To high of prices on new boards

[wawrzon]

if you look at this site latest threads:
http://www.amigacoding.de/index.php?board=5.0
the proposal is to define neo amiga approach as (as far possible) hardware independant fpga implementation and leave the hardware to the dedicated fpga hardware vendors, currently sockit board is beinng meant as price/performance most justifable platform, no custom low volume hardware can beat.

somewhat linked to that there is appraoch of a single developer trying to provide amiga fpga accelerators at very moderated prices:
http://www.majsta.com/
in intention to ignite a community based project.
i think there are welcome developments.

[psxphill]

Are you serious? You don't program on the pc because some of the development software is expensive? That's crazy!

Especially as most development software on the PC is free.
 
Even Microsoft produce free products
 
http://www.microsoft.com/visualstudio/eng/products/visual-studio-express-products
 
Although they seem to intentionally suck at protecting their paid for products to the point they might as well be free too.
 
The Visual Studio 2013 is currently free as well.
 
http://www.microsoft.com/visualstudio/eng/2013-downloads
 
If you hate Microsoft or something then use mingw/msys etc.

[nicholas]

http://www.samsung.com/global/business/semiconductor/minisite/Exynos/platform_partners_platform.html
 
this looks good http://hardkernel.com/renewal_2011/products/prdt_info.php?g_code=G135235611947

These things are nice but like I said, by the time a port to one is finished it will be out of date and most likely out of production.

[nicholas]

Are you serious? You don't program on the pc because some of the development software is expensive? That's crazy! It's like saying: 'I don't own a car, because some cars cost $200000.', or 'I don't wear shoes, because some shoes cost $3000'. Really, that's just nuts!

Lol

The GNU Compiler Collection is available for Windows, can't get more free that.

[majsta]

Yesterday I had desire to write something about the topic but I give up but now I see that someone mentioned me so here we go. This is just my point of view and my experience. Here I will try to explain how stuff works in real engineering world and production of hardware and then I will try to explain why we can't go this way. Remember I m not an engineer and don't intent to become one.

Updating classics Amiga hardware:
I noticed that someone said that Amiga community is not so small. Well if we just consider that there are 10 000 active people in Amiga community and that economy says that 10% of those people are interested into new product you have 1000 people who would buy some product. Let's consider another thing that we don't have just one model of Amiga that we have lot of them so we need to cover lot of different versions but if we decide to cover just one version we need to again to drop for 10%. So now you have 100 potential customers in best case. So let's talk about the prices of parts. If you consider buying some compatible CPU at that quantity it could be impossible to order so low quantity. Seller will agree but price will be so high that your final product because of the price will reject another 5% of interested people. In other words people will start comparing your product and performance against much faster and cheaper PC peripherals and that is first mistake. Include free work in all of this because calculating how much an engineer earn per hour of work could drive this low quantity to unbelievable price.

Let's consider me starting Amiga 1200 FPGA accelerator project.  
There are no trapdoor connectors and my conversation with every manufacturer was basically like this. Me: Can you produce large quantities of this connectors. Manufacturer: Yes we can how much? Me: 1000 pcs Manufacturer: It is too small quantity  price will be high between 10 and 15 USD/pc

So first I need to buy 1000 connectors because that is lowest number you can order at most of the manufacturers. I need to invest 10 000USD in just trapdoor connectors and in the same time I will not sell 1000 boards. not to mention that I can buy PCI connectors for 0.20USD/pc. Now, FPGA if we targeting 300Mips we need to buy very fast and big FPGA and that FPGA can cost 150-300USD, again we can't drop the price because we are ordering just 100 of them because we won't sell more than 100 accelerators. So like you can see, not to mention 10 other components needed we are at huge investment and in the same time at huge price for potential buyers.

Creating new hardware
First problem is prototype board and one or two those boards can cost up to 3000USD, again quantity is the problem. Again serious engineers can earn 100 000USD per year and you need to assembly a team so you need to spend much money just to pay them.

After investigating all of this I made some decisions.
1. open source everything and people will join, experienced one will teach newcomers
2. test tons of parts and find optimum regarding price-quality,spend your money to test them all so we will not end up in situation others waste their time and money
3. earn money for living from something else don't expect to become rich from creating Amiga hardware
4. remove design dependence from all parts that will not be manufactured in future
5. publish all news in process of your investigation,again, not let people waste their time

So we are now not depended from MC68K series CPU, we are not dependent of specific FPGA model or anything else.

[ElPolloDiabl]

I don't agree with the open source option. Open source will work if there are bounties. Linux has paid developers because they can make money doing tech support.

[wawrzon]

@ElPolloDiabl

I don't agree with the open source option. Open source will work if there are bounties. Linux has paid developers because they can make money doing tech support.

absolutely wrong! for instance aros development is basically continously going on without any bounties. bounties cover particular demands that are not along the path the developers would invest their time in otherwise, but the main development is not funded anyway. this said i dont mind if there was more reward towards the developers, but it would never fullyy reward their contribution. so regular business cannot be the motivation anyway.

@majsta:
thanks for refreshing and at the same time open. motivating and reasonable input.

[JimDrew]

Majsta, you need to look into some better sources.  You can get connectors made overseas for next to nothing (by comparison to the U.S.).  You pay for the tooling and pin fixture.  That can be a few thousand dollars.  After that, you run as many as you want.  They will probably be a dollar or so, depending on the number of pins.  So, for 1000 pcs, I would be very surprised if you reached a total amortized cost of $5.00 per connector.

PCB protoyping is also very inexpensive now days.  Do a search on alibaba.com, globalsources.com, etc. for PCBA suppliers.  PCBA means both PCB and assembly.  You would probably use a 4 or 6 layer board for the FPGA accelerator, and those are pretty simple to make.  For two layer boards, I don't even get electrical testing done.  If a company can't build a two layer board without issues, you should choose a new source.  It's not much of a stretch to get 4/6 layer boards done.  When you get into the 8+ layer boards, you need to really start watching your reject ratio.  The suppliers I use currently build 12+ layer boards as a common daily build.

Getting BOMs quoted correctly is key.  When your BOM calls for a resistor, cap, LED, diode, regulator, etc. do not give an actual part number as a reference - instead, give a specification and then request the exact datasheet for the part that is quoted in your BOM.  I have found if you give a reference like a Digikey, Mouser, Newark, etc. part number that the price for your part is suddenly much higher than it should be.

ANY part that you are worried about being a counterfit - such as a FTDI, Microchip, Atmel, ST, TI, etc. buy it directly from the manufacturer.  There are a LOT of knock-offs.  Save money on the glue logic, PCB, and assembly, and don't skimp on the CPU, FPGA, PLD, etc.  You can't afford to risk a knocked off chip coming back to haunt you later.

[NovaCoder]

Let's face the fact that nobody is going to make much money out of a classic FPGA accelerator board or any classic hardware for that matter.   Even Jens has said he's struggled to make much return on investment from his recent products.

Someone is only going to make such a board for love, lots of funky love.   That said, it might be possible to at least break even and (maybe) even make a (small) profit if it was done right.

My suggestion would be to go for a semi-open-source approach to the FPGA core design.    You'd offer up the code and people could hack away and use that code to create BETA cores but control what code is good enough to make RELEASE cores.   There are lots of very clever people who are still in the classic Amiga community who also have FPGA skills they would contribute to such a project if it was managed properly.

[cunnpole]

Let's consider me starting Amiga 1200 FPGA accelerator project.

I'm pretty sure I remember Jens saying he had already done the tooling to create the connectors and was willing to sell to other manufacturers such as yourself

[psxphill]

These things are nice but like I said, by the time a port to one is finished it will be out of date and most likely out of production.

It depends on how quick you port. As long as you get a board with open source Linux drivers then a port shouldn't take any more than a month per board.
 
All "new Amiga" hardware is out of date.