I want a bunch of slots in my pcb, to separate out various
experimental tiles. We'd assemble and then cut bits apart. I wonder if
I can just say "slot out dumbbells with 50 mil router" instead of a
mess of dimensioning.

https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1

That would give me clean precise edges too, for edge-launch
connectors.

Gotta ask some board houses.

Last board I did this way, I sawed it up after assembly. The saw
vibration made the little EPC GaN fets jump off the board.

https://www.dropbox.com/s/8sdi4gd5o0brpnz/Z498_combo.jpg?raw=1

--

John Larkin Highland Technology, Inc

The best designs are necessarily accidental.

On Wed, 30 Jun 2021 09:03:49 -0700, jlarkin@highlandsniptechnology.com
wrote:

>
>I want a bunch of slots in my pcb, to separate out various
>experimental tiles. We'd assemble and then cut bits apart. I wonder if
>I can just say "slot out dumbbells with 50 mil router" instead of a
>mess of dimensioning.
>
>https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>
>That would give me clean precise edges too, for edge-launch
>connectors.
>
>Gotta ask some board houses.
>
>Last board I did this way, I sawed it up after assembly. The saw
>vibration made the little EPC GaN fets jump off the board.
>
>https://www.dropbox.com/s/8sdi4gd5o0brpnz/Z498_combo.jpg?raw=1

I like a combination of routed sots and V-grooves. You get smooth
edges and snap-apart boards.

Not so much dimensioning since the V-grooves have to go edge-to-edge.

--
Best regards,
Spehro Pefhany

On Wed, 30 Jun 2021 12:16:18 -0400, Spehro Pefhany
<speffSNIP@interlogDOTyou.knowwhat> wrote:

>On Wed, 30 Jun 2021 09:03:49 -0700, jlarkin@highlandsniptechnology.com
>wrote:
>
>>
>>I want a bunch of slots in my pcb, to separate out various
>>experimental tiles. We'd assemble and then cut bits apart. I wonder if
>>I can just say "slot out dumbbells with 50 mil router" instead of a
>>mess of dimensioning.
>>
>>https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>>
>>That would give me clean precise edges too, for edge-launch
>>connectors.
>>
>>Gotta ask some board houses.
>>
>>Last board I did this way, I sawed it up after assembly. The saw
>>vibration made the little EPC GaN fets jump off the board.
>>
>>https://www.dropbox.com/s/8sdi4gd5o0brpnz/Z498_combo.jpg?raw=1
>
>I like a combination of routed sots and V-grooves. You get smooth
>edges and snap-apart boards.
>
>Not so much dimensioning since the V-grooves have to go edge-to-edge.

These sundry little boards tend to not fit conveniently into a
rectangular cutout grid. When I plan to saw, it takes a bit of
thinking about the cut sequence. The dumbbells would let me put slots
anywhere.

Cutting out some residual corner webs wouldn't be a big deal for a few
prototype circuits.

My next board proto will be 4 layer on Isola Tachion 100 microwave
laminate. I'll include some test traces to measure risetimes and
impedances.

--

John Larkin Highland Technology, Inc

The best designs are necessarily accidental.

On Wednesday, June 30, 2021 at 9:03:57 AM UTC-7, jla...@highlandsniptechnology.com wrote:
> I want a bunch of slots in my pcb, to separate out various
> experimental tiles. We'd assemble and then cut bits apart. I wonder if
> I can just say "slot out dumbbells with 50 mil router" instead of a
> mess of dimensioning.
>
> https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>
> That would give me clean precise edges too, for edge-launch
> connectors.
>
> Gotta ask some board houses.

Cut rates are better if the router bit is wider than the board thickness; 2mm, 3mm, or 0.125"
are likely to be preferable. Thin cuts are slow ones with a router.

On Wed, 30 Jun 2021 14:55:11 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

>On Wednesday, June 30, 2021 at 9:03:57 AM UTC-7, jla...@highlandsniptechnology.com wrote:
>> I want a bunch of slots in my pcb, to separate out various
>> experimental tiles. We'd assemble and then cut bits apart. I wonder if
>> I can just say "slot out dumbbells with 50 mil router" instead of a
>> mess of dimensioning.
>>
>> https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>>
>> That would give me clean precise edges too, for edge-launch
>> connectors.
>>
>> Gotta ask some board houses.
>
>Cut rates are better if the router bit is wider than the board thickness; 2mm, 3mm, or 0.125"
>are likely to be preferable. Thin cuts are slow ones with a router.

I wrote up a description of the board I want to do, including the
dumbbells, the material (Isola Tachyon 100) and the stackup, and
Bonnie is sending it to a few PCB houses to get their reaction.

We'll give them the option to use Isola on the L1-L2 dielectric, and
Isola or FR4 on the others. But we have a warpage spec too.

On 1/7/21 2:03 am, jlarkin@highlandsniptechnology.com wrote:
>
> I want a bunch of slots in my pcb, to separate out various
> experimental tiles. We'd assemble and then cut bits apart. I wonder if
> I can just say "slot out dumbbells with 50 mil router" instead of a
> mess of dimensioning.
>
> https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>
> That would give me clean precise edges too, for edge-launch
> connectors.

I've done that, just leaving little corners to snap off.
2mm router is about the smallest that can cut at full-speed, so is
cheaper than smaller sizes. Add a V-groove to make them snap well. Some
folk drill a few small holes to delineate the snap line instead.

> Last board I did this way, I sawed it up after assembly. The saw
> vibration made the little EPC GaN fets jump off the board.
>
> https://www.dropbox.com/s/8sdi4gd5o0brpnz/Z498_combo.jpg?raw=1

There's a risk of flexing the board and cracking MLCC capacitors too.
They tend to invisibly fail short circuit, which can make things
interesting when you power them up.

Clifford Heath.

On Wed, 30 Jun 2021 14:55:11 -0700 (PDT), whit3rd <whit3rd@gmail.com>
wrote:

>On Wednesday, June 30, 2021 at 9:03:57 AM UTC-7, jla...@highlandsniptechnology.com wrote:
>> I want a bunch of slots in my pcb, to separate out various
>> experimental tiles. We'd assemble and then cut bits apart. I wonder if
>> I can just say "slot out dumbbells with 50 mil router" instead of a
>> mess of dimensioning.
>>
>> https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>>
>> That would give me clean precise edges too, for edge-launch
>> connectors.
>>
>> Gotta ask some board houses.
>
>Cut rates are better if the router bit is wider than the board thickness; 2mm, 3mm, or 0.125"
>are likely to be preferable. Thin cuts are slow ones with a router.

For some reason 2.3mm comes to mind. Maybe that's the width of the
path, not the bit diameter. At least in metals you don't want to just
plow through, you want to keep the finished cut on one side of the
cutter (conventional or climb).
--
Best regards,
Spehro Pefhany

On Thu, 01 Jul 2021 16:30:51 -0400, Spehro Pefhany
<speffSNIP@interlogDOTyou.knowwhat> wrote:

>On Wed, 30 Jun 2021 14:55:11 -0700 (PDT), whit3rd <whit3rd@gmail.com>
>wrote:
>
>>On Wednesday, June 30, 2021 at 9:03:57 AM UTC-7, jla...@highlandsniptechnology.com wrote:
>>> I want a bunch of slots in my pcb, to separate out various
>>> experimental tiles. We'd assemble and then cut bits apart. I wonder if
>>> I can just say "slot out dumbbells with 50 mil router" instead of a
>>> mess of dimensioning.
>>>
>>> https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>>>
>>> That would give me clean precise edges too, for edge-launch
>>> connectors.
>>>
>>> Gotta ask some board houses.
>>
>>Cut rates are better if the router bit is wider than the board thickness; 2mm, 3mm, or 0.125"
>>are likely to be preferable. Thin cuts are slow ones with a router.
>
>For some reason 2.3mm comes to mind. Maybe that's the width of the
>path, not the bit diameter. At least in metals you don't want to just
>plow through, you want to keep the finished cut on one side of the
>cutter (conventional or climb).

I'm using pretty-good microwave-level edge-launch SMA connectors, so I
need the board edge to be very smooth and accurate. So I can't
v-score. Just 10 mils or so of connector to edge error affects the
impedance.

On Thu, 01 Jul 2021 15:32:26 -0700, John Larkin
<jlarkin@highland_atwork_technology.com> wrote:

>On Thu, 01 Jul 2021 16:30:51 -0400, Spehro Pefhany
><speffSNIP@interlogDOTyou.knowwhat> wrote:
>
>>On Wed, 30 Jun 2021 14:55:11 -0700 (PDT), whit3rd <whit3rd@gmail.com>
>>wrote:
>>
>>>On Wednesday, June 30, 2021 at 9:03:57 AM UTC-7, jla...@highlandsniptechnology.com wrote:
>>>> I want a bunch of slots in my pcb, to separate out various
>>>> experimental tiles. We'd assemble and then cut bits apart. I wonder if
>>>> I can just say "slot out dumbbells with 50 mil router" instead of a
>>>> mess of dimensioning.
>>>>
>>>> https://www.dropbox.com/s/bmopwenwccp51jv/Dumbbells.jpg?raw=1
>>>>
>>>> That would give me clean precise edges too, for edge-launch
>>>> connectors.
>>>>
>>>> Gotta ask some board houses.
>>>
>>>Cut rates are better if the router bit is wider than the board thickness; 2mm, 3mm, or 0.125"
>>>are likely to be preferable. Thin cuts are slow ones with a router.
>>
>>For some reason 2.3mm comes to mind. Maybe that's the width of the
>>path, not the bit diameter. At least in metals you don't want to just
>>plow through, you want to keep the finished cut on one side of the
>>cutter (conventional or climb).
>
>I'm using pretty-good microwave-level edge-launch SMA connectors, so I
>need the board edge to be very smooth and accurate. So I can't
>v-score. Just 10 mils or so of connector to edge error affects the
>impedance.

Mixing the two reduces (but doesn't eliminate) the constraints:

https://i.imgur.com/9VbfPkB.png

Here the connector is along the smooth edge and the V-grooves hold the
PCB to the tooling strips.

--
Best regards,
Spehro Pefhany

On 01/07/2021 9:30 pm, Spehro Pefhany wrote:
> For some reason 2.3mm comes to mind. Maybe that's the width of the
> path, not the bit diameter. At least in metals you don't want to just
> plow through, you want to keep the finished cut on one side of the
> cutter (conventional or climb).
>

I think 2.4mm (3/32ins) is a common default cutter size.

piglet