did they deliver your tiny boards?

The mailing is no problem; I can be patient. :-)

Thanks for the pointer on tin snips; I'd read about that one too, but it seems very unlikely with a populated board (lots of flex, it'd seem).

For the tolerances we need, having them pre-tabbed/scored by machine is looking to be the only practical option, with in-house pre-scoring with a v-bit running a distant second place.

Thanks, Richard

V-groove would be the way to go. Very little waste (you do have to keep traces away from the very edge of the board) and you can assemble and perhaps even test the little b*ggers before you have to break them apart. No tooling, and only a tiny cost per board (perhaps less than the cost of the PCB area otherwise lost to kerf).

Shear is theoretically possible on an assembled SMT-only board but requires a lot of unused space around the cut on the component side for the clamp to come down.

The only way I'd consider a saw is if I could find an appropriate 'table saw' type configuration with a $$ solid carbide slitting saw blade. Something like this:

I think there are purpose-made depanelizing saws, but I've never had occasion to use one.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Something's impelling me to say that, if you go the V-score route, don't break them apart with only your fingers and thumbs - find some kind of fulcrum or blade or something to set them on and just push down.

Probably pretty obvious, now that I think about it. Oh, well. :-)

Thanks, Rich

Fibreglass board will wreck a bandsaw blade very quickly...

In production we stuff the boards while still panelized. They are scored. We then use a "pizza cutter", which is a rotary diamond saw on a fixture that can slide back and forth along rails, to depanelize along the score lines.

As a bag of tiny PCBs :-)

Olimex's edge tolerances are not that great. In fact that's actually my main complaint with them - the edges are pretty rough.

Budgie's comment about Futurlec was pretty interesting. I've used them as a parts supplier (though their online catalog is awful, at least last I checked) but I had no idea they did PCB manufacturing. Have to check that out - might be a bit quicker than Olimex.

Don't forget the old-fashion way. In case of single side boards, I use Press n' peel (the blue stuff) buy a paper phenolic copper clad 12x12 for about $12.00. Iron on the Press n' Peel. Put in in the old Ferric Chloride, and etch the boards myself. You can score the paper phenolic boards with a stanley knife and break clean. Gives you 144 sqare inces (a lot of small boards). I just did 28 small controller boards last night .75" x 1.5" in an hr time

• .00 board. For low power boards, I still like the phenolic copper clad better, you can cut it drill it, file it, and your toold stay sharp forever. Never lost a drill bit on these! see
  formatting link

Stefan V

Good info, thanks. These guys are local too. I'll keep that in mind in case this can be one-sided (i.e., larger). As-is, it'll need 2-sided with PTH.

Thanks, Richard

Yeah, I hadn't thought about the issue of the dust generated during cutting too. By the time I rig this, it'll be cheaper / safer / easier to just pay the board house.

We're definitely planning to assemble and test on the larger board, then snap off for packaging. What drives your suggestion for V-score over tabbed routing? I don't have enough data to weigh cost yet, but is there much difference in the waste? (Routers seem to be in the 0.12" range.)

As for unused space around the perimeter, even reserving a 0.1" border would take a huge bite out of the yield. So anything that minimizes the perimeter space will be welcomed.

Do you know board houses that'll do v-score or tab routing for proto-volume, and a reasonable price? I'm working through my list slowly (it's hard to find this detail info), and Olimex is the closest thing to reasonable so far. Volume is a problem, because at this size

2-3 boards makes a huge batch of devices, and the one-time fees make the per-piece costs high.

Thanks, Richard

I'm planning to use standard thickness, which IIRC is 0.062". Good food for thought.

Yes, I'd very much like a sketch, thanks.

After all these comments and some noodling, it looks like tabbed routing is a strong option, maybe with the indent technique mentioned by John Popelish. Instead of a whole strip of throwaway, see if I can make the tab to snap on both sides, recessed from the edge to hide the jaggies. Separate the first break by hand, and remove the second stub with pliers. Like this:

| -ooo- #1 | |-----/ ||| \\-----| tab |-----\\ ||| /-----| | -ooo- #2 |

The question then becomes how small to make the tab and how close to space the holes in the tab (balancing ease of snap vs. sag during assembly), which I expect will be a matter of experimenting. I imagine with parts this small you can't go less than one tab per side, unless I'm missing something.

And of course, where to get these made in small proto quantities...

Thanks! Richard

Good pointer - they have very good pricing ($99 for 200 pcs)! I wasn't aware they did PCBs. Do you find their routing tolerances are good?

boards

Yes, this is SMT on both sides, and I expect it'll be much easier to paste and pre-heat / solder as a larger panel. One-offs could be managed, but panelizing seemed cleaner.

Have you run into problems de-panelizing, or just not interested since you have a good one-off technique? How is your holder constructed?

Thanks, Richard

There is basically *no* waste with V-groove, provided you don't intend on putting parts and traces right out to the edge. 60 mils is usually enough. I have not tried it with >2 layers, you might need a bit more for that.

Most moderate/small volume PCB suppliers will do it for a nominal charge. Not so many prototype companies.

Best regards, Spehro Pefhany

--
"it\'s the network..."                          "The Journey is the reward"
speff@interlog.com             Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog  Info for designers:  http://www.speff.com

Well, I have to give kudos to Futurlec - I dropped them a note and already got a reply. They will do v-scoring instead of separation, at no additional charge. That looks like a winner - thanks!

Yep. One of the reasons I'm looking at panelizing - trying to work through a process that'll be easier to farm out when the volume supports it.

Cheers, Richard

[snippage]

V-scoring is nice, but SMT parts near the score lines can be damaged, especially if the board thickness is 0.031" or less.

I like using tabs with three 20 mil drill holes in each tab along the board edge. I can post a sketch of the tab w/ drill holes if you want. Easy to break off and the jaggies aren't too bad. Easily touched up with a bit of sandpaper. Hmm, perhaps v-score the tabs.

--
Mark

Just add that to the cost of cutting the boards apart. Don't use a big (12" or 14") bandsaw, get one of the $89 8" tabletop models, so you can use a thinner blade that will cut a smaller kerf. Also, consider a scroll saw, also a very small kerf, but perhaps not as fast. Also, in any of these cutting methods, use a dust collector, and don't breathe anything put into the air that the dust collector might not get.