Vintage Computer Festival(s)

2022-08-14 Stefan No Comments

Hi there, it’s been a while. Somewhere down the line life happened and wait, not one, not two, but three Vintage Computer Festivals have happened since my last blog post. Three that I participated in, that is.

Seeing people that are not projected on square glass pieces was more time consuming than anticipated, and so typing text into this little square glass piece had been down prioritized. After a distant panic of one of my readers about “All good things…” meaning an end of this blog, I did promise some updates.

Last year I had signed up for a booth at VCF east to show off some of my little adapters, like the ROMY redesign that I had made as a starter project. I got a bit of cold feet last minute as three tiny PCBs on a large table are everything but impressive, and so – as there were no other Amiga booths present – decided to just bring a few machines. It turned out to be a great event and I got to know a lot of the local vintage community in person, including the amazing Liza Loop, who told me about the BBQ parties she had with Jay Miner back in the day. These stories are what we are here for, at the end of the day. There is a lot of technology out there, and a lot of PCBs everywhere, but the history of our technology tells us the tale of Silicon Valley and its dreams.

In October Chris Hooper and I made it up to AmiWest in Sacramento. It turned out to be a lot smaller than I anticipated (and I anticipated for a good 30 some years here, this was my first AmiWest), and a lot better. I had just picked up an A4091 SCSI controller on my travels to Germany before, and we were chit-chatting that with the Dave Haynie Archives and this board, we could indeed do a reproduction of the board. That’s crazy talk. Or is it now.

It was a busy winter for Chris and me, and by April I traveled to VCF East with three reproduction A4091 boards in my bag and equipped with a slide deck to give a talk about this project.

All good things …

2021-03-21 Stefan 4 Comments

… come in threes, they say, or “Aller guten Dinge sind drei” in my native tongue German. And so I set out to make a third and final attempt at reliably delivering great 68040-to-68060 adapters. Remember, earlier in my adventure I built one of these adapters with a PCB made by OSHpark. I was so excited, that I sold it on eBay and started to make another one. And failed. And another one. And… Well, you get the story here.

In my second attempt I found out, that I had forgotten to refresh the fill zones in KiCAD – What a rookie mistake, that left half a dozen 68060 PGA sockets dead, soldered onto PCBs where VCC and GND were never connected to the mainboard. The good news? No Amigas, and no 68060s were harmed in the production of that episode. But I promised, I would get this working. At least to myself. I had also disliked some of the characteristics of the old design.

I had to dremel the sockets I got into shape because they tried to occupy the same space as the other parts on the PCB, in clear violations of the physical laws of this time-space.
The interconnect between the two boards was oddly misaligned with the rest of the pins, making it impossible to use the substrate of the socket to keep them nice and tidy (and giving another reason for more dremeling. UGH)
Compared to the original design the KiCAD version by richx used single resistors instead of resistor networks. For the sake of space, those could be replaced by nice little 0603 RNs.

An evening with KiCAD and a few more days of waiting on a PCBway delivery, and here it is:

Exciting! New PCBs (with GND connection!)

Finding common GND

2021-02-25 Stefan 3 Comments

Alright, there is some progress on the front of my 68040 to 68060 adapters. In the last month I have made a serious improvement of my lab: a simul focal microscope with double boom stand.

This has vastly improved my ability to see what I am actually soldering, instead of almost flying blind.

This led me to attempt another approach to making 68060 adapters, this time I could maybe be less of a butcher with a soldering iron. And with so much visibility, it makes sense to learn from the others out there. Adam Polkosnik has perfected his way of making 060 adapters and was kind enough to teach me all of his secret tricks, some of which I will document here.

My favorite of these tricks is to use actual PGA sockets instead of cutting 400 single pins out of pin headers. The original 68060 adapter instructions suggested to cut the VCC pins between the upper and lower PCB. A smarter and cleaner way that I learned from Adam, is to use a third PGA socket to keep the distance and separate the 5V and 3.3V circuits safely. This gets rid of a very destructive part of the construction of these adapters.

iKod.se int7 adapter

2021-01-18 Stefan 1 Comment

Don’t you know the feeling when you’re just on cloud 7 with your Amiga?

Maybe you remember the action replay for Amiga 500 which let you scan memory for lives and mods? Or you miss a nice and handy debugger when debugging that latest 68000 assembly of yours?

ikod.se had the right solution for your Amiga 1200 or 4000: an int 7 adapter. It is easily assembled with only a 52pin PLCC socket, 3 1n4148 diodes and a pin header and switch.

The thing is, it’s hardly worth, or even possible, to order anything in single quantities when doing PCBs. The price for 10 might just be the same. So let’s find solace in the soothing effects of soldering

This was one of the projects that I used to practice working with solder paste and hot air. In the picture above I used a little bit too much in some places.

68040 to 68060 – Upformation

2021-01-18 Stefan 9 Comments

Back in the late 80s and early 90s, when the Amiga was relevant as a technical revolution, I was never able to lay my hands on anything beyond an 68000 with 7.14MHz. Years later, around 1997, I acquired an Amiga 3000 with 25MHz as part of one of my university side gigs of porting software from various UNIXes (in this case SCO Unix) to Linux. But a 040 or even 060 was far out of reach until I was well into my first full time job at SUSE Linux when I got a Cyberstorm MK1 that – as we know – doesn’t fit into the A3000. Long story short, the first time I actually ever ran an Amiga with something faster than the 68030 @ 25 was around …. Christmas 2019.

As life happens, and we sometimes overcompensate for our youth’s desires as we grow older, I added one of Chucky’s A3660 rev 1.1 to my collection, (built by Nicolas Baumgardt who lives in Switzerland, but only few miles from where I grew up and hacked on the Amiga back in 1987) that so far consisted of the A3640 that screamed for pad replacements and recaps.

Along the way I learned that the Amiga CPU local bus is really a 68030 bus, so memory access on everything above a 030 has to be translated and is then slightly slower than with an actual 68030. Ouch.

I also learned that the Zorro III slot is not really all that great. Well, Super Buster is not in its current implementation. While the theoretical maximum of Zorro III beats the first PCI standard, the actual implementation is stuck at 8-10MB/s memory access speed. We need an update Dave Haynie! Then our BigRAM+es could be really super awesome!

But back to … back to the 68060. Once I tasted the sweet excess of the Motorola flagship’s raw CPU power, and when I got my beautiful Hakko soldering station, I had to upgrade my 3640 to a 060 processor as well.

Winter is coming…

2020-12-05 Stefan No Comments

I have not been keeping up well with this blog in the last few months. Starting a new job definitely does that to this sort of project. I did spend time behind the scenes on some cool things that are in various stages of completion. Some involve 3d printing with the Prusa i3 MK3s that I assembled this summer. Others are following the dream of creating new and exciting Amiga custom chips.

But overall, it has been harder to spend time in the Silicon Valley garage also for one reason: Winter is coming! And spending time out there for projects has become a lot less attractive.

Today the solution to keeping the endless stream is ideas flowing has arrived. It will help the atoms dancing and the feet warm during nights of soldering. As the year nears the end, and many projects are waiting to be realized, this will heat up the velocity:

Cyber, Cyber!

2020-09-01 Stefan 4 Comments

Or was it “Hyper, hyper!“? Scooter, please help a child of the 80s and 90s out here! This episode is about shuffling hardware, and what it takes to make your Amiga really fast!

I was lucky enough to get my hands on a Cyberstorm MK2 card (and I still have all my organs). I almost despaired when I put the card into my Amiga and nothing happened. Turns out, the Amiga wants J100 jumpers to internal despite an oscillator being present on the card. What a shock and what a relief when dummy me was told by the internet to read the manual that came with the card. Who does that anymore these days?

The previous owner parted with this piece of 90s magic with an 68060 revision 1 at the factory frequency of 50MHz.

That’s no fun, thought this silly old lad, and after some journey to the Nightfall Crew and Cosmos’ Blog I felt the best possible thing I could do is to risk destroying another piece of ancient hardware and overclocking a poor 68060 CPU by no less than 100%.

For that I would have to exchange the 68060 rev 1 with the rev 6 from my A3660 (Thank you, Nicolas from Switzerland). But how to get the CPUs out of their sockets without destroying CPUs or boards. With the A3660 I was less nervous. The card is somewhat solid and I had been able to remove the CPU with my Hakko CHP 7-SA tweezers (6.50USD on Amazon, great investment), but the Cyberstorm MK2 just made me a lot more nervous, as it was starting to bend under the pressure. The Cyberstorm is super packed, there is not much room left and right, unlike the 3640/3660 design. So I found these Intel Overdrive chip removers on eBay:

One on each side, and any 68060 will follow your lead easily.

Everybody gets a second chance

2020-09-01 Stefan No Comments

… and even more obviously, everybody needs a second recap in their quarantined life. After the recap of my A3640 went surprisingly well, and even ripped off pads were replaced and reconnected, I felt confident enough to do a board that’s a little bit bigger and a little bit more involved than that poor little 040 accelerator: My Amiga 4000.

As my Amiga 3640 processor card, the Amiga 4000D board was previously recapped in a job that replaced the SMD capacitors with their through-hole siblings.

I know, I know, another recap story, REALLY? Haven’t we heard enough of that already? Let’s just twist the caps off, the recap trolls on Facebook will tell you. Come on, folks. You are lucky enough to have the chance to operate some 30 plus year old machinery. Think about it for a second. What else that you have today will still be good in thirty years from now? Your iPhone or Android phone? Certainly not. Your MacBook? Your Chromebook? This stuff is going to be in the landfill and long forgotten in much less time. So PLEASE, treat these machines carefully. They are witnesses of a very different time, when reality consisted of the cold war and 1200 baud modems. If you have doubt about what to do, let someone help you. There is a friendly community of folks keeping these machines alive. And if you want to do it yourself, practice with something else first. Something that is ok to break. I heard there are a lot of cheap Ataris out there (just kidding!!)

It’s not visible in the picture, but many of the legs of these caps were not attached to the board anymore

Rock around the clock

2020-08-31 Stefan No Comments

… When I have been playing my overclocking games, I was sourcing a lot of oscillators from eBay. One thing I noticed was that there are a lot more choices for modern SMD oscillators than for the old DIP14 tin cans. As a matter of fact, I found 50 and 100MHz oscillators in all sizes quite easily, but what if you want to go up gradually, say 105 or 108 MHz? Maybe I looked in the wrong places, but I was out of luck.

A solution had to be found, and that was to put a rock under the clock to make it fit. And of course I was just looking for a project simple enough to try and build something with KiCad.

Being a hardware developer in 2020 is almost as easy as being a software developer in 1980. You basically need a computer and a lot of patience. Once my KiCad design was done, the first version was quickly uploaded to OSHPARK. This place is fabulous. Want to try some hardware really quickly? Get three boards of decent build quality and great looks for two bucks in the mail, with no delivery fees. It’s almost paradise.

The order process is simple and small design are not discouraged by 15 dollar delivery fees

Das Labor

2020-08-22 Stefan No Comments

Ok. I have to make a confession. I’m really awful at keeping organized. And if I let my inner engineer engineer out for an evening or two, it can get pretty messy and it starts looking like my cat Katze’s bed, a wonderfully cozy place full of toys and no space to be:

Alright, I admit another thing. My space looks way worse. That’s mostly because Katze has not figured out how to order toys on the internet (yet):

Hazard Warning! Don’t do this at home. And don’t show this to your children.

So on Friday evening I just couldn’t stand the mess anymore and I stopped by at Home Depot for a solution, which I found in a nice 10in wide wooden panel that the friendly man with the giant saw cut into three pieces, two of them 10in long and one 5ft.