I just updated the heatsink on one of the CPLDs on the TerribleFire 1260 in my Amiga 1200.
The heatsink I used was modified so that the compact flash adapter I had could be fitted on the TF1260. And it was so ugly.
Not that I would see the heatsink when I used the computer. But it still bothered me. So I ordered a new heatsink and installed it instead.
And tadaaa suddenly my Amiga 1200 is dangerously close to being finished!
N2630 is a great 68030 CPU card you can build yourself for the Amiga 2000 or A2000EATX project. You can find more information about the N2630 project on the website here.
There is a model for a bracket for the N2630 that you can 3D print. I did that but it was too flimsy and eventually broke in two pieces.
I was planning on ordering a 3D metal print of the bracket from JLC but heard of arymanx on Amibay who had made a small batch of metal brackets for the CPU card. In this hobby, you have to act fast, or the chance might never come up again so in less than a week a shiny metal N2630 Amiga 030 CPU card bracket arrived to me from UK.
If you are interested in a metal bracket for the N2630 too, feel free to contact arymanx directly about it on AmiBay.
Its just a bracket, but is a great reminder of that it is the small details that makes a DIY build great.
I mounted the bracket to the N2630 with two screws and nuts. The nuts got a drop of blue Loctite to hold them secure. Fitment was excellent.
Not only does it hold the card better in the case it also looks better.
Pay no attention to the nut wedged between bracket and case. The CPU slot I used when building my A2000EATX motherboard was taller than a regular CPU slot used on the Amiga 2000 making the CPU card sit further out from the motherboard.
I had two summer projects this year that I recently finished. One was an Acill A4000D replica motherboard and the other was a Hese made A4000+ Alice A4000D CR motherboard replica. You could say two brothas from different mothas or something…
Both motherboards worked fine when doing basic test runs, however they both failed to run with a BFG9060. All I got when running them with a BFG9060 060 CPU card was a black screen.
Anyway, as usual building them up was pure pleasure from start to finish. I even enjoyed desoldering the Acill A4000D motherboard from a few passives and pin headers someone else had a false start with. I mean, off course you want a full set of pin headers soldered to your 400+ small parts PCB (not really).
The motherboard was all built up and looking good in shiny red matching a BFG9060 with a mystery (rev. 1) 060 I built last year.
As good as it was looking it failed to run with the BFG9060. It was working fine with the A3630 CPU board I built earlier this year, but not with the 060 card. That reminded me that I had the same problem with my Alice A4000D motherboard a couple of weeks ago.
The Alice A4000D has an 030 CPU on the motherboard so it does not need a CPU card. It was working fine with the on board 030.
I got the suggestion to try a new delay line as that could be the problem the CPU card failed to run. As the A3630 has exactly the same CPU as the one mounted on the motherboard, it is difficult to say if the A3630 was running or not.
If you are in this hobby you have to grow a passion for trying all hardware combinations to find the solution. In this case, the reason for the Amiga 4000 getting a black screen when running a BFG9060 was the PSU as everything seemed to run fine with another PSU than the SFX one I had in the A4000D case.
I changed the SFX PSU that previously was working fine with the EXACT hardware setup I am running here (but different Amiga 4000 motherboard) with a small ITX PSU and suddenly DiagROM worked fine and detected the 060 CPU.
This lead me to believe that maybe the PSU was not pushed too hard as some ATX PSUs fail to run if there is a tiny load on them (or something like that). So I decided to do a final test by adding some cards to my A4000D and try the SFX PSU again.
Unfortunately it did not start with the SFX PSU and 060 card even if I loaded the machine with all Zorro slots filled, including adding an old 3.5″ harddrive.
So next step is to get a new SFX PSU. But at least I know I have to working motherboards!
Update! 20251215
Apparantly it was the Kickstart that was the problem.
You may have seen my previous post where I did a test run on my ReAmiga 3000 that I finished earlier this year. If not, check out my post about my ReA3000 that I built in April here.
While I did do a test run in DiagROM previously I did not have a daugherboard for it so I just whent through the usual tests (successfully) and called it a day. This time I did a more comprehensive test session where I tested a Amiga 3000 daughterboard, a BFG9060, a GottaGoFaZt3r 256MB Z3 memory card, programmed and replaced the logic chips and also replaced some ICs that I had to use adapters for previously.
While my ReAmiga 3000 started up fine in DiagROM it failed to run with a Kickstart rom. I was puzzled about this since it was running fine in DiagROM. No matter what I tried I ended up with a black screen. Problem was solved by adding the daughtercard. Now Kickstart boot screen came up.
I am using a Matze A3000DB, more info on it here. This is just like a regular Amiga 3000 daughterboard but it also has I2C functionality so you can monitor temps. More info about I2C here, the added functionality is based on that project. So to be honest, I have not really fully understood I2C, perhaps I will do a deep dive in the future. I have an CPLDICY card I built last year and its neat to see temps. I am thinking this could be interesting to monitor if running a tight A3000 case with bad cooling.
Another note, this is a version of the Matze A3000DB with some added functions by kavanoz & CDH, see more here about this specific version of the daughterboard.
A GottaGoFaZt3r 256MB Z3 memory card was inserted and it was detected by DiagROM.
Recently I built an A4000D that failed to run with a CPU card. That motivated me to test my older builds with a CPU card to make sure they work properly with a faster CPU. The only issue I had was to figure out how to jumper the motherboard, then it was smooth sailings.
DiagROM identified the 060 and also the fast ram. Now I need to do a 6 hour fish render and stability will be tested (something to do for the future).
If you look at the BOM for the ReAmiga 3000 you can see that Chucky recommends replacing the 74F646S with 74FCT646. Currently 74FCT646 can only be found in SOIC-24 from Mouser and Digikey and not in DIP so you need a SOIC-24 to DIP adapter to use them. While the adapter worked fine I just felt it would look better to run DIP chips instead. I had an UTsource order going for my A2386SX order in the pipeline so I added eleven 74FCT646 chips to that order from UTsource. Why 11 when the A3000 only needs 9? Well two of them is going to my A2386SX boards.
The 74F646S chips actually runs hot, especially when there is 9 of them, that is something I noticed when using them from my donor machine. So it was a nobrainer to replace them with cooler running chips.
As I put in sockets for all the 74FCT646 chip adapters I am thinking of removing the sockets in the future and solder the chips directly to the motherboard for an even cleaner look. The ground plane in an A3000 is brutal though, not sure I wanna wrestle with this board desoldering stuff again. Desoldering the KEL 200 pin CPU slot was a nightmare.
There are four logic chips on the Amiga 3000. About 15 years ago I had the oppportunity to buy a broken Amiga 3000 cheap because it had a broken display output. Turned out it was because one of the logic chips was broken, so it was an easy fix to just replace one of the chips with a new one.
So to future proof this ReAmiga 3000 build I replaced the logic chips I got from the donor A3000 with modern alternatives and got the JEDEC files to program them with from here.
I hope I can find a case for this build, they are difficult to come by but occasionally you can find one. Actually wish I had one now as I would like to set it up to a running system!
C’est la vie…
I just finished building this beautiful purple PCB BFG 9060 accelerator card for the Amiga and thought I would share a picture of it. As it was mostly already finished it just took an hour to solder on the KEL connector, fan header and voltage thing (regulator I think?).
I finally get to use the original spec Motorola 040 CPU heatsink that came from an old 040 C= 3640 CPU card I had 20+ years ago. Just have to get some thermal tape, will also add heatsinks to the CPLDs. This card will not be overclocked as it is fitted with a rev 5 68060.
I would like to take the opportunity to thank everyone involved in making this project a reality!
Here are some links related to BFG 9060:
Today I built a replica of the A3630 030 CPU card. The A3630 CPU card came with non CR Amiga 4000D models (the card works on the A4000T and on the A3000D/T also). The CPU card replica PCB is made by Chucky, read more about it here if you are interested in getting one yourself!
I got an (original Commodore made) A3630 in 2005, mainly for testing purposes. It was placed in storage for 15 years and I failed to check the capacitors on it. Sure enough the capacitors had leaked, the board had a fishy smell to it.
When I recapped the genuine C= A3630 board one solder pad lifted off the PCB (very common if capacitor have leaked for a long time). Sure, that is not a big issue, can be fixed quite easy with a bodge wire. But I thought rebuilding the A3630 it with a new ReA3630 PCB was a better option.
I decided to use the CPU and the KEL male edge connector from my current card instead of wasting a new mint in box unobtainium KEL connector from my stash on such a low end card. I removed the CPU from the card by using a hot air rework station. I used a desoldering gun to remove the KEL connector by desoldering each of the 200 solder points for it.
As you can see it is a quite simple card, especially if one do not mount the socket for the FPU. Since my main usage of this card is for testing purposes, the 33MHz FPU and crystal will be used on my ReA3000 instead.
I bought a 68060 revision 6 (71E41J) recently for a very good price. The rev. 6 CPU can usually do a really good overclock, often reaching 100MHz or more, which makes them very sought after and sometimes fakes are offered as the genuine version of the CPU. As I was already running an 060 in my Amiga 1200 I was not sure if it was worth swapping CPUs to gain more megaherts. The rev. 5 68060 I was running could do a 66MHz overclock but ran very hot. And a 50MHz 060 is not a slow CPU (in the world of Amiga).
I decided to swap CPUs because I wanted to test performance of an Amiga without a graphics card but with a fast CPU and high resolutions (like HighGFX and highres laced). Luckily removing the CPU was not a difficult task, prying on the corners of the socket it was out in less than 5 minutes ready to go into another one of my systems. Once the rev. 6 CPU was inserted into the Terrible Fire 1260 I was eager to do a test. Sure enough, it did not complain anything running at 100MHz!! (keep in mind I had a heatsink on the CPU when doing tests). Next time I will try 106MHz!!
Would I be able to detect any difference between a 50MHz 060 and a 100MHz 060 or would it only be noticeable in a benchmark program?
I was not too sure – When I had overclocked my 060 in my Amiga 4000TX from 50 to 100MHz I could feel that icons loaded faster, drawers just popped up, Workbench was just smoother and more responsive. But I was not sure if it was my imagination or reality.
With that said, I was surprised that my 100% non scientific mean of measuring the difference between a 50 and a 100MHz CPU in the following AGA Amiga Workbench screenmodes highres, highres laced, HighGFX @ 1024×768 (all screen modes in 64 colors) proved that there really was a difference!
Workbench had become even faster and more responsive. Running 3.1 with the stock Commodore 4 color icons, icons did not load anymore, they just popped up, almost all at once, drawers opened up instantly. The system speed by at such a high rate I had only witnessed something like this in WinUAE. What an amazing experience!
I changed background colors to 256 colors and could sense little to no slowdown from a 64 color screenmode. Sure, I have to test this more, but the early results are great, this is what base level Amiga should be like! Lets bring on 200MHz 060 (lets crowd fund it lol).
Before I permanently run this computer at 100MHz I have to figure out a good cooling solution for the CPU. This rev.6 68060 runs cooler than the old rev. 5 I had in it before. But overclocking it makes it run slightly hotter. I suspect its above 50 degrees but not more than 60. I have some ideas on how to cool it down but will have to sleep on the ideas a little more before I pull the trigger on the stuff I need to build one..
I have a Terrible Fire TF1260 in my Amiga 1200. I used to have an Apollo 1240 upgraded to 060 in it and was very happy with that card (despite what other said about the card, it was very stable for me). Time marches on though and the TF1260 started to look more and more impressive, so I built one last year. The TF1260 supports overclocking by software. One comment I read was that the chances of a good overclock improved if you put heatsinks on the two CPLD chips.
Keep in mind that I had a heatsink on the 060 when doing overclocking tests. But the space between the 060 and keyboard is so tight it was very difficult finding a heatsink in my stash that would fit with the keyboard installed. I have actually not solved this yet, thus I just focused on the CPLDs and will try to solve the 060 heatsink some other day, until then it will run at 50Mhz.
The problem with the CPLD cooler is that the right CPLD need a short heatsink if you want to run a compact flash adapter over it. As you can get an impressive speed boost for the CF card by using the IDE port on the TF1260 this was off course something I would need to consider.
I had ordered these 18x18x20mm heatsinks for this exact CPLD before. The left one is the original form and the right one is the heatsink cut down and modified. I dont like to do fabricating as it gets so messy and difficult to get good results with hand tools. I will probably order a new heatsink in the correct height in the future.
Here I am test fitting the heatsink.
And here it is secured to the chip. I use a thermal pad from AlphaCool that I cut to fit myself. The thermal pad from AlphaCool is somewhat thicker than what you usually get when you order heatsinks from Aliexpress. The thicknes is 0.5 mm, the thermal pad is also softer and more gooier than regular flat thermal pad. Once secured with the AlphaCool thermal tape, these pads wont go anywhere, they are secured very firmly. I highly recommend these AlphaCool thermal tape and would use them even on the 060 with a large heatsink without no problem or worry that the heatsink would fall off (even if mounted vertical!)
So here we are, everything fits now. But to use the IDE port on the TF1260 I need to burn a custom Kickstart with ehide.device, more on that when I will get to it.
And about finding a solution for cooling down the 060? I have not got a clue, I will probably need to cut down a thin copper plate and place a heatsink on it offsett from the CPU. Not ideal, but better than nothing. A small heatpipe cooler leading heat to fins over the top of the cards close to the floppy drive would be ideal, but prototyping a cooler like that would take weeks and lots of money. I will see what I will come up with.
I built a BFG9060 060 turbo card last year that I could clock to 100Mhz, this is the second BFG9060 I have built and it is in a cool red PCB color. Thanks to the Xilinx programmer I got I could finally program the CPLDs on it but did not have time testing it fully. Now that I have tested it I can finally confirm that it is working 100%.
Actually, I just ran Quake for an hour with it mounted in my primary Amiga, my A4000TX daily driver, and the card performed just fine. Once I have it permanently installed in one of my big box Amigas I will do a proper 24h POVRay fish render test. But as it is now, I am happy with how it performed!
I got this 68060 chip from Vesalia years ago (probably 12+ years ago), and for a really good price too. It was a deal too good to pass on even though I did not know what to do with it. Well I am happy that I kept it but for all these years I wondered three things:
The answer to those questions are:
While the CPU works, it is a little bit of a bummer that it is a rev 1 as that revision has bugs. But there are work arounds for that so it does not really matter. Unfortunately, as its rev 1, that means no 100Mhz overclock. But to be honest with you, I am glad just having another full (MMU/FPU) 060 on a big box Amiga turbo card.
I was afraid that the CPU I had was going to run too hot so I went into my stash of heatsinks to look for a suitable cooler for it. I found this thin 5v laptop cooler that I must have kept in my stash for more than 20 years. I got it from an old Acer Celeron laptop I found in the trash at a job I had and I was glad I finally found a use for it.
It is a thin radial fan that sits on an aluminium bottom plate that acts as a heatsink. I have not found modern alternatives in this size for a good price so I guess they have become rare these days.
I used thermal adhesive tape (also called “frag tape” BITD) to secure the heatsink to the CPU. Believe me, this thermal double sided adhesive tape has insane stickiness, there is no chance of the heatsink falling of the CPU if mounted vertically.
I can confirm, according to tests done with my finger, that the CPU does not run hot at all with the heatsink/fan on it. I am starting to wonder that maybe it was overkill, but if I end up using it in an A3000D case it might come in handy as that case has poor cooling performance.
I built my second Amiga 2000 EATX last year, it is a beautiful clone of the Amiga 2000 remade into a standard extended ATX format so it can fit a standard EATX case. You can be sure that I will add an article about that motherboard here in the future. The author of that project has also made the n2630 project, its a clone/reimplementation of Commodores 030 card for the A2000, the Commodore A2630.
To be honest, I wish there was a modern 060 or 040 card we could build for the A2000 since I think 030 is just too slow these days – at least for my modern Amiga standards. To be fair while the A2630 was 25 Mhz this one runs at 50 Mhz, so its a little bit faster. 040 runs so hot, but would be a perfect fit for the A2000EATX because of the improved case airflow (and having an 040 in one of my machines could be interesting to compare performance with 060 machines).
Anyways, I should be glad with what I have and I am thankful that we can build this cool modern 030 card for our A2000 computers. And the n2630 is such a nice compliment to the A2000 EATX motherboard having IDE/CF slot built in.
I decided to build my n2630 card earlier this year with 128 MB fast memory. There was an option to go for 256 MB fast memory also but thought it was not really worth it, maybe I will regret it in the future. Otherwise this build was very straight forward. It uses three CPLDs and six memory chips.
There was some parts that was difficult to find, first it was the 030 CPU socket. I actually got the parts number for one that was availble on Mouser after asking around, surprised about that. The CF slot at the back of the card needs to be the right one to fit solder points on the board, ask on the Discord if you wondering what type you should get if you are building one. Other than that it was an easy build. There are also two eproms you need to have and program, but as usual, most stuff that is rare can be found with some patience.
I have since mounted an FPU to the card and I also got a 3D printed bracket so I could secure it in my regular ATX tower that was converted to EATX. I ran out of black pin stripes that is why I used yellow and green (it was also a joke about obsessing over the color syncing of hardware).