Epic! 8-bitguy uses 1 weird trick to detroy rare prototypes!

[Migraine Box]

What is the general consensus about Naomi?

View attachment 5812240

I have no idea who this person is and I would prefer to stay that way.

That being said, if @AmpleApricots is impressed with how active the new Amiga hardware and software scenes are, his head would spin if he knew how active the Acorn and ZX Spectrum scenes are for hardware and software respectively.

Oh, you Brits and your fancy home-grown 8-bits, how I envy thee. Closest I own is a Timex Sinclair 1000, but oh dear christ.

Last time I peeked at the Speccy world, there were enough in-production replacement parts that you could effectively build a brand new 1:1 replica, sans ULA. It's unreal how much love that little machine gets.

Good to hear the Acorn side is getting some more love. Been lucky enough to mess around with a RISC PC at a convention, and man what an incredible machine. Shame Acorn's ARM machines are pure unobtanium here in the US.

 

[Pee Cola]

Last time I peeked at the Speccy world, there were enough in-production replacement parts that you could effectively build a brand new 1:1 replica, sans ULA. It's unreal how much love that little machine gets.

One thing the ZX Spectrum has going for it is that replica cases and keyboards are readily available. I know the C64 is getting close now that replica cases are out there and all of the custom chips have been reproduced, but there's still no brand new C64 keyboard out last time I checked.

Another thing the ZX Spectrum has going for it is just how many ways there are to fill said case:
- you can buy a reproduction of the original main board and fill it with new parts including a reproduction ULA;
- you can buy a Harlequin 128 kit like @AmpleApricots and recreate the ULA with logic chips;
- you can buy an emulator board with an ESP32 on it, such as the ESPectrum; or
- you can buy a ZX Spectrum Next board (N-GO or similar) and use it in one of the OG Speccy modes

Yet no matter which way you go, it'll still feel like a real ZX Spectrum thanks to that case and keyboard (and the keyboard isn't THAT bad).

Objectively it's one of the less capable 8-bitters, but they're easy to reproduce and understand. Especially if you go down the Harlequin path, as you can see each part doing its thing. Vintage Speccies are also easy to fix and are still quite cheap in some parts of the world (tho not the USA obviously).

TBH I grew up with a C64 but I have a lot of love for the Speccy.

Good to hear the Acorn side is getting some more love. Been lucky enough to mess around with a RISC PC at a convention, and man what an incredible machine.

The RISC PC and Archimedes were great machines. Whilst it's not quite the same as a vintage RISC PC or Archimedes, RISC OS is still under active development for the Raspberry Pi.

 

[The Velvet Spoon]

Replacements for all Commodore 64 parts have been available for a while now:

 

[AmpleApricots]

Harlequin 128 kit

If anyone reading this goes down the route with blank PCB and no parts kit of the older issue 2D and finds the schematics and not much else - a few random thoughts: IMO do U39 HCT or even LS, if you populate it at all. (It's for the joystick port) Actually using all HCT logic might work better. If going with HC, go with a CMOS Z80 (Z84C00). I could imagine that the sound chip could be wonky in some constellations, if you have trouble with the sound, it might not be the chip itself. Beware the Chinaman, as he likes to reprint such parts. Don't get hung up on the oddly specific choice of outdated transistor for the keyboard matrix, an easier to get part is SA1015 or really any general purpose PNP transistor. If you have trouble sourcing TL712, (it's for the tape drive) for this application you can build it discretly without too much trouble, look at the datasheet.

The part kits feel overpriced but on the other hand since COVID there's been a massive price inflation and general lack of availability that just slowly seems to get better, so if you consider saving a lot of time and probably pretty much needing to order from different vendors with shipping costs if you source your own parts etc.. maybe not.

The keyboard thing would be relatively easy to solve these days. Because of mechanical keyboard autism and cheap china PCBs you can get a mechanical keyboard replacement done relatively cheap with whatever switch you can imagine. There's a lot of DIY in the mechanical keyboard sphere and you can get inspiration from self-built keyboards there. It's mostly a game of patience laying out the keyboard matrix and taking the right measurements so the replacement pcb fits into the original case. Another problem could be non-standard sized keyboard caps, so it all wouldn't fit into the original case properly or have the wrong labeling if you can get an off-the-shelf kit to somehow fit. That said, there's at least one service that lets you do custom sets with even your own labels/fonts you can design and upload (with laser etching) with which you could do a proper labeled layout. They're not even that expensive, about fifty bucks. I actually wanna set up an A600 semi-permanently I have in a custom case right now and am gonna go that route on that thing, since the original keyboard is horrid for my sensibilites onwadays. This if of course generally spoken, some cases won't fit anything else.

ESP32

That thing seriously blurs some lines. I want to set one up with zimodem firmware to use with my old systems, since that feels somehow more appropiate than networking them. I might tinker one into an old actual modem case.

 

[Migraine Box]

Replacements for all Commodore 64 parts have been available for a while now:

FPGAs are still funny to me. They're an absolute blessing for sure, but the idea of using a device dozens if not hundreds of times more powerful than the original machines they're being placed in just to emulate a single component is just inherently silly.

Those keyboards though...I know someone irl who did one of those 100% "new" C64 builds and used the same keyboard design but the keycaps almost never happened iirc. He was an early backer for them and they took aaaaaaaaaaaaaaaaaages to ship, and I don't even know if they're still in production. Modern keyboard replacements for anything are weird anyways mainly due to keycaps. There's one for the TRS-80 Model 4, but you need to fabricate a new metal mounting plate if you want it to fit within the original bezel of the computer, or else you're left with a very naked keyboard assembly sitting in a void in the case. Even then it still uses MX switches, so you can't re-use the original caps, which are very tall and sculpted that nothing in-production matches, not even SA, so it'll never look right. Eh.

 

[The Velvet Spoon]

FPGAs are still funny to me. They're an absolute blessing for sure, but the idea of using a device dozens if not hundreds of times more powerful than the original machines they're being placed in just to emulate a single component is just inherently silly.

Yeah, it does seem kinda stupid when building a machine from scratch to use individual FPGAs for each chip. That's where something like the Ultimate 64 is a more sensible option, and I would like to see something similar for more systems.

 

[AmpleApricots]

I still have my old Ultimate drive thingie. Got it way back in the day, I remember when it was all Hobbyist and you basically had to write Gideon privately. Funny to see a professional website now. Makes sense tho.

FPGA solutions for all individual ICs of a system never made that much sense to me. There's some usual suspects for some systems that like breaking and are widely known but otherwise it just doesn't happen that often or ever really. I never bought this weird narrative that this stuff is old and basically can fail any day. Your average custom IC is a semiconductor on a *huge* (compared to what we have now) manufacturing process. Most of them don't surpass temperatures of 45C in hours of operation. If they're well treated and maintained (e.g. no hotplugging, get a good power supply, replace failing caps etc.) they probably will outlive their owners, even if they were used every day which they usually really aren't. That there's no dire need of spare parts for many of the home computers you can see yourself by googling the prices of individual custom ICs. They are usually still pretty cheap. There just isn't that big a demand. If you're worried, stock up on spares.

There's a bigger problem for conserving I'm not sure is addressed as much. Some of these old computers and especially 3rd party extensions (everything from amiga accelerators to ISA cards) contain programmable logic which usually was proprietary IP and which original source code probably doesn't even exist anymore. The programming will fade with time, manufacturers of these ICs usually gave a retention period of about ~20 years which, safe to say, all of this stuff has surpassed, by about a decade or two by now. Eventually all this stuff will "forget" how it works. Contrary to e.g. eproms, you usually can't just read them as they're protected against that (to protect that IP) but have to be reverse engineered. There's been some method to coax the programming out of GALs by reversing voltages, but that's pretty much all I ever heard about. Granted, some of this stuff is a weekend project to reverse engineer (e.g. some simple address decoding) some other is more akin to designing the entire thing from the ground up. Since everyone and their grandmother seem to design expansion hardware for some of the more popular systems with much less limited and much more comfortable to work with modern logic, I could imagine many just don't care about preserving old 3rd party extensions. Easier to start from a clean slate with parts that are actually still available I guess.

 

[Mikoyan]

FPGA solutions for all individual ICs of a system never made that much sense to me. There's some usual suspects for some systems that like breaking and are widely known but otherwise it just doesn't happen that often or ever really.

Really I agree with you but I don't think that's how it's turned out.

On a bunch of the 8-bits, 74 series logic, which should be on a very large, low-stress process dies all the time. Most everything on a C64 (the one I know best), depending on exact build date, dies. The CPUs are pretty darn solid, the PLA is a trainwreck - but a well-reverse engineered one. The SID is a mess but that had to be on a weird process since it's practically a hybrid. VIC II is a hot runner and also tends to die, that's where my talk of werid franken-emulators came from.

You bring up the thing I love too: Every time I see some piece of equipment with a giant FPGA at the center of it, I think 'what happens when those start dying?' They seem remarkably stable so far but that is unlikely to last.

 

[AmpleApricots]

dies all the time

Yeah not wrong when I think about it. There was a worldwide 74 series logic chip shortage in the 80s, so MOS spun their own for a while to service their own needs. These are super unreliable and notorious for dying and in the C64s (and some floppy drives) I picked up, I basically replaced all these on sight. I don't remember what the problem with the PLA was, something with the passivation process in the early ones I think? It basically rotted from the inside. I have some 82S100 programmed to be PLAs from way back in the day. (these were the very first FPGAs pretty much. Same vintage as the C64 so it doesn't count, heh) I don't remember if these production problems affected some of the other ICs too. Screen RAM also had a tendency for dying, you could replace it with a later, more reliable CMOS part though. Same goes for the regular RAM which sometimes is defective, yes. That's another long term probem; RAM ages and DRAM from that time can have reliability problems by now as a result, but I noticed that failures are often clustering around the same manufacturers and dates. If you're not opposed to some tinkering you can use cheaply available SRAM in those though. Can even make it coin-cell battery backed and bankable, if you want a challenge. I remember somebody drilling and temperature probing the original NMOS SID and it had a die temperature in the 90s Celsius. One of the few chips I actually witnessed dying during using it myself, right in the middle of Edge of Disgrace. In the old style C64s, I'd personally put heatsinks on all of these. Commdore sorta did this on some revisions, in the cheapest style possible. These early C64s were wild. I have I think about seven of these and they're all very, very different one the inside. Some had an after-production (but still "officially issued" by Commodore) fix applied to it for port protection. The used components were east german. Before the fall of the wall east german. Classic "we'll use the cheapest we can get our hands on" Commodore. These serial board errata fixes btw. were often work-from-home jobs for bored german housewives.

Generally I'd recommend to use the late revision C64 II though, those you'd usually (but not always) find in the newer style cases. A lot of these problems are resolved there. PLA, some logic and color RAM disappeared into one ASIC, SID and VIC are HMOS, barely get warm and the whole thing is a lot more reliable. I also have a C16 and a plastic C128D. The C16 works but has a similar long-term reliability problem IIRC.

Every time I see some piece of equipment with a giant FPGA at the center of it, I think 'what happens when those start dying?' They seem remarkably stable so far but that is unlikely to last.

Some of the early ones still have external ROMs from which they get programmed on power application and which of course are easy to back up, but yeah, not always. the PALs you find on a lot of early PC hardware will probably last because these get programmed by basically burning fuses away, I'm less optimistic about the later ones and it stands to reason that some stuff already failed.

A way back (and I have to be very vague now because it's a good way to dox myself inadveriently) I came across an engineer who did such 3rd party hardware in a company, working in our common, unrelated field. We got to talk and exchanged emails later since I happened to own that hardware and he was very touched that he met somebody who still cared and tinkered around with this, his early work. (This was before the retro craze, for most people this was junk) He was kinda bitter about having bet on the wrong horse in the fledging computer market and left it soon altogether afterwards. He also sent me some schematics of that hardware which I think you can't get on the internet to this day. He admitted to me something I had guessed for a long time in regards to that hardware - that it had a pretty severe stability problem that would always eventually end in a crash of the system. (This apparently happened because of cost-cutting and he hated his boss for it) I lowkey tried to convince him to send me the source code of the logic, but no dice. A few months later he mailed me out of the blue and asked for my address. I gave it to him and recieved a few programmed ICs for that board in the mail soon after which fixed the problem. That hardware is still around and still usually buggy. While I found him very nice it's amazing that even after all these years he just decided to sit on the source code, probably taking it to his grave. Huge pity. Wouldn't even have been a legal concern, either. In most companies though, that stuff probably was just thrown away.

 

[Basic Block]

There's a bigger problem for conserving I'm not sure is addressed as much. Some of these old computers and especially 3rd party extensions (everything from amiga accelerators to ISA cards) contain programmable logic which usually was proprietary IP and which original source code probably doesn't even exist anymore. The programming will fade with time, manufacturers of these ICs usually gave a retention period of about ~20 years which, safe to say, all of this stuff has surpassed, by about a decade or two by now. Eventually all this stuff will "forget" how it works.

This is a big concern for the stuff I collect and repair. I've seen numerous PAL/GAL failures. The GAL failures I've observed have all been damaged outputs rather than degradation of their programmed contents. I try to backup whatever programmable logic I encounter but 9/10 times they have their security bits/fuses set. While the combinatorial stuff can be reversed easily enough, I don't know of any easy solutions for registered PALs and GALs. CPLDs like the Altera MAX 7000 series scare me even more. I either need to start spending quality time with a logic analyzer or roll the dice on sending these things to a cracking lab in China.

On a bunch of the 8-bits, 74 series logic, which should be on a very large, low-stress process dies all the time. Most everything on a C64 (the one I know best), depending on exact build date, dies.

In my experience 1980s Fujitsu 74LS logic seems to have a much higher failure rate than anything else. They tend to fail quite obviously too (non-tristate outputs float).

 

[AmpleApricots]

GALs

At least for some GALs you can try to first apply a voltage to the programming pin (+7V-+16.5V, I think it was fairly random what exact would work) on e.g. an GAL16V8 (Pin 2, this puts the GAL in "edit mode") THEN the normal +5V voltage on VCC, legend says this should bypass the security fuse and let you read the GAL like normal, e.g. with a minipro. This is fairly timing critical and needs some trial and error. (don't apply the programming voltage directly though, I'd recommend 4.7 - 10 kOhm) If you want to try this, practice on unimportant GALs you maybe filled with some test and fused yourself first. Also be careful how you apply that voltage with your specific programmer, so you don't blow the programmer up. I've never tried this myself so I don't have the slightest idea if it actually works, some people are very confident that it does. If you try it, please let me know. This will only work on GALs, not PALs, not even PALCEs.

The manufactures of GALs and PALs used to offer copy services where you'd sent them a "master IC" they would copy. They always explicitly stated that you should set the security fuse if you want the copies also to be security fused. Since I don't think they reset the fuse under an electron microscope, there probably is a way.

Otherwise yes, a lot of that old hardware is probably more or less fucked if you don't want to rebuild it.

 

[Enig]

Adventures in 90s Apple before Jobs got bought back in, a fun update.

It took me weeks but I finally got Mac OS 7 to let me use Stuffit and not implode using the BlueSCSI transfer window. All it took was hilarious amounts of trial and error:

1) Run Stuffit 5.1 (pretends to run but quits out without even an error message)
2) Install Stuffit 4 and run it (starts running but crashes with "error -120")
3) Move all the programs and games I need for it into a 500MB .img file using a Powerbook G4 running OS 9.2.2 (BlueSCSI transfer program shits its pants and load loops at 7% transferred)
4) Move it all in dozens of 10MB .img files (Disk Copy 6.3.3 declares they're all corrupt and don't have checksums, pisses its pants)
5) OPEN THE FUCKING DRIVE IMAGE FOR MAC OS 7.6.1 DIRECTLY IN DISK COPY ON OS 9 AND RAM ALL THE SHIT IN (somehow OS 9 thought that drive images launch in an image viewer at first and then it managed to fuck the resource files up and leave its own FINDER.DAT files exposed, haha now it's all on there but broken)

In the end, know what the ultimate fix was? Downloading a 500MB drive image with System 7.5.3 preinstalled meant for the OpenSCSI project. The entire problem wasn't with disk images, or Stuffit, or even the SCSI bandwidth or the BlueSCSI adapter itself. Mac OS 7.6.1 is just such a corroded piece of shit that somehow the act of running on a 68030 platform makes it shit its pants, if anyone tells you to just use 7.6 on an old Mac call them a retard outright because there is a reason they slammed OS 8 down on the market the first chance they could.

 

[Pee Cola]

There's some usual suspects for some systems that like breaking and are widely known but otherwise it just doesn't happen that often or ever really. I never bought this weird narrative that this stuff is old and basically can fail any day. Your average custom IC is a semiconductor on a *huge* (compared to what we have now) manufacturing process. Most of them don't surpass temperatures of 45C in hours of operation. If they're well treated and maintained (e.g. no hotplugging, get a good power supply, replace failing caps etc.) they probably will outlive their owners, even if they were used every day which they usually really aren't.

I think it depends on how good the quality control was on that IC when new. MOS was notorious for sloppy QC, which is why the C64 is well known for PLA and 74LS logic chip failures. Some of their chips (ROMs, CPUs) were less bad, but I've seen more 6510 failures than I have 6502 failures. MT (Micron) RAM quality control also sucked. Guess which brand of RAM Commodore often used on the C64?

The 264-series Commodores (C16, 116, Plus/4) were notorious for CPU and TED (ULA/PLA thingy) failures. They used an oddball CPU (7501/8501) which is no longer in production but there's an adapter board allowing the use of a 6502 or 6510. A TED replacement seems a little way off though...

Over on the ZX Spectrum (why do I keep bringing that up?), the ICs fail due to a combination of heat coming off the 7805 DCDC converter, shoddy Sinclair power supplies, low quality caps and the fact many were owned by kids who had no clue how to look after their machines and would either code or game for hours on end almost every day.

Then there's the BBC Micro, whose video ULA was notorious for running very hot, though nobody at Acorn realised this was a problem until the Beeb gained popularity in Australian schools. Computers switched on all day, summer temperatures reaching 40C and there was no such thing as air-conditioned classrooms in the 1980s...

The best example of your point is the early Amstrad/Schneider stuff (CPC, PCW). The custom chips in those almost never fail, which is ironic considering that Amstrad's IBM Compatible PCs were unreliable piles of garbage.

VIC II is a hot runner and also tends to die, that's where my talk of werid franken-emulators came from.

At least there's now a viable VIC II replacement (Kawari). As you probably know, it uses a small FPGA and FPGA stands for 'field programmable gate array'. The main differences between a modern FPGA and a vintage gate array chip (such as a PLA or ULA) is that an old-timey gate array had to be programmed at the time of fabrication whereas an FPGA doesn't, and FPGAs tend to have more gates in them than the old gate arrays did.

Obviously not all modern custom chip replacements are FPGA, as a smaller and cheaper CPLD will do the job just fine (and are the reason there are myriad PLA and SID replacements now).

Generally I'd recommend to use the late revision C64 II though, those you'd usually (but not always) find in the newer style cases.

I agree. The short board C64C is objectively the best widely available C64 variant, and they're relatively cheap on account of lacking the breadbin form factor nostalgia. Though if you're in Europe, there's always the C64G if you want short board reliability wrapped in an Amiga-coloured breadbin.

 

[Kalsarikännit]

Adventures in 90s Apple before Jobs got bought back in, a fun update.

It took me weeks but I finally got Mac OS 7 to let me use Stuffit and not implode using the BlueSCSI transfer window. All it took was hilarious amounts of trial and error:

1) Run Stuffit 5.1 (pretends to run but quits out without even an error message)
2) Install Stuffit 4 and run it (starts running but crashes with "error -120")
3) Move all the programs and games I need for it into a 500MB .img file using a Powerbook G4 running OS 9.2.2 (BlueSCSI transfer program shits its pants and load loops at 7% transferred)
4) Move it all in dozens of 10MB .img files (Disk Copy 6.3.3 declares they're all corrupt and don't have checksums, pisses its pants)
5) OPEN THE FUCKING DRIVE IMAGE FOR MAC OS 7.6.1 DIRECTLY IN DISK COPY ON OS 9 AND RAM ALL THE SHIT IN (somehow OS 9 thought that drive images launch in an image viewer at first and then it managed to fuck the resource files up and leave its own FINDER.DAT files exposed, haha now it's all on there but broken)

In the end, know what the ultimate fix was? Downloading a 500MB drive image with System 7.5.3 preinstalled meant for the OpenSCSI project. The entire problem wasn't with disk images, or Stuffit, or even the SCSI bandwidth or the BlueSCSI adapter itself. Mac OS 7.6.1 is just such a corroded piece of shit that somehow the act of running on a 68030 platform makes it shit its pants, if anyone tells you to just use 7.6 on an old Mac call them a retard outright because there is a reason they slammed OS 8 down on the market the first chance they could.

This post gave me a good laugh. Not because I’m laughing at you, but because getting those old files onto a newly refurbished Mac is ALWAYS an adventure.

 

[AmpleApricots]

CPU and TED (ULA/PLA thingy) failures

Yeah, something like that. I basically know nothing about the C16 but I think I do have spares. One of these days I have to go through my 8 bitters and see what is actually working and what needs some maintenance. These were thrown out like crazy and I have a rather high number as a result.

Micron

Which is funny considering how big they are now. Another such candiate later on in PC hardware was Vitelic. Stylized V on the package. If I came across defective drams, it was almost always this brand.

7805

The older C64 revisions suffered from this too. If you want to do something good for a system of that vintage, throw the linear regulator out and put in a high quality switching mode drop in replacement. (If there is a big ceramic resistor in the area of that 7805, remove that too) You'll reduce the load on the power supply by a lot and they barely will get warm if at all. Also a good idea if you intend to put something more elaborate on the expansion port. 1A-1.5A (at 5V) is usually enough for these 8 bitters. Speaking of that, considering the nature of USB-C PD I'm surprised people haven't made USB-C adapters for all these systems and are still fuzzing around with custom-built power supples. That's my plan for the A600 when I put it up as more permanent machine, I'm gonna give it an USB-C power plug with internal power supply.

Spoiler: Lots of Amiga spergery, read at own risk

The A600 was pretty much the most pointless Amiga made and was an amazing example of Commodore's mismanagement for the time but with modern needs and sensibilities, it's an alright machine. ECS chipset, easily expandable to 2 MB of chipram, tiny, IDE, PCMCIA slot. The latter is a lot more useful now than it was back then, You can stick a CF card (Or SD card, with adapter) into it and basically use it like a card reader/external drive which of course is super convenient. I was also considering going with an A1200, as AGA is an improvement (18-24 bit color palette, max. 256 colors vs, 32 out of 4096, 14 Mhz chipset, partitally 32 bit) and I have accelerators for my A1200s but that's somehow not what the Amiga was at for me. Also it'll take a while before something like the RGB2HDMI will exist for those color depths.

I ended up getting an accelerator for mine, since 3.1 Workbench usage on the original CPU can be sluggish sometimes, especially with no FastRAM. There is a lot of fancy and very fast accelerators you can buy nowadays for the A600, but for an OCS/ECS system IMO they are kind of pointless. You'll always be weighted down by the slow chipset. The few number crunchy operations (3D rendering, compiling of big programs, internet browsing, playing mp3s, large file I/O) which aren't too chipset dependant will also be slow and a poor experience on everything besides the PiStorm and do you really want to do this stuff on a real Amiga? I don't.


(picture not mine)

There is zero documentation on it besides a short wiki page in polish and an equally polish youtube video. Considering the communities hunt for the longer benchmark bar I doubt it's very popular and indeed, there is very little talk about it. It's a nice and simple design though, the CPU is clocked at 28 Mhz and it has about 12 MB of usable RAM. (which is a lot for an Amiga, I would've been happy with 4. Interestingly, it can add an additional 1,5 MB "SlowRAM/Trapdoor RAM" at the appropiate address. This won't be "true" SlowRAM at all though and indeed be just as fast as the FastRAM) There is a CLPD, some RAM, 3.3V regulation, and a 68SEC000, which is the 68k for embedded applications, but for all intents the same CPU as the original 68k in how the Amiga uses it. At 28 Mhz, it's about double as fast as the 14 Mhz EC020 CPU in the A1200, since that CPU has no FastRAM. U6 (the small chip next to the CPU in the corner) is the ICS501M, which is a clock multiplier. What this expansion does is take the 7 Mhz that the socketed 68k gets from the system, and then multiply it up to approx. 28 MHz. That way we stay synchronous to the chipset. If you turn the turbo off, all it does is run the SEC from the 7 Mhz. (this was probably designed this way to make the RAM still usable, it would have been more complex to make it available to the original CPU and they're functionally the same anyways at the same speed)

The SEC was used in the Minimig, which was an Amiga reproduction from the 00s that had the entire OCS chipset realized in a single FPGA but used this processor as CPU. In these days people found out that you can clock the SEC quite agressively, and in some minimig revisions it runs at 50 Mhz, which is a lot considering the part is rated at 20 Mhz. This puts it about at the speed of a slower 030, with none of the extra instructions and 32 bit bus goodness though. Since a static design of the 68k was kinda trivial in the late 90s, and this was intended to be for embedded applications in a wide variety of situations, it makes sense that the chip is quite robust in it's margins. Motorola's max. ratings were always quite conservative. Although this chip is already overclocked here, it stays at room temperature. That said, for an Amiga part it is in a nice sweet spot of speed. My installation with patches, commodities and all boots in about 4 seconds from IDE, progams load almost instantly and the "killer apps" of the day like Brilliance, Deluxe Paint, FinalCalc, TurboText etc. work nicely and you can easily multitask with all of that RAM. It exactly where it helps some of these applications and games to get over that lag some of them had without running everything too fast or feeling "ridiculous"(for the lack of a better word) and since it has no cache, when all happens from chipram the speed gain is very minimal. You rarely have to disable the turbo, even when playing old games directly from floppy.

I then got curious and measured out what the two clock multiplicator settings pins are connected to, and sure enough - they are not directly connected to GND which is all they would need to run at 28 Mhz (4x setting) but to two distinct pins of the CLPD, which might either mean the designer intended to leave the speed the CPU can be set at open not to pcb manufaturing time but to CLPD programming time or made them software controllable in a hidden way. Time to poke around a bit.

(There's something I dub 'the Amiga effect' where the entire OS is just super inviting to tinker around with. Since I set the machine up on my workbench I drew a new mouse cursor, made parts of an old timey font and doodled around in Brilliance. It's just engaging and inviting that way. I plan to do a few things with it when I have my faux modem built.)

 

[Mikoyan]

PCMCIA slot

PCMCIA-mania back then on desktops was a weird thing. As you say it's great for a CF adapter now, but back then I didn't get the point. IBM went big into that too with the PC 300/700 and other systems. I'm just not sure what it was supposed to be for at the time, other than data exchange with a laptop.

 

[Migraine Box]

PCMCIA-mania back then on desktops was a weird thing. As you say it's great for a CF adapter now, but back then I didn't get the point. IBM went big into that too with the PC 300/700 and other systems. I'm just not sure what it was supposed to be for at the time, other than data exchange with a laptop.

A lot of early digital cameras, PDAs, and MP3 players used PCMCIA cards/adapters to transfer data, using the likes of Compact Flash cards or proprietary memory cards (fuck you sony I still hate magicgate). Then there was the weirder stuff like the Iomega Clik. I guess you could've also used it for ethernet or wifi if you didn't want to sacrifice a PCI slot.

 

[Susanna]

PCMCIA as a proto-USB strikes me as an odd choice, when SCSI already did that, cheaper.

 

[AmpleApricots]

IBM

Yeah my PS/2 system had such an adapter card, complete with software controlled locks for the pcmcia cards so you couldn't pull them when locked. The SRAM cards were the only way to get FastRAM into an A600 for the longest time. They also worked as battery backed storage.

The PCMCIA slot on the Amiga is only 16 bit and was included before the PCMCIA standard was even finished. It is only compatible to 5V devices and lot of PCMCIA hardware won't work in it. That said, it is possible to use both WiFi and Ethernet with an Amiga this way. I think there's even WPA2 support now. Is it useful though? Meh.

 

[BIG BILL HELL'S]

David has a new video up, this time a little history lesson about the 6502 processor.