Issue 8 discussions...

Nagging hardware related question? Post here!
User avatar
Dave
SandySuperQDave
Posts: 2390
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Issue 8 discussions...

Postby Dave » Wed Mar 20, 2019 11:49 pm

Sooo, to kick things off, would people be terribly offended if all the 7400 series logic was SMD instead of DIP?


User avatar
NormanDunbar
Super Gold Card
Posts: 694
Joined: Tue Dec 14, 2010 9:04 am
Location: Leeds, West Yorkshire, UK
Contact:

Re: Issue 8 discussions...

Postby NormanDunbar » Thu Mar 21, 2019 7:42 am

Morning Dave,

I suppose that question depends on who is having to do the soldering ("soddering" for the USA ;)

I've done some hand soldering of SMD stuff, and it's a tad finiky, but no tombstones. (Yet!)


Cheers,
Norm.


Why do they put lightning conductors on churches?
If at first you don't succeed, don't take up skydiving!
If you think your job is pointless, remember, there's someone in Germany who fits indicators to BMWs.
User avatar
Peter
Aurora
Posts: 841
Joined: Sat Jan 22, 2011 8:47 am

Re: Issue 8 discussions...

Postby Peter » Thu Mar 21, 2019 8:33 am

Especially if only larger, easy to solder SMD footprints like SO are meant, that's absolutely fine for me. But I'm interested in high functional and mechanical equivalence to the QL.
For my taste, it should be able to run 128K RAM if configured, support microdrives, and unchanged operating systems. I also like the idea of re-seating most socketed QL chips to this board.
For me, the Issue 8 board should cover those traditional QL features, looks and add-ons the Q68 can not offer.

If I designed the board myself, I'd be in temptations to improve and change a lot of things. But I think most of those temptations should be resisted. Make it a true QL, very last Issue! :)


User avatar
Cristian
Super Gold Card
Posts: 618
Joined: Mon Feb 16, 2015 1:40 pm
Location: Veneto

Re: Issue 8 discussions...

Postby Cristian » Thu Mar 21, 2019 1:01 pm

Peter wrote: But I'm interested in high functional and mechanical equivalence to the QL.

Me too!


User avatar
Dave
SandySuperQDave
Posts: 2390
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Issue 8 discussions...

Postby Dave » Thu Mar 21, 2019 5:29 pm

NormanDunbar wrote:I suppose that question depends on who is having to do the soldering ("soddering" for the USA ;)

I've done some hand soldering of SMD stuff, and it's a tad finiky, but no tombstones. (Yet!)


All the SMD components would be mounted for you. Mostly, it's DIPs changing to SOICs. The reliability would be there, so socketing for easy changes is unneeded. It also gives a bit more flexibility in the design - though it is nice using DIP components because the pins penetrate all four layers. They always hit +5V and GND without tracks and vias. The passives are never smaller than US 0805. The smallest component on the board is a TSSOP5 single inverter gate.

I'm in Texas, so "soldering" introduces a new vowel that doesn't exist in British English. I can't think of an English word that has the exact sound. It is close to the au in "launder"...


User avatar
Dave
SandySuperQDave
Posts: 2390
Joined: Sat Jan 22, 2011 6:52 am
Location: Austin, TX
Contact:

Re: Issue 8 discussions...

Postby Dave » Thu Mar 21, 2019 5:50 pm

Peter wrote:Especially if only larger, easy to solder SMD footprints like SO are meant, that's absolutely fine for me. But I'm interested in high functional and mechanical equivalence to the QL.
For my taste, it should be able to run 128K RAM if configured, support microdrives, and unchanged operating systems. I also like the idea of re-seating most socketed QL chips to this board.
For me, the Issue 8 board should cover those traditional QL features, looks and add-ons the Q68 can not offer.

If I designed the board myself, I'd be in temptations to improve and change a lot of things. But I think most of those temptations should be resisted. Make it a true QL, very last Issue! :)


It IS a temptation. Feature creep is the enemy. The catch is there are essential changes, and the work to do those is 90% of the work for the new feature. IT'S A TRAP! Currently, the main work seems to be focused on eliminating the 8301's stranglehold on the system. It occupies the bus far too much of the time. Its generation of DTACKL is also problematic. Since it wouldn't play nice, we're taking all its toys away.

Currently, it looks like the machine will have Minerva at launch and SMSQ/E shortly thereafter. It'll have 1MB onboard, with the bottom 128K shadowing everything important, including ROM. I've been quite intrigued by an idea you explained in a post a couple of years ago, which would greatly simplify booting and occupying shadow RAM in the ROM areas. The main thing I am stuck on is that it would require some bootstrap code. While that could be written, it might be done in a way that wasn't Q68 compatible, and then it would cause an unneeded fork in SMSQ/E. Nasta is working through the logic in that area right now, and I think has found a brute force way using GALs.

It should have a 128K mode. It does support microdrives and QLNET under Minerva. If the system is supplied with a slightly faster clock, we will modify Minerva so timings are correct.

My plan is to provide assembled boards. There CAN be a kit, but I'd still want to have the SMDs pre-mounted, since I'd have the solder paste stencil right here.

The main area of discussion right now is the video subsystem. It's a possibility we might move the whole thing including 8301, internal video RAM, 245 and 257s to a daughter card, so the entire video system could be upgraded later.

I do have a number of 32K dual port SRAMs, which would completely free the CPU from video timing constraints. This could more than double the speed of the QL and remove a lot of timing restrictions. It would also prevent Minerva screen 1 from working. All it would require is demuxing the 8301's DRAM bus handling.

Hmmmm.


Nasta
Gold Card
Posts: 323
Joined: Sun Feb 12, 2012 2:02 am
Location: Zapresic, Croatia

Re: Issue 8 discussions...

Postby Nasta » Fri Mar 22, 2019 1:31 am

Mostly this is about fixing stuff, and if features are added, the main point is to support SMSQ/E and get the most speed out of existing hardware (this actually comes from supporting SMSQ/E) - so most of these features are not immediately visible and if yes, completely transparent and compatible.
At the moment there are two possible solutions to boot SMSQ/E, if it's from an already running OS like Minerva, the procedure is almost the same as GC/SGC. There is also a direct bootstrap option, which is a bit more involved and may end up fairly similar to Q68.
A code fork (hopefully minor) will have to be made because some drivers will be missing for obvious reasons - no such hardware.


User avatar
Peter
Aurora
Posts: 841
Joined: Sat Jan 22, 2011 8:47 am

Re: Issue 8 discussions...

Postby Peter » Fri Mar 22, 2019 9:58 am

Dave wrote:Currently, it looks like the machine will have Minerva at launch and SMSQ/E shortly thereafter.

The explicit mention of these two bewilders me a little. I hope the board remains so close to the QL, that unchanged QDOS works also. (And I think you mentioned that earlier.)

Dave wrote:I've been quite intrigued by an idea you explained in a post a couple of years ago, which would greatly simplify booting and occupying shadow RAM in the ROM areas.

We probably talked about a Q68 style bootloader that pulls a ROM image from SD card and then triggers a hardware reset. Porting the Q68 loader to a QL with shadow RAM should be possible, but I see danger of breaking the principles that I hoped to find in this mainboard. E.g. I hoped for existing original ROM sockets and the boot process unchanged (at least as an option).

Dave wrote:It does support microdrives and QLNET under Minerva. If the system is supplied with a slightly faster clock, we will modify Minerva so timings are correct.

If it can not run at original speed and with unchanged QDOS, the board would lose it's special "retro" charme for me. Which is of course a matter of taste.

Dave wrote:The main area of discussion right now is the video subsystem. It's a possibility we might move the whole thing including 8301, internal video RAM, 245 and 257s to a daughter card, so the entire video system could be upgraded later.

Video is a problem also for myself - so I understand this is tempting. On the other hand, if the mainboard goes for
- Higher speed and changed timings
- OS changes as a consequence
- Shadow RAM
- Virtualized ROMs
- Direct SMSQ/E support
- New video controller
- Direct SD card support
it gets closer to the question of a Q68 derivative in QL form factor. Since the Q68 will soon support QL network, it is mostly lack of microdrives and matrix keyboard which keeps it from doing a similar job. At potentially lower cost and a number of further improvements. It comes down to the question: Is this mainboard going to be a "pimped" machine or a rock solid, true, final QL? I vote for the latter.

Dave wrote:I do have a number of 32K dual port SRAMs, which would completely free the CPU from video timing constraints. This could more than double the speed of the QL and remove a lot of timing restrictions.

Again I hear "speed change" ;)

To be clear: I just write about my personal taste for this mainboard - if it goes a different route I will still be very happy if the project succeeds!


Nasta
Gold Card
Posts: 323
Joined: Sun Feb 12, 2012 2:02 am
Location: Zapresic, Croatia

Re: Issue 8 discussions...

Postby Nasta » Fri Mar 22, 2019 10:14 am

Peter wrote:
Dave wrote:Currently, it looks like the machine will have Minerva at launch and SMSQ/E shortly thereafter.

The explicit mention of these two bewilders me a little. I hope the board remains so close to the QL, that unchanged QDOS works also. (And I think you mentioned that earlier.)


Don't worry, the default configuration is exactly an old QL except for one difference (which could also be disabled), of having 1M of SRAM on board - and it can use the original ROMs. The main difference in ROM sockets is that they will be quality turned pin ones :)
Everything else is transparent. The timing is changed in the sense that everything can work ~zero wait but the default clock is still 7.5MHz exactly the same so MDV, NET all work excatly as before.
All the enhancements (actually most are optimizations of existing stuff and bug fixes) are about 'preparation', so they stay 'in the background' and can be used by future software, and yes, if the 1M RAM is enabled, it should be able to run SMSQ/E even starting as JM/JS.
The one thing we are debating at the moment is screen 1 support, so opinions on this are welcome.


User avatar
Peter
Aurora
Posts: 841
Joined: Sat Jan 22, 2011 8:47 am

Re: Issue 8 discussions...

Postby Peter » Fri Mar 22, 2019 10:48 am

Nasta wrote:The main difference in ROM sockets is that they will be quality turned pin ones :)

Fantastic! My taste for this mainboard is very high quality, even if I have to pay more.

Nasta wrote:The timing is changed in the sense that everything can work ~zero wait but the default clock is still 7.5MHz exactly the same so MDV, NET all work excatly as before.

That's good - although I'm not sure what will happen to some favorite games. If speed increases too much for normal gameplay, some way of slowing down would be important.
(E.g. the Q68 can add hardware waitstates for speed tuning.)

Nasta wrote:The one thing we are debating at the moment is screen 1 support, so opinions on this are welcome.

Opinion: The Q68 supports screen 1. A QL mainboard should not fall behind the Q68 in compatibility.
Practical relevance: Very small.



Who is online

Users browsing this forum: Pr0f and 8 guests