Amiga 4000 Repair Part 2: RAM issues are solved!

[@SyphistPrime]

Someone finally did it on KZbin. They actually changed a RAM stick's capacity manually. I've been waiting to see this done for a long time. Thank you for showing it can actually be done.

[@georgechambers3197]

The reason for 16 Mb was I used it for graphics work. That's why it had a Toaster and the other cards. I'd like to see some of that stuff running again too. I was never a game player and never used any of my Amigas as game machines. 😁 Thank's for getting it running again it's really good to see!

[@johnpernia5777]

Thank you for sharing with the world George! It was great to be able to see Adrian work his magic and produce such great videos with it.

[@chrisharvie-smith486]

Couldn't Adrian have just left the 8 MB modules alone and used sim slots 1 & 3 for 16 MB ?

[@only257]

George Chambers 🥳cool

[@JCMayPE]

I had an 2000 with a PP&S 040 accelerator and 12 MB of RAM, a lot for a college kid at $40/MB. I was doing a lot of DTP. PageStream was my jam.

[@FryNOR]

My father taught me that you should count slowly to eleven when turning off electronics before you turn it on again. Great video, thank you!.

[@bryan2garcia]

Glad to know I was not the only one

[@MrWaalkman]

Is that 11 in decimal, or binary?

[@OzRetrocomp]

@@MrWaalkman there are 10 types of people: those who understand binary and those who don't. :D

[@gertsy2000]

Yep agree. He should do all the tests again but this time wait 10 secs after turning it off every time. I think you will find it's reliable. @14:05 Yep.. Your older RAM is still active when you turn it back on so quickly. Well done in sorting it out thougg. And modding the 8MB to 4MB Simms. Looks great.

[@00Klingon]

MrWaalkman There are 10 types of people: Those who understand binary and those who don’t.

[@Mclaneinc]

Superb work as always Adrian, the 4000 is a strangle old beast, my friend used it for MIDI stuff, he decided to get a SCSI drive for it which in those days was horrendously expensive. It arrived, he picked up up and dropped it on to a concrete floor by accident. He was literally in tears....Not only was it dead but it had a mark so he could not send it back as DOA. A lot of money later he was working again and with out a long story showed Jean Michel Jarre the software he used, Bars and Pipes which Jarre had never seen as he was using an ST (I think). True story.. (JMJ was here for the Docklands gig and his friend was the councilor in charge of overseeing the gig, knew Kevin was a fan and brought him around to his house, short version of the story)

[@thereallantesh]

It amazes me how finicky the RAM issue was. I was super impressed with how you converted the 8 megabyte simms to 4. You should print some custom labels to put on them indicating what they are now. It may help the next guy 25 or 30 years later when he is working on it.

[@TechToTunes]

It should still make sense. Using the chip codes and the solder points that determine the size, it should be enough to figure it out.

[@RJiiFin]

@@TechToTunesBut why leave that as an exercise for the next guy instead of, as was suggested, printing some labels that would immediately tell what's been done?

[@bearmatic]

Thorougly impressed by your knowledge of chips and their functions, and even more by your persistence, logic deconstruction and fault-finding progress of the RAM issue. Phew!

[@NJRoadfan]

A lot of pre-Pentium era machines have compatibility problems with 60ns EDO 72-pin SIMMs. You need to find a good batch of standard 70ns FPM 72-pin SIMMs to have on hand when working with these old machines (386, 486, Macs, Amigas).

[@twocvbloke]

Hmm, I never knew that about earlier types of RAM having those solder points to configure them, learning something new every day... :D

[@RavenWolfRetroTech]

Adrian, you are testing my resolve to work my way through the computers I had in chronological order. The C=64 is done but I'm not sure I can wait to get a 128 up and running before doing an Amiga. The BEST computer I ever owned! A great video series that is bringing back so many memories!

[@rCRTEr]

I remember sitting in front of my Amiga 4000/030 what a wonderful time that was! I ran my own BBS on it for many years!

[@Shifter-1040ST]

Even I cheered with you when that A4000 finally worked well. And I'm an Atari ST guy. ;-)

[@user-wj9xq7ig2v]

Those Atari engineers did a great job on the Amiga

[@BAZFANSHOTHITSClassicTunes]

I had the A1200 from 1993 to 1996. Awesome computer, awesome times.

[@gremblobeans5004]

i know this is an old video and i doubt you'll see this but man you do some killer work, keep it up!

[@andybasstbn]

My six-year-old son wants to find an old Mac to recap thanks to you!

[@HoboVibingToMusic]

I'd not let your son touch a machintosh due to the CRT. Maybe an apple 2 series of PCs, but a machintosh is a no no for a young kid. :|

[@vitonildo]

@@HoboVibingToMusic The kids wants to. It doesn't mean he found an Old Mac and it is already opened in front of the child with a charged CRT. Cool down.

[@TheoPantazi]

That's awesome, but please don't let him start a company call Skynet later! 😂

[@Rv-Tech]

Then He should put the right mask, to avoid breathing the Flux fumes that are not good at all for childs

[@kaczan3]

I'm so sorry that your son became an Apple fanboy.

[@Schooner316]

Big thumbs up for modding memory! I have never heard of anyone doing that before. Resourceful is an understatement! 👍👏

[@TheHandOfFear]

Thanks for recapping the re-capping.

[@John-uc6gb]

I know this an older video. I'm trying to catch up with missed episodes. This is still amazing. Great video. Thank you.

[@sniperviperman6400]

awesome repair and that nostalgia game: another world at the end makes it the cherry on the cake.

[@johnpernia5777]

Great memories!

[@wojiaobill]

That Amiga interface looks so slick. Wish I had one of these growing up.

[@dcfuksurmom]

You can very easily recreate this interface on Linux, there are plenty of simple desktop environments and window managers out there. Wouldn't mind seeing a modern take on that UI, I might attempt to duplicate it.

[@a4000t]

If you remove the cross brace holding the slotboard,you will find a notch in the slotboard to run the cable thru. You can also run the cable thru the gap at the back where u ran power,theres just enough room. 4M low profile simms can be had off ebay for about $2ea or less. I find it a good idea to tin the fresh,cleaned traces that were previously corroded with solder,it can sometimes bridge a microscopic crack and add a layer of protection in any case before coating it.Did i hear you say EGA? LOL

[@barthonhoff5547]

a rule of thumbs in the 80’s was: Wait 20 seconds before switching the computer back on. Never had any problems

[@JasonHalversonjaydog]

yeah i was taught the same thing, not to flip it off and immediately back on again. give it a few seconds

[@6581punk]

@@JasonHalversonjaydog Gives the charge in the RAM time to discharge.

[@adriansdigitalbasement]

This only Applied to some computers like the Apple II where it looked for a specific piece of memory to bet set to know if you were doing a soft-reset or a hard reset. If you power cycled too quickly it thought you wanted a soft-reset and that would create unpredictable results. (See my Apple II repair video.) This issue doesn't affect computers like the Amiga that don't even have a soft-reset capability. (soft reset preserves memory contents.)

[@tw0469]

Yeah, I agree. I mean, the ram get's initialized after Power cycling anyways, so I doesn't matter if data is still present or not.

[@timrb]

I read this somewhere when setting up my A500, you should wait 20-30 seconds before turning back on. I also get this error when I turn off and on too quickly. It might be in the manual.

[@tostumpen]

A very nice and informative video. Love the old computers both PC's and Amiga alike! And what a beautiful track the "Synthwave 4" was! I really hit home. And no wonder;it's produced with three of the most legendary electronic instruments of all time: Juno 106, MiniMoog and TR-909. A true experience to listen to!

[@curtiswillie3249]

I really enjoyed seeing the first two parts on restoring an Amiga 4000. It brought back memories of my experiences with the Amiga 500 and 2000 that I had. Back in the day, we used the 500 to run a BBS service. It had an expansion slot and we added memory and a hard drive via an expansion box. My 2000 was configured to video work with a digiview connector like the one that Stewart sent you in the Apple ][+ series. It also had a camera and color filter. I had software to split and combine the 4 different images. I also had a NTSC video card and an HDTV device that allowed me to output real HD. I look forward to seeing the rest of the restoration. There was also a 2000 at work that we used as a video titler. It had removable hard drive platters (A syquest) that was horrible as the unit was frequently used, and they did not stand up to frequent moves.

[@louisnarcisi8691]

The really strange thing is, I have the exact 4000. With the exact grey slots. That does the same exact things ... As I speak right now, I'm about to go through it again. I've ordered new ram all around. The difference is I had 3.1.4 roms in it. For the hell of it, I've ordered some regular 3.1's like you have, which are here. Plus I have the gotek and the software you have to test the ram. It's all been professionally recapped. Some of the other Amigas, they do not have those exact grey slots. When I install all the ram, both sets even the new set, I get 2 chip, 4 fast. One of the new fast ram chips turns the screen yellow. However, when I do not use that ram, I do not get a yellow screen. Sometimes green flash, but never hangs. You had mentioned the chip ram was 8 I believe? I watched this so many times, I am going by memory now. Mine is actually 2 meg. Curiously, that may be why you are getting the results you are? Because the 2 meg only addresses 1 ... However, when they are all put in the slots. It measures 2 meg chip, 4 meg fast ... I need to think about this more. In case I can't get it. I've ordered a bigramplus card from individual computers. My 4000 is perfect, I hate to waste it. Much doesn't run with just that, 4 fast. 8 as you said is plenty. Also, I need to look at the ram charts. But I received the new ram from someone adept in Amiga stuff. Any input would help, great video!

[@drright71]

You can swap the oscillator on the A3640 CPU card for a few more MHz. I had my A4k up 36.3MHz. Your mileage may vary depending on the CPU.

[@noth606]

Stumbled across this, and the nostalgia is off the charts for me. Used to have an A4000/040 setup a long time ago, with Picasso graphics and a high res multimode monitor. Sometimes I think of getting one again but it goes away as soon as I focus more on the idea, I have enough other retro gear with no practical use as it is, last thing I'd need is more of it.

[@pauledwards2817]

Thank you for a fabulous video. I feel so grateful and dumb after watching this video. I have boxes and boxes of 72 Simms and spent many an hour going through with a magnifying glass looking at chips to identify them to work out the size of the Simm. Doh, so obvious that they could have a size sense looking back. I even have some with dip switches to change it. Once again thank you.

[@davefarquhar8230]

Glad to see you got your 4000 back up and running. That superfast boot time off the compact flash sure makes me wonder what might have been, if the Amiga had continued development. I had a 2000 in the early 90s and loved that machine.

[@plixplux]

Watching a true master in action. Nice!

[@NoobixCube]

If you check out Retro Man Cave’s recent Amiga restoration, he put a modern power supply inside the case from the old one. Big advantage in stability of the output, and not having to recap it. Also has the hidden bonus of the new power supply being 100-240V tolerant. See Perifractic for what happens when you mix up which voltage the power supply wants...

[@UpLateGeek]

Very interesting stuff with the RAM chips. I've been staring at 4M x 4 DRAM chip datasheets for months, and they all have the same SOJ-24/26 pinout with the two centre pins missing, so I saw those and assumed they were 4M x 4 chips. Never even looked at 256K x 4 chips. The only other DRAM chip datasheets I've been looking at are the 4M x 1 and 1M x 4 chips, which all have a SOJ-20/26 pinout with the 3 centre pins missing on each side. And obviously they've got different pinouts, so while they fit on the same footprint as the larger capacity chips (albeit with missing pins), they definitely won't work. In any case, the other thing you could look at is the refresh cycle of the chips. Not sure whether the same applies for the 256K chips, but certainly the 4M chips can either be 2K or 4K refresh, which means it either uses 11 rows and 11 columns for 2K, i.e. 2,048 x 2,048 = 4,194,304 words; or 12 rows and 10 columns for 4K, i.e. 4,096 x 1,024 = 4,194,304 words. For 256K chips that would be 9 rows and 9 columns or 10 rows and 8 columns for 512 or 1,024 cycles, i.e. 512 x 512 or 1,024 x 256 for the full 262,144 words. Might be interesting to check the datasheets for the memory chips you've tested to see what refresh cycle they use. In my investigations into 2K vs 4K refresh memory chips on IBM PC compatibles, I never found a conclusive source that explicitly said whether 4K refresh chips are compatible. Prior to the pandemic I had planned on sourcing both and testing it for myself, but not only have a lot of the chip listings disappeared off eBay, the chips that I did buy still haven't shipped from Finland since most international freight routes have been suspended indefinitely, and the only 286 board I've got that's not in storage seems to be faulty so I can't test them properly anyway. I've got a Mac SE/30 and Classic II that are working, so I can use those to test the SIMMs, but that's not going to tell me whether IBM PC compatibles can use 2K and 4K refresh chips. Anyway, I always find these Amiga videos interesting, since they weren't really a thing here so I never used them back in the day. It really seems like they were at the cutting edge of what the best hardware of the time could do. At least until the x86 machines caught up.

[@droopysloopy]

Older ram had a much longer data retention than newer ram, wich means that it took longer for the ram to be cleared at power down. The more chips are on the ram sticks the greater the problem. Faster (60 vs 70 ns) also helps in this regard. Always wait at least 10 seconds after power off before turning the power back on.

[@cardigansrule]

I was going to say. Even on my 1MB Amiga 500 if you flicked the power on and off too quickly it would crash (or lead to Bad Things). 10 seconds is a while, but may be needed for older computers with high (for the era) RAM capacity. At least 5-6 seconds should always be allowed, even on a modern computer. Also? I cringed a lot, just for the power supply, the entire time he was flicking the power On/Off like that. Honestly I was kinda surprised he didn't check the power supply 1st, that would be where I would start with any computer this old (but I'm not an expert). The caps on those go bad too... I realized he checked the voltage on the PSU and it was good, but old/marginal caps can cause intermittent problems too.

[@8_Bit]

I have a small collection of bulletins that were shipped with Amigas that said to leave the computer off for at least 30 seconds while power cycling. They must have got a lot of complaints to prompt that!

[@bf0189]

Amiga seemed ahead of it's time and one computer I've always wanted to mess with! Great job restoring it and troubleshooting it.

[@jeremiefaucher-goulet3365]

That reminds me a RAM quirk (was it early Apple-II's?). If you turned them off and back on too quickly there was still some memory retained that didn't disappear while the system was unpowered and affected how the system booted (warm start).

[@Starchface]

It was a crude hackish way of detecting cold vs. warm start. The memory location theoretically could have the value for "warm start" there randomly when you turn the machine on. It could have been done properly with a few cents in components, but that wouldn't be Apple would it?

[@00Klingon]

I used to have fun rebooting a computer while playing a game and then loading up the HGR2 graphics mode right away and see that sometimes, some of the video memory was retained from the game.

[@thomasrotweiler]

Had a Sinclair QL back in the day, it was used quite heavily. Found when testing software, debugging that a 10 second wait (or longer ) was needed to for the memory to clear otherwise you could get random stuff still there when you re-booted.

[@milk-it]

Good work. Highlighting those idiosyncrasies with the RAM was helpful, and explaining the details of the RAM was very useful, too.

[@pcm720]

Thanks for yet another great video, Adrian! Making two more RAM modules by cloning the original NEC PCB and using those four chips you removed would make an interesting project. Not sure about the quotes for PCBs though, edge connectors can make them quite expensive to manufacture...

[@garthhowe297]

Re powering on to quickly. I think it had more to do with hard drives... they were bigger, had a lot of inertia, and drew a lot of current, and apparently didn't like being powered back on, while spinning down. Another awesome troubleshooting video... three thumbs up.

[@basvanharen2904]

I just cant seem to get enough about Commodore videos, thanks! Good to see this one has been saved. Thick thumbs up!

[@francoisfritz198]

Excuse me, but for the last 30sec, I don't heard you: just looking the out of this world intro😅, In France we call it '' another world'' one of my favorite games. As always good video! Take care

[@minombredepila1580]

And another excellent video from ADBasement Productions...

[@dLLund]

your persistence paid off. well done. might have been interesting to try the unmodified double-sided 8mb sims in non-adjacent slots, maybe slots 1 & 3 for example. take care n stay safe.

[@ceebee23]

oh the Amiga™ ...always a feeling of "what could have been...." the 4000 was the drool machine of my youth

[@Thunk00]

What a beautiful machine, inside and out.

[@martinbuckley260]

Absolutely outstanding!

[@m9cuss]

finally an amiga video, please do more and also atari st as well, i`m one of those rare types that loved both.

[@MrSatellitehead]

Adrian you are a machine!! Thanks again for all the videos.

[@danielson9579]

Good idea replicating the ecs and aga problem in U.A.E 🙂👍

[@terosaarela4555]

Nice work sorting out the RAM issue! There is a cutout on the top edge of the A4000 daughterboard where you can slip the IDE cable through under the bracket.

[@josko50]

I made a similar comment. I wondered if maybe that's exactly what the notch was for!

[@alerey4363]

Your last game at the end of the repair is self-describing your journey with that ultra picky ram! You have out-of-this-world sherlock skills to find the problem! Great job, keep it up!

[@jerkerivarsson938]

Great vid, man! I love people taking care of our Amigas!

[@doug834]

Who in the world would ever give Adrian a thumbs down? These videos are awesome and I love every one of them.

[@Nukle0n]

He used to encourage people to give thumbs down as well as thumbs up., Any engagement is good in the eyes of KZbin.

[@tomr3422]

it must be the Atari and nintendo crowd.

[@tstahlfsu]

Jerks. That's who.

[@AndrewTubbiolo]

I gave him and you a thumbs up.

[@thecaptain2281]

It's just people wanting to be contrary.

[@kevinwright7931]

I like that you replaced that loud fan with a quitter one. I would do the same thing. Some computers what the ram to operate a certain frequencies and will accept several different frequencies, you need to have them matched up.

[@Homemade-Blurb]

I am quite amazed with the hardware knowledge you have. Amazing 👍👍👍

[@spartonberry]

Can't help but :D when "I want to recap exactly what I've done" is one of the first things said. :)

[@MindFlareRetro]

A great series! Such a brilliant idea reconfiguring the RAM modules that way. I am embarrassed to say I have an A4000 buried away in its original box somewhere. I really do need to get my act together and dig it out for servicing.🤦‍♂️ Excellent content, Adrian. I always look forward to your videos.👍

[@SuperVstech]

MindFlareRetro so cool... I would love to find a 4000 so I can complete my collection... lemme know if you want to trade it for a different model, or otherwise move it.

[@MindFlareRetro]

@@SuperVstech Thanks. I will keep you in mind if my situation changes, but I do require this A4000 for my Amiga collection too 🙂. I assume you are in the US or Canada? Personally, I am collecting only native NTSC Amiga models. I managed to track down an interesting A600 last year -- video at some point. I am still on the hunt for an A3000 and C128d, but I am in no crazy rush.

[@SuperVstech]

MindFlareRetro i have both a c128D, and an amiga 3000

[@SuperVstech]

MindFlareRetro and, yeah, I’m in NC

[@SuperVstech]

MindFlareRetro I also have two A1000’s

[@awilliams1701]

I've had a lot of trouble with my C64 since I got it. The 1541 U2+ that I have for simulating disks was unreliable. Even the slightest touch can mess it up. Also the video is glitchy and only got worse. After watching your videos I got some deoxit. The 1541 U2+ works PERFECTLY now. No issues there. The video glitches are still a problem. Most likely logic chip issues at this point. A 10 print goto program will turn the text black randomly. It actually looks kind of like fire as it scrolls up the screen. But yeah deoxit did help a LOT!! So thank you for being so obsessive with it. lol

[@ProfessionalArmourer]

He should have used Deoxit in the memory sockets right off and not wasted so much time, I'm a retired IBM engineer, I lived off the stuff.

[@alasdairlumsden670]

Amazing video - thanks for making these. Nothing quite like coming home after an exhausting day and finding a new one. So relaxing and restorative to watch. Well done on downgrading the 8MB SIMMs, never seen that done before! Fascinating regarding the jumper pads.

[@dennissmithjr.5370]

I really love your repair videos, I have been learning a lot, thanks.

[@fredrikpeterson1987]

Thank you man for saving old retro Amiga its worth saving them and have some fun to iam old user of amiga 500 and the A600 + Amiga1200 back then i tink before 1988-89 last amiga i own and i remeber when i did a siwtch to the kickstart from orginal to kickstart 2.1 on amiga 500 :) its was scary back then to take the chip out. but i was able to sucess

[@BollingHolt]

Good job! Pretty cool how you modified those SIMMs. I've never heard of anyone doing anything like that before LOL. That's a good-looking machine, too. It cleaned up nicely!

[@nielsroetert]

Pure genius, converts a double sided DIMM to a single sided one, how cool.

[@DavePoo2]

23:52 - Interesting to note that those simms with the NEC ram chips worked better. Commodore used NEC ram chips on the Amiga 600 & 1200 boards

[@Megatog615]

I remember doing fingers crossed every time I booted up my V-Tech Lazer 486SX/2 in the early 90's. Seems like older stuff, even when it was new, had a little bit of trial and error.

[@aposenteiComBTC]

Man, I love videos like this on your channel! I love the details and the troubleshooting process. Keep doin videos like this!

[@Elkarlo77]

I am looking at your "bad" Ram Modules from the Amiga. I spotted some black points on the contacts, it maybe the Camera, but one Question: Did you use a soft eraser on the contacts? Corrosion on the Contacts is not uncommon especially on the old "silvery" contacts where the Pins were on the contacts and current was flowing they get black. Using an very soft Eraser (transparent or white NOT coloured) will take out the corrosion, but will destroy the factory coating against moisture. So normaly i give wonky Modules a gently eraser treatment, till they are uniformly shiny, then an alcohol treatment to protect them against new corrosion. With this treatment they are good for 5-6 years, eraser only 3-6 month, as corrosion creeps faster in. It may be the Camera angle but at 10:09 the right module shows Pin corrosion were the pin from the socket was pushed against the contact and current was flowing. From above Pin2 4 and clearly 6 shows it. This would be a typical case for eraser Treatment.

[@msvaughan]

I seem to remember that the older memory modules don't instantly lose its memory, there is a lag before it completely clears, as if it retains a charge.

[@markwhatever256]

im quite sure that ram will be really slow, not as slow as chip ram, but A4000 is optimum when the RAM is on an accelerator daughterboard. Enjoyed the video :)

[@more.power.]

Thank you Adrian excellent information o the Amiga 4000.

[@OzRetrocomp]

23:11 Watch that little pixellated screen atop Adrian's parts drawers carefully as he powers up the A4000.

[@basvanharen2904]

software faillure😂

[@JohannesLauesen]

I had exactly same behavior on my A1200 with accelerator card and fast mem. The Chipmem is SRAM that is purposly cleared. Fast mem is DRAM and it is basically capacitors with a charged state... powercycling might not be long enough for it to clear. Don't think your machine is broken, but some DRAM's just need more time to clear.

[@PaulinesPastimes]

Excellent video. So much effort on your part to solve those problems. Brilliant stuff and really nice computer. Cheers

[@Renville80]

Intermittent failures are always a pain to chase down. One way to try and aggravate an intermittent problem to track it down would be using heat to warm up a few suspect chips at a time, and failing that, we would then try cooling the same parts, using canned air (the propellant, specifically).

[@DarkAlaranth]

My brother had one of these and from memory, it did the yellow screen error sometimes when the memory slots were fully populated. I seem to recall it had to do with faster memory and fully populated slots that cause that error, but if I remember correctly, it only caused the issue on a quick power cycle for his machine. I seem to remember the A4000 being finicky about fast memory back then. Be nice to see it equipped with more memory though (Even if it doesn't need it.. :-P)

[@nadeembackus2741]

I think you are an amiga convert :-) welcome to the family!

[@damolin77]

Old Tech Reborn awesome job Adrian

[@detalite]

Often happens that 72 pin SIMM with presence detect are only compatibile with HP PC. I have a few of those and they have unusual memory modules with asymmetrical row and column adresation. As example MB8117400A is standard memory chip with symmetrical adresation used in most of SIMM modules (RAS A0 to A10, and CAS A0 to A10), and MB8116400A chip with asymmetrical addrestion (RAS A0 to A11, and CAS A0 to A9).

[@BartechTV]

I used to use 72-pin FPM SIMMs from a Silicon Graphics Indy in my A4000, it worked perfectly.

[@richardkelsch3640]

The RESET capacitor in the RESET circuit hasn't had enough time to discharge. The best way to avoid power on issues, is to just hit RESET before powering on, as this forces an immediate discharge of the capacitor, and forces a cold boot. Anyone that had these machines when brand new, would tell you that immediate power cycling will give you bizarre behavior. You must WAIT before powering back on. This was the case for most computers in the 80s, not just Amigas.

[@curtdawe]

Can this type of issue occur on a 128DCR? I've had strange hangs and whatnot when rebooting... any thoughts, Richard?

[@richardkelsch3640]

@@curtdawe Like I said, it can occur on any retro machine. Just hold reset when power cycling and it should never again cause an issue.

[@altareos]

"I don't need a full 16MB, what am I gonna use that much RAM for" Proverbs from another time...

[@Mclaneinc]

Lol...Very true...My first Amiga HD was a sizzling 80MB, and I never filled it up :)

[@michaelblair5566]

640K ought to be enough for anybody!

[@bdhale34]

Now I'm over here sitting with 32GB of it wondering if I should have bought more...

[@PiddeBas]

@@bdhale34 Same

[@GdotWdot]

There are systems with 1 TB-ish of RAM now. It runs hot, and gives no benefit to your everyday software, but it's there.

[@seancurtin6103]

I seem to remember either the 3000 or 4000 would throw a yellow screen if you tried booting it without the slot riser board installed. There were bus terminators on there that it relied on for clean bus signals.

[@Manawyrm]

Cool video! If you're up for a hardcore challenge you could build custom PCBs that can hold the desoldered RAM chips and give you the full 16 MB of Fast RAM :)

[@lemmingsbeest]

Many thanks for sharing the amiga testkit disk in the link... (i dit not know that it exsists) I like all your video's.... I watch them all!!!!!!!

[@gallgreg]

Great video! Nice troubleshooting!! Hoping there is a Part 3 to wrap up the clock issues?? I remember getting trapped like you with "slightly" incompatible RAM! (Can't find anything wrong with it in any tests, yet it has weird issues until it is replaced!)

[@adriansdigitalbasement]

It'll be down the road most likely -- for now I have moved onto some other machines. :-)

[@danielson9579]

I fixed this problem once by putting in 3.1 rom chips as it adds a delay to stop these sorts of problems.

[@stevencamp6824]

Love these videos!

[@Psychlist1972]

Best episode yet.

[@dungeonseeker3087]

2 years late but IIRC you can totally change the big box Amiga's memory configuration through Workbench. Hold down both mouse buttons then power on, keep them held for a few seconds and it will boot to a hidden menu, I'm 99% sure it has a memory option page where you can change RAS/CAS latency and overall speed (though its been a long time and its possible I'm misremembering, the memory menu might actually be from an accelerator card?)

[@jtveg]

Great work. ✔️💯🏆 Thanks so much for sharing. 😉👌🏼

[@nowaymangoshtomuchna]

this is the tricky thing about the yellow screen... it can be a hardware issue but also a software issue, it could be that the ram is to slow to really reset itself but it could also be the clock jumpers being set wrong and a multitude of issues... i would try reflowing and resoldering all the pins... i know its a sucky thing to do but it might help... also a very toughrou cleaning can help brah... funny thing is... for me it was the kick rom that was damaged :p

[@jgingiloski]

Adrian, I've got a feeling that the issue could be either of the lower two ram sockets, rather than an incompatibility. Have you tested your two 4MB ram sticks in the lower two?

[@devil5051000]

I noticed that you had not installed the daugterboard while testing. That is necessary, because there are resistor packs at the top of the daughterboard and they terminate the bus. If the bus ist not terminated bus errors can show up, maybe leading to instabilities. The Amiga 3000 wasn´t even booting without an installed daughterboard and only showed a yellow screen instead.

[@EnderMalcolm]

My friend got a Dell workstation, actually two of them, with dual socket setups and really good Xeons. Each machine only had one CPU though, so he moved the cpu and ram from one machine into the other to make a really powerful workstation. However, very randomly, there was a broken solder trace on the BIOS chip of the machine he put everything into. It refused to boot if the computer was cold, but if you let it sit in the crashed boot for a while and warm up, it would then boot just fine. And restarting the computer also worked fine. A very interesting and random problem. The crack must have been so tiny, for the heat from the CPUs to be able to expand the solder and make a good conection.

[@eDoc2020]

If I had that happen I would just move the cpus and ram all into the other workstation. Intermittent problems are the worst.

[@CharlieShu86]

Amazing work :) I wish to have that machine too :) The best wishes goes to you Adrian. Greetings from Poland !!

[@krnlg]

Hey Adrian, nice video as always! Some thoughts: It seems unlikely the RAM check would just look for random data ( that wouldn't really be testing! :) ), but if it just looked at the memory a bit at a time then it would always get back the same data it put in. It just wouldn't physically be at the location it thinks it is. i.e. If it writes 256kb of data (or whatever) then reads it back, then writes another 256kb and reads back, gradually moving up the address space, everything would look fine for it. Because it wouldn't be going back to the lower addresses and then seeing the incorrect data which got overwritten by its higher-address writes. I don't know much about the reset yellow screen issue, but it wouldn't surprise me if you're just not waiting long enough when power cycling. Seems likely your original RAM chips were perfectly fine. It could probably be explained by just tolerance differences between the two different types of module. When turning it on and off within like a second its probably normal for it to screw up, no?

[@shawnmulligan3471]

It could verify a lot of random data easily without storing it by just storing the 2 or 4 byte seed used to initialize the random number generator. Usually it would use the clock for the initial seed, store that, initialize the random number generator, write out 256KB of random data, then reseed the number generator with the original seed, and read the memory back, checking the random numbers it generates against what it generates again. The reason that this PROBABLY won't catch memory that's repeating is that if the amount of RAM it tests at once is less than the amount of RAM that repeats, it's not going to see a problem. If the RAM repeats on 1MB boundaries and it's always checking 256KB at a time with offsets of 256KB, it's never going to encounter the repeated memory in a way that affects the test. However, if those were really 1MB chips being read as 4MB, I'd expect it to be mapped as 1MB repeated 4 times, 1MB repeated 4 times, etc. This would give you only 1MB of contiguous reliable 'fast RAM' in the OS, and it would probably very quickly crash if you did anything fancy. I feel like they are probably larger chips (the presence detect on the RAM would also make sense if they were 16MB sticks according to the diagram shown).

[@shawnmulligan3471]

Another option for testing with random data is to calculate a running CRC (hash is better but Amiga is probably too slow for that to be feasible) of the written data. Then when reading it back just CRC the incoming data and compare it. If they match you can be reasonably sure that the data you read was correct.

[@GarthBeagle]

Adrian, can you please link the JEDEC memory doc you found? Thanks

[@eformance]

SOOO is also the PD value for 60ns 16MB modules.

[@mrchrome5403]

brings back the days of my childhood these amigas and commadors playing on Lemmings and Super Frog some of my best memories :D ty adrian keep it up pal

[@kd5byb]

FANTASTIC!!! Love the custom-made RAM sticks - fabulous!!! :) :) :) :) :)

[@alextrusty2585]

Trick with SIMMs was awesome.