The 8-Bit Guy's new 65816 based computer

[Elminster]

I saw that. Not really an update. More of a throw it away and start a new project. I was following the facebook group but got bored, jus back to watching the occasional updates now.

[BigEd]

Not having seen any of the facebook discussion, this update did look to me like an update! Some ideas had been had, and explored, and then there was a simplification. It's now looking like a 6502 project with a 16 bit address space and a relatively clean memory map, with heaps of banking. There's no local video memory, which keeps things simple, and the video engine is an FPGA now.

Of course it's not too hard to put an 816 in place in such an 8-bit system and get some advantages from that, without using the extended addressing, and that's mentioned towards the end of the video.

There's no project ever which has made everyone happy - seeking consensus in a crowd is a bit of a wild hope. The best you can do is get some input, some ideas, some feedback. Then build it, and see if it flies. (In this case commercial viability as a new platform seems to be part of the picture, which makes it rather ambitious.)

[Kazzie]

I'm of a similar opinion to BigEd, as i hadn't followed any of the intermediate discussion either. It may not be a direct update to the initial design, but it's certainly an update on the progress of the project. And it still looks promising, to me. The current direction seems to be 6502-based processor surrounded with off-the-shelf dip-chips, plus a custom-designed video processor chip (with the issue of audio still to be resolved). And it's not too dis-similar to the Beeb in that respect.

[Elminster]

My comment was more based around the requirements of being a remake of c64 at around £50 with no FPGAs.

Now has fpga, more vic 20 than c64 and it is £200, probably down to £100.

So none of the original requirements (that I remember without going back to original video) remain. Hence comment about being pretty much a new project.

It looks good, didn’t say it didn’t.

[BigEd]

Well indeed, requirements is quite a strong word! And desiderata is a bit obscure... "some initial thoughts" might be what was needed, but that's three words.

[Elminster]

Maybe. But that is just my view, but I won’t hold them to it contractually don’t worry.

What do people think in cost? £200 I probably wouldn’t buy, £100 I might, £50 I probably would.

Edit: although physical size might be deciding factor for me. I.e. my lack of space.

[Andrew_Waite]

There is now an emulator available for the 'Commander X16'.

https://github.com/commanderx16/x16-emu ... gl7A33TFtk

Getting BBC BASIC, or Jonathan Harston's BBC BASIC for the Commodore 64 up and running on the 'Commander X16' would make for a nice project.

http://mdfs.net/Software/BBCBasic/C64/

[davidb]

Getting BBC BASIC, or Jonathan Harston's BBC BASIC for the Commodore 64 up and running on the 'Commander X16' would make for a nice project.

http://mdfs.net/Software/BBCBasic/C64/

Get to work, then!

[julie_m]

There seems to be a fundamental difference in the way computer graphics should be represented and rendered on either side of the Atlantic!

British computers such as the ZX Spectrum, BBC and Amstrad CPC all used simple(ish ..... there were undoubtedly complicating factors like out-of-order address lines, but you could get mathematically from a set of co-ordinates to a location in display memory) bit-mapped displays, whereas the American machines were based on detailed sprites overlaid on chunky backgrounds.

[Ramtop]

I think the common factor for UK machines was the use of ULAs, and consequently the display logic had to be fairly simple. Whereas several of the US manufacturers had either their own ASIC production facilities or could afford to hire one.

Commodore had access to their own chip fab, so they could get away with much more complex designs - if I recall correctly about 60% of the die area of the C64's VIC-II chip was used to provide sprites. TI also built their own custom sprite-based video chip.

Personally I was never much of a fan of sprites on 8-bit machines as they always seemed to be used to try and cover up some performance bottleneck, like the slow CPUs in the C64 or TI machines and the register-based video memory access on the MSX. I'd rather have a simple bitmap display and a fast CPU, Speccy-style.

But I suppose sprites are a good feature for a machine primarily aimed at teaching people about computers. Getting something to move on the screen by POKEing a few registers is obviously much easier than writing a software sprite routing.

[roland]

I wonder if we can ever get the source code of that Vera video board and see if we (actually Hoglet ) can merge this into the AtomFPGA's. Since the Atom2k18 has an option to add a 512kB fast ram (just like the 1 MHz Beeb FPGA adapter) it could use that as video memory. So beyond CLEAR 4 a lot of additional modes

[hoglet]

I wonder if we can ever get the source code of that Vera video board and see if we (actually Hoglet ) can merge this into the AtomFPGA's.

I was wondering that as well, but I don't expect they will release the source.

Dave

[Elminster]

I wonder if we can ever get the source code of that Vera video board and see if we (actually Hoglet ) can merge this into the AtomFPGA's.

I was wondering that as well, but I don't expect they will release the source.

Dave

At least until they have sold a few and made the development costs back I should think. They look like they are going the whole hog (no pun intended) and I expect that will cost a pretty penny.

[Ramtop]

Presumably it'll show up on Kickstarter; David Murray raised over $100,000 for Planet X3, so he should easily be able to get enough to put the board into production. The source for the video FPGA is an important factor, though. If the developer goes off in a huff (as so often happens on these kind of projects) then the whole project would be set back quite seriously.

[Elminster]

Presumably it'll show up on Kickstarter; David Murray raised over $100,000 for Planet X3, so he should easily be able to get enough to put the board into production. The source for the video FPGA is an important factor, though. If the developer goes off in a huff (as so often happens on these kind of projects) then the whole project would be set back quite seriously.

I think he said after the hassle with Kickstarter on X3 he was not going to do one, but I could be misremembering that.

Edit: This is what I was thinking off

https://youtu.be/wCx32lrBSNQ

Where he says now Kickstarter for x3 is done he might open source the code, and also how little he enjoyed the kickstarter process. Don’t think he says he won’t do it though.

[Ramtop]

His experiences with Planet X3 did seem to be a bit of a nightmare. I wouldn't have wanted to be under that kind of stress.

The money required to put a large board full of through-hole parts into production is not trivial, though, so do I wonder where it'll come from if not Kickstarter. Even a small initial batch, say 250 units, could easily run to $40K+ just for parts and assembly. It's a capital intensive process, and probably quite scary if you're paying the bills out of your own pocket.

[Elminster]

Kickstarter or equiv is the most common way for retro project. I wonder how else you could do it. I can think of:

- Partner with a company.
- Open hardware project (but this seems very unlikely in the short term).

[gordonDrogon]

Kickstarter or equiv is the most common way for retro project. I wonder how else you could do it. I can think of:

- Partner with a company.
- Open hardware project (but this seems very unlikely in the short term).

I have 2 projects right now (one a 6502/65816) that I'd like to package and sell - as kits. Ready made poses many issues to do with the various CE, Compliance testing and so on that makes it hard to do from a hobby point of view (at least in the UK) so that's where the pain of Kickstarter can help by making sure you actually have enough funds to do it all - but just selling kits seems OK for the likes of the RC2014 project and Ben Eaters breadboard CPU and 6502 projects....

-Gordon

[BigEd]

oops - crossed in post

One nice thing about through-hole is that you can sell as a kit. So no assembly costs. But there will be support costs, and returns, from people who really don't know what they are doing.

At the other end, once you're all surface mount - which Dave Murray seems to suggest as a user-friendly endpoint when he has volume - you can get assembly houses to do the work for you. Referring to
https://www.seeedstudio.com/blog/2018/0 ... explained/
we see that a $50 BOM turns into a $200 cost. So the end-user price would normally be 2x to 3x that. Which means to hit a $100 end-user price, you need no more than $8 of components, if I have my sums right. (For a viable business. Hobbies have slimmer margins! Even negative margins!)

Good point about licensing and compliance!

[Elminster]

Good point about licensing and compliance!

If I remember rightly if you want someone to test for things like interference, they are the people with the $100,000 scopes and spectrum analysers, so can't image that will be cheap.

[BigEd]

It's a fundamental thing, that the ridiculously low cost of electronics today is all down to scale. If you take away the scale - which almost all retro projects do - you're left with a very different price structure. Even 1000 units is not a volume production run, to the factories which make stuff in volume. Bunnie Huang has written a lot of good stuff on this.

As a rule of thumb, a US designer is better off assembling PCBs in the US for volumes less than 1k, and they don’t start seeing clear advantages until perhaps 5k-10k volumes.

[Kazzie]

One nice thing about through-hole is that you can sell as a kit. So no assembly costs. But there will be support costs, and returns, from people who really don't know what they are doing.

As Acorn and Science of Cambridge found out first-hand!

[Ramtop]

Good point about licensing and compliance!

If the final product is anything like the board shown in the video I have a feeling compliance issues may kill off any chance of it being sold as a fully assembled item in the EU. Stand-alone consumer devices can't be legally sold without the CE mark, and you can't show that without also being RoHS compliant. A board containing NOS non-RoHS chips isn't going to fly.

A possible work-around would be to sell those non-RoHS chips as a separate purchase and provide sockets on the board for them. Legally a bit shaky, but probably feasible for the sound chips at least, because the computer will function without them.

[Elminster]

How does that work for assembled kids on tindie.com, doubt they go through any sort of compliance. Or is it because you are using an ‘assembly’ service, I.e. the creator in their shed?

[Ramtop]

I haven't studied the relevant documentation for some years, so I'm a bit hazy on the specifics, but my recollection of the regulations is that an assembled kit is legally no different from a pre-built product. If you're selling assembled kits they really should be CE compliant (with a few exceptions). Of course Trading Standards, who enforce compliance for consumer goods in the UK, have little interest in chasing someone selling a few home-built widgets on Tindie. They're busy trying to stem the flood of dangerous goods coming in from China.

There are exceptions to the compliance requirement for some items that are intended to repair or upgrade a pre-existing device. There was also an exemption that parts and upgrades for pre-RoHS equipment did not need to be RoHS compliant. Not sure if that's still the case, however.