The shim:
a) make one shim as you've made it, but you'd want to reduce the size of the diamond pads to match the new rotator board

b) make one shim with regulator metalic through holes in place of the pads. No pads.

Sphere478 wrote on 2023-01-03, 22:52:

It might be possible to do a offset thru hole. I don’t find it as elegant in this case though. And it would definitely be a full redesign.

I think it is best to create a part so more people can assemble it and with a greater degree of success. Elegance would only come afterwards in opinion. I still have my doubts as to my ability to assemble this.

Sphere478 wrote on 2023-01-03, 22:52:

It would be even more cramped on the inner layers with the larger vias(now being larger plated thru holes for pins) also physically larger obviously.

The least space consuming would be to have the solder pads as the vias on the existing design.

Sphere478 wrote on 2023-01-03, 22:52:

The only thing we can do better I think would be more layers. Ground planes are easy. […]
Show full quote

The only thing we can do better I think would be more layers. Ground planes are easy.

I could go 8 layer and spread out the doubled up traces. That might help. But going 8 layer to add ground doesn’t make a whole lot of sense. We would want more like 12 layers.

I think this 6 layer smd design is the best of cheap, organized, and compact.

It’s not great for signal interference though.

It looks like 6-layer and 8-layer will be the same cost, so doesn't matter.

Sphere478 wrote on 2023-01-03, 22:52:

I don’t think there is going to be a 180* version.

Here is why: It will have to have a high layer count or a bunch of uneven random spaghetti traces everywhere like shown earlier in thread.

Maybe it will be doable with pads in place of the vias. Perhaps look into auto-routing?

Sphere478 wrote on 2023-01-03, 22:52:

It makes sense to simply try and bga stack this. It will give more distance between traces and cost will be low. Since the orders are already 5 per order you already get 2 180* versions per order.

I don't follow. The BGA stack you've presented looks to be a way to make blind vias. I'm not sure how this alters trace distance. You must be referring to another BGA stacking system. I think you mean to stack two of the existing 90 degree rotator boards together w/BGA to make a 180 degree. This is similar to stacking two assembled 90 degree units, but with less PGA conductor length involved. I have strong doubts I will be able to get this BGA assembly to solder properly. I would think that doing a direct re-route on the rotator board for 180 degree would yield the shortest trace lengths, at the expense of more time involvement.

files

layers

shims

plan for 90* is:

socket
Shim
90 degree rotator
shim
pins

-

plan for 180* is:

socket
Shim
90 degree rotator
90 degree rotator
shim
pins

If this doesn't work we add ground planes and go 12 layer.

testing for shorts to via is pretty easy pad in same location top and bottom will have continuity if short.
if you want to test for continuity to pads around specific pad that can be done. basically you should find no continuity on one side of the socket to any other pin on that side. only from one side to the other 90 or 180 degrees offset for every pin.

plan for 270* is:

socket
Shim
90 degree counter clockwise rotator which I have yet to make.
shim
pins

If 6-8 layer doesn't work, the project is dead in the water until JLCPCB is able to produce cheap 12-layer boards.

The png file contained within the "pads and holes shim" zip does not show any pads. Is this correct? Perhaps the image needed to be turned over for the screenshot?

I'm not feeling confident with the PCB smashed sandwich approach for the 180 degree. I looked up the technique required to do this and it is too involved for me at this point. Maybe someone else reading this wants to BGA assemble these?

Looking at the BGA process, one would ideally use a reflow oven, although it was suggested that someone with sufficient experience can do it by hand. Unless the JLCPCB's come pre-balled (can they?), someone would need to order BGA balls and place them on the pads one by one with tweezers and add flux paste. Then pre-heat the board/balls/flux so that the flux holds the balls in place. Now add solder paste to the non-balled side, assemble the unit, and try to match the reflow profile.

I only have experience with hot air soldering onto the paste/pads themselves. For the approach at hand, heating would need to penetrate though the whole PCB and the user is blind. I'm used to seeing what I'm doing with solder paste. I do not have the time or energy at this time to become proficient in BGA balling. At first, I thought I could just use solder paste without balls, but have since realised this won't work.

Without a 180 degree single rotator board, I am finding my motivation waning.

The shims should be correct

I made one with metal plated holes and one with metal plated holes and diamond pads

Give it a shot.

I think maybe do the two rotators. Then immobilize them somehow, and then do the ends.

Maybe it makes sense to just tin them a little and heat them down to the rotator then solder the pins on by hand one by one then push the plastic on.

Idk. It’s gonna be difficult to assemble though for sure. Sorry 🙁

Jlc may actually be able to assemble these… ask them?

You may be able to make your own solder wire cutter to make your own metered size solder balls

Working with hot air station for several years, I can say, this is not enough. What you need is board preheater to bake the board to get ball layers bonded. Do it one layer at a time then finish installing pins and components with hot air.

Cheers,

pentiumspeed wrote on 2023-01-04, 22:46:

Working with hot air station for several years, I can say, this is not enough. What you need is board preheater to bake the board to get ball layers bonded. Do it one layer at a time then finish installing pins and components with hot air.

Cheers,

Could a 3d printer bed be utilized? Or must it be hotter?

Feiopa, perhaps a local electronics repair place can help?

Lead solder melts at 188C on average and very nice easily liquefy. Easy to do with 200C-230C hot plate.

No-lead solder is worse and does not liquefy easily above 217-220C and easily melts socket and connectors made of plastic if you are not careful with temps and working with these tend to be around 300 to 350 C with hot air and bit less using hot plate. I hate these no lead solder when I work with components and replacing connectors takes hellish of heat to come off.

Always use no clean flux on everything. I use Amtech 559. Smells nice. Amazon have these from this seller only, sells 30gram tubes and in tubs.
https://www.amazon.ca/NC-559-V2-TF-no-Clean-T … 387&sr=8-6&th=1

Cheers,

Would you be willing to order a batch and try it Pentiumspeed?

Sphere478 wrote on 2023-01-05, 01:14:

Would you be willing to order a batch and try it Pentiumspeed?

Hi there, how I do this?

Thank you and cheers,

pentiumspeed wrote on 2023-01-05, 19:23:

Sphere478 wrote on 2023-01-05, 01:14:

Would you be willing to order a batch and try it Pentiumspeed?

Hi there, how I do this?

Thank you and cheers,

Download the latest file go to jlcpcb.com upload it select the appropriate thickness for the shims and probably 1.6 or 2.0 thickness for the rotator you would have to ask Feiopa how thick the shims should be. My guess is 1.6 also. And raise the pins/socket out to make it flush

Ordering from them is pretty simple.

Other than color and PCB thickness there really isn’t a whole lot that you have to worry about on the order selection screen

Did anyone end up ordering?

The no replies probably means no. I didn't order either. The expectation of a BGA solder job for a socket rotator is asking a bit much.

I decided that having to stack two 90 degree units was about as undesirable as my existing 20 stacks of PGA sockets. I primarily need a 180 degree board.

To be honest Feiopa I have serious doubts as to whether or not this 90° rotator is going to work without ground planes but the success of your pga tower gives me hope 🤣. But trying to stuff 180° in there and expecting it to work without ground planes phew, 😓 that means every trace travels past by twice as many others and those traces are twice as long and closer together because of pcb thickness and crowding. A bga stack mitigates some (a lot) of that and since they are ordered in sets of 5 anyway is also cheaper.

To be clear I am highly suggesting that we add ground planes to the 90. But we can try it as is.

Again, there are multiple ways to do this, but this takes a long time to draft out and this is the best version of the 90 I came up with. (So far)

It is possible I could make something similar to this for 180 but I honestly think that bga stacking this is probably better than what that would end up being.

Anyway,

Another idea is a offset 90 with thru holes.

It would be taller and complicated to assemble for 180.

1000 ways to do this, and probably few actual good ways among them.

I don't think any hobbyist is going to BGA solder these. It is too involved. I feel it best to let that go.

I think running the traces through the central region is the way to go, not around and between the core pins. You can route the traces in the shortest possible path this way. Doing the slight offset, as pointed out earlier, may clear up some surface area. A through-hole approach, if possible, may also be suitable for the masses.

There is no need to rush anything, or continue work if you aren't interested. It looks like JLCPCB has removed their promotion for 6 and 8 layers.... then brought it back a week later. hmm...

I also managed to buy twin sister of feipoa's Daewoo AL486-D that is Panda 386V (the same motherboard but one is branded for Daewoo) - both are from ECS
https://theretroweb.com/motherboards/s/ecs-panda-386v
https://theretroweb.com/motherboards/s/ecs-al … 6v-d-for-daewoo

My goal is to run the best possible CPU that this mobo is capable of... and to do so I'll need a rotator too... So I'll happily support this project any way I can.

Batyra, do you know someone who is familiar with, or interested in, doing PCB layout? Your PM led me to believe that you may know someone. If yes, they can probably get a start with sphere's gerbers. There should be sufficient detail in this thread about my original proposal.

I think we should strive for through-hole top, and through-hole bottom, similar to how the SXL2 interposer was laid out. Or possibly through hole bottom, surface mount top socket. There should be a different PCB for each orientation, that is, a PCB for 90, 180, and 270 degrees. We want to reduce noise, trance length, inductance, parasitics, etc as much as possible, so we'd want Vcc and GND planes. We'd probably need 8 layers. Up to 8 layers is still cheap on JLCPCB. It is OK to exceed the 36x36 mm footprint of the PGA132 if needed.

We left off in this thread with a plan Sphere was pushing: a BGA design that uses only a 90-degree board. 180 deg would be two stacked 90 deg boards. 270 deg would be three stacked 90 deg boards. The BGA design is unrealistic for hobbyists, it increases trace lengths for 180 deg and 270 deg orientations, and is missing GND/VCC planes. For the Panda 386V, you'd want a 180 deg or maybe 270 deg rotator, but the existing design, assembled by a hobbyist w/out BGA will contain too many parasitics for your application. You can probably only get away with 90 deg rotator though.

Below is what I assembled some time ago. Adding the filtering caps helped getting the board to at least turn on.

90 degree rotator:

90 degree rotator with added caps:

If you wanted to sandwich two of these together, 180 degree would look like this:

And of course, 270 degrees:

I have run tests with the SXL2 at 85 MHz using these rotators. Only 90-degree worked, but I feel it is on the edge because other assemblies of mine would not work at 85 MHz, but would work at lower frequencies. In the rotator photos, you see one with red markings on it - this unit didn't work at all. So I manually soldered all the pins; you can see the solder looks like a mess on it. Still didn't work. Then I swapped motherboards, and it worked. Red flag. The existing rotators are too noisy.

As you are after maximum performance on your VLB 386 and would need 180 degrees, we would need a redesign of this project - individual PCB's for each rotation and VCC/GND planes. With a single 180 deg. PCB , you might be able to get away with high CPU frequencies. Maybe even the existing surface mount pins would work out.

Yes! I’ll ask a friend of mine to register (if he is not registered already) and dive deeper into this topic and post.

For now I was asked to get some socket like the one you have.