Re: modifiable backplane with sockets?

On 5/28/2025 2:55 PM, Christopher Howard wrote:

How are you planning on supporting these cards? What sort of masses
involved? Any connections to the opposite end of the card (that would
add to the stress on such a "socket")?

 I was imagining that they would be fairly small cards, something like
5mm wide by 6 or 7 mm long, with lightweight components only on the card
— small ICs, resistors, and such like. Certainly much smaller and
lighter than something like a graphics card. Nothing on the opposite end
of the card.

Yikes.  And you want them removable/repluggable?  The connectors may prove to
be bulkier than the boards!  (the example I provided likely is!).
If you were NOT expecting to disassemble/reassemble, I might have just
suggested DIP parts carriers plugged into WW DIP sockets -- which could
even exploit a 2-dimensional layout instead of "all in a row".
But, even using machined-pin carriers, these will be pretty fragile
for reuse.

My original thought was to have the cables to the patch panel(s) hook
directly into a socket as well, but maybe I could have a shorter cable
going from that to some kind of plug in port on the chassis.
 

Are you certain that (analog!) signal integrity won't be an issue with
this approach? The frequencies and signal levels involved (as you
likely will be severely constrained as to HOW a signal can be routed
from "socket A" to "socket F")

 I'm working always at low frequencies, like 100 Hz or less. The
simulations usually display on an oscilloscope at around 1 second per
scan, or slower.

OK.  It wasn't clear from your initial post.

Are your modules going to use printed wiring? I.e., does the "plug"
that mates to these sockets mount *onto* the card? Can you use
fingers, instead (more costly to manufacture; why pay a PCB house to
plate them when you can purchase a "plug/socket" that is already
plated)

 The plug would mount onto the card, unless I was using just contacts
that slide in. I am open minded to ideas.
 I don't really want to use printed wiring, preferring to
experiment/design/build as I go. But I'm not really sure a practical way
to do that outside of a breadboard. Maybe I can come up with some kind
of generic printed board that gives me some flexibility on how I solder
on ICs and wires.

If you are using perfboard, you will be constrained as to the locations
and placements of devices (based on the "holes").  If you can stick
with DIP packages, this could be OK.  SMT might be a stretch to fabricate
like this (esp if you didn't want to be forever "patching" it)

How durable do you want the finished assembly to be? I.e., are you
just trying to use it to mock-up designs? Or, are you intending to
*deploy* a system thusly assembled? How fussy can you be in assembling
(and disassembling) such a configuration -- how durable do the
connectors need to be?

 It would just sit on my desk. I'd need to be able to slide in/out or
(dis)connect the module cards without anything breaking. After assembly,
I would mostly be interacting with just the patch panel, not the cards,
but I'd want to be able to take cards out to tweak them, or replace
them.

How about splaying the backplane out "like a circuit board" instead of
like a set of card slots?  Imagine "postage stamp" modules arragned
side by side, row after row (i.e., in a small grid) such that you
could see and access all of the components WHILE it is assembled.
Flip it over to do point to point wiring, as required.
There, you might be able to come up with a different mechanincal
design for the modules -- one that mimics a DIP (with a row
of pins on each "end") and plugs into a set of contacts
similarly arranged.
This does the "major, intermodule" wiring on the backplane and still
leaves you the freedom to tweak any interconnections WITHIN a
module AND OBSERVE THE ENTIRE CIRCUIT WHILE POWERED.
Size is just a tradeoff between how delicate you want the modules
to be and the number thereof.

wouldn't need to be part of a "backplane PCB" but could just be
fastened to a support member spaced appropriately from its neighbor(s)

 Ah, okay, so I could just screw that onto some support structure, then
wire wrap to those long posts. Assuming the posts are strong enough to
endure wire wrapping.

The real downside risk would be if *a* post snapped while the others
were intact.  You'd have to remove the entire connector just to
restore that post's functionality.
[If the entire "site" was unused when the break happened, you just
replace the whole connector.]