Beginning of very long journey

[Gaiyan]

Lets say... fixed motherboard with a few essential updates.

Are you planning to sell a kit or pcb or make your mods available?

[krupkaj]

This is really great. Thanks for your work. Do you want to make Atari version of Amiga PCB explorer?

[1st1]

Hi, interesting prohect, but where to get the chips?

[mpattonm]

RAS/CAS memory signal decoupling caps beeing over 40mm away from its pins on the memory board socket. With almost 80mm traceroute from the memory chips they are basically useless.

[viking272]

Am enjoying these updates! Lots of questions like most but will hold fire for the moment. A new board physically compatible with the old would be amazing, plus a new accessible case (another thread) would be a dream.

[mpattonm]

C217 (CAS3 signal) has track distance over 95mm from memory chip. That would probably get 1st place in dumb-ass design awards.

[wongck]

C217 (CAS3 signal) has track distance over 95mm from memory chip. That would probably get 1st place in dumb-ass design awards.

Back in the 80s probably all routing is done manually. So it's probably not so bad.

[mpattonm]

Back in the 80s probably all routing is done manually. So it's probably not so bad.

By 1990-1991 I believe at least OrCAD already had an autorouting module as an add-on, I vaguely remember playing with it on high school. Anwyway, first step in PCB design, before you even start start routing, is component placement. And that is where they failed in this case.

[joska]

I think you're a bit unfair here. Remember that the Falcon030 really wasn't a finished product, but a prototype that someone deemed "good enough" to manufacture while there still was a market for it.

[PeterS]

I wonder how much of the design/routing was "copied" from the STe or TT to help reduce their development time ?

Are there any features that you could omit or replace with something simpler/better ?
I know you would want to avoid feature creep but it might make it simpler in terms of component availability.

For instance use a 1GB card even if it can't address all of it or use SATA instead of IDE.

[mpattonm]

I think you're a bit unfair here. Remember that the Falcon030 really wasn't a finished product, but a prototype that someone deemed "good enough" to manufacture while there still was a market for it.

Indeed. As I have mentioned before, its appearent all across the board, at every level.

[mpattonm]

For instance use a 1GB card even if it can't address all of it or use SATA instead of IDE.

I do not want to dissapoint you, but SATA is most likely not going to happen. By the way what do you feel as limitation of Falcon PATA implementation? To me, this seems like one of the least problematic chapters.

[EvilFranky]

Might be completely out of scope for this project but would it be a serious amount of work to remove the proprietary memory interface and replace with SIMM sockets?

[PeterS]

For instance use a 1GB card even if it can't address all of it or use SATA instead of IDE.

I do not want to dissapoint you, but SATA is most likely not going to happen. By the way what do you feel as limitation of Falcon PATA implementation? To me, this seems like one of the least problematic chapters.

It's parallel so I presume it has a lot more tracks on the pcb. The sata components might be smaller, cheaper. Drives are more readily available.

Just suggestions.

[joska]

IDE is an integral part of the Falcon custom chipset, so replacing it does not make sense.

[mpattonm]

Might be completely out of scope for this project but would it be a serious amount of work to remove the proprietary memory interface and replace with SIMM sockets?

Thats something I am working on right now.

[mpattonm]

IDE is an integral part of the Falcon custom chipset, so replacing it does not make sense.

Exactly. And adding PATA/SATA adapter on board brings no advantage over external solution for ones who really need it. Just added cost and complexity.
Unfortunatelly, same goes for SD card interface.

[mpattonm]

As you all are well aware I guess, Falcon video connectivity is based on D-sub 19F/M connectors and these are nowhere to find already. So I am thinking to replace it with something more viable.
I am currently considering these options:
a) 20 pin miniD connector: https://www.tme.eu/cz/Document/caf2d678 ... %20THT.pdf Pluses: Guarantied availability. It is about 2/3 of DB19 width and about the same height, so mechanical compatibility with current chassis is guarantied. Enough pins to keep full Falcon video interface intact and yet add one ground connection. Minuses: higher price, would require all new VGA, SCART... adaptors to be made, these would be pricey too.
b) dual row D-sub 15F: https://www.tme.eu/cz/Document/afe87211 ... 634585.pdf Pluses: Guarantied availability. Consistent look with rest of Falcon ports. Supercheap. Minuses: Would require to ditch some of GND pins, but they are mostly redundant - for example R,G,B grounds are all connected to common GND inside Falcon anyway. There is 6 GND pins in DB19 interface, 2 would have to do - and would do. Again, all new adaptors would be required, but cheaper this time.
c) D-sub HD 15F, the ordinary VGA connector Pluses: Available, cheap, consistent look, woud allow you connect VGA displays directly. As VGA conn is much narrower then DB19, there could aso be some smaller connector, such as mini-DIN, placed right next to it. Such a conn could provide signalling missing on VGA conn. If I calculate correctly, there could also be enough space for 2 miniDIP switches to allow use to toggle video output between VGA, RGB and mono.

Opinions, anyone?

[mpattonm]

Just a proof of concept for now...

[Knezzen]

As you all are well aware I guess, Falcon video connectivity is based on D-sub 19F/M connectors and these are nowhere to find already. So I am thinking to replace it with something more viable.
I am currently considering these options:
a) 20 pin miniD connector: https://www.tme.eu/cz/Document/caf2d678 ... %20THT.pdf Pluses: Guarantied availability. It is about 2/3 of DB19 width and about the same height, so mechanical compatibility with current chassis is guarantied. Enough pins to keep full Falcon video interface intact and yet add one ground connection. Minuses: higher price, would require all new VGA, SCART... adaptors to be made, these would be pricey too.
b) dual row D-sub 15F: https://www.tme.eu/cz/Document/afe87211 ... 634585.pdf Pluses: Guarantied availability. Consistent look with rest of Falcon ports. Supercheap. Minuses: Would require to ditch some of GND pins, but they are mostly redundant - for example R,G,B grounds are all connected to common GND inside Falcon anyway. There is 6 GND pins in DB19 interface, 2 would have to do - and would do. Again, all new adaptors would be required, but cheaper this time.
c) D-sub HD 15F, the ordinary VGA connector Pluses: Available, cheap, consistent look, woud allow you connect VGA displays directly. As VGA conn is much narrower then DB19, there could aso be some smaller connector, such as mini-DIN, placed right next to it. Such a conn could provide signalling missing on VGA conn. If I calculate correctly, there could also be enough space for 2 miniDIP switches to allow use to toggle video output between VGA, RGB and mono.

Opinions, anyone?

Option C sounds like the best one out of that lot

[shoggoth]

I agree with option C. It makes sense to migrate to the actual standard connector for VGA. The external clock line is needed by peripherals such as Screenblaster, but I guess another way to achieve that is to put an oscillator on board instead, preferably a programmable one.

[mpattonm]

Looks like we are on the same page then.

[mpattonm]

Ever wondered why U56 chip (combel) looks so oddly soldered? This caught my eye on every single motherboard I have ever seen, I even remeber having that thought back in 1998. Well here is why...
Entire Combel footprint is skewed vertically and bottom pads are off by 0,3mm from top ones. With just 0,5mm pin pitch, this basically mean pins 1 ot 52 would have been completelly misplaced off their pads, sitting right in between them. Actually, close to/partially on wrong pad. I wonder how they managed high volume productin with a f*up like this.

[calimero]

Implement all Falcon signals on VGA connector and no problem - if someone need old DB19 it can make VGA <> DB19 adapter.

E. g. I am using SM124 sometimes with my Falcon (I have original Atari DB19 <> DIN13 adapter for this).

[mpattonm]

I imagine the factory ECN for the Combel mispaced pads was looking something like:
a) stop the pick and place machine
b) open the feed tray cabinet
c) locate Combel feed tray
d) take first in line Combel chip from the tray
e) bend all legs 1 to 52 to the left by 0,3mm with tip of utility knife. do not go over this value
f) place modified Combal back to the tray
g) turn the machine back on, let it populate one board
h) repeat from a)


jokes aside, they probably had to force chip maker to produce Combels that had a pins bent from the factory.