New MiSTer hardware board

[noel]

Hard to tell what is critical unless you try to port the cores to a new board.

Ok it looks like I need to try the us02 module before going any further.
I will receive a us02 dev kit next week, it's the start of a great adventure

[WolfgangBlack]

I thought the DE-10 nano was essentially subsidized making it cost far less than it should.
Making different add on boards seem more logical than trying to make some all in one board that will end up
costing a lot more than a standard MiSTer for little to no advantage.

[jorge_]

I thought the DE-10 nano was essentially subsidized making it cost far less than it should.
Making different add on boards seem more logical than trying to make some all in one board that will end up
costing a lot more than a standard MiSTer for little to no advantage.

A full MiSTer setup is well over $300 once you get everything you need, and price and capability aren't everything, there is also form factor to consider. The biggest problem with using a DE10 platform is the straight jacket it puts you in for case options and i/o placement. There are clever ways to get around this, but a true MiSTer-compatible board in a more consumer-friendly format would be welcome by many people.

Certainly no need to discourage somebody from trying!

[Sorgelig]

A full MiSTer setup is well over $300 once you get everything you need

This is very subjective. You may spend even $1000 if you consider some expensive case, elite gamepad and other things. It doesn't change the fact you only need de10-nano + SDRAM as a bare bone to play all cores.
Custom board won't make it cheaper. And the way you expect it will be $600+

[jorge_]

This is very subjective. You may spend even $1000 if you consider some expensive case, elite gamepad and other things. It doesn't change the fact you only need de10-nano + SDRAM as a bare bone to play all cores.
Custom board won't make it cheaper. And the way you expect it will be $600+

It wasn't a criticism. I was just pointing out that many (most?) people spend nearly $400 to get the DE10, I/O board, SDRAM, USB hub, case, and wifi and bluetooth dongles, so an all-in-one solution doesn't have to be *that* inexpensive to compete. MiSTer is worth every penny, and it's one of my favorite possessions.

I agree that the FPGA size we need is not quite cheap enough yet, but eventually I would love to see a layout with all of the above functionality built in, and USB port up front where they belong

[Sorgelig]

The key point is that you DON'T have to spend that much. Less than $200 is enough to start to play with MiSTer. And then later as you will get better picture how you want to use it, you can add more money to the MiSTer. Also it's DIY project so you can choose to make addon boards yourself and it will be way cheaper. If you choose DIY path then DE10-nano with ALL addon boards will be much cheaper than $200.
Entry price $600+ is complete different class and area. While some people are ready to spend that much from start, it's not a good price to popularize the project. And we are talking just about the board itself. Some fancy plastic case will add much more on the top.

If you go to electronic hobby, then today it's much more boring than in 80x-90x. Everything is inside big chips which hard to solder by a newbie. And it's more complicated when it comes to BGA chips. So it's not like before when you will need just bare soldering iron. You have to have a lot of expensive equipment and big budget. And it's more predefined now as chips are pretty much define the functionality.
I wanted to make MiSTer as a fusion of new and old technologies where you can do something yourself and be rewarded by result.

[WolfgangBlack]

Certainly no need to discourage somebody from trying!

By all means, if you have the interest do it up.
After all, that is how we got MiSTer.
Just wanted to make sure you know a DE-10 is sold for less than they should cost.

The key point is that you DON'T have to spend that much. Less than $200 is enough to start to play with MiSTer. And then later as you will get better picture how you want to use it, you can add more money to the MiSTer. Also it's DIY project so you can choose to make addon boards yourself and it will be way cheaper. If you choose DIY path then DE10-nano with ALL addon boards will be much cheaper than $200.

That's how I did mine and printed a case. I did eventually buy the 7 USB port board and printed the bottom to make it work.
Still using the 32mb memory I built and have a few more I can sell reasonably if anyone needs one.
It's a great project and you'd have to pry my MiSTer out of my cold, dead , hands!
Right now it is a Time Pilot '84 arcade machine!

[andrezheng]

Is xilinx xc7z030 a good candidate or not? 800MHz cpu, 125K LEs... and the price is around 60$.

[noel]

I just received the Exor SoM development kit !

EXor-eval-kit.png (1.36 MiB) Viewed 10241 times

I have a lot of work to do until the end of September but I can't wait to start looking at what we can do with it.

There is a lot of work to be done on this kit to run Mister's linux and a first fpga core, knowing that there is no HDMI or VGA output ... but hey it is only the beginning of the adventure

[antibolo]

While the subsidized low cost of the DE-10 Nano makes everything else a non-starter as far as price goes, I think we should all consider the very real possibility that the DE-10 Nano might eventually be discontinued. I'd say it would make a lot of sense to explore alternative options (preferably an open design) before that happens.

[noel]

Some fresh news! We have done a lot of testing and we are in the process of developing a hardware prototype in ITX format using KiCad. It's a long process and we are currently in the process of seeing if we can put a CycloneV directly on it. Using KiCad to route DDR3 and SDRAM will be an interesting challenge!

One of the problems encountered is being able to use an HDMI output because you have to be an HDMI adopter to be able to buy and use these components, and it costs between $ 5,000 and $ 10,000 per year... At first we will use a DVI output to stay on the legal side on the subject.

I will come back to the subject soon with more information, if you have any suggestions please do not hesitate!

[Duffygag]

Nice exercise but I personally would spend my resources differently

[noel]

Nice exercise but I personally would spend my resources differently

It's a little crazy project, isn't it? However, having a Cyclone V card for Mister in mini-ITX format and open hardware, that seems to be a good idea. I've already started so why not go all the way now

[Duffygag]

It is! Best of luck mate

[noel]

Thanks Duffygag!

I just created the major components and hooked up some signals to see what happens and for now it's going in the right direction.

1613666523_1742_18022021_462x461.png (18.44 KiB) Viewed 8803 times

I'm going to create the Cyclone V now .... I'll share some progress soon !

My current question is whether I should focus on remaining 100% pin compatible with the DE10 Nano fpga core pinmux or on the contrary take advantage of this new design to give more flexibility ...

[noel]

I will go on with Cyclone V directly instead of the Exor module. The first important and most difficult step after the creation of the FPGA component will be to do the DDR3 routing with KiCad. Now we will have to be patient, the next tens of hours may be concentrated on this part of the layout !!

Luckily the new features of kicad 6 are really promising. Just look how nice the ratsnets color feature is !

1613844700_1911_20022021_1122x569.png (88.41 KiB) Viewed 8649 times

To be honest, I don't think that it will be doable with less than 8 pcb layer total. Evaluation board stakup is 12 layer thick in comparison.

[andrezheng]

I will go on with Cyclone V directly instead of the Exor module. The first important and most difficult step after the creation of the FPGA component will be to do the DDR3 routing with KiCad. Now we will have to be patient, the next tens of hours may be concentrated on this part of the layout !!

Luckily the new features of kicad 6 are really promising. Just look how nice the ratsnets color feature is !
1613844700_1911_20022021_1122x569.png

To be honest, I don't think that it will be doable with less than 8 pcb layer total. Evaluation board stakup is 12 layer thick in comparison.

I found some interesting replacement. My work will go on.

[noel]

Impressive work ! I really like the way you done it I also choose to route on a 6 Layer stackup with 3 routing layers. Here is my stackup :

1614014652_1824_22022021_522x327.png (25.47 KiB) Viewed 8532 times

I just finished to validate my 80 ohm diff / 40 ohm single calculation for outer and inner layers. Now i'm ready to route.

As I plan to release this design in open source hardware, feel free to contact me to help me designing my board. I prefer the mini ITX form factor and I will use KiCad as it doesn't make sense to release opensource hardware that only relies on software that no one could afford.

For my part, I don't have a solution for the price of the fpga. We'll see later once the board is ready. There is also the problem of the hdmi because you have to pay $ 6000 per year to be able to buy the components !!

[grizzly]

DVI port instead of HDMi to save 6k$ per year is probably fine with most people that will buy this.
DVI to HDMI cables is cheap and easy to buy and the signals is the exact same (as far as i know anyway) DVI does not officially have sound but can be added.

The two small issues i see is.
1, It´s a bigger connector that will take up more pcb/case space, can be an issue but probably not.
2, The dvi end of the cable is bigger/longer/bulkier so people can think they are uglier, and they do take up a bit more space on the outside of the case.

You can probably find a supply for the HDMI stuff and a cheap Chinese factory to build them, but is it worth taking the chance?
Just look at the SD2SNES that got in trouble just because the name and had to change it.

[ExCyber]

While DVI-D would be bigger than HDMI, DVI-I should be smaller than HDMI + VGA, at the cost of requiring (passive) adapters for both of those. DVI also doesn't carry CEC signals. While it would be cool to have CEC support, lacking it altogether would still be an improvement over DE10-Nano screwing it up.

Just look at the SD2SNES that got in trouble just because the name and had to change it.

Yes, but as far as I know a name change was also all that happened, because it was presumably a trademark issue. It's not like it was redesigned to use USB drives instead of SD cards.

There are a few vendors out there that have taken a similar approach to HDMI, i.e. putting an HDMI-compatible socket on their product and never actually advertising it as "HDMI".

This is, of course, not legal advice. I'm just describing a thing that I've seen some vendors do.

[noel]

There are a few vendors out there that have taken a similar approach to HDMI, i.e. putting an HDMI-compatible socket on their product and never actually advertising it as "HDMI".

You can't purchase hdmi transmitter components unless you're a HDMI adopter (See https://www.analog.com/en/products/adv7535.html)

I agree with you on the DVI-I. As shown on the architecture bloc diagram I propose, I plan to implement DVI-I as a first step (https://www.patreon.com/posts/motherboard-v-47744392). It will output the same signal than HDMI but it will not embed the audio signal.

[ExCyber]

You can't purchase hdmi transmitter components unless you're a HDMI adopter (See https://www.analog.com/en/products/adv7535.html)

I guess that's a relatively new thing, because I'm sure I remember ADV7511/13 being normally stocked products at major distributors a few years ago. It looks like it might also be specific to Analog Devices products; NXP transmitters seem to still be available, and the datasheets only say that you need a license to use them "in equipment that complies with the HDMI standard". The main set of part numbers I see seems to be rated for only 150 MHz pixel clock rather than the 165 MHz rating of ADV7513 or TFP410, though.

[noel]

NxP transmitters seem to still be available, and the datasheets only say that you need a license to use them "in equipment that complies with the HDMI standard".

Analog devices and Silicon image devices are locked if you're not an HDMI adopter. I did not plan to use the NxP parts as they seems to be tagged as obsolete (https://www.arrow.com/fr-fr/products/td ... conductors). You can still buy the NxP components but I don't know for how long.

I added the TFP410 part in my current design as its not an HDMI transmitter but a DVI one. So there is no restriction to use it for now.

[pgimeno]

I'd love to have a design based on an FPGA device that can be programmed using open source tools. Apparently there isn't much open source software in that category, though.

[noel]

I don't think there is plenty of alternative to program the Cyclone V using open-source tools. It seems that if I replace the Cyclone V by Artix-7 fpga + external CPU, it would be possible to use the SymbiFlow open source toolchain. As the current PSX core development is based on Artix-7 fpga dev board, it seems that this target worth a look, but we are starting to move away seriously from the DE-10 Nano platform