Building, testing and hitting problems with a CHEAP Amiga 600 ACCELERATOR

[@CRG]

Oh dear, I've only just noticed how loud and annoying the game music is during Frontier and Monkey Island. Apologies for that. I've been looking at the original board again and have made some progress, fairly sure I know the cause of the problem now and I hope to have a 2nd channel video done for it in the next couple of days.

[@Voyager_2]

Really appreciate your tenacity. Looking forward for the diagnose on the second channel. Well done!

[@unfies]

19:25 I woulda suggested fine grit sand paper on glass / mirror for this.

[@8antipode9]

Good job on the soldering, I definitely feel your pain. I don't solder SMD that much, but I think the last thing I did was a PiStorm and it took me a long time to get everything soldered completely.

[@CRG]

Thanks, I'm usually OK at smd soldering but this board I just struggled with. I've been toying with the solder paste and it's maybe a little easier to working with. It'll be covered in the next video.

[@amurtigress_mobile365]

@@CRG Maybe it helps to pre-tin pins of SMD ICs when conditions are as stupid as with the CPU there. This has helped me on more than one occasion overall...I realize tho that with the extra layer of tin on the pins the chip may not be perfectly flat on the pads below.

[@davepauljones]

Use a bamboo skewer to brush alone the IC pins, to check if they are soldered correctly, doesn't leave any marks on the pins, and the bamboo flex's enough not to damage the pins !

[@jcanodk]

Just an idea... The cap under the crystal might be shorted due to the crystal case touching both sides of the top of the cap.

[@minombredepila1580]

I guess some of us were shouting at the screen at that very moment

[@FranktheSquirell]

i remember when i bought a 600, fella in the shop was amazed i didn't want the 1200 lol, i only wanted it for word processing so it did the job. another fine video there too 👍👍

[@Jamesfo8rf]

Thanks for the video. It’s a really interesting project. Hopefully these will be available to buy in kits.

[@RetroSegaDev]

Patience of an absolute saint with that soldering!

[@otis7359]

You could bend the pins inwards of that m68k before soldering. It will give you some necessary spacing. Just press the chip on all four sides against e. g sharp edge of a table. (Just vertical part, not the tips of pins).

[@CRG]

Good idea, I'll try that next time I've got the same problem.

[@g4z-kb7ct]

@@CRG Yeah it's a technique I use a lot. One step to do at the end of the bending is sit the chip on something that is perfectly flat and push down on the chip. This will align all the legs so they are all in-line. If you don't do that you'll have a _really_ hard time soldering the chip because the legs will be all over the place. When manufactured, the legs are all flat and in-line within 0.1mm as per the specification but bending them by hand messes that up so you need to re-align it by pushing down on the chip while it sits on something flat. You may need to repeat the bending and pushing until you get enough of a gap between the chip leg and pad end.

[@Boris_Amiga1200]

Great work Glen! Respect!

[@wskinnyodden]

Hmmm, did you insulate the bottom of that crystal before slapping it on top of that SMD Cap with the metal slapped right on both cap pins?

[@CRG]

Nope and you are onto something. Although it's not the only issue. Will all be covered in the 2nd video.

[@ChrisEdwardsRestoration]

Lets not forget about the A630

[@amigaretro1985]

I bought the latest version that has a 40 MHZ 68882 FPU, works nicely and is relatively quiet. I also have the non-FPU version that was sold previously. You can never have too many Amiga accelerators :)

[@CRG]

Silly me, apologies

[@morantaylor]

Its a hell of a lot more expensive.

[@amigaretro1985]

@@morantaylor It is, but gives you 64 MB of RAM and an 68030 with an additional FPU. I built an similar project a while back for the Amiga 500 that clocked the 68000 to 14 MHz and added 1 MB of Ranger Mem. Like mentioned by CRG in this video, it really isn't enough Memory to help much with Games or Applications that need it.

[@ChrisEdwardsRestoration]

@@morantaylor 220 isnt expensive

[@g4z-kb7ct]

Another way to make the socket shorter is to rub it on a sheet of fine sandpaper. This will ensure the ssurface is nice and flat and you can make very fine adjustments more easily.

[@CRG]

Yeah seen a few others recommend that process too. Will keep it in mind for next time.

[@g4z-kb7ct]

It's strange why the PCB designer didn't just take the german design open-source 68EC020 accelerator made for the A500\500+ and re-work it to use the square CPU. That's a really amazing accelerator for the A500 with 16MB true fast RAM and a very fast IDE controller built in.

[@CRG]

Sounds cool but would great increase the complexity and cost of the board. The designers aim here was to make it as simple as possible and granted while there are a few minor issues in his design, I think he did a great job.

[@ste76539]

You'd be better using fine sandpaper, taped onto a flat surface to modify that socket top edge - much more control and the result would be much neater and cleaner with less chance of borking it.

[@CRG]

Thanks for the tip, I'll be sure to try that next time.

[@Voyager_2]

@@CRG A 30x30 mirror tile with some 400 grid sandpaper is indeed the way to go!

[@dazzathedazza]

I might be wrong on this one, but I thought that ranger/slow ram, although only accesible to the CPU only ran at the same speed as the amiga's chipset as it still has to be accessed via agnus, hence only "real" fast memory (accessed directly by the CPU) accessed at the CPUs native clock speed. That might explain why you're not seeing the speed up you'd expect from things like frontier.

[@CRG]

The ranger ram on this card sits local to the cpu. It's added as slow ram but is accessed only by the cpu at its clock speed. That's why we get the speed increase in sysinfo.

[@patzik1910]

I’ve build fastram expansions for the a600 that use the same kind of socket method. There a pain… some brands work better than others. Best way I found to sand them down without damaging the pins, is just to remove all pins from the socket, then sand them down which allows you to sand them just that tiny bit further and then re-install the pins in the socket. A bit of work, but it works better, I suspect connectivity is your issue (next to that cap underneath the crystal).

[@g4z-kb7ct]

4:20 Use magnification and simply hold the part with the point of some tweezers and the chip won't move. With the same hand that is holding the tweezers also hold some solder. Touch the iron to the solder to pick up some of it on the tip and solder the corner of the chip. The added flux on the chip will make the solder flow. Check the position and adjust if necessary then solder the opposite corner. It isn't that difficult with the correct technique ;-)

[@kilosierraalpha]

My money's on a bridged pin or a disconnected pin. Another possibility is that xtal is shorting that cap that it sits on top of. PCB manufacturing fault is not something I think is likely.

[@CRG]

I have since got it working and your partly correct. It's not a PCB error anyway.

[@RetroRecollections]

A nice bit of kit for the price!

[@miked4377]

nice job glenn......i dont really get why that other board isnt working. .other than its bad....that sodering was super tough..like you said try the components on a new board....still agood job given the circumstances...

[@drramtop1576]

That PCB just screams the designer is inexperienced. Long pads are critical to hand soldering QFP and TSOP packages. I put very long pads on my PCBs because I build hundreds of them per year and can't afford to faff about chasing soldering issues. Second, that cap under the crystal is terrible design. It'll short on the crystal's case, which is grounded. It looks like the designer tried to cram everything into the smallest possible board, which isn't really necessary here. Also, the tantalums are just smoothing caps for the voltage regulator. Some regulator manufacturers recommend tantalum caps, some don't, but ceramics will work just fine in any case.

[@charlesdorval394]

I was going to mention the capacitors. Some datasheets have values for both, some manufacturers datasheet are better than others for the same parts

[@CRG]

I'd have to agree as the footprints for most parts made it extremely difficult to solder, especially the cpu. I've since moved the cap under the crystal to the bottom as it was causing an issue but although not the only issue I've found.

[@minombredepila1580]

I do not have a 600 to check with, but I wonder if the holes on the accelerator matches the two "possible holes" in the board, meaning screw and bolt could help to keep it tight, avoiding any dodgy connection issues.

[@Breakfast_of_Champions]

68k boards notoriously lack decoupling caps, maybe your three expansion boards are just overloading the little thing. The CF card could also be introducing just enough parasitic capacitance to bring the whole house of cards down. And that socket mounting is wild, every little bump can make it jump off the CPU.

[@CRG]

You're absolutely right and (spoiler alert), but that's essentially my conclusion in what will be the second part to this. I've tried adding more capacitance but it hasn't fixed the issue, although granted just throwing a few more caps at it isn't exactly the most scientific approach.

[@kathrynevans6849]

I had similar problems when daisy chaining multiple interfaces on the 48k ZX Spectrum, back in the day. Individually, the processor could handle the extra load added by a single interface, but when using multiple addons (Opus Discovery, Multiface 1, homebrewed sound sampler, and more), put together the extra load and capacitance on the address and datalines meant the CPU couldn't switch them fast enough to remain in spec. A temporary solution was to fit a faster CPU that could drive the lines harder. I wonder if the failure when adding the CF card is related to bus driving issues? And of course that cap under the xtal osc is a design flaw. Shorty shorty.

[@zombunzedb960]

man i love the monkey island soundtrack one of the best tracks ever made

[@jeffknopp9576]

You mentioned that you changed the socket several times. Is it possible that you may have damaged the pcb (especially the inner layer connections of the socket) of the 4 layer PCB? The only way to find out for sure is to buzz all the socket connections per the schematic. A tedious task.

[@CRG]

I thought that myself but in haven't. I do have the first board now working. It'll be covered in a 2nd channel video later this week.

[@stephenwhite506]

MC68SEC000 can overclock quite well, however it comes in 8, 12, 16 and 20MHz. They can often overclock to 3x.

[@mattsqwrl]

That speed on Frontier Elite did seem very slow, maybe slower than my stock A1200. When I tried it with a 28mhz 68030 with 8mb fast ram, it ran pretty smooth, maybe 30+ fps.

[@g4z-kb7ct]

Yeah fast RAM is the key, he didn't have enough. The designer should have added 8MB and made it all fast RAM. Back to the drawing board for version 2.0 lol!

[@jamespalmer5960]

My original thought would of been after all the usual checks, would the cpu lanes be attached to IDE and timing. I have an a600 8mb fast, 1.5mb slow card, 1 mb chip expansion running rom 3.1, WB 3,1 1997 version. It runs a grate no issues running as a standard 68000 gets 7 Drystones more than a stock machine so a miniscule noticeable improvement. Have been wanting to upgrade to 020 but the prices are so high I might as well buy a a1200 and get that Alice aga as well in the price. Even though the a600 was a ECS machine there was not limited software that used that mode OCS and AGA was the most common.

[@borayurt66]

I really like this thing, although I hate the method of upside-down socket to tap into the CPU bus and the idea of using a CPU at 2x overclock, the versatility of having the Kickstart (or any other ROM) on RAM is intriguing enough to try one... Only if I had an A600... Is there a A500+ version of this board?

[@g4z-kb7ct]

There are MUCH better open-source A500 / A500+ accelerators around. I have one with 68020 and 16MB true fastRAM with built in IDE controller that I built myself. Works incredibly well and super compatible with everything.

[@borayurt66]

@g4z-kb7ct What is it called?

[@g4z-kb7ct]

@@borayurt66 I replied, YT insta-deleted it. Sorry, YT doesn't want you to know.

[@borayurt66]

@@g4z-kb7ct Don't put a link, just write what it is called, I'll find it with Google

[@Neruomir]

Miss-aligned pins or a short is my bet as that is what is usually the culprit when it comes to manually installed surface mounted components. That or inferior Aliexpress scam/knockoff/QC reject components or a borked PCB with iffy trance tolerances.

[@beatadalhagen]

Looking at the CPU pins vs pads I can't help thinking the footprint was meant for a socket, though at the wrong scale.

[@endotherm]

You may be right, I believe they solder on on the inside. Attaching either that CPU or a socket to the board would best be done by applying solder to the pads and building it up a bit first, then placing the component/socket on it, then heating it with hot air (or a special hot plate) until it melts. The pins will tin themselves, center on the pads and attach to the molten solder pool creating a proper electrical connection.

[@g4z-kb7ct]

No that CPU must solder to the board as a socket is not available. This is a common amateur PCB designer mistake. He simply made the board too small and thus there was not enough space for the chip so he had to squeeze the pads smaller to make it fit. The proper way to do it is to simply make the PCB 2mm larger. There are no space restrictions in the A600 and a board 2mm larger won't make any difference to operation but will allow the CPU pads to be larger.

[@markramsell454]

Your accl card and IDE interface all use the address bus like every computer invented. If the accl card is holding a bit low, and it's an upper bit things like some devices won't show up, weird software errors will occur, and you could see half your memory device. Look for a stuck address bit in the upper half of the bus.

[@TzOk]

Haven't you shorted the cap underneath the TCXO by its metal cover?

[@kathrynevans6849]

I did wonder myself, why he didn't put kapton tape on the crystal oscillator

[@g4z-kb7ct]

6:00 That's a design fault. The board designer should have made the pads longer. My guess is he either used a footprint for some other chip that is not the correct one because he was too lazy to make a custom footprint (there definitely won't be a 68000 QFP footprint for that chip in Kicad) or didn't actually check it properly, or he may have used solder paste and a reflow oven which would work fine with short pads. Edit: Another possibility is the space available is very tight and he made the footprint as small as possible. The correct way to solve that issue is to make the footprint larger and then use alternative routing (not run the trace near the edge of the board that forces the pads to be short) or make the board 1-2mm larger. The board can easily be made larger there is no reason to make it that small and a couple of millimeters larger won't hurt but will allow for more space to make the footprint larger. This is your typical amateur PCB designer mistake, seen it hundreds of times before. I'm usually taking these types of projects and improving them and re-releasing it and have been doing that for many years. Maybe I'll get hold of a PCB, reverse and fix it ;-)

[@CRG]

Ive been chatting to one of my KZbin friends about this board and he is going to undertake a redesign on his channel.

[@TheSudsy]

Did u get a solder hair on the cpu etc?

[@TzOk]

MLCC would be fine as decoupling capacitors. Sometimes I replace old tantalum caps with MLCCs.

[@fra4455]

Great video

[@Adrian_Finn]

And you're sure that the crystal oscillator is not shorting the ceramic cap that's sat underneath it?

[@endotherm]

I would have modified the CPU socket by putting a sheet of sandpaper on a flat surface and rubbing the plastic away slowly until you just touch the pins. It will ensure a flat finish and remove the same amount of plastic all round at the same time. This will permit the best contact while not straining the socket, which aren't the strongest at the best of times. I'd probably look at starting with 120 grit, anything coarser (smaller number = coarser grit) will be too aggressive, anything finer might be really slow. I'd really give your soldering a deep inspection to really look for a defect -- a tiny blob or bridge which is nearly invisible can cause issues like that. Otherwise, try with a 20MHz crystal and go up by 5 or 10 from there, just to ensure that particular CPU can handle it. Even though they are marked 20MHz, you might find most are capable of overclocking to 40. The manufacturer, during testing, obviously started detecting problems with SOME chips in that batch at anything above 20, and thus labelled the whole batch at 20 MHz. You might be lucky with some, but others won't handle overclocking nicely.

[@g4z-kb7ct]

You can't use a 20MHz oscillator. The CPLD is programmed to use specific timing and requires the specific clock speed it is programmed for. He said there is a CPLD code for 30MHz but he didn't have the oscillator and it didn't even boot at all with 40MHz oscillator. Using any random oscillator is not possible unless the board designer also makes the matching CPLD file for it.

[@fu1r4]

38:03 I vote for human error and not manufacturer defect ... Even if you move all the components over to another board it could still be a human error on the first board.

[@crumplezone1]

I bought 20 of those CPLDs from aliexpress and they were fakes so it really is hit and miss now with those chips, but great you got one working bud, also the furia with the new firmware is a great choice or even an A630 03/50 acc is available too

[@CRG]

I've built the 5 if them now and found two bad ICs. Not the CPLDs though, they are all fine. It really can be a lottery ordering from ali but it does save a lot of money. I think i got the 5 ram chips for the price of one on mouser. Of course with mouser at least you know it'll work.

[@paulladdie1026]

I have a Furia in my A600

[@noth606]

Not to be a knob or anything, but why did you not pretin the IC pads a bit instead of wiping them with the braid? The way I do IC's like this is to pretin the pads so they have a slight layer of tin on them, place the chip, tack in the corner legs to the tin already there, then tin up the iron tip and "bead roll" the legs. By bead roll I mean use the tip of the iron "wet" and add tin as you slowly go from leg to leg to leg. You can do it with a bead of tin if you feel lucky, the name comes from that but it's a PITA to roll an actual bead the whole time, you get the same effect if you keep the tip "wet" so it has tin on it all the time. It's also fairly quick to do it this way, you're done in 5min tops if you roll on through it. If you get a short/bridge disregard it and roll on down the side of the chip, then clean the bridges up with a wick afterwards. I used to do a lot of assembly and rework, also with hot air, and had time to mess around and try different methods, this is what was the easiest and cleanest for me. What you get is the legs, or the "foot" of it is coated with tin and so is the pad and it seems to be the best "bond" like that. Also liberal application of flux is recommended hehe. - just to be clear, cleaning the pads isn't a bad idea, but I think leaving them "dry" is, in particular if you have to take a break and so on, you can get oxidation or weird stuff make the pad not want to wet with tin later. Not sure exactly what does it but it's happened to me several times if I cleaned a board and left it "dry" for some time.

[@g4z-kb7ct]

That's the 'bad' way to do it. I'm guessing you have been watching Sorin lol! The way he did it is correct. Remove all solder and sit the chip _flat_ onto the board. Having the chip sit on a mountain of solder and hoping the legs of the chip will drop down is a wishful thinking. Doing it the correct way is easier and faster too and when done correctly, works first time.

[@TonyHamlyn]

What about solder paste and a hot air gun, is that a possibility?

[@g4z-kb7ct]

@@TonyHamlyn Same thing. Solder paste is simply tiny balls of solder and flux mixed together.

[@TonyHamlyn]

@g4z-kb7ct but I mean couldn't you apply paste to the pcb, drop the ic into it, and heat it with a heat gun, wouldn't it solder the pins more reliably that way, the wicking as the flux helps to flow the embedded solder and maybe it flows enough to stop pin to pin shorts? I have just started surface mount stuff myself, and bought some paste to try this method out (previously doing it with similar strategy as this vid, by soldering 2 corners and fluxing the whole thing up well and working with the iron, which, as we saw in this vid, doesn't work so well when you can't get any heat onto the pcb pads).

[@g4z-kb7ct]

@@TonyHamlyn Yes you can use solder paste and heatgun if you want and it will work fine. Be sure to tap parts slightly and watch them bounce back into the correct position to know that the chip is correctly reflowed. Note solder paste only works for smaller chips or chips with coarse pitch pins (maybe 1mm minimum). If you try to reflow using solder paste and hot air on a large chip with 208 pins and with 0.5mm pin spacing you will not succeed. The correct way to solder those is by hand as I described in my first comment.

[@scramble45]

could have just use a hot plate and solder paste

[@tomekrv942]

When I bought my A600 I also wanted accelerator but later I started to think do I really need it because graphic chip will always be the same like in old A500 so this will not change my A600 into A1200.

[@TzOk]

Get rid of this clear "cup" from your microscope (it is only a friction mount, just pull it down with a slight twist). It is useless in this application. I believe it has been copied from a handheld USB microscope, which you could put directly onto the observed object (like a coin or stamp). When a microscope is in a stand, it only limits the space and disturbs the light.

[@CRG]

Ahhh, i didn't know ow that comes off but it'll help greatly if it does. What i really need to do is stop being lazy and capture the image from it into OBS so I can show you guys clearer. Will try to do better next time.

[@monsjakob]

Can someone help me find the cpld and ram - I have looked on Ali and lcsc but can’t find them

[@CRG]

These are the cplds I bought a.aliexpress.com/_Ew4j2b As for the ram, the ICs i got were the cause of my problems so maybe find another supplier. If you Google "CY62167 aliexpress" you'll get a few sellers.

[@firsteerr]

get yourself practicing with the paste i use a directed hot air gun and it works a treat much much easier

[@g4z-kb7ct]

Hot-airing a cpu will dramatically shorten its life. Don't copy Sorin's poor techniques lol!

[@Bushuu]

have you got the link to the github?

[@CRG]

Whoops forgot to put it in the description. github.com/PR77/A600_ACCEL_RAM

[@레트로메니아-r9o]

cpu mc68ec000 type problem not pu type usefull

[@wimwiddershins]

I thought your microscope brand was Doom.

[@stephenwhite506]

The PCB layout is quite poor. Even though they went to the expense of using a four layer stack up, they then create large via dams in the ground plane that could have easily been avoided. The cap under the crystal could have easily been placed on the underside. They even placed a ground via next to the crystal's 5V pin! The RAM is a 45ns part, this is almost at the limit for 40MHz.

[@DaveVelociraptor]

20 minutes into a 40 minute video and Glen says let's power it on and see if it works.... 😢 I'm quite enjoying the whodunnit aspect of this though

[@CRG]

Things rarely ever go as planned

[@abx42]

I'm guessing a DOA part.

[@grommdk]

but why only 1½ megs of RAM? - it need to be atleast 8. Its seams a bit odd to me..

[@CRG]

I think this was done to simplify the logic needed. The 1.5mb sits in the ranger ram space, that address range can only accommodate 1.5mb. I do agree though, 8mb of Z2 fast would have made it a lot more usable. Would increase the complexity and cost though.

[@stephenwhite506]

They are using a 2MB static RAM to sit in the 1.5MB ranger RAM spot in the memory map. It can't be 2MB because the custom chip registers are located in the last 0.5MB. So they use the unmapped space in the RAM chip to store a mapped version of the ROM. RAM in this area does not need to be auto configured as the kickstart ROM will check for it. If they were to add an additional 2MB SRAM it would need to be mapped into Zorro2 memory space and require auto config logic. If they were to use an 8MB DRAM memory chip to fill up the Zorro2 space they would also need extra logic to keep it refreshing.

[@muzicmolehead3943]

Hi crg molehead here trided a old hard disk than a sd disk maybe that may work better all the best muzicmolehead

[@g4z-kb7ct]

20:54 One thing that is curious is the RAM should be fast RAM. Slow RAM is connected to the chipset and fast RAM is connected to the CPU. That accelerator should have the RAM connected to the CPU so why is it configured as slow RAM????? I think the board designer needs to go back and do a bit more thinking lol

[@CRG]

The designer made it like this on purpose to keep the project simple. Going to fast ram increases the complexity of the logic required.

[@salmon85]

nice

[@BrianLamtman]

I will take it..

[@infopackrat]

It needs a new version with more fastram

[@TzOk]

There is a very similar accelerator, but with 11MB of true FastRAM - it is called xT and sold by Archi-TECH.

[@SanguineBrah]

There's no mystery here. It's got to be dodgy Ali parts

[@Cybernetic_Systems]

Good job, but the easy way to solder a QFP chip on such a tight footprint is to tin all the pads, and then use hot air to solder it. It’ll even self align for you.

[@Cybernetic_Systems]

@ rubbish, if you heat up the board first, use the right size square tip on the hot air station, and don’t go above 300c, it’s quick and safe. I haven’t been watching anyone, I was an electronics tech back in the 90’s. This is not new technology.

[@Cybernetic_Systems]

@ LOL, I’ve been soldering since the early 90’s and studied electronics at college, I know what in doing.

[@CRG]

I'll have to be honest and say I've never had much luck soldering like that. But I'm just self-taught, so no doubt am "doing it wrong" to those who have a formal education in it. I know i struggled a bit with this one but for the most part I can get on ok and enjoy hand soldering as part of the hobby. In the 2nd channel video related to this board, I had a go at solder paste and it's maybe a little easier to work with. Not sure if I'll use it long term though cause as I say I enjoy my hand soldering.

[@Cybernetic_Systems]

@@CRG I’d never accuse anyone of doing it wrong when it works, I’m just advising there is a simpler method. I don’t recommend using solder paste unless you have a PCB stencil to control the application precisely. Then it works really with a reflow oven or hot plate. Hot air will work too, but it’s not as efficient at heating the entire PCB evenly. Btw, I still hand solder one offs cause it’s fun, but only when the footprint is big enough to do so. When you get to QFN’s it’s just not worth the effort.

[@g4z-kb7ct]

You should consider getting better quality sockets rather than the cheap chinese crap. On the Individual Computers ACA620 or ACA630, once it is pushed onto the CPU it is *extremely* difficult to remove.

[@CRG]

In fairness to them the sockets have been fine. The issue though is down to other Chinese sourced parts.

[@craiggilchrist4223]

What a head screwer.

[@DonnyHooterHoot]

Amiga? Dead garbage! Bury it, it stinks.

[@CRG]

sniff... no idea what you're talking about... smells fine to me.

[@DonnyHooterHoot]

@@CRG I still have my original Amiga 1000 motherboard.

[@CRG]

Maybe dig it up then before the worms get to it 😂

[@DonnyHooterHoot]

@@CRG It's not buried, it's on my man-cave wall, along with my Amiga 2500 motherboard. Properly buried and DEAD!!!!!

[@CRG]

Sounds like you once had a nice Amiga collection. If you still by chance have the rest of the machines maybe assemble and sell them. Even in a dead state they are worth a good few hundred. But there's also nothing that can't be repaired 😉