Complete working transistor-scale replica of the classic MOS6502 microprocessor | Hacker News Reader

Complete working transistor-scale replica of the classic MOS6502 microprocessor

monster6502.com

140 pointsadrian_mrd38 comments

MOnSter 6502 creator here for any questions...

Such a beautiful board! Now do an 8080 :-)

I'd rather an 1802 or 6809.

At the first day on my first professional programming job in 1980 or so, my new manager told me that I would be writing a brand-new 6502 macro assembler, in 6502 assembly language, to replace the slow & limited one supplied by the vendor (Ohio Scientific).

Under his guidance I wrote the assembler, and even earned a raise after simulating the hashing function in FORTRAN and finding it lacking.

I still have very fond memories of that job and of the 6502, and would be interested in purchasing a replica as soon as you go into production!

Any chance this will be commercially available either as a kit, a finished product or just the breadboard?

Are you nuts enough to abandon transistor-level compatibility and do a BJT design?

I had a single page reference card (plastic) for the 6502 that had everything you needed to know on it -- it was awesome! It was used many times while hacking around on the Apple ][+ computers we had in grade school. I miss the days when computers could be completely understood by a single person.

Looks like there is a scan of it online: http://www.atarimania.com/documents/6502%20(65xx)%20Micropro...

This would make some incredible, if expensive, wall art for those of us who got started on the 6502.

Would be great to display this next to a real delidded 6502 to show the level of integration. Then tell people "that was in the 1970s" and explain how big their phone chip would be. Best visual of scaling ever.

An other thing which would be interesting is a 6502 in a modern process, the 6502 had already had such an update in the 8502 (a smaller and less power-hungry 6510). Or possibly a machine-code-compatible version, if the die shots have revealed known issues which could be fixed by redoing the logic.

Or at least an FPGA with a 6502 burned in (like the 65F02) but I'm not sure it would be smaller as i assume the FPGA has a lot more hardware than the 6502 (even needs).

WDC still sells the 65C02 (uses CMOS logic instead of NMOS): https://www.westerndesigncenter.com/wdc/chips.php

One fix in this variant is that all reserved/undocumented instructions are NOPs, they don't do anything unexpected. Another one is that you can stop the clock entirely and it will not lose track of what it was doing earlier.

DC to light! Well .. DC to medium-wave broadcast frequencies anyway.

They would make a million of them one wafer, and then struggle to find them. :)

Why do they expect it to be so expensive? ("It is definitely not cheap to make one of these. We are currently estimating the cost at between $2k - $4k.")

With 4769 components on the board, that would suggest a cost per component of over 40 cents. Looking at JLCPCB prices for SMD assembly and components, I would have expected an order of magnitude less.

I have wondered the same but I think SMD wasn’t as cheap a decade ago as it is today for one-off designs. I wonder what a modern day jlc version would cost.

If they do manage to sell it in the $200-400 range, I know what's going to be hanging on my wall.

Heck yeah!

But actually, I'm thinking a nice framed picture might be pretty nice.

Me too! Honestly, ever since I saw that years ago I wanted one. I don't even like the 6502. I would rather have the ARM1 (http://www.visual6502.org/sim/varm/armgl.html) but man, oh, man I love this idea. This is the cheapy version of all of that: https://hackaday.com/2022/10/12/the-2022-supercon-badge-is-a... though it's all faked and emulated through an MCU. Cool, but not as cool as real discrete logic.

from https://monster6502.com/ : "the cost is actually dominated by the component and assembly costs of an extremely large number of tiny components, each of which is individually quite inexpensive. "

Yup. And the cost of debugging the board because of a mistake in the assembly.

I saw that, but that doesn't line up with the prices I saw, that's why I was confused. As I said, it'd imply over 40 cents (for part + assembly) per component.

don't forget assembly costs. very few boards have such a large number of components, and they tend to be big and expensive server boards...

larger boards are also more difficult to work with in general. JLCPCB plays a few tricks to keep the costs down on the smaller boards, but once you go past certain limits, the price goes up significantly.

I did consider assembly cost, that's why I was confused.

Specifically JLCPCB claims "$0.0017 per joint" for PCB assembly, which would be around 10k joints for the transistors, and another 5k for the resistors, LEDs and other parts, so around $25 + the parts cost.

Some other PCB manufacturer (edit: PCBway) I checked also didn't seem to charge more than a few hundred per fully populated board (for the board + assembly).

It's of course possible that that JLCPCB pricing goes way up with such parts counts, but the other manufacturer did ask for count + size before producing the automatic quote.

Edit: Didn't realize you were the project author - thanks for the response! I guess that's what hides behind "Final price is subject to our review" then. Still, looking forward to seeing this in person

Most of the components are leds and transistors (and some resistors) and those are (still) dirt cheap... you can get whole microcontrollers for ~3 cents a piece [0] (if you buy a few thousands). PCBs are comparatively cheap, few tens of dollars, depending on the process, and with pick-and-place, assembly is also relatively cheap.

so yeah.. maybe they're making them in some more expensive country or use some exotic components.

[0] https://www.youtube.com/watch?v=VYhAGnsnO7w

At 50 kHz, can you understand anything from looking at over 300 LEDs? Or is it just the impression that things are happening too fast? Which isn't really the case for 20 times slower than the original 1 MHz.

Very surely loops should show a rhythmic pattern. I am sure I have seen equivalent project that run on the browser...

Check e.g. Visual 6502 - http://www.visual6502.org/JSSim/index.html

Would be fun to watch this hardware alongside a sorting algorithm visualization like this: https://www.sortvisualizer.com/quicksort/

from back when computers had really small memory, I've seen software that did sorting on the user screen because the screen at that point was extra available free memory!

At 1Mhz you could tell certain things, like looping or crashed by looking at the front panel of my Altair 8800.

Yes. Some years ago I was debugging a Z80 start-up problem (ADA MP-1 guitar pre-amp). By probing the address lines, I was able to deduce that it's in a tight loop.

It turned out to be a dumb circuit design around the Z80's NMI. I don't remember the details now, but there is a circuit that is like a low voltage (brownout) detector that triggers NMI, where the firmware branches to a loop. The idea was likely this: to prevent trashing the settings if that happened during a write cycle (setting being stored). Anyway, this NMI would happen on startup, during the voltage ramp-up.

I deduced it purely from reading the schematics that, hey, this circuit could activate NMI close to power up.

The fix was to just cut the trace to the NMI line with a sharp instrument; presto, no more random lockups on power on.

The kicker was having debugged all this, I found a 1989 dated document from Ada which confirmed everything and recommended the same fix. I think had that document all along but it was hard to understand what it was getting at. and take note, if you didn't already understand the problem.

at 50KHz, the LEDs mostly appear to be on, albeit at different brightness levels. but with the clock rate set much lower (say, below 50hz) it looks a lot more animated.

look at that photo on the top, simply a piece of art. just incredible.

Was the design leaked or published or reverse-engineered? Of course today no vendor should have a problem with that kind of secrets becoming public.

The author mentions the Visual 6502 project in the credits section at the end of the page, which is (was) a long-running reverse-engineering effort to map the chip at the transistor level: http://visual6502.org/

yes, it is based on their reverse engineering work. i hand-copied every transistor from their netlist into a schematic and went from there.

wow, this is gorgeous. I've had similar ideas, but the execution here is just phenomenal.

Great piece of work!

Looks beautiful

Reply on news.ycombinator.com