problem is likely that for a Super Socket 7 board you need 4 layers, socket 370 is also 4 layers, and i think socket 462 is 5-6 layers. It's just way more complex, though I wouldn't object to a VIA MVP3 with the 686B, i think ATA100 ect would be nice, or a KT266A based Slot A board, heck maybe an Nforce 2 Ultra SlotA board

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that is just a labour of love.
I have a rather simple four layer (inner layers are only ground and power planes, no signal lines) socket 3 board here that fits ATX cases and which I'd like to clone at some point. All components are still avaliable new or NOS meaning it can be done. But it's so much work to first carefully unsolder, measure and then document all components just to be able to copy the PCB. It's not happening any time soon.
If you make your own design instead you will also have to write a BIOS which adds another hurdle.
I don't think this is a project that will be done at the request of the community. At some point somebody will just make the board they want and present it to us/have us watch it come together. That's the way >90% of successful complex projects come together.

Doornkaat wrote on 2021-11-23, 08:51:

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that i […]
Show full quote

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that is just a labour of love.
I have a rather simple four layer (inner layers are only ground and power planes, no signal lines) socket 3 board here that fits ATX cases and which I'd like to clone at some point. All components are still avaliable new or NOS meaning it can be done. But it's so much work to first carefully unsolder, measure and then document all components just to be able to copy the PCB. It's not happening any time soon.
If you make your own design instead you will also have to write a BIOS which adds another hurdle.
I don't think this is a project that will be done at the request of the community. At some point somebody will just make the board they want and present it to us/have us watch it come together. That's the way >90% of successful complex projects come together.

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or several of you guys together could do this.

I kinda figured someone with bios skills like chkcpu might be able to modify the bios from a existing board to work with the project board. Seems much more possible than making one from scratch.

Yeah, if someone were to sell these, surely they would have a decent price tag but with enough cool features and upgrades it might be worth it.

If it were a open source project we could pass around the pcb gerber and source the ICs on our own and have local bga solder shops/phone repair shops put them together for us or do them ourselves

Sphere478 wrote on 2021-11-23, 11:10:

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or seve […]
Show full quote

Doornkaat wrote on 2021-11-23, 08:51:

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that i […]
Show full quote

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that is just a labour of love.
I have a rather simple four layer (inner layers are only ground and power planes, no signal lines) socket 3 board here that fits ATX cases and which I'd like to clone at some point. All components are still avaliable new or NOS meaning it can be done. But it's so much work to first carefully unsolder, measure and then document all components just to be able to copy the PCB. It's not happening any time soon.
If you make your own design instead you will also have to write a BIOS which adds another hurdle.
I don't think this is a project that will be done at the request of the community. At some point somebody will just make the board they want and present it to us/have us watch it come together. That's the way >90% of successful complex projects come together.

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or several of you guys together could do this.

I kinda figured someone with bios skills like chkcpu might be able to modify the bios from a existing board to work with the project board. Seems much more possible than making one from scratch.

Yeah, if someone were to sell these, surely they would have a decent price tag but with enough cool features and upgrades it might be worth it.

If it were a open source project we could pass around the pcb gerber and source the ICs on our own and have local bga solder shops/phone repair shops put them together for us or do them ourselves

What was trying to say is this:
-It can 100% be done. No argument.
-It's labour intensive if it's a clone and also requires skill/experience if it's a new design.
-It won't currently be profitable because very few people will buy at a price that represents the amount of labour that went into the project.
-Anytime a similarly complex project came together it wasn't because somebody else requested it but because somebody was capable of doing it and wanted it for themselves.
This means two things:
-Demand is not relevant for this kind of project. (It won't pay.)
-What is relevant is that somebody who is capable of doing it wants to do it for themselves.

Conclusion: Time to nut up or shut up.😉 If you want those motherboards you need to either aquire the skills and design them yourself or motivate somebody else by offering them money to design them for you. The latter could be crowd sourced btw and is the most goal oriented approach imho.👍

On another note: Maybe I'm a bit cynical here and I totally get that you're enthusiastic about the prospect of new boards. But it bugs me how this type of thread always reads a bit like "I'd like new [things] with certain features so somebody else please design them for me and open source them so they're cheap."
Why is it never: "I know there are people here who can do this. What do you need from me to make this happen, how can I make it worth your while?"
Also if you've ever worked on a project like this you'll find people tend to start criticising your choices, demanding more features and generally get pretty nitpicky with what you provide to them cheap or even free. I'm not innocent of this either!
If I was capable of designing a new mainboard I would certainly not make it a community project. I'd make the board I want with the features I like and sell bare PCBs or PCBs that have all components presoldered that JLCPCBs' SMT service offers.

I hope you don't view this as a personal attack or anything. Again, I like your enthusiasm for new projects! 👍
But I hope you understand why I think threads like this are not a productive approach towards getting new boards done. In my opinion the best way is either doing it yourself or - if you (like me) lack the skill or time to do so - pay to have a mainboard professionally designed. Maybe split the cost between yourself and other enthusiasts.
Considering the size of a project like this asking for others to just do it so you can have it feels ... inappropriate.

No hate, no dis, just my two cents. If anybody has a different opinion please discuss.

If the market for reproduction computer parts becomes big enough, there will be a supply. It probably won't happen that way, but then again, if you'd told me back in 1984 (when I bought my first beat up Camaro) that beginning in 2014 you'd be able to build a complete 1967 Camaro by stating with a VIN plate, I wouldn't have believed you.

Not to dissuade people, but to put things into perspective here; many of monotech's are based on other people's open source work, including the NuXT. Basically what he did for his idea was to shrink multiple boards people made (including a CPU board) down to an all in one ATX board. So just to get the NuXT, it is based on multiple people's work, and the hours he put in making another ATX board. This didn't happen all that quickly -- it is a culmination of many things.

In terms of what is feasible, I have seen up to 386, and possibly a low grade 486 hobbiest variants floating around already. I have only done a very limited amount of PCB layout, but my understanding is that when bus speeds start going past 25Mhz, then the effort increases on bus timings and noise. So I would say a fast 486/pentium is quite a bit of work yet to do. Although the vortex is a nice alternative in a non-traditional socket, as it simplifies things quite a bit 😀

Then there is still quite a used market for anything later, which functions pretty well for what people still do. So between what has already been done and with the used marked still going strong with anything not really done, and the time it might take, I think the effort is possibly not worth it.

My college school EE lab prof had laid out a generic “socket” with a small amount of logic that allowed a single socket to both mechanically and electrically deal with PGA132 through socket 370 in the same socket as a thought experiment (looked like a socket 370 socket converted into a breadboard)
He laid out how only a couple pins could be used to safely sense the cpu type and soft jumper.
He demonstrated why he believed a ppro+ Chipset should be able to support non-p6 cpus like socket 7 pentiums with a small amount of logic and a custom bios.
Oh and smart people don’t install the cpu in reverse :0

He believed the interface differences between all x86 cpus 386-> celeron were trivial and could be handled by a unified chipset with a small amount of programmable logic

386 + ddr5 and pcie just so you minimize the quantity of vintage gear required

Would only be a million plus to make it market ready, which Elon Musk is into unnecessary modern vintage retro devices?

Wonder if he is still alive?

IMHO the biggest constraint is the availability of suitable chipsets. You need something which is properly documented (datasheets) and still able to be purchased in decent quantities (min 100) from a reputable source, and for an affordable hobbyist board anything in a BGA package isn't workable. Even the NuXT -- which is an amazing piece of kit -- relies heavily on the Faraday FE2010 chipset for most of its functionality and guess what? It's nigh on impossible to buy any more FE2010 chips! The available supply has been exhausted. Now all that's left is fake remarked counterfeit chips from China.

And anything beyond perhaps a Socket 7 board is just not happening, because that's when AMD and Intel started using their own proprietary sockets and the chipset vendors stopped providing useful/complete datasheets. Not to mention everything was BGA by that point.

kdr wrote on 2021-11-23, 20:20:

IMHO the biggest constraint is the availability of suitable chipsets. You need something which is properly documented (datasheets) and still able to be purchased in decent quantities (min 100) from a reputable source, and for an affordable hobbyist board anything in a BGA package isn't workable. Even the NuXT -- which is an amazing piece of kit -- relies heavily on the Faraday FE2010 chipset for most of its functionality and guess what? It's nigh on impossible to buy any more FE2010 chips! The available supply has been exhausted. Now all that's left is fake remarked counterfeit chips from China.

And anything beyond perhaps a Socket 7 board is just not happening, because that's when AMD and Intel started using their own proprietary sockets and the chipset vendors stopped providing useful/complete datasheets. Not to mention everything was BGA by that point.

Maybe it is possible to replace the chipset with an FPGA? Then just add isa/pci slots, memory and cpu sockets 😉

janih wrote on 2021-11-23, 20:48:

Maybe it is possible to replace the chipset with an FPGA? Then just add isa/pci slots, memory and cpu sockets 😉

Sure, but FPGA brings along a whole different set of problems, and an FPGA big/fast enough (and with enough I/O pins) to handle a Socket 5/7 CPU is invariably going to be in a difficult BGA package. Oh and none of the currently-available FPGA families support 5V I/O pins any longer so you'll have to add an immense number of level shifter ICs to the board.... and that's ignoring the fact that nobody has yet coded up a full x86 compatible chipset in Verilog or VHDL.

It absolutely can be done but I don't see enough folks out there that'd be willing to fund the cost of developing a brand new "modern" Socket 7 board!

Doornkaat wrote on 2021-11-23, 15:31:

What was trying to say is this: -It can 100% be done. No argument. -It's labour intensive if it's a clone and also requires skil […]
Show full quote

Sphere478 wrote on 2021-11-23, 11:10:

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or seve […]
Show full quote

Doornkaat wrote on 2021-11-23, 08:51:

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that i […]
Show full quote

If you are aiming at an attractive price point you're never getting back the time you spend on such a project. Stuff like that is just a labour of love.
I have a rather simple four layer (inner layers are only ground and power planes, no signal lines) socket 3 board here that fits ATX cases and which I'd like to clone at some point. All components are still avaliable new or NOS meaning it can be done. But it's so much work to first carefully unsolder, measure and then document all components just to be able to copy the PCB. It's not happening any time soon.
If you make your own design instead you will also have to write a BIOS which adds another hurdle.
I don't think this is a project that will be done at the request of the community. At some point somebody will just make the board they want and present it to us/have us watch it come together. That's the way >90% of successful complex projects come together.

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or several of you guys together could do this.

I kinda figured someone with bios skills like chkcpu might be able to modify the bios from a existing board to work with the project board. Seems much more possible than making one from scratch.

Yeah, if someone were to sell these, surely they would have a decent price tag but with enough cool features and upgrades it might be worth it.

If it were a open source project we could pass around the pcb gerber and source the ICs on our own and have local bga solder shops/phone repair shops put them together for us or do them ourselves

What was trying to say is this:
-It can 100% be done. No argument.
-It's labour intensive if it's a clone and also requires skill/experience if it's a new design.
-It won't currently be profitable because very few people will buy at a price that represents the amount of labour that went into the project.
-Anytime a similarly complex project came together it wasn't because somebody else requested it but because somebody was capable of doing it and wanted it for themselves.
This means two things:
-Demand is not relevant for this kind of project. (It won't pay.)
-What is relevant is that somebody who is capable of doing it wants to do it for themselves.

Conclusion: Time to nut up or shut up.😉 If you want those motherboards you need to either aquire the skills and design them yourself or motivate somebody else by offering them money to design them for you. The latter could be crowd sourced btw and is the most goal oriented approach imho.👍

On another note: Maybe I'm a bit cynical here and I totally get that you're enthusiastic about the prospect of new boards. But it bugs me how this type of thread always reads a bit like "I'd like new [things] with certain features so somebody else please design them for me and open source them so they're cheap."
Why is it never: "I know there are people here who can do this. What do you need from me to make this happen, how can I make it worth your while?"
Also if you've ever worked on a project like this you'll find people tend to start criticising your choices, demanding more features and generally get pretty nitpicky with what you provide to them cheap or even free. I'm not innocent of this either!
If I was capable of designing a new mainboard I would certainly not make it a community project. I'd make the board I want with the features I like and sell bare PCBs or PCBs that have all components presoldered that JLCPCBs' SMT service offers.

I hope you don't view this as a personal attack or anything. Again, I like your enthusiasm for new projects! 👍
But I hope you understand why I think threads like this are not a productive approach towards getting new boards done. In my opinion the best way is either doing it yourself or - if you (like me) lack the skill or time to do so - pay to have a mainboard professionally designed. Maybe split the cost between yourself and other enthusiasts.
Considering the size of a project like this asking for others to just do it so you can have it feels ... inappropriate.

No hate, no dis, just my two cents. If anybody has a different opinion please discuss.

https://github.com/WannaBAcoder/PCB-Powerwalls#readme

I designed this and critiqued it, while someone else did the drafting. I would be happy to help within my limits! 😀

janih wrote on 2021-11-23, 20:48:

kdr wrote on 2021-11-23, 20:20:

IMHO the biggest constraint is the availability of suitable chipsets. You need something which is properly documented (datasheets) and still able to be purchased in decent quantities (min 100) from a reputable source, and for an affordable hobbyist board anything in a BGA package isn't workable. Even the NuXT -- which is an amazing piece of kit -- relies heavily on the Faraday FE2010 chipset for most of its functionality and guess what? It's nigh on impossible to buy any more FE2010 chips! The available supply has been exhausted. Now all that's left is fake remarked counterfeit chips from China.

And anything beyond perhaps a Socket 7 board is just not happening, because that's when AMD and Intel started using their own proprietary sockets and the chipset vendors stopped providing useful/complete datasheets. Not to mention everything was BGA by that point.

Maybe it is possible to replace the chipset with an FPGA? Then just add isa/pci slots, memory and cpu sockets 😉

Oh that’s brilliant.

Imagine a universal retro board that had several different sockets on it…

Sphere478 wrote on 2021-11-24, 03:41:

Imagine a universal retro board that had several different sockets on it…

That would be cool. The https://opencores.org/projects/ao486, used in MisterFPGA's ao486 for example, implements these:

• ao486 processor,
• IDE hard drive that redirects to a HDL SD card driver,
• floppy controller that also redirects to the SD card driver,
• 8259 PIC,
• 8237 DMA,
• Sound Blaster 2.0 with DSP and OPL2,
• 8254 PIT,
• 8042 keyboard and mouse controller,
• RTC,
• standard VGA.

But of course it is not that simple as stated by kdr, the i/o voltages and complexity for something like an socket 7 motherboard is on another level I guess.

janih wrote on 2021-11-24, 07:19:

[…]
Show full quote

• ao486 processor,
• IDE hard drive that redirects to a HDL SD card driver,
• floppy controller that also redirects to the SD card driver,
• 8259 PIC,
• 8237 DMA,
• Sound Blaster 2.0 with DSP and OPL2,
• 8254 PIT,
• 8042 keyboard and mouse controller,
• RTC,
• standard VGA.

Wow, that's actually quite comprehensive - didn't realise that ao486 was more than just a CPU soft core! Could serve as the starting point for a decent Socket 3 chipset. You'd definitely want to aim for a 486-class motherboard as a stepping stone towards anything faster / more complex. I think you can fit (parts of) the ao486 core in a cheap hobbyist-friendly iCE40 chip and still achieve speeds of 40-50Mhz, so it's a realistic goal.

But of course it would probably be a lot of work to get an FPGA interfaced with an actual vintage 486 CPU using its local bus protocol -- there is a fair bit of signalling involved to arbitrate between devices on the bus. (Since you have CPU, L2 cache, memory controller, ISA bridge, plus one or more VLB cards.)

Sphere478 wrote on 2021-11-24, 03:41:

Doornkaat wrote on 2021-11-23, 15:31:

What was trying to say is this: -It can 100% be done. No argument. -It's labour intensive if it's a clone and also requires skil […]
Show full quote

Sphere478 wrote on 2021-11-23, 11:10:

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or seve […]
Show full quote

Yeah, I was figuring a limited run, and a complex project, but from the talent I’ve seen around here I think some of you or several of you guys together could do this.

I kinda figured someone with bios skills like chkcpu might be able to modify the bios from a existing board to work with the project board. Seems much more possible than making one from scratch.

Yeah, if someone were to sell these, surely they would have a decent price tag but with enough cool features and upgrades it might be worth it.

If it were a open source project we could pass around the pcb gerber and source the ICs on our own and have local bga solder shops/phone repair shops put them together for us or do them ourselves

What was trying to say is this:
-It can 100% be done. No argument.
-It's labour intensive if it's a clone and also requires skill/experience if it's a new design.
-It won't currently be profitable because very few people will buy at a price that represents the amount of labour that went into the project.
-Anytime a similarly complex project came together it wasn't because somebody else requested it but because somebody was capable of doing it and wanted it for themselves.
This means two things:
-Demand is not relevant for this kind of project. (It won't pay.)
-What is relevant is that somebody who is capable of doing it wants to do it for themselves.

Conclusion: Time to nut up or shut up.😉 If you want those motherboards you need to either aquire the skills and design them yourself or motivate somebody else by offering them money to design them for you. The latter could be crowd sourced btw and is the most goal oriented approach imho.👍

On another note: Maybe I'm a bit cynical here and I totally get that you're enthusiastic about the prospect of new boards. But it bugs me how this type of thread always reads a bit like "I'd like new [things] with certain features so somebody else please design them for me and open source them so they're cheap."
Why is it never: "I know there are people here who can do this. What do you need from me to make this happen, how can I make it worth your while?"
Also if you've ever worked on a project like this you'll find people tend to start criticising your choices, demanding more features and generally get pretty nitpicky with what you provide to them cheap or even free. I'm not innocent of this either!
If I was capable of designing a new mainboard I would certainly not make it a community project. I'd make the board I want with the features I like and sell bare PCBs or PCBs that have all components presoldered that JLCPCBs' SMT service offers.

I hope you don't view this as a personal attack or anything. Again, I like your enthusiasm for new projects! 👍
But I hope you understand why I think threads like this are not a productive approach towards getting new boards done. In my opinion the best way is either doing it yourself or - if you (like me) lack the skill or time to do so - pay to have a mainboard professionally designed. Maybe split the cost between yourself and other enthusiasts.
Considering the size of a project like this asking for others to just do it so you can have it feels ... inappropriate.

No hate, no dis, just my two cents. If anybody has a different opinion please discuss.

https://github.com/WannaBAcoder/PCB-Powerwalls#readme

I designed this and critiqued it, while someone else did the drafting. I would be happy to help within my limits! 😀

I'm sure you would help! I'm sure if you were able to you'd make the whole board yourself and open source it.👍 I hope it is understood I didn't mean to question your character in my rant.
But considering the small contribution people of our skill level could make toward the total effort it takes to make a new mobo I wouldn't feel comfortable asking others to do it for free (or for what is basically just a tip).

Idea: Are there already KiCad footprints for the relevant NB/SB chips and the CPU sockets? If not you could make them. If you want to make this a community project this may be a first step. 😃

Doornkaat wrote on 2021-11-24, 11:17:

I'm sure you would help! I'm sure if you were able to you'd make the whole board yourself and open source it.👍 I hope it is unde […]
Show full quote

Sphere478 wrote on 2021-11-24, 03:41:

Doornkaat wrote on 2021-11-23, 15:31:

What was trying to say is this: -It can 100% be done. No argument. -It's labour intensive if it's a clone and also requires skil […]
Show full quote

What was trying to say is this:
-It can 100% be done. No argument.
-It's labour intensive if it's a clone and also requires skill/experience if it's a new design.
-It won't currently be profitable because very few people will buy at a price that represents the amount of labour that went into the project.
-Anytime a similarly complex project came together it wasn't because somebody else requested it but because somebody was capable of doing it and wanted it for themselves.
This means two things:
-Demand is not relevant for this kind of project. (It won't pay.)
-What is relevant is that somebody who is capable of doing it wants to do it for themselves.

Conclusion: Time to nut up or shut up.😉 If you want those motherboards you need to either aquire the skills and design them yourself or motivate somebody else by offering them money to design them for you. The latter could be crowd sourced btw and is the most goal oriented approach imho.👍

On another note: Maybe I'm a bit cynical here and I totally get that you're enthusiastic about the prospect of new boards. But it bugs me how this type of thread always reads a bit like "I'd like new [things] with certain features so somebody else please design them for me and open source them so they're cheap."
Why is it never: "I know there are people here who can do this. What do you need from me to make this happen, how can I make it worth your while?"
Also if you've ever worked on a project like this you'll find people tend to start criticising your choices, demanding more features and generally get pretty nitpicky with what you provide to them cheap or even free. I'm not innocent of this either!
If I was capable of designing a new mainboard I would certainly not make it a community project. I'd make the board I want with the features I like and sell bare PCBs or PCBs that have all components presoldered that JLCPCBs' SMT service offers.

I hope you don't view this as a personal attack or anything. Again, I like your enthusiasm for new projects! 👍
But I hope you understand why I think threads like this are not a productive approach towards getting new boards done. In my opinion the best way is either doing it yourself or - if you (like me) lack the skill or time to do so - pay to have a mainboard professionally designed. Maybe split the cost between yourself and other enthusiasts.
Considering the size of a project like this asking for others to just do it so you can have it feels ... inappropriate.

No hate, no dis, just my two cents. If anybody has a different opinion please discuss.

https://github.com/WannaBAcoder/PCB-Powerwalls#readme

I designed this and critiqued it, while someone else did the drafting. I would be happy to help within my limits! 😀

I'm sure you would help! I'm sure if you were able to you'd make the whole board yourself and open source it.👍 I hope it is understood I didn't mean to question your character in my rant.
But considering the small contribution people of our skill level could make toward the total effort it takes to make a new mobo I wouldn't feel comfortable asking others to do it for free (or for what is basically just a tip).

Idea: Are there already KiCad footprints for the relevant NB/SB chips and the CPU sockets? If not you could make them. If you want to make this a community project this may be a first step. 😃

I’ll survive 🤣. 😀

Sphere478 wrote on 2021-11-24, 03:41:

Oh that’s brilliant.

Imagine a universal retro board that had several different sockets on it…

I can. 😀

If we made a motherboard what chipset would it be based on?

Seems like we would want one that had a lot of new old stock laying around or one that had a lot of motherboards that chips could be harvested from.

Ali V or mvp3 seems like the one for socket 7?

Unless there are new old stock of sis 540 floating around somewhere

An i 815 board with slot-1 ... Drooling

H3nrik V! wrote on 2022-12-25, 22:52:

An i 815 board with slot-1 ... Drooling

You mean like the Abit SH6? Yes, it's unobtainium, but compared to the price something new in small run will cost, paying eBay premium when one does come up will be cheap by comparison.

Being marginally more realistic, the biggest use-case I see is to offer an ATX-form factor 486 board with modern I/O, so people can hook up USB mice & keyboards. In fantasy-world I'd love for a real So3 socket, VLB and PCI slots (hell, EISA too while I'm at it), but being somewhat feasible, one of the Elan/Vortex/ZFMicro SoC designs with VGA either integrated, or with a nice compatible (available...) chip onboard, some RAM onboard and then only a few ISA slots for real sound cards or other stuff would make a lot more sense, particularly as 486 boards get rarer and rarer, but so do the needed controller boards, peripherals and knowledge how to use them - see how many topics there are about people who are lost at how to get their nice new keyboards and mice to play nice with ancient serial & DIN stuff.