exxos's Atari, Amiga & retro forum

I thought I would start a thread about this.. basically "whats next". I am just "thinking out loud" with all this currently.

Currently the H4 is near completion. No doubt there will be a H5 revision fixing any issues on the H4. Though I hope the H5 will at least get flash and IDE built on.

As people know, Icky is working on a case as we want to one day built a complete machine. Though with a lot of parts being obsolete it's not so simple to produce a complete machine.

There is the aim to replace the blitter,mmu,glue with the suska cores. We may lose some compatibility (I hope not to much) but its the only direction we can really go in as there is no "new" supply of those chips. It then becomes a issue of all the IO connectors which are hard to find, like the DB19, ROM etc.

So I envision 4 directions for the project.

1) A "legacy" board. The "Hx" series which will become the "legacy" board. Will basically be the H4 as it is now but a more final board rather than a development machine. So that series will conclude in the near future where the "final step" would be to build on some goodies such as flash and IDE. This series will be a updated and fixed board over the original STF. So rather than people fighting with endless "bugs" on the original machines, they can swap chips over to the Hx series and "fix all in one go". There are a huge amount of jumpers/headers on the H4 to allow a future FPGA dev-board to plug in which will replace the blitter,mmu,glue. Though we really need the help of a experienced FPGA coders to help development with the FPGA stuff.

2) A "gaming machine". With a custom case (maybe the size of the STM) Will still have floppy and internal PSU. But only have monitor output on the rear (with RGB or FPGA shifter for HDMI output). Will likely use suska cores for the main IC's. No other connector will be there. This will solve sourcing problems and build costs. Basically a "minimal" STF machine which can play games. Likely flash-rom and IDE will be built on and likely just 1x 68K expansion socket for a booster. As we can do a custom case, we can make the board half size (kinda like the Alpha was smaller) and reduce costs with the smaller board size and slightly smaller case if possible. So a complete gaming machine, where you don't have to spend huge amounts of time and money fixing a STF then having to go through it all again trying to get a hard drive working etc. Everything will be ready to go "out of the box".

3) A "Super ST". This will be based on the "legacy" Hx series, but adapted for use with suska cores to replace the main chips. Again we may lose some compatibility, BUT, With a new chipset we can crank the speed up to maybe 32MHz. Higher speeds will break stuff anyway. It would be nice if we can down clock back to 8MHz for a fully compatible machine, but I think this is unlikely.

4) We go full FPGA and make a cheap as possible machine based on Suska or MiST. We would basically be re-inventing the Suska / MiST. Though as these machines exist anyway, I think it would be a somewhat pointless direction to go in. It also kinda goes against the remake project goals. Where I wanted to build a STF machine which had all the bugs and issues fixed with the option of easy upgrades.


In terms of when a "new board" will appear. Well, myself and Icky have a huge amount of work ahead with flashy-clock, IDE, boosters,cases,FPGA stuff etc etc . So this will keep us busy for the next 12 months easily. We may produce a "Fixed/updated" Hx at the end of next year, though this depends on demand and other projects progress. Speculation of building on flash & IDE is still not decided. In some respects, having these as a plug in option can keep the buyers inital buy costs down and can just plug in the stuff they want. Also assuming a fully built machine, having stuff "plug in" makes debugging faults a lot easier. Though it would be nice to have flash & IDE built on the board ultimately. So there is a lot to think about and consider.

I have been looking at the H4 a lot in relation to moving stuff about in the future which would be the H5 board.

The plan would be to remove the falcon RAM board and move the MMU & SIMM more towards the shifter. The GLUE would get relocated more where the MFP is, as a lot of lines from GLUE are going right across the board. In fact some Atari boards have the GLUE located there.

CPU & BLITTER would get relocated to the front of the motherboard next to flashy clock basically. ROM IC would likely be removed and have a integrated flashy-clock instead. One huge pain in all this, is the PLCC & SMT MMU & GLUE . I think a next build I wouldn't bother with the SMT ones. I think pretty much all the resistors and caps would go SMT versions, along with all the TTL logic chips as well.

Though its possible the H5 would never get done if Icky and I can get the MMU,GLUE,BLITTER FPGA working over the next year.

Currently the H4 is revision F, and @Icky is working on revision G. This sees all the current H4F fixes but also icky is moving the resistors and caps to 0805 SMT parts. The idea being to get some generic SMT parts preassembled. I may well replace the logic chips with SMT as well. Currently there are no plans to produce that board unless there is enough demand.. I think as its not much different than the current rev F boards that nobody would be up for another run.

I am thinking the ultimate direction may be to redo the PCB to fit a STM type case. As we are looking to 3D print cases, the smaller the case the better. We think we can still fit the floppy in a STM type case.. This would be the "legacy" design.
We may also strip that design back for use as a cheaper gaming machine.
.

The FPGA will be a add on board for the H4x series which icky is working on. Because suska isn't cycle accurate, the FPGA based machine will be 16mhz or faster system only. Should cycle accurate cores appear in the future, it would only be a firmware update.

Happy to back you again for the next run / idea / creation.

You could always go to Gotek internal butkeep the external floppy port for real floppies with that built in drive switch. The oled could somehow be included in the case badge?

I'd also back a new design...but for me it would have to still run some of the old sequencers/MIDI packages.

That's one longer term aim for me with the H4...to fit an accelerator and see what still works.

I'd expect cubase 3/score & Logic to be ok...less sure about Notator/Cubase 2 (but that's partly why I need to test my collection... after finishing the H4(s)... :roll:

As long as some run then I'm fine...

Happy to back again ;) I'm all for compatibility with original machine. Fixing things that were broke on Mongrel ;)

I'm basically thinking out loud constantly with all this but I'm trying to make people understand what is going around in my head continuously..

We are basically at a crossroads between speed and compatibility. We basically cannot have both..Not on a single machine that is anyway..

I am not talking about adding a accelerator onto a basic 8Mhz machine either. I'm talking about a full proper 32Mhz bus... I mean sure you can still put a 030 on it or whatever... Because the replacement chipset is needed for high speeds, there will be no more 8mhz mode.

This is why the motherboard is called the "mongrel" because it is a mash up of all sorts at the moment. I have actually rode I think volumes to Icky about the future and direction to move in and it is far from easy.

But one thing I think has basically been decided, and as pretty much known already, is the FPGA board to replace the MMU,GLUE,BLITTER with the suska cores. This way the end-user can decide if they want the classic or accelerated version.

In terms of the 16MHz mod some did on the Alpha.. Still a half topic of discussion.. I am leaning towards not doing this on the H4 because I know it will be a lot of time messing about for basically a dead-end project. I think the time would be better spent working on the FPGA cores In getting them working at 16MHz then 32MHz.

What I would like to do is totally redo the H4 PCB & layout.. but that would be some months of work. But that would be the classic machine final... The problem is this could not double up as a faster machine which can accept the FPGA board.

At the moment all the main chips are squashed up together which makes it easy to plug in a single FPGA board. but the routing for that setup is not particularly great. But in order for me to be happy with the whole thing, the chips would need be located in different places where a single FPGA board could no longer be used...Using multiple FPGA boards is just adding expense and complexity from no real reason. So it would basically at the moment come down to redoing the whole layout just for the sake of it.

But this is why I am thinking of a STM sized board... with the thought that one day printing the smaller cases would be a lot more beneficial for a complete machine.. This is what I was doubling as a "classic or gaming machine". With the problems of obtaining various connectors, I think they would just simply not be fitted on a gaming machine and neither would the expansion sockets. Basically stripping it down to the minimum for a stock machine. Anything I would likely add would be a internal IDE drive so people can easily load games on it from hard drive. It would still have a floppy drive interface, but does not necessarily mean people have to use it or even fit the drive.

So basically the H4 would evolve into a more FPGA based system which is faster... But also branch off into a de-evolution into a more basic classic machine. It is really those two directions I can visualize.

The problem with a FPGA based system, is as mentioned many times before, how far does it go before we end up recreating MiST/suska etc. my own personal roots have always being for maximum compatibility as there is simply no new "classic ST". As people know these classic machines are just becoming unrepairable these days. So salvaging them for their souls for a new motherboard is the only way this classic line will live on I think.

As I will just not be able to work on 2 motherboards at the same time... I am basically thinking that the next "H4" would basically be the final board of that series.. It is geared up for testing out various add-ons which will keep us all busy for the next few years easily I think. When things do eventually quiet down a bit I can start thinking more about the classic series of machine.

Does the "Next Generation" machine have to keep the original ST(m) footprint at all?

Surely a small, squarish pizza-box with keyboard socket would be easier? (Mini-MegaST?) Fit an STFM keyboard in an external case of its own and make up a lead to fit between the two. Those should allow for even smaller 3D printed cases too.

+1 for the enhanced version. I'm not sure I would ever go for a classic machine as the H4 fits this bill for me in it's current standard capacity.

Happy to test both though I guess?

Also +1 for the change of footprint idea but that's entirely up to you @exxos obviously. Whatever you choose needs to be Atari ST in style and not just another basic black box with an FPGA in it, as you say, theres the Mist already for that. (Already have one.)

Until then, let's hack the H4 into some kind of Falcon killing beast of a machine.

PhilC wrote: 12 Mar 2020 22:11 Also +1 for the change of footprint idea but that's entirely up to you @exxos obviously. Whatever you choose needs to be Atari ST in style and not just another basic black box with an FPGA in it, as you say, theres the Mist already for that. (Already have one.)

Indeed, that's why I suggested a smaller (possibly) Mega ST style case, with internal PSU and all the usual ports (probably).