The humongous Sandy Electronics thread...

[RWAP]

The ability to do a quick FORMAT ram1_mdv1 to make a fast dump of a microdrive to ram disk....

But that is a function of RAMPRT.

Ah - now who has the rights to that ??

I have emailed Tony Tebby about some of his QJUMP stuff, but no response unfortunately...

[prime]

Only problem that I can see is this.

Adding 512K RAM to bring the machine to 640K is trivial and could be done with a single SRAM and a little bit of decoding. Adding more than that gets hairy

For example the Sandy SQB when plugged into the QL will map it's rom at $C0000-$C7FFF (asuming disk rom and TK2). However there will be a hole at $C3fC0-$C3FFF that has the I/O space that is used to talk to the disk / parallel / mouse.

With the SQB this could I guess be partially mitigated by forcing SP1..SP3 to 1 on the through con which would move the ROM area up to banks $E & $F which should give you an extra 224K.

To get the full extra 256K I think you might have to start doing some address translation on the RAM card, not impossible but it may get messy.

Cheers.

Phill.

[1024MAK]

Given that Dave only talks about using 512k SRAM chips, and what Nasta has previously posted in this forum about RAM memory upgrades, I think it is more likely that they are on about using TWO 512k SRAM chips with some clever decoding (using PLDs or a CPLD) to shadow the QL internal DRAM and to leave holes in the SRAM memory decoding for the QL ROM space, QL internal I/O space and for expansion card address space.

Mark

[tofro]

The Trump card apparently does a number of interesting things in its initialization code to determine whether certain memory areas can be filled with RAM or not. For this, it maps its own ROM temporarily into the QL onboard IO space and looks for ROM in the complete expansion space. Wherever it encounters the $4afb0001 ROM signature, it will not page in RAM. Some details for that can be found in the MAME source codes here (thanks for that )

http://mamedev.org/source/src/mess/drivers/ql.c.html

Tobias

[Dave]

Great - some new 768K expansions (not 896K until you count in the QL's internal memory) would be ideal and as you say could breathe new life into some of the older disk interfaces.

Ah, but it also replaces the internal memory with faster external memory, so it is also re-provisioning that 128K.

Could I ask for the ability to add an EPROM with Toolkit II on it on the board, which would provide 99.5% of the functionality of a Trump Card then if you had an old simple disk interface. Remember that there are not many EPROM cards left out there for plugging in the ROM expansion slot, so people need a way of running Toolkit II from EPROM (yes, I know they can load it into memory from disk, but that is not the same, as it does not provide the timing required for the enhanced network capabilities of the EPROM version).

So what is the 0.5% of the Trump Card's facilities which would be missing?

The ability to do a quick FORMAT ram1_mdv1 to make a fast dump of a microdrive to ram disk....

I'll talk with Nasta and see if he can work that in - there's certainly room in the GAL and on the PCB. However, I'm wary of feature creep. However, I suspect it would still not help with the network issue because it would still be faster.

[Dave]

The Trump card apparently does a number of interesting things in its initialization code to determine whether certain memory areas can be filled with RAM or not. For this, it maps its own ROM temporarily into the QL onboard IO space and looks for ROM in the complete expansion space. Wherever it encounters the $4afb0001 ROM signature, it will not page in RAM. Some details for that can be found in the MAME source codes here (thanks for that )

http://mamedev.org/source/src/mess/drivers/ql.c.html

Tobias

A situation which can be simply avoided if we make an assumption that this will either be the only card in the system, or that if there is a card, it will be plugged into this card and will only contain a single ROM image, in which case we can override it into the top memory slot because we control the SP0-3 lines...

[Dave]

After discussing the TK2 issue with Nasta, I'd like to add the following:

TK2 would not have problems in OUR RAM expansion because it works as fast as ROM does

This works even for the 512k version as long as TK2 is LRESPRed. For the 896k version with internal RAM speedup, it does not matter - code will run at maximum speed anywhere, even if loaded into the screen

The 896k version will contain 32k of RAM at $10000..$17FFF unless the peripheral relocation to $10000 is activated (in this case it will have 16 of RAM at $14000..$17FFF). This can be used to emulate ROM, a ROM image (or two in case the full 32k is available) in there will be initialized by Minerva. It also survives reset since it's SRAM, and will keep the contents as long as there is power.

There will also be a plentiful supply of Minerva upgrades which also contain TK2.

[QLvsJAGUAR]

The ability to do a quick FORMAT ram1_mdv1 to make a fast dump of a microdrive to ram disk....

But that is a function of RAMPRT.

Ah - now who has the rights to that ??

RAMPRT has been released as freeware some weeks ago by Marcel Kilgus (maintainer of many Tony Tebby software). It can be found in the file qpac2_supp.zip on this webpage http://www.kilgus.net/smsqe/qpac2.html

[Dave]

All the pre-orders for battery adaptors and Minervas will be shipping out this week.

[RWAP]

There will also be a plentiful supply of Minerva upgrades which also contain TK2.

Just wondering what these Minerva + TK2 upgrades will be - remember the ordinary Minerva uses up the QL's standard 48K ROM space, and does not include Toolkit II normally....