Gonna need a net list and footprint. Cool to use the footprint from the adapter we made?

This also will take a while. Btw. But I can get started soon.

You can take Vcc and GND locations from any 386 pinout datasheet, e.g. Intel's http://datasheets.chipdb.org/Intel/x86/386/da … ts/23163011.pdf

However, all other pins, even those which are N/C should be rotated accordingly. This is because other chips, like Cyrix, use these N/C pins.

alpha sketch 1

how are we doing this? smd? thru hole with slight offset?

verify socket on right has correct vcc/gnd plz. are you able to open and see nets? ( I can check the link you sent later. no time now.

sorry. only a few fleeting moments to work on this. will return later. 😀

I'm unable to open those KiCAD v6 files. Is there any feature of KiCAD v6 that you need to use which is not available in v4 ?

Three options:

1)
One would be just to do through-hole with a slight offset.

2)
Other option is psuedo-surface mount. When I went looking at some of my interposers which use surface mounted pins, they like to use a 2.0 mm thick PCB. I'm guessing they have closed bottom pits for the pin's vias, which parhaps go down 1 mm or 0.5 mm? Not sure what they are called, perhaps "false vias"? I wouldn't be opposed to trying out this false via method if it can be done in 4 layer or less. You could also try using these false vias for the female end of the adapter as well, but it would mean that we'd have to use row headers for the female side as well, rather than a proper PGA132 socket. No big deal I think.

3)
Use all through-hole with 2 PCBs connected together.

I'm thinking #2 might be the easiest and cheapest if these false, or sunken vias can be realised.

For the footprint, I was hoping it to be no larger than a standard socket. Some of my sockets are 35.5mm x 35.5mm, while others are 37mmx37mm. So I think 37 mm x 37 mm is a good starting point. If it starts breaking the ability for 4-layer, perhaps start to increase the footprint. If the footprint gets too big, then more layers needed.

They are called blind vias.

I won’t be in front of a computer with kicad again for some days to come. But I will return and will be on here in meanwhile from phone 😀

I am unsure about using v4.0 it seems that you would want the pcb made in the newest version? Not being able to open it on your end, perhaps you have a junk computer you are willing to load up with windows 10/11?

Anyway. Kicad can make a blind via, which is super cool. The problem is that jlc can’t manufacture it.

If you find a manufacturer who can let me know their board setup specifications.

But I think it is a mistake to make the board so specialized. Even though the blind vias is a wonderful idea.

I think SMD or offset is the best option.

This is going to require a lot of layers with offset. Again, we need to do a bunch of drafts to research best routing and see that it is even possible. But with smd, it should be a lot more organized.

It sucks, with smd we are still going to need vias. Next to the smd pads. And with offset we will have two thru hole pads. Which will take up a lot of space.

You keep bringing me bowls of spaghetti to organize 😂🍝

Can we upload high res pics of oem versions to analize?

They cannot do blind vias, but they do *tented* vias - that is vias with soldermask on top. This may be relevant instead of trying to cover them with silkscreen like yall were trying to do with the other thing

Most people, myself included, seem to still be using kicad 4. New thing bad 😁

I don't use Windows any longer for day-to-day use. My retro machines have up to XP on them. I don't have a Windows 10 installation anywhere, nor do I own a Windows 10 licence. I also don't think I have any machine that will reliably run Windows 10. I had hoped that Windows moving away from Windows 7 would have forced more users to transition to Linux. If you go through that 3 month pain barrier with learning Linux, customising, and getting your hardware working smoothly, there's a strong chance you'd be glad that you transitioned.

If there's no tangible benefit of KiCAD v6 for these hobby designs, perhaps you can have a parallel installation with KiCAD v4. Others, aside from myself, will benefit from this.

OSHPark also doesn't support blind vias, but PCBWay does. I decided to do a price comparison of JLCPCB vs. PCBWAY for the SXL2 interposer board. WOW! JLCPCB $86 for 10, PCBWAY $368, with a note: "Tip: Final price can be negotiated". LOL! PCBWay operates their pricing scheme like eBay sellers now.

I don't think a surface mount is the best approach because I'm not confident I'd be able to line up the rows of machine pins sufficiently. Slight offsets of either position or perpendicular angle will create cascading problems for final insertion. Also, prying the device out of the socket will likely break the solder joints.

I will send OSHPark an inquiry to see if they have eventual plans to support blind vias.

EDIT: I have no OEM samples for this adapter. Another user mentioned to me once that Evergreen made these rotator boards, but I haven't seen any photos online.

Okay, it would help greatly to see how it has been done before. Otherwise we (well, I) face a long process of trying many different non precise approaches before finally sitting down and actually building the design that worked best in trial runs.

It seems like there may be some sort of pattern that can be used to do this. Where like every forth row routes on a specific layer or something going around radially

Idea..

It might be easier to do 180* than it is to do 90*

I can do some experiments

Will 180 be unhelpful?

Sorry, I don't know anybody who owns the Evergreen rotator boards. I don't think an image ever made it to the internet.

All three orientations are equally important because there was no standard placement of the PGA132 on motherboards, nor was there a standard orientation for where pin 1 went.

For one such motherboard I have setup at present, the 180 degree orientation is the one it needs. However, I have run into this issue on a few motherboards.

Very well, our mission is clear.

Not one, but three…… boards.. 😳

Wonder if two boards could be set up with a rivet in the middle and a target pattern between. Rotate to desired angle, heat up to solder bga style.

Would probably work with a low number connections but I bet too many on this. 🙁

I think needing BGA soldering would loose 99% of the customer base.

I think 180 degrees would be a good starting point.

Agreed.

I’ve been pondering this in my head. I think this may end up being even more of a mess than your other interposer.

Imagine this image but all the legs of a 132 package. And every via goes to all layers. So they all have to be routed around. 6 layers means possibly 6 vias for every signal if we are to use direct twist routing. And somehow it has to all be flat.. 😳

This may get easier if able to route around the socket, which I am sure we can. But dude, this is even more nuts than your last one. 😂

I can give it a try though. Nothing like a good ole puzzle.

Desperately seeking ideas though.

Btw, planes may not even be possible with that many signals packed all over the place.

I’ll know more once I start throwing traces down.

My idea is to connect each pin to a circle, and thus each circle -from small to large- allows it to turno 360°.
Like a series from the center to the edge.

It'll cost a lot to produce. Imagine a vinyl, each pin has its groove and the interposer above makes connection to it.

Kind a like my rotating BGA idea and honestly, I do think that would work if we can make the circles small enough but as for physically soldering that forget about it

Have you thought about making this out of bodge wires about 1 inch tall then filling the inside with epoxy?

Honestly, I think that would be better than using a PCB for this specific problem.

The wires would be all spread out and insulated and isolated whereas if we try to cram it all into a PCP they’re just going to basically almost be touching every other one

Yeah, we really might have to take this into the third dimension 🤣

I’ll keep brainstorming

Sphere478 wrote on 2022-11-06, 01:36:

Kind a like my rotating BGA idea and honestly, I do think that would work if we can make the circles small enough but as for phy […]
Show full quote

Kind a like my rotating BGA idea and honestly, I do think that would work if we can make the circles small enough but as for physically soldering that forget about it

Have you thought about making this out of badge wires about 1 inch tall then filling the inside with epoxy?

Honestly, I think that would be better than using a PCB for this specific problem.

The wires would be all spread out and insulated and isolated whereas if we try to cram it all into a PCP they’re just going to basically almost be touching every other one

Yeah, we really might have to take this into the third dimension 🤣

Yes, it could be done but the costs would be through the roof.
I saw that in some 360° machines, with way fewer contacts.

Feipoa is asking for the impossible.
Maybe a Escher?

Nexxen wrote on 2022-11-06, 01:55:

Maybe a Escher?

🤔

Don’t look into the truth tortoise’s eyes!!