Wow. Lots of discussion on the ROM module idea.
Yes, I have a ROM module laid out. I haven't looked at the design in a couple of years though so there could be some issues:
1) There might be a few details left to fix.
2) I may have it on an 8" X 10" panel with a bunch of IIfx SIMMs.
3) I had a hard drive failure a while back which wiped out a bunch of rocket engine press-form designs. I thought that I always made backups of my designs, but if those are gone, some of my circuit boards could be gone. I need to check this.
However, none of those is much of a problem. It doesn't take very long for me to lay one of these out. I am helped on the ROM layout by the fact that the SIMM socket is the same (IIRC) as the IIfx RAM SIMM. So I already have a part created for the edge connector size and pins and latch holes on the edges. And I know that this part works, because my IIfx SIMMs work.
Yes, the 63 mil vs 47 - 50 mil is the biggest problem. In fact, it bit both Gamba and I. We were working merrily along each doing our part to make that SIMM and Gamba emails me to say he finished the assembly and he can't get it into the ROM socket. At that point it had pins on both sides of the board.
So he filed/ground down the back side of the SIMM until the edge connector was .5" instead of .063". This works because it's a SIMM and not a DIMM.
And yes, one of the biggest things holding me back was that all the cheap circuit board fabrication options I could find were for "standard" .063" board.
So I was looking at having 100 - 200 of the board made and then I could populate them as needed. However, at the time that I actively looked at that, the price would have been ~$600 for the lot and the chances that there wouldn't be enough people who want one to pay back the cost seemed high.
Enter SeeedStudio, as bigmessofwires just presented. I've seen their offer before and it looks great:
http://68kmla.org/forums/viewtopic.php?p=148796#p148796
I just hadn't considered it in connection with the Mac II family ROM project. Thank you bigmessofwires. That should make it doable.
An academic issue, but perhaps of interest is that regular board fabrication has fallen in price a bit, or at least there are more special offers available. I think that having a hundred boards made, might cost more like $200 these days, because 4pcb.com keeps offering board specials with no NRE costs.
On a small run like this the NRE (non-recurring expenses) is what really kills a project.
As T80 surmises, I am discussing two different surface mount designs. There are through hole PLCC32 sockets. So it could be a through-hole design. However, if one made it through-hole, that would preclude soldering the chips directly to the board. For folks who just want a 32-bit clean ROM for their SE/30, chips soldered to the board is preferred.
On the other hand, surface mount PLCC32 sockets are a pain. If sockets are the goal, it's nicer to do through-hole.
On the gripping hand, surface mount PLCC32 sockets would be easier than through-hole if I was doing toaster oven soldering. But I'm not doing that yet. It's a dream. Maybe even a goal. But it's not on the horizon.
I suppose I could lay it out as surface mount pads with holes drilled in the middle where through hole could be mounted. I'm not certain, but I think the foot prints may be compatible.
What do people want? Hand assembly? Toaster assembly? Sockets only? Hybrid surface mount/through-hole layout?
Is there someone else who could do the assembly if I designed the boards?
And is there anyone else who is gung-ho to lay their design out immediately? It will take me some weeks, possibly months to get to this. I still need to finish the PEx ROM before I do anything else...
So, I'm willing to work on this. But I'm also willing to step aside if others can get to it faster.
Oh, and T80, very cool about the two temperature activation for the solder. One could always use leaded and lead-free solder. They have about a 20 - 30C difference in melting point. But one would need a good controller on the oven. I found a link a while back that pointed at some controllers for a little under $100. I was going to whip one up out of a microcontroller and SCR, but when it's all said and done, buying an existing product for $100 is probably cheaper than developing one's own solution, even if it looks like the DIY should be cheaper. Experience says it won't be...
Edit: T80, I just noticed that I missed your question about home-made vs. fabrication. In recent years all my thought has gone to commercial fabrication. For a while there I was going to do home-made. I was going to use the toner-transfer method:
http://www.pulsarprofx.com/
but ultimately decided it was too much trouble.
That's why I had the ROM SIMM laid out on an 8" X 10" panel with IIfx SIMMs (and some 30 pin SIMMs, IIRC). The toner transfer paper lends itself to doing 8.5" X 11" panels, or slightly smaller. I found some .050" double sided copper clad board in about those dimensions.
Making the boards sounds easy. Here's where it gets to be too much trouble: Vias.
One is left having to hand drill all the vias and then solder a wire through the board and onto each side. Alternatively, one could build a CNC-type drill (one fellow built one out of an old plotter). Once the holes are drilled, and before the board is etched, one could squeegee electrically conductive ink onto the board and then blow it through the holes, so that it coats the inside surface of each drilled hole. Then drop the whole board in a copper sulfate tank and electroplate until the ink-covered holes are coated in copper.
Setting up to electroplate is a pain. That conductive ink is either hard to find, or ridiculously expensive.
By the time I looked at all the problems that vias were going to cause, I decided it was better to just go with commercial fabrication.
If I had a workshop, where I could set up a permanent etch tank, and electroplating tank and associated stuff, then maybe....
And then there's solder mask. It's not strictly necessary, but it certainly reduces errors when assembling (soldering) the boards. It's a photosensitive laminate. Well, I already have a laminator to make the toner transfer method efficient, but then one needs a photo-exposer thingy to get the solder mask pattern onto the solder mask laminate. And at that point, one may as well abandon the toner transfer method and go to a full-on photo-resist method of etching the boards.
But again, the vias are a pain.
So, commercial fabrication all the way. There's a point at which one's volume is large enough to make it cheaper to buy/build all the stuff to do it at home, and I think that point is where one goes into business as a PCB fabrication house.