SBC kit for education

[GARTHWILSON]

The first post says,

Quote:

The idea of this topic is to discuss the feasibility of a DIY singleboard computer for educational purposes. As it have been already discussed in the UK media, the lack of understanding concepts of technology has had a negative impact on the society, the environment and the computer sciences/engineering education. Benefits of such kit would be:

- entry point and low barrier to technology

- profound understanding of computers, embedded systems and programming concepts

and I like that!  In another topic I mentioned I had 110 6512's, and the reason is that in the late 1980's I, too, wanted to make a board for schools and hobbyists, and these were on sale for pennies on the dollar.  I never got around to making and marketing a board.

Many years later 6502.org and the Delphi forum came on the scene (before there was the phpBB forum here on 6502.org), and every so often someone would re-introduce the topic of making a board that everyone could enjoy.  (This was before PCBs got so cheap for small quantities.)  The discussions would get more and more outlandish, with people wanting to add all the things of the latest high-end PCs (that's only a small exaggeration), and it never got done.  After everyone kind of forgot about it and forgot their hostilities, someone with their head on relatively straight would unwittingly start the cycle over, proposing a simple board, and again it would gradually go from a hobbyist/workbench/educational thing back to a consumer item that would take a team of engineers untold man-years to complete.

Then Daryl (8BIT here on the forum) quietly made his own board, the SBC-2, and said, "Hey everyone, I have this board ready to go.  Anyone want one?" and I think he got orders for 50, without advertising or approaching any schools.  I think that with a little help here, and after getting your feet wet, you could do a similar thing, go to a few schools' science or computer departments and talk to the professors who have some say in the curriculum, and get things going.  If you have a desire to do that, I have no doubt they would find it appropriate, and you'd be successful.  They, too, might be unhappy with how other boards on the market insulate the student from the bottom level which someone has to understand to make the top levels work.  In fact, maybe you would want to just see if Daryl could supply you the already-designed boards (which he did a good job on, BTW) and you get them into the schools.

[BigDumbDinosaur]

GARTHWILSON wrote:

The first post says,

Quote:

The idea of this topic is to discuss the feasibility of a DIY singleboard computer for educational purposes. As it have been already discussed in the UK media, the lack of understanding concepts of technology has had a negative impact on the society, the environment and the computer sciences/engineering education. Benefits of such kit would be:

- entry point and low barrier to technology

- profound understanding of computers, embedded systems and programming concepts

and I like that...the Delphi forum came on the scene (before there was the forum here on 6502.org), and every so often someone would re-introduce the topic of making a board that everyone could enjoy...The discussions would get more and more outlandish, with people wanting to add all the things of the latest high-PCs (that's only a small exaggeration), and it never got done. After everyone kind of forgot about it and forgot their hostilities, someone with their head on relatively straight would unwittingly start the cycle over, proposing a simple board, and again it would gradually go from a hobbyist/workbench/educational thing back to a consumer item that would take a team of engineers untold man-hours to complete.

Deja vu all over again.

Quote:

Then Daryl (8BIT here on the forum) quietly made his own board, the SBC-2, and said, "Hey everyone, I have this board ready to go. Anyone want one?" and I think he got orders for 50, without advertising or approaching any schools...maybe you would want to just see if Daryl could supply you the already-designed boards (which he did a good job on, BTW) and you get them into the schools.

Daryl's SBC-2 design would fulfill the role of a 6502-based "entry level" kit/unit that exposes the machinery directly to the experimenter, yet has the ability to stand on its own as a functional device. I could probably do something similar with my POC unit, although it doesn't have a built-in console (a dumb terminal or terminal emulation software running on a Mac or PC would be required).

Whatever the case, I think it needs to be very clear that this is an entry level device, is powered by a 65C02 or 65C816, has a modicum of RAM and I/O and has enough of a functional ROM in it to get to boot and talk to a console device of some sort. Beyond that, it should be left to the individual experimenter to add to it as desired.

[alkopop79]

It does. Back to the original topic.

[alkopop79]

I like Daryl's design a lot. In fact, either that or the UK101 would be perfect for the kit. I don't think you can build an SBC simpler. I tried to build a Cosmac Elf and the glue logic bit is mind boggling, has twice the ICs.

[whartung]

I'll just summarize my 2 bits.

Whatever components end up being used, they're effectively worthless without a curriculum to coincide with it. This is a combination of applicable theory, design agenda, assembly instructions and experiments.

I don't think it should be a PCB. I think it is more interesting and more accessible as WW or breadboard. Populating a PC board is not interesting, that is an exercise in soldering. Physically wiring the bits together adds more to the process, since the user must think about each wire, why it's there, where it's going. It's the difference between cutting and pasting code in to an editor and typing it in your self. The physical process of assembly is educational in its own right, and simply soldering chips in to a board is too high level and skips large aspects of the design.

Assembly should be done in testable, provable stages so that there is confidence that things are working as assembly progresses. Things like making sure your power is working properly, that your clock is working properly, that any decode logic is working properly, that the CPU runs (someone mentioned hardwiring NOPs and watching LEDs blink -- that's an excellent experiment), etc. etc. So that when you finally get that last components in to their sockets, not only are you confident it's not going vanish in a whiff of ozone and a curl of blue smoke, but that it's going to function. And if it doesn't, you have a good idea how far back you need to go since it "worked last". I fully expect assembly stages to possibly undo work in the previous stage that was their solely for testing (like the NOP wiring...).

Finally, some interface experiments to work with the "real world" via either blinky lights, relays/SCRs/motors, simple LCD displays. thermometers, etc.

A 6502, a static RAM chip, a VIA of some kind, an EPROM, a MAX-232 and some discrete decode logic is simple, it's functional, and it's easy to build. It has all of the major components laid out in a 3D, functioning block diagram.

An EPROM should be provided with the necessary software, I don't believe this is a software problem, so the software should be effectively solved until the machine is running. Once it's running, the students can program it for the interface experiments. I think it's fine that the students may well need to reprogram the EPROM throughout the assembly and testing process. It would be more interesting if a single EPROM could contain all of the software for all of the stages, with perhaps some jumpers telling it which particular stage to run to avoid the necessity of needing an EPROM programmer.

At the end, there should be some advanced experiments documented with more components: adding a keypad and display, adding more RAM, adding another VIA, moving the components to a permanent PCB and housing, etc.

To me, refinements to this design would head in the direction of making it cheaper and more fool proof for a student, especially a solo student, to participate. Simply, if you're going to change the design, provide the answers to "is it cheaper" and "is it simpler" and why.

It would be exciting to see if the combination of components and tools to assemble and test this minus a host PC, a USB -> Serial cable and the EPROM programmer could be done for < $200. It would be so awesome if this can be done without a scope.

So feel free to update.

[8BIT]

alkopop79 wrote:

I like Daryl's design a lot. In fact, either that or the UK101 would be perfect for the kit. I don't think you can build an SBC simpler. I tried to build a Cosmac Elf and the glue logic bit is mind boggling, has twice the ICs.

My first computer was the Netronics ELF II. Yes, I tried for a long time to figure out how to simplify the glue logic, but if you want the bootstrap mode, you need it all.

It may be possible to tweek the SBC-2 a little to make it even better. Drop one of the VIA's, add a simple text video and pc keyboard decoder, put in a newer USART. It could then operate independently and still allow a simple path for downloading test programs via RS-232.

The nice thing about it is that you can re-program an EEPROM in-system with a little care. It would cover a lot of the basics, including several types of I/O, simple device decoding, and assembly language programming.

Daryl

[GARTHWILSON]

whartung, those are definitely good considerations. It makes me wonder though if there's enough time for building. Lab time might not be enough, and although other classes expect that the student spend that much time or more in homework, I'm not sure how well students could be expected to do without help being available, whether in the library, dorm, home, or wherever they do it. I would like to think it would work. Some professor will need hundreds of hours of experimenting with students (to find out what's reasonable to expect) and with preparing the curriculum before the first class. I hope someone gets it going. At least there wouldn't have to be an (E)EPROM programmer for every student. A few programmers for each class should be plenty.

Our labs in 1981 with the AIM-65 were mainly software. I got a dabbling in gates a little before that on my own, and more later, but never had any training in computer hardware in school myself. I picked it up so gradually (over a period of about five years) that I hardly remember how.

[alkopop79]

Let me add, I don't teach at school (used to) and the kit is not for schools but for individuals to order and assemble by themselves.

[alkopop79]

whartung wrote:

I'll just summarize my 2 bits.

Whatever components end up being used, they're effectively worthless without a curriculum to coincide with it. This is a combination of applicable theory, design agenda, assembly instructions and experiments.

I don't think it should be a PCB. I think it is more interesting and more accessible as WW or breadboard. Populating a PC board is not interesting, that is an exercise in soldering. Physically wiring the bits together adds more to the process, since the user must think about each wire, why it's there, where it's going. It's the difference between cutting and pasting code in to an editor and typing it in your self. The physical process of assembly is educational in its own right, and simply soldering chips in to a board is too high level and skips large aspects of the design.

Assembly should be done in testable, provable stages so that there is confidence that things are working as assembly progresses. Things like making sure your power is working properly, that your clock is working properly, that any decode logic is working properly, that the CPU runs (someone mentioned hardwiring NOPs and watching LEDs blink -- that's an excellent experiment), etc. etc. So that when you finally get that last components in to their sockets, not only are you confident it's not going vanish in a whiff of ozone and a curl of blue smoke, but that it's going to function. And if it doesn't, you have a good idea how far back you need to go since it "worked last". I fully expect assembly stages to possibly undo work in the previous stage that was their solely for testing (like the NOP wiring...).

Finally, some interface experiments to work with the "real world" via either blinky lights, relays/SCRs/motors, simple LCD displays. thermometers, etc.

A 6502, a static RAM chip, a VIA of some kind, an EPROM, a MAX-232 and some discrete decode logic is simple, it's functional, and it's easy to build. It has all of the major components laid out in a 3D, functioning block diagram.

An EPROM should be provided with the necessary software, I don't believe this is a software problem, so the software should be effectively solved until the machine is running. Once it's running, the students can program it for the interface experiments. I think it's fine that the students may well need to reprogram the EPROM throughout the assembly and testing process. It would be more interesting if a single EPROM could contain all of the software for all of the stages, with perhaps some jumpers telling it which particular stage to run to avoid the necessity of needing an EPROM programmer.

At the end, there should be some advanced experiments documented with more components: adding a keypad and display, adding more RAM, adding another VIA, moving the components to a permanent PCB and housing, etc.

To me, refinements to this design would head in the direction of making it cheaper and more fool proof for a student, especially a solo student, to participate. Simply, if you're going to change the design, provide the answers to "is it cheaper" and "is it simpler" and why.

It would be exciting to see if the combination of components and tools to assemble and test this minus a host PC, a USB -> Serial cable and the EPROM programmer could be done for < $200. It would be so awesome if this can be done without a scope.

So feel free to update.

[Dajgoro]

alkopop79 wrote:

I don't think you can build an SBC simpler.

Well actually you can.
There exists a chip that has ram, i/o and timer, it is the mythical RIOT(6532) chip.
You can practically build a computer with 3 ic(cpu, riot, rom), and some minimal glue logic.
The Atari 2600 was built in a similar way, but it has only 128 bytes of ram.
I actually have the RIOT chip, and i thought about building such a sbc.

[GARTHWILSON]

Quote:

Quote:

I don't think you can build an SBC simpler.

Well actually you can.
There exists a chip that has ram, i/o and timer, it is the mythical RIOT(6532) chip.
You can practically build a computer with 3 ic(cpu, riot, rom), and some minimal glue logic.

"Minimal" is right! I re-checked to see if you could go with none at all, but you do need one inverter, to invert A15 to the ROM's CS\. A15 without inverting can go to the 6532's CS2\. The 6532 data sheet is on this site at http://archive.6502.org/datasheets/mos_6532_riot.pdf.

[Dajgoro]

GARTHWILSON wrote:

Quote:

Quote:

I don't think you can build an SBC simpler.

Well actually you can.
There exists a chip that has ram, i/o and timer, it is the mythical RIOT(6532) chip.
You can practically build a computer with 3 ic(cpu, riot, rom), and some minimal glue logic.

"Minimal" is right! I re-checked to see if you could go with none at all, but you do need one inverter, to invert A15 to the ROM's CS\. A15 without inverting can go to the 6532's CS2\. The 6532 data sheet is on this site at http://archive.6502.org/datasheets/mos_6532_riot.pdf.

Actually you could do without it, but then you would have danger zones in the adders range, since the rom and riot would collide. If you don't want the inverter as a ic, you can build it using the BFR90 transistor and 2 resistors, as i did because lack of space on my sbc.

Edit: Except, if you live in a country like mine where is it almost impossible to get an 4 pin oscillator, then you need the 74ls04, 3 gates for the oscillator, one for the address inverter.

[GARTHWILSON]

Quote:

Actually you could do without it, but then you would have danger zones in the adders range, since the rom and riot would collide.

—not just danger, but non-operation, as there would absolutely be bus contention.

Quote:

3 gates for the oscillator,

With a 65c02, you don't even need an external oscillator. You can put a crystal or RC right on the processor pins. See the 2nd and 3rd diagrams on http://wilsonminesco.com/6502primer/ClkGen.html.

[BigDumbDinosaur]

Dajgoro wrote:

alkopop79 wrote:

I don't think you can build an SBC simpler.

Well actually you can. There exists a chip that has ram, i/o and timer, it is the mythical RIOT(6532) chip.

I'd stay away from silicon that hasn't been produced for years. If you design a DIY kit around something like RIOT you run the risk of one day not being able to furnish a complete kit because the RIOT is now unobtanium. We need to be careful here in that parts are used that are readily procured through distribution, not old stuff sold through one-off eBay auctions. That means no NMOS parts like the 6532. The lone exception should be the W65C22N sold by WDC, which is a bridge to NMOS, although actually a CMOS device.

Although it is desirable to keep it simple, simplicity should be overdone, as the challenge in learning how it all works will be lost.

[BigDumbDinosaur]

Dajgoro wrote:

Edit: Except, if you live in a country like mine where is it almost impossible to get an 4 pin oscillator, then you need the 74ls04, 3 gates for the oscillator, one for the address inverter.

You can generate a perfectly symmetric clock signal using a comparator (e.g., one half an LM393), along with a few passive items. Maximum frequency is limited but wouldn't be an issue in a basic DIY kit.