Hi all,
I'm doing some work that needs feedback from the community.
I places a backplane on a card on the left side of the QL for four slots. In no time I found it deeply aggravating and unworkable. My left hand/wrist would constantly bump it as I was typing, and any function keys and the left three inches of the main keyboard were difficult to use comfortably.
I have three approaches available to me.
The first is to length the card in the QL and move the entire expansion chassis further to the left. It seems this must be at least 4". With the backplane, eurocards and enclosure this makes for a 11" protrusion on the left side of the QL. It's so long....
The second option is to rotate the expansion chassis so it plugs into the back edge of the expansion card plugged into the QL, and extends 7" back from the back edge of the QL. This cuts down the length a lot. It does push the QL forward or the monitor back a bit, if you have limited space. Especially with any connectors coming off the port edge of the cards at the back.
The third (a) option is to end the QL card almost flush with the left edge of the QL. An Mplane-style backplane can move the bus to behind the QL, facing right back along the back edge of the QL, past the ROM port and stopping right before the second joystick port. Cards mount horizontal, inserted from the right.
The third option (b) could have the cards vertical, connector edge facing backwards. Cards would be inserted from rear to front, upper side of PCB facing right.
The second and third options would suggest including a custom designed riser. The QL could be flat on the desk, or the rear might be raised. If raised, the chassis would be up in the air. A wedge shaped insert could hold it securely.
The other issue with options two and three are that the chassis needs to be at least 4" tall. It can hold four cards in the 4" model, with the backplane being easily swappable to add up to theoretically 16 cards. It supports a mix of L Cards (Legacy, QL 8-bit, 5V, 7.5 MHz) and X-Cards (Extended, 8/16/32-bit, 3.3V, 7.5 MHz, 15 MHz or CPU speed, active termination).
All options can accept any standard Eurocard sized card in a legacy slot. The card will fit if it's shorter or longer. It just means it would be pushed inside or stick out a bit. Some floppy interfaces became a bit wider after exiting the QL. These will not fit in the rails of the case.
I have built a 4x L Card version (4L) and a 2x L-Card, 2x X Card (2L2X) backplane so far. I am waiting on PCBs for a 4L4X backplane. The extended bus uses a 164 pin connector. I am working out an open source logic to provide for anyone wishing to use the extended bus. The idea there is that anyone can have a pre-tested known-good set of equations, CUPL, Verilog and VHDL available to incorporate into their own designs easily.
The point of the last two paragraphs is to show that the backplane could be up to 13-16" long. The slots have .75" spacing. I haven't settled on if cards should be Eurocard sized (100x160mm) or 100x100mm. There isn't a large saving for 100x100 as the size discount would never apply. The one end always has 7mm occupied by gold fingers. I'm leaning towards 100x160 if only because it leaves more options open, eg, using it for a 68060 CPU farm *lol* using the bus mastering features.
One big benefit of the now-behind-me health issues is I have worked on this with a freakishly sparse budget. This has forced me to design it using a careful series of choices that make it quite inexpensive to produce. The entire backplane system including buffers, local power regulation from USB-PD (you probably already have a tablet charger) are cheap, and a 4-slot backplane. The system is modular. At any time the 4-slot backplane can be replaced with another.
I will also be ordering some 8-port all Legacy (8L) cards with onboard power regulation. Switchable between 5V and 10V VIN depending on your current system set-up. Default is 5V. Can be set per slot.
Before I commit to ordering PCBs to make 25 units and 25 of each backplane option, I need to know which option 1-3 (including 3a and 3b) people prefer. You know your desks and the scourge of things being weird shapes. Choose one option only, please. Really think about the choice, because it has a significant impact on a lot of future work by several people. Once I have a clear preference shown, I will show photos of an actual working system. Just got to get the black finished PCBs ordered and delivered.
Expansion chassis
Re: Expansion chassis
I wish to mention a couple of extra things. I didn't include them in the first post because it was already getting a bit overloaded with information.
Both "Option 3" choices obstruct the ROM port. I would place a ROM port duplicate socket, 100% physically and electrically identical, on the left side of the QL on the backplane turn-around card if one of those options wins. The chassis case has a ROM-sized stub that locates in the QL's original ROM port to physically stabilise it. This is surprisingly sturdy.
The chassis spacer card can be an active buffer card or a 30 MHz, 8 MB RAM, 2MB flash card with two "Q68+" compatible podule slots. Cards work on the expansion system and the Q68. Which will also get a Q68+ expansion chassis supporting four devices. This card is part of the expansion system and will also be upgradeable to larger CPUs and memory as these finish development.
Open Source. All pinouts, signal specs, equations, schematics, timings, etc. of the expansion chassis systems will be fully open source. The CPU/RAM card will become open source upon my death, my loss of interest/absence for one year, or when I conclude my interest in making a particular model. I will deposit all the data with a trusted community member that will be able to release the files based on my say so, or if I lose the ability to give consent due to illness or death.
Ooof. Got a case of the morbs.
Both "Option 3" choices obstruct the ROM port. I would place a ROM port duplicate socket, 100% physically and electrically identical, on the left side of the QL on the backplane turn-around card if one of those options wins. The chassis case has a ROM-sized stub that locates in the QL's original ROM port to physically stabilise it. This is surprisingly sturdy.
The chassis spacer card can be an active buffer card or a 30 MHz, 8 MB RAM, 2MB flash card with two "Q68+" compatible podule slots. Cards work on the expansion system and the Q68. Which will also get a Q68+ expansion chassis supporting four devices. This card is part of the expansion system and will also be upgradeable to larger CPUs and memory as these finish development.
Open Source. All pinouts, signal specs, equations, schematics, timings, etc. of the expansion chassis systems will be fully open source. The CPU/RAM card will become open source upon my death, my loss of interest/absence for one year, or when I conclude my interest in making a particular model. I will deposit all the data with a trusted community member that will be able to release the files based on my say so, or if I lose the ability to give consent due to illness or death.
Ooof. Got a case of the morbs.
Re: Expansion chassis
Drawings, because some people don't have a mind's eye.
- My drawing skills are on vacation.
Re: Expansion chassis
Hmmm. Left, right, front, back is all good. What's bad is: A rigid connection between QL and "expansion box" is the crux - This makes the whole thing practically non-moveable - I'm spoiled by separate, moveable keyboards, and a conundrum of rigidly interconnected boxes will become a hassle.
Did you think about a flexible connection between QL expansion board and "slot box"? Obviously would need a bit of a boost of the QL's signals - I have no idea if this would even be possible.
Did you think about a flexible connection between QL expansion board and "slot box"? Obviously would need a bit of a boost of the QL's signals - I have no idea if this would even be possible.
ʎɐqǝ ɯoɹɟ ǝq oʇ ƃuᴉoƃ ʇou sᴉ pɹɐoqʎǝʞ ʇxǝu ʎɯ 'ɹɐǝp ɥO
Re: Expansion chassis
The TI-99/4a is a good example of this
If you expand the system with ALL the “sidecars” then you’d probably need a desk about 8+ feet wide
But the Peripheral Expansion Box allows multiple expansion cards to be fitted in a single box and that’s connected to the system with a flat “hose” (though a more flexible cable would be a better solution)
Since I got back into the QL and looked into how to have multiple cards attached I’ve thought a QL PEB would be an expansion game changer
If you expand the system with ALL the “sidecars” then you’d probably need a desk about 8+ feet wide
But the Peripheral Expansion Box allows multiple expansion cards to be fitted in a single box and that’s connected to the system with a flat “hose” (though a more flexible cable would be a better solution)
Since I got back into the QL and looked into how to have multiple cards attached I’ve thought a QL PEB would be an expansion game changer
Re: Expansion chassis
A limitation exists because this system needs to work at much higher speeds than a BBQzL. A ribbon cable would trash signal quality and introduce a painful amount of crosstalk. Even with the best design practices the system needs active termination. The only way I could make it work now and into the future was by using a 4-layer bridge board. The CPU/RAM card is 6-layer. It needs to be isolated from the QL electrically as much as possible. The backplane has been tested on an Atari Falcon - which is the level of performance this needs to reach, eventually.
I owned a TI 99/4A expansion box. I loved it in principle and in practice. Clean, integrated, worthy of imitating in many ways. My choice here for future systems is to place the CPU in the expansion unit, and only run BBQL speed signals to the QL. The chassis works with the bridge card or standalone. It really is designed to grow and to last.
I was asked in PMs about the new connector. I chose one that meets these criteria: affordable, at least 100 insertions, high pin count, low cost, assured long term availability. The final requirement is a maximum length of 100mm for the connector housing, so it would be mechanically compatible with Eurocard dimensions. The few choices here got whittled down to only one. PCIe 16x slots are here for the long hail, and below 60 cents in even quite small quantities. The cost they add to cards by enforced PCB requirements is lower than the additional cost of the next best option, which was TX24 120 pin connectors. The other difference is 164 contacts instead of 120. That allows a whole private bus on the backplane so cards can DMA each other without disturbing the main bus.
I have a few thousand DIN connectors to use up. Legacy cards will be well supported.
I owned a TI 99/4A expansion box. I loved it in principle and in practice. Clean, integrated, worthy of imitating in many ways. My choice here for future systems is to place the CPU in the expansion unit, and only run BBQL speed signals to the QL. The chassis works with the bridge card or standalone. It really is designed to grow and to last.
I was asked in PMs about the new connector. I chose one that meets these criteria: affordable, at least 100 insertions, high pin count, low cost, assured long term availability. The final requirement is a maximum length of 100mm for the connector housing, so it would be mechanically compatible with Eurocard dimensions. The few choices here got whittled down to only one. PCIe 16x slots are here for the long hail, and below 60 cents in even quite small quantities. The cost they add to cards by enforced PCB requirements is lower than the additional cost of the next best option, which was TX24 120 pin connectors. The other difference is 164 contacts instead of 120. That allows a whole private bus on the backplane so cards can DMA each other without disturbing the main bus.
I have a few thousand DIN connectors to use up. Legacy cards will be well supported.