Any need for a MiniMig 3.0?

[Illuwatar]

After reading about the success with the FPGAArcade project and the development of NatAmi, I start thinking of maybe quit the line of MiniMig designs. The 2.0 is completed, fully tested and with a few happy owners (I have basically sold out all units I had, more could be ordered if there are enough interest). The FPGAArcade is the biggest competitor to the MiniMig and if he can bring down the end customer price below €250 for a working board, then I can't compete price-wise even with the 2.0 MiniMig. In my plan, there is an improved 2.0 that integrates the Yaqube ARM-controller and more SRAM, but is there any idea to continue with it?

[Illuwatar]

There were some replies here...

And most of the do have one common thing: a lot of added features. As I see, the FPGAArcade and NatAmi would cover most of these at the end (and with expansion slots, the missing ones could be added later). No need for me to get in to that with a third project that ends up almost the same as the FPGAArcade (better to buy one of these boards and save my time for other projects). My intention with the MiniMig 2.0 (and up) was to keep the compability with the Dennis core (the 2.0 runs the core from his webpage unmodified), just add to the surroundings what was missing on the MiniMig 1.1. All these new designs (including my Mini MiniMig) requires modification to the FPGA core in some way (and with the ARM, new loader firmware too). The board had to be hand solderable too, which rules out BGAs (that is required for larger FPGAs).

Anyway, I might concentrate on the Mini MiniMig instead. I need anyway to redesign it as some parts can't be bought anymore. And there is an interest of a compact/portable MiniMig. The ARM could easily be integrated into it and doubling the SRAM isn't that hard. With the RGB I/O from the FPGA exposed, it should be possible to hook it up to a RGB-type TFT...

[Illuwatar]

That is the only way for me to build the prototypes as I'm lacking the ability to solder non-legged components (no hot-air, no ovens...). I think that mass production of the boards would be cheaper too with this types of parts (if someone would do that).

[Illuwatar]

mikej: what company are you using for your boards? If they are able to put the BGA in place for me as a step in the manufacturing process, there will be new possibilities for me to both extend and shrink the design.

I did a quick look at Digikey for the Spartan 3 models available. There are two bigger models (1000k and 1500k gates) possible. They should be good plug-in replacements for the XC3S400 if more gates and I/O are needed. Or I can turn the idea all around and use a BGA version of the current FPGA and shrink the design even more (compared to the Mini MiniMig). Time for 6-layer board maybe...

[Illuwatar]

It seems by reading this that the wish is for a small FPGA-board with a large FPGA (gate-wise) and 16 MB SDRAM. Basically, a generic FPGA lab platform with a few added features needed for the Amiga to be playable (Audio, video and ports for controllers), squeezed down to minimum size. If someone could take care of the SDRAM adaption and the softcore, there is a chance to make a really tiny "MiniMig" (the massive changes to the core invalidates it as a real MiniMig as it won't run the native Dennis core). The Big Rig approach is taken care by the Replay and NatAmi, so I will not compete with them.

[Illuwatar]

FrenchShark:

I have a new board in the work.
What I can tell you is :
- Bigger FPGA
- More memory
- The PCB fits in my hand
- It is cheap

Any more details about that one (especially how you solve the last point)?

Regarding size, my current Mini MiniMig is 110 x 65 mm in size. It is built as a module that requires external connectors or adapters to be fully useful as an Amiga. One of the reasons for it was to be able to make a portable version. But if we can ditch the hardware CPU and go for a softcore and use BGAs for both FPGA and SDRAM, the thing could be even smaller. 4/6 layers are required and parts on both sides would be necessary, but the thing could be really tiny...

The MiniMig 2.0 could be built rather cheap if there are a manufacturer around that could do the work for me. The PCB itself is just 2-layer and there are no BGAs in the design.

[Illuwatar]

countzero wrote:
i'ld say a minimig with yaqube's arm board added in and possible fast ram would be great. in mini itx form of course. I think the would idea of a portable amiga is pointless cause it requires many hours of hair splitting design and in the end no body will spend >100$ to a hand held device to play turrican on the way to work. hand held devices need more functionality.

lets keep things simple and move step by step. there's no functional AGA core or soft 68020 core in the horizon yet. so a minimig based on yaqubes improved ECS, fast ram, and yaqubes arm board would be in order. (of course in mini itx form, original minimig form would be pointless too cause many people would have to spend fortunes again for custom cases)

That is basically my first intentions for the Mini-ITX MiniMig 3.0. 8MB SRAM (not SDRAM), the same FPGA as the original and the ARM-design from Yaqube integraded on the board. I think it would be possible to pull that design off on a 2-layer board too. Adding expansion connectors for CPU-bus would be possible too. But it seems like waste of time as the scenario looks today.

wolfchild wrote:
@Illuwatar: Maybe it's time to move to other projects, such as daughter boards for the Replay ;-)

You might be right. I have several other projects in the queue anyway. And where can I get a Replay to play with... ;-) :-)...

[Illuwatar]

That is not a problem that can't be solved... :pint:  :pint:  :pint:

By the way, the ARM, is it coded in C(++)? Otherwise, I have to learn a new assembly language (again)...

[Illuwatar]

I think I will have a plan for this (finally). The Mini-ITX line will be put on hold as it is now (the 2.0 is the final incarnation of the classic MiniMig design). The FPGAarcade/Replay takes it to the next level. But just withing the last weeks I have got a few requests and questions regarding the Mini MiniMig. As my design is already outdated, I will improve this one instead. In some applications, a module is more suitable than a full-fledged board with connectors and all that stuff. As it is basically a FPGA-module, there are no requirements to run Amiga on it.

So, my plans are these (to fulfill at least some requests here):

* SRAM replaced with SDRAM (16 MB or more?).
* LVDS instead of RGB (many native TFT panels interfaces directly via LVDS).
* Yaqube ARM-controller instead of PIC.
* Larger FPGA (Spartan 3E), BGA package would be nice, giving us up to 1.6M gates to play with.
* Additional I/O (within the limits of the module size - size is prio 1 in this project).

This design will require a new core to run as both the RAM-interface and LVDS must be implemented. Also, should I keep the 68000 or is there a softcore good enough (even more space and cost savings)? I also have to find someone to put the BGA on the PCB...

I don't know if this is of any use, but it would be fun to pull this off. Some help would be required as I'm totally lost regarding Verilog/VHDL coding. Also, some ideas how to design SDRAM and LVDS interfaces would be helpful (as I have seen, the board layout is more critical for these compared to plain SRAM and analog RGB).

By the way, the current stock of 2.0 boards are empty but there is a plan to order new PCB's. Any interested in having one?


Edit: a 32M x 16 133MHz Micronas single-chip SDRAM and a Spartan 3E with 1600 000 gates will be the base for it - the work with making symbols have been started...

[Illuwatar]

My idea is to make a tiny module that could be put in to whatever design a FPGA with accessories is needed. It could be a base for a compact, portable MiniMig console or even control the ignition of an engine...

The Mini-ITX line I was planning were basically voted down and I can see the reason. No idea to make three almost similar designs/ideas (Replay, NatAmi and my Mini-ITX MiniMig). But the module idea does not rule out ATX/mATX/Mini-ITX board designs either. Actually, it could be a rather clever design: Make a base board without any "brain" where the module plugs in to get the ports and accessories you like. Document the module and all interfaces well, make it Open Source and create a platform for development. New FPGA's coming in? Just replace the "brain". Need an other CPU? Design your own module...

Basically this idea:



But in Mini-ITX format (or ATX/mATX if it is prefered).

[Illuwatar]

mikej - I haven't seen a final end-customer price for Your design (and the adapter boards) yet (or I have missed it). So, it is actually hard to see what I compete with. But I can't think that a module without expensive connectors and less features would be more expensive. And I still think that something with a width of 170 mm is too large in some designs. There are some interest in the module I did already produce (but it can't be built anymore as some parts are obsolete). Anyhow, when there is a final Replay out, I will certainly buy one with all the expansion things you may throw on it. Must sell some more 2.0's first...

The idea of placing two 16M x 16 SDRAM's side by side for a 32-bit RAM-datapath sounds tempting. At least something new...

[Illuwatar]

Ok - write me down on one board and include the additional modules to it. I may be interested in that shrinked down version too when you have it done.

I have been studying your latest schematics and I think there are not much for me to add actually. Specially if you make a shrinked down version too. If you manage to drop the final unit price below €200, then it is no need for me to do more in this area regarding the hardware. The current 2.0 will continue to live and I may look into a 2.1 version where I replace the PIC with the Yaqube ARM, no other changes done. It should be possible to build these well below €100 if I don't have to do it by hand...