19:18 "I decided to do what we usually do...dig deeper." I love the commitment and thoroughness. This is one of the reasons Noel's videos are so satisfying.
This is the first video of yours I watched, and the first time seeing an Amstrad repaired. You did an absolutely incredible job showing us how the video signal was generated, I've never seen someone deep dive into a repair while explaining things so clearly. I have rarely learned so much in 30 minutes. Outstanding video!
I really like how you don't ignore symptoms cue fixes but back them up with data. Your attention to bias in data shows a very educated understanding of engineering and trouble shooting. Bravo!
Haha! Apart from the personal fondness I have for the CPC, peculiar is what makes these quirky computers interesting for me 😃 I mean, the TI99 is *really* peculiar, and that's where its charm lies.
Absolutely fascinating Noel. It's awesome to see that a faulty RAM chip is not good enough, you need to understand exactly why it behaves as it does because of that fault. Totally brilliant.
Good video as usual. I'm glad I'm not the only one that sometimes sees a hardware problem and chases down a trail only to find out that it was one thing I didn't expect. Funny how you can get focused early and then not see it until later.
Thank you! And to be fair, that's totally fine as long as you keep checking your assumptions. The thing I'm most proud of in this repair is that I resisted the temptation to desolder anything until I was sure it wasn't working correctly (unlike a certain C64 repair I did a while ago 😃).
@@NoelsRetroLab I went down a 128 repair for a week. Replaced some expensive parts. Turned out to be a single bad trace on multiplexer. I think we've all been there ;-)
Yay! The "gatorade" is back. It wasn't on this video but rather the first Spectrum video, you kept talking about the "gatorade" which was of course baffling to me as I had zero knowledge of the speccy internals. But about halfway in you showed a block diagram with the gate array/ULA was marked, but gate array was still pretty far from gate array in my mind, but over a few seconds I figured it out. So now the ULA is forever the gatorade at least in my twisted mind.
I really love your deep-dive videos. Though I'm now working on IT, I started my career in electronics. As an intern in my first job (at 16) the department of the company I work for designed Z80-based computers to use in industrial environments so tinkering with those old 8/16 bit computers is something that has a great appeal to me. Too bad here in Brazil the prices for anything like that are too rich for my pocket.
Really glad to hear you're enjoying them! I have heard about the retro situation in Brazil! In general retro is getting more expensive everywhere, but that sounds ridiculous from what people have told me 😖
Please keep making videos, they are really well done and enjoyable. Between you and Adrian, I get my electronics fix in a week. One thing I'd like to see, is someone find out why these old computers were designed the way they were. Cost is obviously top priority, so hearing the solutions around that problem would be fascinating. Such as why the CRTC and Gate Array weren't combined into one chip!
The CRTC is a stock component made by Hitachi, Motorola and others (?) and can be found on PC graphic cards of this era. The GA is what makes the CPC a CPC and does a bit more than video output; maybe compare it to the PLA on the C64 (with some Vic2 mixed in) :-). See also neuro-sys.github.io/2019/10/01/amstrad-cpc-crtc.html
In addition to what Thomas said, if the CRTC was combined with the Gate Array... then we would need many more pins! They did that later when they went to ASICs, but I guess using that level of technology, this was much cheaper that way.
And it is fascinating to compare the US 8 bit micros to those developed in UK. The C64 for example has lots of custom chips (VIC2, Sid, PLA, CIAs) whereas the design of the 464 is really simple (no custom parts except for the GA). The 64 basic sucks, an OS is almost non existing. The CPC basic and AMSDos are way better! But the C64 has HW sprites, the CPC doesn't.
@@thomaskalbe6018 That would make for an interesting video in itself. A general look at those sorts of differences. The Amiga 500 (and 600 etc) is full of custom chips, though later than the C64. It has weird quirks that the CPU isn't even on the same bus as the RAM. Anyway, thanks Noel. Never surprised to see a reply from you. You do great at interacting with your community.
On SCART at least the lines are usually terminated with 75 Ohm resistors in the display IIRC, so yeah, that'd drop the thing quite a lot when plugged in.
I love your CPC video series! Years ago, with my cousin, I developped Fugitif, a french adventure game. It was really a great journey into learning computers. I recently rebought a CPC 6128 to dig again in this fantastic world, and I will have look at all these cool tools, like Dandanator, M4, etc, I which I had at the time. Thanks again for all the great informations you give in your videos, it brings me back 30 year back :o) And it gives me the desire to restore some CPC!
Really glad to hear that. If my videos make someone more excited about using and restoring their old computers, mission accomplished 😃 And wow, Fugitif looks amazing! I never played it back in the day, but I wasn't using the Amstrad as late as 1991. I'm going to have to check it out.
Another great video Noel! Diagnosing signal errors is always a field of rabbit-holes, especially when everything is inter-dependant like in these machines with ULAs. Great bit of detective work. Just goes to show that it pays to fix things as you find them, even if it seems unrelated to the main problem.
Im glad i found your video, my 464 is exhibiting the same symptoms as #4 and #5 intermittantly (either one or the other) it used to do the white square sometimes but now its only ever one of those two. Ill buy a bag of ram and some sockets and change the lot seeing as i dont have a scope anymore, thankyou so much
Today I learned a new geography lesson. I never knew about the city of Gijon until I saw your map. Several wikipedia links later, I'm now a seasoned expert. :) My geographical overview of the world has improved.
I'm quite curious to know Noel's background (if he's happy to share it). When I first started watching his videos from his accent I'd have guessed he spent quite a bit of his life in Northern Europe, Netherlands, North West Germany or Denmark areas
@@MartinSuper7 I assumed he was Northern European as well with heavy exposure to American English, so I figured perhaps military/NATO. I latched onto his channel because he shows his probe points on a sub schematic of what he's measuring. It's a great quality of his vids.
This was fascinating to watch! Very in depth but also very well explained. Amazing to think that the problem was caused by one faulty RAM chip! Pleased to see you figured it out in the end though, after a bit of a wild goose chase :-)
Imagine how cool you can be if after watching this you'd be able to tell some Amstrad head the number of RAM chip that's busted just by looking at garbage on the screen. "Yeah, I know that lines, it's number three"
You know, I was planning on making a comment like that "If you ever see the screen rolling that way, you can instantly say, oh that's bit 5". Unfortunately bit 4 ruined my plan by also causing the screen to roll. Grrr... 😃
Super vid Noel, REALLY like your deep dives, I always like to know why something went wrong not that just it did and now it's fixed, so appreciate the extra effort there. I bet the missing-ram-chip-screens-shot will be freeze framed by CPC repairers all over the world!
Hi Chris! Glad to hear that. I'm never quite sure what level to go to while still keeping things interesting. But I guess that if it's interesting to me, chances are it's interesting to other people as well 😃 But I'm really glad to get that kind of feedback anyway. Cheers!
Excellent video as usual, Noel - thank you! I may have mentioned this before - sorry if I have! - but may I suggest that the "probe" icon on the schematic is placed so that it does not cover the label of the pin you are probing? I suppose that a red dot or a small arrow would work anyways, I can understand that changing the transparency of the schematic picture would take time. Looking forward to your next video!
Thanks! Yes, I remember your suggestion but I didn't come up with anything that worked really well. Maybe I can just use a smaller probe graphic and put it on the inside of the chip. I'll try that next time. Cheers!
@@NoelsRetroLab No problem and sorry for mentioning this again. What about a red rectangular shape which highlights the pin AND the label? A smaller probe/arrow from the inside would work as well I think.
Great Video. Very thorough explanations. Really like how you showed exactly where to put the oscilloscope probe and what signal we should be getting. This makes me even more excited for when I get mine.
Great as ever... Is that a late issue short board but with an early gate array? The more 8 bit systems I restore the more love I have for the design and build of the CPC464.
That's right! A bit of an odd combination, isn't it? At least it has room for both kinds of GAs. Personally, I'm a fan of the original long board. That thing is soooo roomy to work on! 😃
Always a pleasure to see you fix a CPC (my favorite). Your explanation of the video signal generation is awesome. It is great that you could follow the issue down to where in the ROM code it must have ended up to corrupt all the CRTC registers. Makes me want to modify my simulator to support damaged RAM bits and reproduce in the debugger where it goes step by step :-)
Do it! I'd be very curious about it. Are you still working on your emulator by the way? I noticed it doesn't run on 64-bit Mac. Maybe I need to get the source and build it myself. Is there a Github page by any chance? I wasn't able to find the source even though the web page says you're going to open source it.
@@NoelsRetroLab I'll try it and let you know. Yes I still improve CPC++ on my ample spare time :-) It's on bitbucket at bitbucket.org/bricerive/cpc/src/master/ I need to update the build instructions but it currently builds on my Mojave mac. I should also make a release at some point.
interesting...this might be the issue with mine not displaying anything on my RGB to VGA board... (cant test it on an original monitor coz i dont have one!) thanks Noel.. your vids are always top notch!
Do you have access to a Dandanator, M4, or some kind of ROM board? If so you can pop in the Diagnostics ROM and you should be able to sort it out (unless your edge connector was dirty like mine was). It also beeps as soon as it starts, so if nothing else, if you hear that you know that there's a lot of stuff working already, even if you get no video out.
@@NoelsRetroLab unfortunatly no access to a Dandanator or the like... and the top of my edge connector is pitted bad! ...and i dont have the tape deck either...(its just the original Z70100 board!)..so no sound out! (guess i should bodge up a speaker!) ive sorta given up on it :( i think ive ripped some vias on the CRTC...and my eyes are going bad!... ...i wish i had access to a working cpc i could probe around on... you wouldnt be interested in trying a repair? ill send it over to ya from Aus! (i was hoping to get the board working so i could make up my own input for it...it was just missing the gate array when i purchaced it.. ive replaced that but still cant get anything out on the RGB to VGA board..even tho there seems to be signals coming out of the CPCs RGB+Lum+Sync) ive even contemplated buying one of the clone/smaller form factor boards and using the chips from my board
Ok, going out om a limb here: I'm stating that this is a stellar video, as always. Don't make me come back and edit this comment, Noel! 😜 Edit: Goddammit! I had to edit because I didn't say that I made my statement before viewing the video, obviously. Don't make me edit it again! 😁
Excellent detective work! I really enjoyed the deep dive aspect of this video ... especially the scope work. Thank you for taking the time to explain the interrelationships of the chip functions and data flow (I was barely hanging on in places ... but I followed 😉). The graphics were very helpful.
@@NoelsRetroLab, a lot indeed and I plan on going back and watching them all :) I finally got my hands on an Atari 800xl and a 130xe but haven’t even had a chance to test them. Hopefully soon, I can play.
I noticed you use a desolder pump, me too. However they take the use of both hands, I bought an Iron that has one built into it and boy it works great. They are 30W and cost about 16 quid on ebay. Great video btw ^^
Thanks! I use both kinds of desolder pumps: the manual ones for single things, but then I pull out the desoldering gun for whole ICs. It's such a useful tool!
The problem of how the CPU and the video hardware fight over the RAM is always interesting to me because every personal computer of that era had its own solution.
Absolutely! That's actually one of the main distinguishing features of computers at the time: separate RAM, interleave access, or blocking CPU are the main general approaches. Honestly, I'm not sure which one I would pick if I were to make a similar computer today.
@@NoelsRetroLab There's also the simple but ugly approach that my Ohio Scientific used. It just blanked the video output with an AND gate whenever the CPU was accessing the video RAM. Whenever you wrote to the screen, lines would briefly disappear which was better than the sparkles that would have been displayed.
Yeah, that's frustrating. I'm really annoyed that YT got rid of email notifications a few years ago. That's why I ended up setting up a mailing list in case people want to receive notifications that way: noelsretrolab.com/
I've seen this with bad RAM chips before.. it's important they stay off the bus when they're not supposed to be there... but i have seen faulty chips that always hold their output at some bad level regardless. Have definitely seen this happen on those Samsung 4164's in ZX Spectrums
The green one had slightly better resolution, but some games were downright impossible with it. My parents bought me the green version and the game Trailblazer. After my dad tried playing it a bit and realizing that some of the color tiles looked identical, he went back to the store and swapped it for a color one 😃
another great video your knowledge is supreme ...... ohh them smaller mother- boards are so fragile..i have seen that with the power sockets on a few of them i normally put a new capacitor in and bend the leg round.. any advice on a ""cheap" "O scope" for use with these computers (Amstrads/spectrums/ maybe commodore at some point) ? i have got a couple 464's at the moment with that dreaded 'square' of doom
Hey, that's a cool trick to fix the power track and doubles as a nice voltage stabilizer. I like it. I don't have any recommendations for a cheap scope. Some people have good results with PC USB ones but I never used them. My scope isn't super expensive and I even bought it second hand, so it ended up being around 150€ I think. Totally worth it, but I understand it's still a lot if you aren't going to be using it all the time. You can try just using a logic probe that just tells you if there's activity in a line. They're probably like 5-10€. And if you have that square of doom, put in the Diagnostics ROM and it'll probably narrow it down for you. Good luck!
I didn't. Was it swollen? I noticed it was a bit dirty around there, but it wasn't the cap leaking. Electrolytic caps almost never fail on the CPC main boards for some reason, so I hardly ever look at them.
@@NoelsRetroLab OK that's an insane coincidence. He's currently working on a 1972 HP PC whose PSU shorted and sent 13V into the 5V rail and as far as he fixed it, the current state is exactly the same as you said there - the machine executes a subroutine, but when it returns, the return address is corrupted. I genuinely thought this was a nod to CuriousMarc! Be warned though: His videos are entertaining, very informative (he makes learning fun) and even thrilling - he ends most episodes with a cliffhanger so if you start watching his channel at 10pm, it could happen that you won't go to bed until 3am...
@@senilyDeluxe i just watched his latest vid on that yesterday... ...looks like Marc might get some hints from this vid!...he was thinking it was a DRAM refresh issue... seems it could be a completely bad RAM chip!
@@WacKEDmaN Good idea. Usually these 4116 are pretty sensitive and die easily. But they can fail in a way that single bits drop back to zero quicker than refresh can catch them. But mind you, the machine's self test tested the entire RAM and found no faults.* Also he checked various refresh-related signals on the RAM PCB and found them to be way out of whack compared to a working board. So the board is definitely not refreshing the RAM the way it should. * TL;DR when testing RAM, don't just write a bit, read a bit, but fill the entire RAM and then read the entire RAM. I've encountered so many machines that pass their RAM test even though the RAM is flaky because of bad RAM test routines. Sure, it'll find dead chips, but it won't find flaky chips. (but I've also encountered ones that did it right and flagged chips that otherwise would have passed - you can see it on the screen, it clears and then a few erratic pixels start to appear - the simple RAM test would pass that and you'd have to find the bad RAM yourself)
19:57: It bothers me slightly how you're saying here that it has "two bits set to zero". I guess you could say that on that particular nibble, the _second_ bit (from the LSB) is set to zero, but unless there's something else you're not showing us, this isn't "two bits".
What I meant by that is that each byte will have 1 bit (the 5th from the LSB). Since the return address is 2 bytes, popping that 16-bit address off the stack will return it with two bits set to zero (or one, or incorrect in general): The 5th one of the lower byte and the 5th one of the upper byte. I didn't show it graphically with the second red circle because that one was already zero, but maybe I should have to make it clearer.
If only they made it so the CRTC could start its first fetch at a specified address or better yet, allow the address to be changed mid frame. This would have allowed for cheap. fast and smooth vertical scrolling. They make the same dumb mistake in the Atari ST.
There's no doubt the CRTC could have been much better, but the one there isn't horrible. I believe you can specify the starting address, although maybe not mid-frame though. You can tweak a lot of parameters.
Hi, I had a memory problem. I poked the memory and found that 49151 memory address failed. I guess that is upper memory. Is that the 4 top chips on your vid? Your help would be really helpful. Save me taking all chips out. :)
not really possible to be a burn at that point with 5v supply and those resistors it would never get hot. resistors never really go short circuit inside and if they did the burn would be in the middle of the resistor. a high resistance joint there would just drop voltage and reduce the current.
No, but the patter would be totally different since it would cause to jump around to different places. And then different combinations of bits would cause different patterns, and whether they failed low or high... it's a crazy number of combinations (256 to be precise 😃).
Horrible nasty little buggers, those RAM chips are. They cause all sorts of problems. Why do they fail? Did you replace the other RAM chips as well. They might fail in the future.
I didn't. It's a fair amount of work and the traces on the Amstrad CPC boards are pretty delicate, so I'd rather deal with them as they fail. The time I'll do that is more than half of them are faulty, then I figured there's something really wrong going on and swap them all.
Yes, totally agreed about the YT notifications. I don't even have them on on my phone so I pretty much don't get any. I'm really bummed they removed email notifications a couple of years ago.