An earlier version had pockets sized exactly for the flash and the two pins for TP1 & 2. The current version just has a single big pocket leaving a wall all around the perimeter to be more generic and allow other stuff, and I moved tp1&2 to get inside that border. (and the border is marked in the cad file on a non-manufacturing layer)
The depth of the pocket is 1.2mm. This clears the flash chip and through-hole legs if they are trimmed, and most ordinary smd parts, but a lot of parts are thicker than 1.2mm of course. The flash chip is a bout 1.1. This leaves 1mm thickness for the plastic. If you have parts going right to the edges, and I can well believe it would be non-trivial to move things around to pull in from the edges enough to clear that border, I think the thing to do is just customize the carrier and see if it still works. I'd be happy to try it. I'd modify the model and order a print and have it mailed to you. I see a few possible conditions, and things to try. If you have components that go right to edge or even beyond, it would be easy to just cut out those walls and only have a flat sheet on the bottom. The result would be very flexible, even floppy, but only when empty. It should still snap onto a board and the combined module might feel perfectly solid. The two ends would essentially be holding themselves in place by tension, where the snap wedge on each end is like a hook and the thin sheet going across the bottom is like a bungee cord. It's all floppy by itself, but fine once the two ends are hooked over the edge of some other rigid structure. In fact that's already partially the way it works now. Another possible issue is if you have any back side components that are taller than 1.2mm. If it's just a few things and not the majority of the surface, we can just poke holes in the bottom no problem. If there are a lot of components but only a little thicker than 1.2mm, the sheet across the bottom is currently 1mm thick. Shapeways can't go any thinner than 0.7mm and I think it would become pretty weak and maybe pull apart if I tried to go that thin, but we could try it and see. If there is so much stuff on the bottom that there is just not enough volume left for plastic either in the form of a sheet or in the form of a couple of rails to connect the two ends, then maybe we can flip over and snap the pcb in from the bottom and have the connecting plastic on the top. ... no then I wouldn't trust it to pull the pcb back out of the socket. Anyway, I would be happy to see what could be done. It's not too hard to try a few things. The result would probably be a custom carrier for that pcb, not a generic one that handles that pcb and others, but that's ok. I also have edge cut dimensions that provide polarity prongs from the pcb itself, but that is kind of annoying to use. You still need an extraction ribbon and spacer, and then the prongs have to poke through the ribbon, which is not only ugly but weakens the ribbon and I wouldn't want to give that to someone else and trust it never to break on them. And that still leaves the possiblity of putting the pcb in upside down. The prongs only enforce one dimension, and make the ribbon annoying enough to maybe not bebworth it. I toyed with maybe a couple loops of fishing line instead of a ribbon. It works but it's more work to make. You have to loop it at couple times, and tie it together to make a closed loop, and melt the knot. But It's an option. I have had pcb's made that worked so the router path is ready. The old commercial roms that used this method had a big ceramic dip28 eprom on top, and the extraction ribbon ran under the chip instead of under the pcb. Another possibility is, the width of that bottom tray is just a dimension I picked to make it impossible for an empty carrier to become trapped under the pins in the socket. It might be possible to push that a little. Maybe even a lot. We could disregard that concern and just say don't disassemble the module and then put the empty carrier in the socket. You could also just glue them together for ones you produce. Or maybe the empty carrier will be weak enough that if it ever gets stuck in the socket by itself you can just flex it or break it and get it out without hurting the socket pins. Like if it just had a couple of rails instead of a full bathtub. As for the edge contacts, I just wanted a shallower/flatter arc touching the pin in the middle instead of a half circle wrapping around the pin. I don't claim it necessarily corrects a problem, just that it's what I want. Actually I think I'm still not quite done there. I think the hole could be still bigger and moved a little further out, leaving an even shallower/flatter arc for the actual contact. I think I did create more usable pcb real estate by doing that though. I can sneak more traces between the cpld and the edge than before. -- bkw On Sat, Aug 10, 2019, 8:20 AM Stephen Adolph <[email protected]> wrote: > Brian, > > I get that this carrier improves the look of the install but I dont see > how it improves operation. It doesn't improve contacts in my mind does > it? I've done so many installations of rex now...I think the spacer and > ribbon approach works fine. I agree that I have a lot of practice though. > > REXCPM has uses most of the secondary ie bottom side for parts. Like > REX2. So it won't work with this carrier. Also the flash and cpld are > placed right at the edge. > > What is the board dimension you need? I will have to reduce size by the > sounds of it and move parts. Not sure if I can, depending. It would mean > pretty much every track re layed out. > > Can you make a design that leaves pretty much all of the secondary side > available for TSOP48 and VQFP44? From what I saw you have a cutout for the > flash only, in the center. > > > Thx Steve > > > > > On Friday, August 9, 2019, Brian K. White <[email protected]> wrote: > >> That modified rex pcb and matching carrier I was working on are now at a >> usable state. >> >> There is a pcb which is REX1 electrically, with only some layout and edge >> cut changes. >> >> There is another pcb with the same outline that uses the same carrier, >> but has only a simple eeprom on it. I'm calling that Teeprom. >> >> There is a github with both the carrier 3d model and an empty template >> version of the matching pcb, to make new circuits using the same pcb and >> carrier. So hopefully this is reusable. >> >> By now I have received and tested multiples of both pcbs and the carrier, >> and everything is dialed-in pretty well finally. >> I have several rex's and teeproms and a bunch of carriers all of the >> current/latest versions the same as what's currently orderable on oshpark >> and shapeways, and the boards all fit perfect in all the carriers, and the >> carriers & boards all fit and work in my couple M100's, T102, and T200. >> >> I overhauled the build and flash directions a lot. Filled in more >> deatails with actual steps as if going from scratch, expanded some steps >> into their own separate docs but linked right where you need them. Could >> still do more but it's just better than before. Simplified and organized a >> little better (I hope). >> >> http://tandy.wiki/REX >> http://tandy.wiki/Teeprom >> http://tandy.wiki/Pcb2molex8878 >> >> http://shpws.me/RngB >> https://github.com/aljex/pcb2molex8878 >> >> https://photos.app.goo.gl/zynHU4LSpd5BZBYC9 >> >> >> I'll get the rex and teeprom full sources up in a github or something >> soon too. The .kicad_pcb files are already available via oshpark, but you >> want the rest of the kicad files too if you were going to work on them. >> >> Anyway, it's at the point now where I'm saying it's ok to order these >> things and they'll work and I'm not still changing them and obsoleting them >> right after you ordered them. >> >> Just for the record, I have the shapeway shop set at 0% markup for all >> the items. I don't want any damned shop but that's just the only way you >> can share the model on shapeways so that other people can order prints >> without downloading and uploading the mesh files themselves. The same model >> is also on thingiverse where you can share without having a shop, but >> thingiverse is mostly FDM printers, and I don't think any fdm printers are >> good enough for this part. But, if you have your own printer and want to >> try it, the mesh file is up there. So the only reason there is a "shop" on >> shapeways is just to get shapeways laser sintered nylon printing. It comes >> out very clean and accurate and strong enough that the part is actually >> usable. >> >> You may notice that the seemingly nicer print options are not enabled for >> this on shapeways. That's on purpose. If you get the part printed in >> multijet fusion, it looks great, but doesn't fit. >> >> Unfortunately, for the carrier, my freecad "source" file, while available >> in the github, isn't actually very "source"-like. It's essentially just the >> final 3d solid shape just like the stl file for printing, not all the steps >> to build it up, nor any parametrics you can edit. But I do plan to redraw >> it from scratch in freecad or better yet openscad sometime. Especially >> openscad will make it awesome for keeping in git. >> >> Also if you wanted, it IS actually possible to modify a standard REX to >> go in one of these carriers. >> >> You would do 4 things: >> >> 1: Cut the polarity pin off the carrier. Doesn't have to be any more >> accurate than just lower than the lip on the bottom tray. Just use wire >> cutters or a knife. >> >> 2: Sand both ends of the pcb (the two short dimensions) down about 0.5mm >> each, or whatever it takes until the overall length is 37.35mm or shorter, >> and doing both ends equally. This will eat into the jtag pin holes a >> little, but actually not enough to hurt anything. You could even still use >> the jtag port if you ever needed to. It doesn't affect normal operation at >> all. >> >> 3: Chamfer all 4 corners of the pcb edge about 1mm each corner. Make the >> very ends narrow enough to get between the vertical walls on the ends of >> the carrier, yet don't chew into the edge contacts at pins 1, 15, 15, 28 >> more than about 10-20%. You can actually graze the ends of those 4 corner >> contacts slightly and they'll still make perfectly good contact. But you >> should be able to just barely not touch the actual half-circle part. >> >> 4: Remove anything glued to the bottom of the pcb. Some glue residue left >> behind would be ok on the pcb surface next to the flash chip, up to the >> thickness of the flash chip, but the surface of the flash chip should have >> nothing thicker than paper on it. >> >> -- >> bkw >> >> >> >> On 7/29/19 9:34 AM, Josh Malone wrote: >> >>> Fred, >>> >>> If this is one of my REX units (shipped with wood spacer and a purple >>> polka-dot ribbon) then my apologies for your issues. I tested the >>> spacer sizing in 100, 102, and 200 units and found what seemed like >>> the right size to prevent over-insertion and "spontaneous removal". >>> I'd have to check my notes, but I sanded my boards down to a precise >>> size which I think was 0.048" (~1.2mm). They do require a very firm >>> insertion into the socket -- you may feel uncomfortable applying that >>> much force -- but you should feel the board snap into place below the >>> catch in the pins. If you don't feel a "snap" then it is possible that >>> the spacer board is too thick for whatever reason. I don't use any >>> adhesive, etc. on my boards to make it easier to repair them should >>> anything go wrong, but I've also had one other issue with a user >>> neglecting to install the spacer which caused damage to the REX and >>> his M100, so I will be changing to adhesive spacers on my next batch >>> (which I'm producing the Tandy Assembly). >>> >>> When you install REX, do you feel it snap into place? Can you send >>> pictures of the installed REX? (You'll have to find a host for them >>> since the list doesn't handle attachments; or send them directly to >>> me) >>> >>> Brian's work with the new REX PCB and the 3D-printed carrier looks >>> neat -- I'm following his progress -- so if you continue to have fit >>> issues one of those carriers might solve your problem, as well. >>> >>> https://photos.app.goo.gl/aNfjvB1XucSehPGB9 >>> >>> -Josh >>> >>> On Sat, Jul 27, 2019 at 8:29 PM Stephen Adolph <[email protected]> >>> wrote: >>> >>>> Hi Fred, >>>> >>>> Honestly my first thought here is ... what software are you running and >>>> have you ever done a total wipe of rex followed by a reload from scratch of >>>> the software? >>>> >>>> These symptoms more or less sound like corruption of the directory >>>> which can happen on older software loads. >>>> >>>> I've tried to catch all the fixes in my last supported software load. >>>> Which is at bitchin100. >>>> ..Steve >>>> >>>> >>>> On Saturday, July 27, 2019, Fred Whitaker <[email protected]> >>>> wrote: >>>> >>>>> I have been having trouble with my REX chip becoming loose and losing >>>>> data as a result. The computer also locks up. I was wondering if anyone >>>>> had >>>>> any ideas about better support for the chip corners, instead of a loose >>>>> piece of wood under it. >>>>> I was thinking of trying to find some small diameter thin stick on >>>>> feet, about 1/8 inch thick. Has anyone had a similar problem? >>>>> The problem will be the same with any of the ROM's on similar boards. >>>>> The Super ROM that I have is on a similar board but it also has a full >>>>> length spacer glued to the bottom of the board. >>>>> Until I can resolve this issue I am putting my REX chips in storage. >>>>> This is only true for the Model 100 and Model 102. >>>>> >>>> >>
