What's the point of the MiniMig?

[HenryCase]

amigadave wrote:
The Minimig is an important first step toward what will hopefully become the greatest chapter of Amiga history - Open Source Amiga Hardware AND (I hope eventually will lead to an) Open Source Amiga-compatible OS!

That's my dream for Amiga's future as well, and a perfect way to continue Amiga's legacy. Think about how many of the greatest achievements of the Amiga have come from the community surrounding it rather than the hardware owners (the ones controlling the purse strings, not the engineers). Considering all that the Amiga community has been through it seems fitting that the hardware should belong to us, and thanks to Dennis' great work we have brilliant opportunities ahead of us.

The price is high right now, but think of the possibilities if we make the time to develop the Minimig. Amiga laptop anyone (easily done with the current version)?

Also, given some extra development we could have a device like the C64DTV (big inspiration behind the Minimig). Okay, so it's not likely to cost less than £30 like the C64DTV due to economies of scale, but an important step towards this new device has already been made (TobiFlex incorporating the 68k core into a Minimig FPGA). Given enough development time I predict it would be possible to have a Minimig in a joystick within a few years that costs around £50-£70. If the memory card feature was kept it could be acceptable VFM, as the FPGA core could be made to emulate other hardware as well. Having said all that, I think Clone-A would be a better solution for this. Watch this vid (skip to 40min mark for current Clone-A plans):

http://video.google.co.uk/videoplay?docid=7945941150233337270

The dream Minimig (for many) would be one that could emulate all Amiga hardware and was linked to a fast CPU (used to run OS4, for example) but as others have said we need to walk before we can run.

freqmax wrote:
Then there's the possibility to do demos coded directly in HDL (like vhdl) and really exploit the fpga power. Or even use the fpga capability to interface with other electronics as logic analyzer etc..

This is an exciting possibility. I wished Commodore had followed through with its plans for incorporating DSPs into Amigas (like the A3000+), who knows what great uses developers would have found for them. FPGAs are one step up from DSPs as they are programmable, so the possibilities are even greater.

[HenryCase]

dammy wrote:

The Minimig is an important first step toward what will hopefully become the greatest chapter of Amiga history - Open Source Amiga Hardware AND (I hope eventually will lead to an) Open Source Amiga-compatible OS! Then we won't have to worry anymore about Amiga Inc., Hyperion, ACK, and all the rest that have let us down year after year after year.

Open Source Amiga-compatible OS?  Think we already have that.  It just needs more Developers.

Dammy

I can't speak for amigadave Dammy, but I was certainly thinking of a Minimig + AROS future.

You raise a good point. The one factor that has held AROS back is lack of developers. It's currently looking impressive now, but just imagine what it would be like if it had strong support from the beginning.

If all these dreams for Minimig are going to come true, we have to realize that WE are the developers. I intend to start learning Verilog (there are loads of great resources for this on the Internet), so that I can make a contribution to Minimig one day. Who else is doing the same?

[HenryCase]

AmiGR wrote:
What are you going on about? There's nothing on the MiniMig that is expensive. The reason they go for $200 is the fact that they're built in single units. You could use a larger FPGA and put most of the peripheral hardware onto that, have it on a BGA package on a tiny PCB and it would cost next to nothing.

As far as I can see the most expensive parts of the Minimig are the FPGA, 68k processor and RAM. Would it be possible to build the RAM into an FPGA?

[HenryCase]

AmiGR wrote:
The FPGA hardly expensive in terms of the total price of the MiniMig. The XC3S400-4PQG208C costs £13 in single unit quantities. The 68SEC000 is £7 and could be replaced with a larger FPGA and an IP core. Dunno about the RAM, the SRAM used is the easy option, I'm sure it could be re-implemented with some dirt-cheap DRAM.

Generally, there are many things that can be done to get the price down but many of them rely on larger quantities.

On another note, has anyone got an XC3S400-4PQG208C or plans to order some? If yes, please let me know, I'll be eternally grateful! :-)

The 68K processor has already implemented in a FPGA core alongside the Minimig code (see TobiFlex's work), so the biggest barrier to reducing price seems to be the RAM. Would using DRAM rather than SRAM really reduce prices by a significant amount?

If you are looking for a seller of the Xilinx XC3S400-4PQG208C within the UK, a quick websearch came up with this result:
http://www.ashlea.co.uk/157.html

At the time of writing this they have thousands in stock:
http://www.ashlea.co.uk/scripts/availability.pl?partID=XC3S400-4PQG208C

I've never used this particular supplier before so I couldn't vouch for their reliability. Why not send them an e-mail for a quote.

[HenryCase]

AmiGR wrote:
Wow, I'll take a look at TobiFlex's work, I was planning on putting a 68k core on the FPGA myself. About the RAM, well, SRAM is generally expensive. If we implement a DRAM controller on the FPGA and getting the timings right, 2MB of EDO DRAM is £2.50-3 in single quantities, at least £2 cheaper than SRAM.

Well, that's certainly worth doing then (changing to DRAM). Anything we can do to make the Minimig cheaper is worth considering, are there any other areas we could make savings (such as incorporating the PIC functions into the FPGA)?

AmiGR wrote:
Oh, cool. It seems I'll have to go through University to use them, though, they do not appear to be selling to individuals.

If you do find out what price you can get through your University, can you let us know just in case there are interested parties looking to purchase the chip. I've just e-mailed Ashlea Components to see if they will sell to individuals and if I get a response I'll do the same. I tried to find the specific Xilinx FPGA through RS Components, Maplin and Rapid, no dice. Are there any other UK electronics suppliers we could try?

[HenryCase]

Message for alexh and AmiGR.

Please both stop this willy measuring type of argument, is getting us nowhere.

Reducing the cost of the v1.1 Minimig is clearly worthwhile, I'll go into ways we could feasibily do this in a minute. However, take a look at the comments made on the subject of the Minimig you'll see that a lot of people are waiting for a more powerful and expandable Minimig. Therefore, considering the size of the Amiga market, v1.1 is not going to be hugely mass produced.

Now onto the ways we can reduce Minimig costs. These are:
1. Find the cheapest sources for components and buy in bulk with other Minimig builders.
2. Look at ways the design can be simplified so it requires less money to purchase the components (this would require a PCB redesign too).
3. Look into companies able to do runs of PCBs and supply parts in one kit, buy in bulk.

You may even be able to find companies willing to assemble short runs of Minimigs, which wouldn't necessarily reduce the cost but would open the Minimig market up to Amiga fans not ready to solder units themselves. I appreciate the work of people like wizard who are doing something about this.

How about those of us who go to University, does your University have electronics assembly equipment?

[HenryCase]

AmiGR wrote:
Yep, sorry, I fear that I started it by snapping at Alex. Sorry about that.

Well I respect anyone who has the cohones to admit when they've done something wrong, so no worries.

alexh wrote:
I am suprised that there has been no ports of other "opensource hardware" to a MiniMig board yet?

It is a shame. How easy would it be to port one of those examples mentioned? I mean is the h/w they normally run on fairly similar?

[HenryCase]

alexh wrote:
At the data rate it was expecting etc. which would be the research bit that takes a couple of days. But you'd implement SNA (memory snapshots) and emulate floppy disks first.

Would it matter? I mean, I've used Spectrum emulators before that have tape 'snapshots' like the *.tap files for the C64 emulator mentioned by freqmax. The games loaded instantly. As long as all the data was loaded into the correct parts of RAM, I don't think you need to emulate the data rate too closely. Check this video out:

http://www.youtube.com/watch?v=VKu9qreiI3E

[HenryCase]

alexh wrote:

Check this video out:

If you notice he quadrouples the clock speed of the spectrum (3.5 - 14) and then quadrouples the playback speed of the audio sample thus keeping the timing 1:1.

Might be a coincidence but he doesnt demonstrate a game with a turbo loader.

Leads me to think that data rate is still important.

You're probably right that the acceleration of clock speed shown in that video needed to be a certain rate to match the tape speed. It shouldn't be too hard to get the basic timing right, and then the speed up the loading process in a similar way.

If a Spectrum emulator is on the cards, here are a couple of programs that may be useful:

http://www.worldofspectrum.org/taper.html
Can be used to see tape loading timings.

ftp://ftp.worldofspectrum.org/pub/sinclair/tools/amiga/SpecDisassemb20b.zip
Spectrum Disassembler for Amiga. Useful if there are any games that don't work.