My God, this is all to good to be lost for ever. Someone has to take the lead on one or many of these projects or we have to find money to give to Nasta (who, I believe is still the best person to make it work).
Also, I offer my help to write the firmware for such a beast. My assembly skills are still quite good. However it would be my first attempt at such a task so help will be welcome! François Le 07-02-19 à 03:17, Rich Mellor a écrit : > Ah - I do have an email here from Nasta about the Ultra Gold Card > project: > > Well, I wish I could give you good news about that project, but I > can't. > It has basically been mothballed at the same stage it was when last > talked > about. I have a number of parts ready and a partial design, but at the > moment, trying to ressurect it would be VERY difficult financially. > not > impossible if it was only down to the money, but what makes it so > is the > complete lack of time. There is, however, a lot of documentation, > which > couldbe used as a good guide to completing the project. > > In essence, when last revisited, the GoldFire (that was the project > name) > spec and documentation was upgraded so that a pair of 68060 CPUs > could be > used. One was designed as optional, but would have been included in > the > prototypes, since I have a number of used ones graciously donated > by Tony > Firshman. These are the 68EC060 version, 66MHz if I recal right. > They were to be coupled to a SO-DIMM SDRAM, 256Mb - the type that > was used > with (now older) laptops. It is still available. This was to be the > standard, and also maximum configuration as in the eman time, RAm had > become sufficiently cheap. On the IO side, there were the usual > floppy and > parallel ports (full bidiractional parallel port), but also PC > style PS/2 > mouse and keyboard connectors, as well as 2 fast serial ports. > These were > all handled by a single Ultra-IO chip, that also comes from the PC > world. > Extras were to be a PC style sound chip and a Ethernet 10MB/s > network chip. > A 2M byte flash ROM was intended to hold the system software (SMSQ/E), > which would, of course, be upgradeable. > All of this was designed and well documented. > The part that was documented but was not fully designed was the singe > large logic chip that conencts all of this into a usable single-board > computer. The reason why it was not done, was that the manufacturer > of the > logic chip had been through a merger with another company, and as a > result > ended up changing all their developement software - requiring from > me a > seizable additional investment to buy it, along with an even bigger > investment in time to learn how to use it (it's a different > approach and > programming 'l;anguage'). To make it worse, the actual logic chip > ended up > being scrapped by the new company, though, fortunately, there is > still a > compatible one available. > The logic implemented by the chip is quite complex. It involved not > only > connecting the SDRAM to the twin 68060 (that was actually the simpler > task) but also being a 'bridge' betwen the VERY fast world of the > 68060 > with the comparatively veruy slow world of the QL bus. In order to > provide > an open path to further peripheral developement (like an Aurora II and > Qubide II), it also incorporated a protocol that could use the > existing QL > bus signals to transparently implement a fully 32-bit data transfer > protocol for newer peripherals, while still providing compatibility > with > the old peripherals, as well as improving their performance. I have > done a > lot of work on this, the protocols and hardware signals are fully > documented, but the logic for the logic chip was not designed fully. > Finally, the GoldFire also was to include a small on-board > switching power > supply with very high efficiency, which made it un-necessary to > have an > extra heatsink (like GC or SGC) as well as adding anability to work > in 9V > or 5V powered bus systems without any alteration or hardware setup. In > fact, the GF would even be able to supply +-12V at a small current for > serial ports external to it, and enable the user to build a low power > system running off of 5V only. This was fully designed and even > tested as > a separate module. However, today it could be made cheaper and smaller > with theu se f more modern components. > > GF was a very ambitious project in a situation where the cash > available > for it's completion was fast dissapearing. Follow-up projects were > also in > the works, as the GF opened up a lot of new possibility. The road map > included 3 key components to building relatively (for a QL) high end > systems using very little power and occupying very littl space. 3 key > components were to be designed next: > 1) New bus backplane system, with a different connector that could > still > use the old boards as well, but added a whole row of ground plane > pins in > order to introduce first class signal integrity and noise immunity, > necessary for fast data transfer. > 2) An Aurora II with dual monitor and 64k color capability, with extra > tricks added to alow quick color space conversions from 24bit, > 8bit, 4bit > or 2bit color, alowing fast compatibility, as well as a means to write > 24bit color stuff for later implementations. There was even an > option to > add a DSP chip with direct video-RAM access to add high-level > accelerated > functions to the graphics. The fact that I even have themain parts for > this project should be an indicator of how serious it was. > 3) Finally, there was also a Qubide II of sorts. The existance of a GF > would have made it quite easy to do, quad ATA ports that could be > used in > a RAID configuration. > > As you can see, this would eventually have led to a complete > abandonment > of legacy hardware, which, as much as it is useful, is a great > burden to > adding new features. > > When it became clear that GF had to be put on hold, I decided to > design a > stop-gap device of sorts, that would alow software developement in an > environment similar to the GF, but on the 'old' QL hardware. This > was the > EtherIDE. It incorporates some of the features of the Qubide II as > well as > the ethernet network hardware from the GF. > This is a fully designed piece of hardware, including logic chips. > What > it's missing are the funds to make the 4-layer boards, as well as > passive > components and conenctors to build them. I have chips for a run of > approximately 50 of these, but no funds to make the PCBs. > However, this piece of hardware has a bottlenek - it relies on the > SGC or > better to work, just like the Aurora. Unfortunately, the SGC supply > has > long dried up, so primairly, this would be of interest to users who > have a > SGC and Qubide, as an upgrade. > In light of what has happened in the QL world in the meantime, it is > probably prudent to replace the Ethernet hardware with an USB host > chip - > in the meantime, these have become available for a fair price, and it > would be a relatively minor redesign of the board. This project, > should it > still be deemed feasible, is ready to be put in production if > someone can > foot the bill for the missing parts and boards. > > Finally, because I am involved with various embedded systems at > work, for > a wgile I also had a QL derived system under consideration. > Although for > the imemdiate future we have decided to go with what we already > have, the > QL technology based system is still under consideration. A big plus is > fairly easy developement, assuming the initial work to port the OS > is done > first. At the 20-eth anniversary meeting in Eindhoven, i gave a > small talk > about this as well as being able to talk at length with Tony Tebby and > Arnauld Nazarian about it. Arnauld wants a small QL machine to > market as a > toy computer, and has some funds available. Unfortunately, this > project > has a built-in potential to provoke a conflict of interest, since a > large > portion of the developement money would be (and has been) injected > into it > by my employer. > This small QL derivative is based around the 68SZ328 'Super > DragonBall' > CPU, which itself is really a 'Turbo' 68000 with many nice add-ons. My > employer has already invested in the basic (and most expensive) > parts of > this system - the CPU, the RAM, the Flash ROM and the USB > controller chips. > It should be noted that this is not a real 'Ultra GoldCard'. > Although the > 68000 core in the DragonBall works at 66MHz and uses a > comparatively fast > SDRAM, and has a full 32-bit address space, a 68060 or even just a > 68020 > at this clock rate would be substantially faster - because this > still is a > 68000, so 16-bit bus as well as no caches etc. By my estimation, it > should > still be about twice as quick as the SGC. > What makes this an interesting project, is that it is a complete small > computer. The chip also has a built-in LCD controller that can also > drive > a CRT (at resolutions similar to and even slightly larger than the > Aurora), in 256 colors, it has numerous peripherals including a CF > and MMC > card reader, USB port, sound output and input (mono), and is also > capable > of directly conencting with RAM and Flash chips without the need for > external logic. > Unfortunately, this versatility also poses two problems: > 1) It is absolutely necessary to use a multilayer PCB for this > design, as > the chip is housed in a truly tiny 12 x 12mm case that nevertheless > has > 192 pins. The PCB uses very fine geometry and at least the > DragonBall chip > has to be assembled by machine as it is impossible to solder it in > a home > environment. In order to make a space efficient design, the board > has to > be designed by hand - not a major consideration since all of the > ones I > did were done this way - until you take into account that this > trades a > signifficant money saving with a signifficant increase in time > needed for > design. And, if you remember the start of this mail, for me time is > at a > premium. > 2) The manual for the chip has about 1000 pages, configuration is > quite > complex and extensive, so some time is to be reconed with till the > OS is > ported or all the signifficant capabilities are supported in software. > Because some of the functions also depend on the multitude of ways the > external hardware can be configures, perusing this manual is also > needed > for the hardware designer. So, more time is needed. Fortunately, I > have > done most of this already. > > Tentative specs for this system were as follows: > - About the size of an Aurora (160x100mm) > - Optional QL style expansion port to asist with developement (Would > eventually be phased out) > - Two CF slots that are arranged in such a way they can either hold > a CF > card or a CF-sized small card with expansion on it (like a IDE > adapter, > modem, etc). > - On-board 64M SDRAM (this is the maximum configuration) and up to > 64M of > parallel Flash storage (this is visible as regular ROM and programs > can be > run from it, though at a speed premium), optional NAND flash chip > can be > added on board for more bulk storage (similar to a built-in USB flash > memory stick) > - On-board graphics, with a digital interface for a small LCD screen > and/or analog out (Aurora style) for VGA monitor (CRT or LCD). The > chip > can do up to 32768 colors but only in small resolutions, and up to > 1024x512 resolution-wise. The idea is to use it in [EMAIL PROTECTED] colors > (Aurora compatible). It should be noted that it does not directly > support > the old QL 4 and 8 color modes so these would have to be emulated > (should > notbe a problem). > - Matrix keyboard interface (8x8 like the QL), using one of the > DragonBall > SPI inputs it should be possible to emulate a PS/2 port in software as > well, but this should peraps better be done in a different way (for > instance by a small PIC or AVR microcontroller) > - Serial port(s) > - 2 USB ports (host or device, so called USB-OTG) > Parts that I have for this: 10 68SZ328 chips, and enough Flash and > SDRAM > for 10 prototypes, as well as the USB-OTG controller chips and CF > sockets. > Needed: some design work and a bunch of passive parts, as well as a > decision on how it is going to be powered. The current idea is > either from > an external 5-7V supply or 4 AA cells (alkaline or NiMh). No charging > function is provided. > Main problem: lack of time (since my employer is interested, it is > possible he would be willing to put up the money). > > So, there you have it. Feel free to pass this on to your friend, or > anyone > who you think is wiling and capable of continuing these projects, I > would > be glad for any help or even just to give them over to someone else to > finish. _______________________________________________ QL-Users Mailing List http://www.q-v-d.demon.co.uk/smsqe.htm
