You know, you're totally right. I should have checked that. I guess I'm not used to the clock being shorted that way. It happens more with the data or address lines, but especially since the Z80 is socketed, I should have totally tested that. Let me check...
All right, I just tested it and that wasn't it. That would have been funny if it was that. It goes to show that I'm so paranoid about these ASIC chips, that sometimes I'm too quick to blame them. But in this case it looks like it was faulty. I also thought about just generating the 4MHz signal myself with some external circuitry off the crystal oscillator, but I think it would be hard to get the perfect synchronization with the other clock signals in the GA, so I don't think that would work. Oh well. Good idea though!
@@NoelsRetroLab My first thought indeed when I saw you jumping to the conclusions so early. I'd use a diode tester to see if there is a hard short from this pin to GND and +5V rail, in both directions. This sometimes gives some insight into what exactly could have gone wrong - is it an ESD protection diode within the chip, or perhaps a passive component somewhere in between that and the CPU?
Hey Noel, there looks to be very much crusty old rosin flux around the asic which sometimes can go conductive and corrosive. Try to clean all that old stuff off the board and check again. I had this happen just with standard throughhole cmos chips and that ASICS pins are less than 1/2 a mm distance to each other so this can be a potential problem.
Hmm... I hadn't thought of that, but it's true that there's a fair amount of flux residue. I'll try cleaning it see if that helps. I'm also toying with the idea of generating the 4 MHz clock signal myself in an external circuit. But I'd have to do it just right for it to work, so I might revisit this console after all. Thanks for the tip.
Thank you! Yes, that's very similar to the Spectrum ULA, although I believe the signal gets amplified a bit before being fed into the ULA, right? It doesn't go straight from the oscillator. It's the same with the regular Amstrad CPC range.
I dont mean to be offtopic but does anybody know of a tool to get back into an Instagram account?? I somehow forgot the account password. I would appreciate any tips you can give me.
@Prince Kameron thanks so much for your reply. I found the site on google and Im in the hacking process atm. Seems to take a while so I will get back to you later with my results.
I have a working GX4000 myself. It's a nice little machine. It's such a pity that the ASIC cannot be replaced with a clone chip like the ZX Spectrum ULA can. As far as I know, a FPGA implementation of the GX4000 doesn't exist. Sadly, GX4000 machines will stop working sooner o later, and the logic of the system will be lost. Anyway, it was great to follow your inspection of the board, even if it was not possible to repair. Thanks!!
Really? I forget the details but often times the signal out of those oscillator circuits is only about 1-2V peak to peak or so. I don't know about this one specifically though. I'll compare it when I get a working GX4000.
@@NoelsRetroLab Pure crystals (with 2 pins) are usually driven at 1 to 2V, because that's a good voltage for the crystal to operate on. On the other hand, crystal oscillators (4-pin boxes, either square like here or rectangular) have the driving electronics built-in, and are supposed to output a TTL- or CMOS-level signal at 5V.
@@NoelsRetroLab These are TTL level signals, so you should see 3.5 V at least. You may see less if the clock is close to the bandwidth limit of your 'scope. The ground point is important at frequencies like this. Use a ground as close to the oscillator as you can. Try to avoid using the pigtail clip lead - a spring-style contact is much better.
@@Derundurel Thanks for the tip. I was familiar with the spring-style contact, but I looked into it and what a difference it can make in the right situation! Off to look for one of those contacts for my probes!
@@NoelsRetroLab 99% of these are XTAL's with TTL output. 39MHz must be visible even on cheap osciloscopes. But on other hand these devices are very sturdy- never seen them dead. The big chip can be bad and load xtal.
Could you dig this back up and create your own clock for the Z80 and see if that is the only component broken on the gate array? Would be very cool to see this machine running.
8:37 they were craaaazy ... guess it was cheaper to use linear regulators than buck converters for the power input stage. 15:32 wouldn't think the oscillator is fine - if it was a crystal, you could expect something like this but a crystal oscillator has a very nice rectangular signal between 0 and 5V. This looks like you have limitted your bandwidth of your oscilloscope - or the oscillator is almost dead. But strange that this 1MHz looked so fine.
As the 1MHz looks fine, I am confident that Noel just enabled the 20MHz bandwidth limit option. 40MHz is severely hampered by it, but 1MHz passes it just fine.
1) upgrade the PSU to support the C4cpc cart (without a new psu you will brick the machine) 2) get the c4cpc cart :) 3)save money instead of silly cart prices
@@fogvarious2478 That's very true. There are some gems out there though. And the modern games being made for the CPC are amazing. They show how great they could have been back in the day!
Another Amstrad falls victim to ASIC failure 😔 Still it's very educational to see your troubleshooting process. I really need to get one of those current limiting bech PSUs. I agree with you on preferring to work on older machines with through hole components - much easier to debug and repair!
Since the cartridge port is so close to the gate array, could mechanical strain from inserting and removing cartridges broken a solder joint? Maybe a few passes of hot air to re-flow the solder on the pins nearest to the cartridge port could help with any loose or intermittent connections?
It could be, but since I'm definitely seeing issues in the clock circuit, you'd think that just checking the inputs, VCC, and ground pins would be fine. But who knows with these all-in-one chips!
uhm, i must ask. Why you didn't inspected board on the bottom? And why you didn't removed this gunk on gatearray's leads? Looking at video even in 360p i can see there either sticky liquid (like cocacola after seasoning xD) or solder flux.
Well, all that big ass chip probably does in this case is to divide the clock from the oscillator down to 1MHz, with a chain of flip-flops, and tap the chain along the way in multiple places to derive the 4MHz and 16MHz or whatever else it needs. It should be very easy to throw something together to divide CLK16 down to 4MHz and feed it to the board. Assuming of course that the only thing that's broken from that chip is the CLK4 output, which is not necessarily very likely... still it might be fun to try it.
Yeah, definitely want to try it. But I wanted to try on a working GX4000 first to ensure my clock divider works, and then try it on the faulty one (and like you said, I fully expect it not to work because I'm sure something else is broken).
Quick comments. the Amiga 1000 came out in 1985, along the Atari 520 ST. The Famicom came out in Japan in july 1983 (almost a full year before the first CPC464). The GX4000 also came out 2 years after the Sega Megadrive.
It did cross my mind, but I was afraid that unless I generate it exactly the same way the GA does, it might be out of sync with the 1Mhz and 16MHz signals that come out of the GA. Also, I wouldn't be surprised if other parts of the GA are faulty as well, so even if I did it correctly, it might not work. Still, it could be a fun mini-project.
@@NoelsRetroLab Yep only just seen this video but if it doesn't generate the correct 1Mhz signal other parts of that chip may also be faulty. I am also thinking something else may be wrong to cause the ASIC and the RAM be faulty in the first place too.
@@p166mx Yes, you're right. Chances are someone plugged in the wrong voltage. Or even the correct power supply that is known to have problems (the GX4000 has some really weird power circuitry to allow two kinds of power sources). I'm getting another GX4000 soon, so I might look at the signal and try to recreate it as a fun project, but I doubt it'll work. I'm sure there are other issues like you said.
The side-by-side power schematic and circuit board spit screen was brilliant! It’s great to see that not every project is a success ... very real, and that’s very encouraging! As the episode went on I was thinking “Not the gate array, not the gate array” 😬🙂. Perhaps sometime you could talk about the role of some of the discrete components on a board like a particular resistor or capacitor? I often wish I’d pursued the electronics I’d learned in high school (where I only learned the basics of resistance-capacitance circuits). Thanks again for the great videos!
Glad you liked it! Yeah, the GA/ULA is always biggest fear when dealing with GX4000 or some of the latter ZX Spectrums. You're right that I tend to overlook the "small" components. I'll make a point to talk about them sometime. It's never too late to learn electronics. I studied Computer Systems Engineering, but I didn't learn "real" electronics until a few years ago when my Amstrad died and decided to repair it :-)
It still doesn't mean the gate array is dead, it is connected to so many functions, it could be one of the reset lines are being pulled low. Or a CAS/RAS line is shorted, the gate array might just be in an unknown state.
that's exactly what I was thinking too. Especially the older these things are the fewer spare parts we can find for them and have to make do and mend...
what would happen if you build some opamp or logic hate clock multiplyer and substitute 4mhz clock or inject exyernal 4mhz clock from another clock oscilator+transistor circuit? or replace z 80 to check if it makes clock shorted
Also there's question wether 4mhz clk going out from generator is "flat" or is it "grounded"? Short to ground isn't as hard to troubleshoot, especially when you have literally a straight line from generator to cpu, challenge may be to lift appopriate lead of generator, but this can be easily circumvented by cutting lead whereever it's comfortable to do so and measuring both directions.
I seem to be addicted to your channel (until I get to Commodore videos, lol.). interestingly, I think the CPC Plus's / GX4000 were the only European systems to have hardware scrolling and sprites, however it was way too late as you mentioned too.
Thanks! Not a Commodore fan, or just too many Commodore videos out there already? I don't think the Plus had any hardware scrolling beyond the CPCs abilities, did it? Not 100% sure, that's why I'm asking. Sprites for sure, but it's amazing that the TMS9918/9929 already had hardware sprites all the way back in the late 70s! So MSX systems using it had hardware sprites, and so did the TI99 (not really a EU system) and the Colecovision.
what makes these chips die? I have been watching you troubleshoot and uncover failures in these machines, just wondering what is making these chips fail? Is there a ghost in the circuitry?
HI Noel, I know it is an older video, but I have to ask: Are you sure it is a crystal and not a crystal oscillator? The ´1.2mV sinusoidal signal on the gate array input doesn't look good at all. I guess the oscillator is just dead and if you replace it you'll find a 5V square wave signal with the desired frequency there at pin 2 of the ASIC.
Did you say the big chip replaced separate chips from an earlier modal? If you know what it replaced you might be able to find small SMD replacements of the separate parts and maybe something like the RP2040 with PIO and get a small pcb built with castellated holes :p
Not exactly. The GX4000 was always designed that way. What I meant is that it's an evolution of the CPC board, and in that board it did use several discrete chips (plus some larger custom ones) and here they're all combined. So it would be pretty hard to replace exactly and besides, who knows what else is broken there. But I still want to try to feed it the clock signal by hand and see if that changes anything. It would be a fun experiment.
I just picked one up that only came with a 5v dc adapter. I think I may have accidentally plugged it into the 11v jack. Suffice to say it didn't work. Is it possible I damaged it by not supplying enough voltage?
Good video :-) Is an FPGA an option to fix this? Opps I'm a bit late to the party I see. I also love my Adjustable power supply, I am using it in making the Ben Eater 8-bit CPU :-) I keep the amps just above what it needs and I have seen the voltage drop sometimes, so I look for my mistake :-D
No FPGA implementation as far as I can tell unfortunately. Hopefully in the next few years those will become cheaper and more common for more computers. Ben Eater is AWESOME. Love his videos and explanations!
I bought one of these when they came out, for about £50. Since it shares the PCB with the full computer version, I thought it would be possible to attach a keyboard, and use the CPC+ cartridge instead of the game one to get a CPC+ that can feed a SCART output, and not need to buy a monitor. I never got round to trying.
I believe you're right and it's possible, but I haven't tried it either. The Plus range actually has a really nice keyboard, so might as well use those if they're available (although they're getting quite expensive).
Wow, had they released the Plus range instead of the CPC it would have been indeed a game changer! Even if they had released it as late as 1987 it would have still been a contender.
The Amstrad Plus computers and GX4000 console could be something really good in 1986, but in 1990 was too late. About the games, all the carts got small roms based games, they could do bigger games like Defender of the Crown, Last Ninja Remix or Maniac Mansion.
Yeah, the problem is that it probably needs to be syncronized to the other clocks in the system. I just got another GX4000, so I'm hoping to do some experimenting with that and see if I learn something.
@@anvz6 Right. I thought you meant generate it completely externally. Yes, that's what I'm hoping to do. The reason I didn't try is that I'd be surprised if that's the only thing failing in that ASIC, so setting it up and have it not work wouldn't tell me much. But now I have a working console, so I should be able to test that circuit and then try it in this one. Should be fun!
@@NoelsRetroLab Yes, there may be some other fults on this chip... but I think it could be a good idea to "patch" the existing chip only bacause it smd and as so much pins...
I think ASICs are great for hardware designers, and are the curse of people like me, trying to repair systems. I don't think this ASIC was ever reverse-engineered, so no luck there. Hopefully someone will at some point.
Yes, definitely still liked the video. But that is the reality of it, can't fix them all. Case looks great. Hopefully you can find an inexpensive board to replace that one and make a really nice console.
Here, have a "like" for producing an honest - and still interesting - video! While I never had the chance to see this console in person I now know its insides and got some osci action, too! ;-)
@@NoelsRetroLab My pleasure - credit where credit is due! Where I live we have a slightly different saying: "Sometimes you lose, sometimes the others win!" ;-)
Be interesting to see if it was just that one issue on that chip not giv your a clock signal by generating the signal with a function generator see if it would of worked like do a doc brown hack in back to the future 3 were he sorts out an issue with a chip that failed
Haha, yeah, I would like to do that in a future video. The problem is that it's not quite that simple because the clock signal needs to be synchronized with the other clock signals in the board. So I'm hoping to first get a working GX4000, see what a healthy clock signal is supposed to be like, and then try to make a simple circuit to generate it. I suspect something else will be faulty in that big ASIC, but hey, it sounds like a fun project anyway :-)
Yes i have that game console only ever seen one game for it burning rubber. I ordered a flash cart from France to play the other games with out spending a fortune on cartridges
What exactly is Chipquik? I was looking at their site and I'm still not 100% sure. It's a low-melting point solder? I've replaced those ASICS before with a hot air gun and some care, but it wasn't the easiest job.
Are you sure about only 5V in? I wasn't aware they had multiple board versions. And yes, totally in agreement about the CPC 6128 :-) Or, if you press me, I guess the 6128 Plus is pretty good too.
You forgot to mention the Amiga 1000 that was released in 1986 and preceded the A500. :) Amstrad actually didn't stop there: three years later, they released the Amstrad Mega PC: which combined a 386 SX and a Sega Mega Drive in one unit.
True, I tend to forget about the Amiga 1000 in general but people keep reminding me it DID come before the 500 (bad numbering, Commodore! :-) ). And also true about the Mega PC, although it wasn't really a console by Amstrad. But yeah, between that and the GX4000, I'm sure they swore never to touch consoles again!
ive only seen these in pictures...never seen anyone handle one...let alone tear it down and (try to) repair it! we have a running joke that the GX4000 is the worst console ever made... but its not 'bad'...it was just late... 2 screws holding the thing together... trust Mr Amstrad!
Yeah, it was waaaaaaay too late. But the case design and attachments were top notch! Especially considering that the Amstrad CPC 6128 is a nightmare of screws.
I remember those being on store shelves and thinking, even as a kid, that can't compete with the NES and Mastersystem surely? I was right. My first machine was an Amstrad CPC 464 which I have very fond memories of and this console was not a whole lot better really.
Thanks. I actually have a non-working GX4000 coming after the virus lockdown is over. Hopefully between the two I can get a working one... unless is that annoying ASIC chip again! :-)
@@NoelsRetroLab Well let's hope it's not that problem again. Looking forward to more videos - highly informative and professional. One of the best retro repair channels out there.
Yes, I want to try that sometime. I'm afraid of two things: a) The clock signal generated isn't in perfect sync with the others in the GA and b) there are more things wrong with the GA than the clock signal. But even so it seems like a fun project, so I'll probably try it anyway :-)
The circuit board look like it's from the 80s. By 1990, I expect to see more surface mount coming in.... Not saying this is "good" or "bad", just an observation.
I didn't think to check that, but I see the GA has a /RESET line. I doubt it because the other clock outputs were working, but who knows. I'll have a look. Thanks!
It's a bit odd that the Gx4000 is hardware compatible with normal Amstrad games which were still being made in 1990, but you just had no way to play them on the thing.
I'm the first one to say that the Plus machines were MANY years late, but I don't think it was realistic to release them in 1984. I think if they had released them in 86-87, they would have been a real contender though.
I had a broken Unipolbrit 2086 from Poland which is a bit rare Spectrum clone. It looked like the gate array was bust. Fortunately it could be fixed by using the gate array from a Timex 2048 which is much less rare. There was no other option than to sacrifice a working computer because of the gate array. Yes, it requires skill to (un)solder a big smd chip, skills I don't have😉
Interesting that they would both use the same gate array! I guess that was a true clone. I have desoldered and transplanted the ASIC chip on a CPC Plus before, so it's definitely doable. When I get back into this project I might have to end up doing that.
@@NoelsRetroLab The Unipolbrit 2086 was basically a Timex 2068 which was slightly modified by the Polish import company. The amazing thing is that both the Timex 2048, which is very much a Spectrum clone, used the same gate array as the Timex 2068, with is a much souped up "Super Spectrum".
A shame you weren't able to repair it, but ASICs have been the death of many a machine. Perhaps one day the decapping of the ASICs will reap benefits and there will be a drop in replacement.
Interestingly enough, the 8-bit market is alive and kicking to this day, so that alone renders the "too little, too late", rehashed, false argument invalid. What killed the 6128 Plus/GX4000 wasn't their so-called "belated" release, it was their backwards compatibility with the original CPC, which lead to the lack of dedicated software for the new systems, as developers back then resorted to porting carbon copy CPC games and calling it a day. Yet, look at Ghosts 'n' Goblins and Sonic the Hedgehog on the 6128 Plus now...plus many more games to come.
The 8-bit market is dead, dead, dead. There's a thriving community around it, but people aren't remotely making money to survive, which is what would keep a console alive. So yes, you can make amazing stuff with it, but at the time nobody wanted to spend their money or effort on a clearly outdate console.
@@NoelsRetroLab The 8-bit market is "dead", yet 8-bit console re-releases in the form of mini-consoles, classics etc. sell like hot cakes. If people are buying the hardware, you can be sure they'd buy the software too. The CPC is seeing a resurgence of physical, commercial releases. People would pay good money for the CPC's contemporary releases (I know I would), it's just that nowadays many good developers like BG are doing it for a hobby, instead of a profession. Said group even turns down crowdfunding offers...so don't be too sure about the "dead" part. "clearly outdated console" 😄 Again you're making the mistake of comparing it with the 16-bit consoles, when the 8-bit market was still pretty much alive when the console was released, even as late as 1992 the market was alive, two whole years after the 6128 Plus/GX4000 were released. The c64-based console and the Master System II were released in '92, for crying out loud... No, what killed the 6128 Plus/GX4000 was the lack of dedicated software, not the "dying" 8-bit market. I was there, I witnessed it all. This system would have been a hit, had it not been riddled with lazy CPC ports.
@@NoelsRetroLab I bought one brand new in 1990 for 99 quid seemed a deal at the time but I was too young to realise the thing was doomed. I got a handful of games and eventually bought a 386 later. Look forward to seeing more gx videos later
it's a huge shame that many of these modern chips aren't socketed, to make up for their complexity, probably due to cost. combining the chips could actually be a great idea, because you'd have to make less socketed slots, you could only have 1 - 3 of them and cover all the complicated stuff. but no one really does that, and nowadays, in laptops, it's even worse. you can't even swap the CPU (which basically holds everything nowadays) in any of them, lest it's some bulky thing with a desktop cpu, and many of those have some complicated GPU that isn't socketed. and even if you had those socketed, if they're prone to failure, they always will be, and no one is able to make proper replacements. maybe some smaller companies would be willing to make somewhat capable computers (just something to play 480p x264 on or something) with lots of expandability and very continuous support. they could iterate on the design of their more complicated, socketed chips if they're flawed while keeping the same form factor. but i'm not sure if something this kindhearted is sustainable. i hope it is.
It looks like they used the latest technology to produce an obsolete game console. I'm sure someone at Amstrad went to work every day knowing this was a terrible idea.
guy had a load of them on ebay , he bought as bankrupt stock.. minty new, and cheap.. wasn't worth fixing BUT the PSU wasn't powerful enough for c4cpc cart (and would brick machine) .. so had to replace that.
@@NoelsRetroLab thankfully I saw the warning, got new PSU's from cpc electronics (ironically enough) .. the guy making the c4cpc sends them quick / good lad.. based in france.
This kind of situation right here with those damned chips that are next to impossible to replace without a full on arsenal of high-end equipment is what had killed my interest in analog electronics and repairs, funny to see that even you feel the same about them!
Yeah, even if you can replace it (which I've done it and it's a chore), there's not much to think about. It's either that chip that's faulty or not. The more discrete chips computers have, the more fun it is to repair them! 😃
nice video!!! i was in a same puzzle with a tv set 15 years ago.... no video output from a video chip ... i changed it and the result was the same - then i changed 3 small electrolytic capacitors around chip and i saw Output ... i saw an electrolytic capacitor near your chip maybe if u change it u will see picture ty for your exellent video
If you can get the chip I would gladly replace it for you for free. I'm in the US. I have all of the surface mount soldering equipment. I'm an electronic technician and I work with smc all the time.
I got one of those back in '98/99 for £5, in box with everything. Played on it once. You're not missing anything. It's more useful as parts. I ended up giving it to a friends 5 year old son. Don't think he played on it much either. LOL.
@@NoelsRetroLab I think you are spot on with your "too little, too late" remark in the video: What is the point of bringing a Z80 based console on the market in 1990? If I wanted a Z80 console at that time, I would have bought the established Sega Master System (Japan: 1985, US: 1986, EU: 1987), just because I knew there were a lot of cool games available, and I could trade games with friends. Does this console have a unique selling point over the Master System? I don't think "some effort" would have been enough to save that device, except if for you "some effort" means being a whopping 3 years earlier to the market.
@@tw11tube Seriously! I don't know how it ever got past the planning stage. Or maybe it was initially pitched and designed as a 16-bit console, and then, somewhere along the way, someone cut costs and stuck a Z80 on it. 🤪
This is one of the worst consoles. It is just what the Amstrad should have been from the get go. I am a big Amstrad fan, but it just was never made for gaming. Only a handful of programmers could actually get anything out of it. That people can do it today does not really count. Anything related to moving graphics was painfully slow. On top of that it seems the Spectrum gets more out of it's AY chip than the Amstrad.
Respect to people who aren't afraid to show their failures as well as their successes!
I would've cut the trace in a reparable place and see if something was pulling that pin low. A shorted protection diode to ground can do that.
What if the Z80 is shorted and pulls down the clock signal ?
You know, you're totally right. I should have checked that. I guess I'm not used to the clock being shorted that way. It happens more with the data or address lines, but especially since the Z80 is socketed, I should have totally tested that. Let me check...
All right, I just tested it and that wasn't it. That would have been funny if it was that. It goes to show that I'm so paranoid about these ASIC chips, that sometimes I'm too quick to blame them. But in this case it looks like it was faulty. I also thought about just generating the 4MHz signal myself with some external circuitry off the crystal oscillator, but I think it would be hard to get the perfect synchronization with the other clock signals in the GA, so I don't think that would work. Oh well. Good idea though!
@@NoelsRetroLab Did you measure the clk16 on pin 117? You could divide it by 4 and use that for CPU
@@petersrensen5972 I did measure it and it was fine (didn't make it on the video though). It might be an easy place to start. Good idea.
@@NoelsRetroLab My first thought indeed when I saw you jumping to the conclusions so early. I'd use a diode tester to see if there is a hard short from this pin to GND and +5V rail, in both directions. This sometimes gives some insight into what exactly could have gone wrong - is it an ESD protection diode within the chip, or perhaps a passive component somewhere in between that and the CPU?
Hey Noel, there looks to be very much crusty old rosin flux around the asic which sometimes can go conductive and corrosive. Try to clean all that old stuff off the board and check again. I had this happen just with standard throughhole cmos chips and that ASICS pins are less than 1/2 a mm distance to each other so this can be a potential problem.
Hmm... I hadn't thought of that, but it's true that there's a fair amount of flux residue. I'll try cleaning it see if that helps. I'm also toying with the idea of generating the 4 MHz clock signal myself in an external circuit. But I'd have to do it just right for it to work, so I might revisit this console after all. Thanks for the tip.
I have the same problem and by doing this I have managed to recover sound! It's more than I expected. Thank you
I just fixed mine because of this comment! Thank you so much!
Great video, and I really like the way you diagnose - real analysis there. So the way the clock is generated is rather similar to the Speccy's ULA.
Thank you! Yes, that's very similar to the Spectrum ULA, although I believe the signal gets amplified a bit before being fed into the ULA, right? It doesn't go straight from the oscillator. It's the same with the regular Amstrad CPC range.
You can think as this chip as a super ula on steroids. It also provides all the superior gfx capabilities of the console
I dont mean to be offtopic but does anybody know of a tool to get back into an Instagram account??
I somehow forgot the account password. I would appreciate any tips you can give me.
@Nicholas Brock instablaster ;)
@Prince Kameron thanks so much for your reply. I found the site on google and Im in the hacking process atm.
Seems to take a while so I will get back to you later with my results.
Just after waking up in the morning and before the first cup of coffee, you could almost mistake it for a ZX Spectrum board 😀
I have a working GX4000 myself. It's a nice little machine. It's such a pity that the ASIC cannot be replaced with a clone chip like the ZX Spectrum ULA can. As far as I know, a FPGA implementation of the GX4000 doesn't exist. Sadly, GX4000 machines will stop working sooner o later, and the logic of the system will be lost.
Anyway, it was great to follow your inspection of the board, even if it was not possible to repair. Thanks!!
Hopefully someone will eventually reverse engineer the ASIC and we'll have a way to make new ones.
Fantastic analysis, Noel.
Glad you enjoyed it 😃👍
The output from that 39.9 mhz clock module should be 5v pk-pk so i don't think it was right.
Really? I forget the details but often times the signal out of those oscillator circuits is only about 1-2V peak to peak or so. I don't know about this one specifically though. I'll compare it when I get a working GX4000.
@@NoelsRetroLab Pure crystals (with 2 pins) are usually driven at 1 to 2V, because that's a good voltage for the crystal to operate on. On the other hand, crystal oscillators (4-pin boxes, either square like here or rectangular) have the driving electronics built-in, and are supposed to output a TTL- or CMOS-level signal at 5V.
@@NoelsRetroLab These are TTL level signals, so you should see 3.5 V at least. You may see less if the clock is close to the bandwidth limit of your 'scope. The ground point is important at frequencies like this. Use a ground as close to the oscillator as you can. Try to avoid using the pigtail clip lead - a spring-style contact is much better.
@@Derundurel Thanks for the tip. I was familiar with the spring-style contact, but I looked into it and what a difference it can make in the right situation! Off to look for one of those contacts for my probes!
@@NoelsRetroLab 99% of these are XTAL's with TTL output. 39MHz must be visible even on cheap osciloscopes. But on other hand these devices are very sturdy- never seen them dead. The big chip can be bad and load xtal.
Could you dig this back up and create your own clock for the Z80 and see if that is the only component broken on the gate array? Would be very cool to see this machine running.
Really good video! Thanks!
8:37 they were craaaazy ... guess it was cheaper to use linear regulators than buck converters for the power input stage. 15:32 wouldn't think the oscillator is fine - if it was a crystal, you could expect something like this but a crystal oscillator has a very nice rectangular signal between 0 and 5V. This looks like you have limitted your bandwidth of your oscilloscope - or the oscillator is almost dead. But strange that this 1MHz looked so fine.
As the 1MHz looks fine, I am confident that Noel just enabled the 20MHz bandwidth limit option. 40MHz is severely hampered by it, but 1MHz passes it just fine.
1) upgrade the PSU to support the C4cpc cart (without a new psu you will brick the machine)
2) get the c4cpc cart :)
3)save money instead of silly cart prices
Sounds like a good plan! Love the C4CPC 😃
@@NoelsRetroLab not overly impressed by many cpc games, as many were lazy spectrum ports. just an 8 bit I had to have
@@fogvarious2478 That's very true. There are some gems out there though. And the modern games being made for the CPC are amazing. They show how great they could have been back in the day!
Excellent vid as always, thanks for making it.
Another Amstrad falls victim to ASIC failure 😔
Still it's very educational to see your troubleshooting process. I really need to get one of those current limiting bech PSUs. I agree with you on preferring to work on older machines with through hole components - much easier to debug and repair!
Looks like amstrad inadvertently released the first retro console.
So true!! 🤣🤣🤣
Since the cartridge port is so close to the gate array, could mechanical strain from inserting and removing cartridges broken a solder joint? Maybe a few passes of hot air to re-flow the solder on the pins nearest to the cartridge port could help with any loose or intermittent connections?
It could be, but since I'm definitely seeing issues in the clock circuit, you'd think that just checking the inputs, VCC, and ground pins would be fine. But who knows with these all-in-one chips!
uhm, i must ask.
Why you didn't inspected board on the bottom? And why you didn't removed this gunk on gatearray's leads? Looking at video even in 360p i can see there either sticky liquid (like cocacola after seasoning xD) or solder flux.
I have no competence in electronic, but I find it very well done.
Well, all that big ass chip probably does in this case is to divide the clock from the oscillator down to 1MHz, with a chain of flip-flops, and tap the chain along the way in multiple places to derive the 4MHz and 16MHz or whatever else it needs. It should be very easy to throw something together to divide CLK16 down to 4MHz and feed it to the board. Assuming of course that the only thing that's broken from that chip is the CLK4 output, which is not necessarily very likely... still it might be fun to try it.
Yeah, definitely want to try it. But I wanted to try on a working GX4000 first to ensure my clock divider works, and then try it on the faulty one (and like you said, I fully expect it not to work because I'm sure something else is broken).
Quick comments. the Amiga 1000 came out in 1985, along the Atari 520 ST. The Famicom came out in Japan in july 1983 (almost a full year before the first CPC464). The GX4000 also came out 2 years after the Sega Megadrive.
Yes, you're right. I always forget about the Amiga 1000. Silly numbering system! 😃
It's well known to me, I bought one new along with Switchblade which is awesome on the GX4000
Nice one! Maybe you could try and make a 4Mhz clock signal by doubling the 1mhz clock two times, using external circuitry.
It did cross my mind, but I was afraid that unless I generate it exactly the same way the GA does, it might be out of sync with the 1Mhz and 16MHz signals that come out of the GA. Also, I wouldn't be surprised if other parts of the GA are faulty as well, so even if I did it correctly, it might not work. Still, it could be a fun mini-project.
@@NoelsRetroLab Yep only just seen this video but if it doesn't generate the correct 1Mhz signal other parts of that chip may also be faulty. I am also thinking something else may be wrong to cause the ASIC and the RAM be faulty in the first place too.
@@p166mx Yes, you're right. Chances are someone plugged in the wrong voltage. Or even the correct power supply that is known to have problems (the GX4000 has some really weird power circuitry to allow two kinds of power sources). I'm getting another GX4000 soon, so I might look at the signal and try to recreate it as a fun project, but I doubt it'll work. I'm sure there are other issues like you said.
The side-by-side power schematic and circuit board spit screen was brilliant! It’s great to see that not every project is a success ... very real, and that’s very encouraging! As the episode went on I was thinking “Not the gate array, not the gate array” 😬🙂. Perhaps sometime you could talk about the role of some of the discrete components on a board like a particular resistor or capacitor? I often wish I’d pursued the electronics I’d learned in high school (where I only learned the basics of resistance-capacitance circuits). Thanks again for the great videos!
Glad you liked it! Yeah, the GA/ULA is always biggest fear when dealing with GX4000 or some of the latter ZX Spectrums.
You're right that I tend to overlook the "small" components. I'll make a point to talk about them sometime. It's never too late to learn electronics. I studied Computer Systems Engineering, but I didn't learn "real" electronics until a few years ago when my Amstrad died and decided to repair it :-)
It still doesn't mean the gate array is dead, it is connected to so many functions, it could be one of the reset lines are being pulled low. Or a CAS/RAS line is shorted, the gate array might just be in an unknown state.
could you not have tapped the 1MHz signal and used a clock multiplication circuit to drive the cpu?
that's exactly what I was thinking too. Especially the older these things are the fewer spare parts we can find for them and have to make do and mend...
what would happen if you build some opamp or logic hate clock multiplyer and substitute 4mhz clock or inject exyernal 4mhz clock from another clock oscilator+transistor circuit? or replace z 80 to check if it makes clock shorted
Also there's question wether 4mhz clk going out from generator is "flat" or is it "grounded"?
Short to ground isn't as hard to troubleshoot, especially when you have literally a straight line from generator to cpu, challenge may be to lift appopriate lead of generator, but this can be easily circumvented by cutting lead whereever it's comfortable to do so and measuring both directions.
These SID Sounds are Amazing in every Video. Where can i get them?
I seem to be addicted to your channel (until I get to Commodore videos, lol.). interestingly, I think the CPC Plus's / GX4000 were the only European systems to have hardware scrolling and sprites, however it was way too late as you mentioned too.
Thanks! Not a Commodore fan, or just too many Commodore videos out there already? I don't think the Plus had any hardware scrolling beyond the CPCs abilities, did it? Not 100% sure, that's why I'm asking. Sprites for sure, but it's amazing that the TMS9918/9929 already had hardware sprites all the way back in the late 70s! So MSX systems using it had hardware sprites, and so did the TI99 (not really a EU system) and the Colecovision.
@@NoelsRetroLab Cartridge games like Navy Seals, Robocop 2 and Dick Tracy do appear to be very smooth when scrolling the screen. At 50 FPS too.
Still an interesting look around a lesser-known system...
what makes these chips die? I have been watching you troubleshoot and uncover failures in these machines, just wondering what is making these chips fail? Is there a ghost in the circuitry?
HI Noel, I know it is an older video, but I have to ask: Are you sure it is a crystal and not a crystal oscillator? The ´1.2mV sinusoidal signal on the gate array input doesn't look good at all. I guess the oscillator is just dead and if you replace it you'll find a 5V square wave signal with the desired frequency there at pin 2 of the ASIC.
The gate array has a blob on the package...?
Did you say the big chip replaced separate chips from an earlier modal? If you know what it replaced you might be able to find small SMD replacements of the separate parts and maybe something like the RP2040 with PIO and get a small pcb built with castellated holes :p
Not exactly. The GX4000 was always designed that way. What I meant is that it's an evolution of the CPC board, and in that board it did use several discrete chips (plus some larger custom ones) and here they're all combined. So it would be pretty hard to replace exactly and besides, who knows what else is broken there. But I still want to try to feed it the clock signal by hand and see if that changes anything. It would be a fun experiment.
I just picked one up that only came with a 5v dc adapter. I think I may have accidentally plugged it into the 11v jack. Suffice to say it didn't work. Is it possible I damaged it by not supplying enough voltage?
No, no. Don't worry. If you put less voltage it simply won't work, but nothing will be damaged as far as I know.
Pardon my ignorance, but couldn't you just cut the clock line and feed it an external clock?
Interesting but isn't their cart lineup a drop in a bucket compared to their disk lineup?
Good video :-) Is an FPGA an option to fix this? Opps I'm a bit late to the party I see. I also love my Adjustable power supply, I am using it in making the Ben Eater 8-bit CPU :-) I keep the amps just above what it needs and I have seen the voltage drop sometimes, so I look for my mistake :-D
No FPGA implementation as far as I can tell unfortunately. Hopefully in the next few years those will become cheaper and more common for more computers. Ben Eater is AWESOME. Love his videos and explanations!
I bought one of these when they came out, for about £50. Since it shares the PCB with the full computer version, I thought it would be possible to attach a keyboard, and use the CPC+ cartridge instead of the game one to get a CPC+ that can feed a SCART output, and not need to buy a monitor. I never got round to trying.
I believe you're right and it's possible, but I haven't tried it either. The Plus range actually has a really nice keyboard, so might as well use those if they're available (although they're getting quite expensive).
If Amstrad had released the plus range from day one, it would have been a C64 killer.
Wow, had they released the Plus range instead of the CPC it would have been indeed a game changer! Even if they had released it as late as 1987 it would have still been a contender.
@@NoelsRetroLab especially if they could have implemented a 6 or 8 Mhz CPU version.
The Amstrad Plus computers and GX4000 console could be something really good in 1986, but in 1990 was too late. About the games, all the carts got small roms based games, they could do bigger games like Defender of the Crown, Last Ninja Remix or Maniac Mansion.
Is that easy to get memory chips? What is the model number?
You could try to generate the cpu click from an external circuit
Yeah, the problem is that it probably needs to be syncronized to the other clocks in the system. I just got another GX4000, so I'm hoping to do some experimenting with that and see if I learn something.
@@NoelsRetroLab it could be syncrohized if you feed it from another clock that is already working.
@@anvz6 Right. I thought you meant generate it completely externally. Yes, that's what I'm hoping to do. The reason I didn't try is that I'd be surprised if that's the only thing failing in that ASIC, so setting it up and have it not work wouldn't tell me much. But now I have a working console, so I should be able to test that circuit and then try it in this one. Should be fun!
@@NoelsRetroLab Yes, there may be some other fults on this chip... but I think it could be a good idea to "patch" the existing chip only bacause it smd and as so much pins...
Great video. ASIC's are great...except when they aren't made any more. I wonder if there is an FPGA replacement?
I think ASICs are great for hardware designers, and are the curse of people like me, trying to repair systems. I don't think this ASIC was ever reverse-engineered, so no luck there. Hopefully someone will at some point.
what if you lift the pin, then x4 the good 1mhz clock?
Yes, definitely still liked the video. But that is the reality of it, can't fix them all. Case looks great. Hopefully you can find an inexpensive board to replace that one and make a really nice console.
Here, have a "like" for producing an honest - and still interesting - video!
While I never had the chance to see this console in person I now know its insides and got some osci action, too!
;-)
Thanks! Yeah, it's the reality of repairing stuff. Sometimes you win, sometimes you don't...
@@NoelsRetroLab My pleasure - credit where credit is due!
Where I live we have a slightly different saying: "Sometimes you lose, sometimes the others win!" ;-)
That was still a cool look at a rare video game console. You can't even find those things in the US.
Be interesting to see if it was just that one issue on that chip not giv your a clock signal by generating the signal with a function generator see if it would of worked like do a doc brown hack in back to the future 3 were he sorts out an issue with a chip that failed
Haha, yeah, I would like to do that in a future video. The problem is that it's not quite that simple because the clock signal needs to be synchronized with the other clock signals in the board. So I'm hoping to first get a working GX4000, see what a healthy clock signal is supposed to be like, and then try to make a simple circuit to generate it. I suspect something else will be faulty in that big ASIC, but hey, it sounds like a fun project anyway :-)
Noel's Retro Lab be a fun video worth watching even if at the end still didn’t work.
At least you have (probably) working Z80 spare 😁
It sometimes happens to not be able to fix something but at least you know whats wrong with it.
i kinda disapointed that you didn't pull the cpu to see if it was pulling down your clock signal #4 at the gate array.
Yes i have that game console only ever seen one game for it burning rubber. I ordered a flash cart from France to play the other games with out spending a fortune on cartridges
use chipquik (SMD4.5NL). Those QFP's come off easily with just a simple soldering iron.
What exactly is Chipquik? I was looking at their site and I'm still not 100% sure. It's a low-melting point solder? I've replaced those ASICS before with a hot air gun and some care, but it wasn't the easiest job.
My UK spec GX4000 only has 5v power in, with no 11v.
I think I prefer the CPC 6128 as a general all rounder for the series.
Are you sure about only 5V in? I wasn't aware they had multiple board versions. And yes, totally in agreement about the CPC 6128 :-) Or, if you press me, I guess the 6128 Plus is pretty good too.
I had no idea it was so small.
Este tío es increíble lo repara todo 🤦🏻🤦🏻🤦🏻
You forgot to mention the Amiga 1000 that was released in 1986 and preceded the A500. :) Amstrad actually didn't stop there: three years later, they released the Amstrad Mega PC: which combined a 386 SX and a Sega Mega Drive in one unit.
True, I tend to forget about the Amiga 1000 in general but people keep reminding me it DID come before the 500 (bad numbering, Commodore! :-) ). And also true about the Mega PC, although it wasn't really a console by Amstrad. But yeah, between that and the GX4000, I'm sure they swore never to touch consoles again!
ive only seen these in pictures...never seen anyone handle one...let alone tear it down and (try to) repair it!
we have a running joke that the GX4000 is the worst console ever made... but its not 'bad'...it was just late...
2 screws holding the thing together... trust Mr Amstrad!
Yeah, it was waaaaaaay too late. But the case design and attachments were top notch! Especially considering that the Amstrad CPC 6128 is a nightmare of screws.
I remember those being on store shelves and thinking, even as a kid, that can't compete with the NES and Mastersystem surely? I was right. My first machine was an Amstrad CPC 464 which I have very fond memories of and this console was not a whole lot better really.
You were not wrong, even as a kid 😃
Always interesting regardless of the outcome. Are you now on the search for another GX4000?
Thanks. I actually have a non-working GX4000 coming after the virus lockdown is over. Hopefully between the two I can get a working one... unless is that annoying ASIC chip again! :-)
@@NoelsRetroLab Well let's hope it's not that problem again. Looking forward to more videos - highly informative and professional. One of the best retro repair channels out there.
get a 4mhz oscillator chip? i guess microchip has some. you can then devide it and get the 1mhz signal synchronized. a lot of work but possible.
Yes, I want to try that sometime. I'm afraid of two things: a) The clock signal generated isn't in perfect sync with the others in the GA and b) there are more things wrong with the GA than the clock signal. But even so it seems like a fun project, so I'll probably try it anyway :-)
The circuit board look like it's from the 80s. By 1990, I expect to see more surface mount coming in.... Not saying this is "good" or "bad", just an observation.
Are you sure the machine is coming out of reset?
I didn't think to check that, but I see the GA has a /RESET line. I doubt it because the other clock outputs were working, but who knows. I'll have a look. Thanks!
Cool Design! , Except front ports
You don't like the front ports? They seem handy and functional.
It's a bit odd that the Gx4000 is hardware compatible with normal Amstrad games which were still being made in 1990, but you just had no way to play them on the thing.
Next new tool, thermal camera. More times than not it shows you the exact problem in less than a second if you suspect a short.
The Plus machines were what the 1980's Amstrad CPCs should have been in the first place.
I'm the first one to say that the Plus machines were MANY years late, but I don't think it was realistic to release them in 1984. I think if they had released them in 86-87, they would have been a real contender though.
we need replacement Gate Array for these and CPC+ line of computers :( FPGA or something :/
Absolutely. Although making it fit the footprint of that ASIC would be an added challenge!
@@NoelsRetroLab older gate array has been documented in CPCwiki
Check D132, in before said watch video or read comments and yes I know that this is an old video, but I am new to viewing this video.
I found one of these in a bin, and then sold it for a few quid at car boot sale.
I got one and it went to the bin.
I had a broken Unipolbrit 2086 from Poland which is a bit rare Spectrum clone. It looked like the gate array was bust. Fortunately it could be fixed by using the gate array from a Timex 2048 which is much less rare. There was no other option than to sacrifice a working computer because of the gate array.
Yes, it requires skill to (un)solder a big smd chip, skills I don't have😉
Interesting that they would both use the same gate array! I guess that was a true clone.
I have desoldered and transplanted the ASIC chip on a CPC Plus before, so it's definitely doable. When I get back into this project I might have to end up doing that.
@@NoelsRetroLab The Unipolbrit 2086 was basically a Timex 2068 which was slightly modified by the Polish import company. The amazing thing is that both the Timex 2048, which is very much a Spectrum clone, used the same gate array as the Timex 2068, with is a much souped up "Super Spectrum".
A shame you weren't able to repair it, but ASICs have been the death of many a machine. Perhaps one day the decapping of the ASICs will reap benefits and there will be a drop in replacement.
Interestingly enough, the 8-bit market is alive and kicking to this day, so that alone renders the "too little, too late", rehashed, false argument invalid.
What killed the 6128 Plus/GX4000 wasn't their so-called "belated" release, it was their backwards compatibility with the original CPC, which lead to the lack of dedicated software for the new systems, as developers back then resorted to porting carbon copy CPC games and calling it a day.
Yet, look at Ghosts 'n' Goblins and Sonic the Hedgehog on the 6128 Plus now...plus many more games to come.
The 8-bit market is dead, dead, dead. There's a thriving community around it, but people aren't remotely making money to survive, which is what would keep a console alive. So yes, you can make amazing stuff with it, but at the time nobody wanted to spend their money or effort on a clearly outdate console.
@@NoelsRetroLab The 8-bit market is "dead", yet 8-bit console re-releases in the form of mini-consoles, classics etc. sell like hot cakes. If people are buying the hardware, you can be sure they'd buy the software too. The CPC is seeing a resurgence of physical, commercial releases.
People would pay good money for the CPC's contemporary releases (I know I would), it's just that nowadays many good developers like BG are doing it for a hobby, instead of a profession. Said group even turns down crowdfunding offers...so don't be too sure about the "dead" part.
"clearly outdated console"
😄
Again you're making the mistake of comparing it with the 16-bit consoles, when the 8-bit market was still pretty much alive when the console was released, even as late as 1992 the market was alive, two whole years after the 6128 Plus/GX4000 were released. The c64-based console and the Master System II were released in '92, for crying out loud...
No, what killed the 6128 Plus/GX4000 was the lack of dedicated software, not the "dying" 8-bit market.
I was there, I witnessed it all. This system would have been a hit, had it not been riddled with lazy CPC ports.
@@NoelsRetroLab I bought one brand new in 1990 for 99 quid seemed a deal at the time but I was too young to realise the thing was doomed. I got a handful of games and eventually bought a 386 later. Look forward to seeing more gx videos later
Many years ago, I got a brand new GX4000 for £5. :). It flopped a bit.
it's a huge shame that many of these modern chips aren't socketed, to make up for their complexity, probably due to cost. combining the chips could actually be a great idea, because you'd have to make less socketed slots, you could only have 1 - 3 of them and cover all the complicated stuff. but no one really does that, and nowadays, in laptops, it's even worse. you can't even swap the CPU (which basically holds everything nowadays) in any of them, lest it's some bulky thing with a desktop cpu, and many of those have some complicated GPU that isn't socketed. and even if you had those socketed, if they're prone to failure, they always will be, and no one is able to make proper replacements.
maybe some smaller companies would be willing to make somewhat capable computers (just something to play 480p x264 on or something) with lots of expandability and very continuous support. they could iterate on the design of their more complicated, socketed chips if they're flawed while keeping the same form factor. but i'm not sure if something this kindhearted is sustainable. i hope it is.
Right. That was definitely their way of reducing costs. That was a trend across the board with all motherboards in the 80s and 90s.
Muy interesante, como siempre
Gracias!
It looks like they used the latest technology to produce an obsolete game console. I'm sure someone at Amstrad went to work every day knowing this was a terrible idea.
Now that's a t-shirt!
Actually the SNES is an 8bit console. It's like saying that the Megadrive is 32bit :-P
guy had a load of them on ebay , he bought as bankrupt stock.. minty new, and cheap.. wasn't worth fixing BUT the PSU wasn't powerful enough for c4cpc cart (and would brick machine) .. so had to replace that.
Right. I've seen some bad GX4000 with blown regulators because of it.
@@NoelsRetroLab thankfully I saw the warning, got new PSU's from cpc electronics (ironically enough) .. the guy making the c4cpc sends them quick / good lad.. based in france.
This kind of situation right here with those damned chips that are next to impossible to replace without a full on arsenal of high-end equipment is what had killed my interest in analog electronics and repairs, funny to see that even you feel the same about them!
Yeah, even if you can replace it (which I've done it and it's a chore), there's not much to think about. It's either that chip that's faulty or not. The more discrete chips computers have, the more fun it is to repair them! 😃
Well I suppose you can't really call the Amstrad emailer E3 a console can you?
nice video!!! i was in a same puzzle with a tv set 15 years ago.... no video output from a video chip ... i changed it and the result was the same - then i changed 3 small electrolytic capacitors around chip and i saw Output ... i saw an electrolytic capacitor near your chip maybe if u change it u will see picture ty for your exellent video
If you can get the chip I would gladly replace it for you for free. I'm in the US. I have all of the surface mount soldering equipment. I'm an electronic technician and I work with smc all the time.
I got one of those back in '98/99 for £5, in box with everything. Played on it once. You're not missing anything. It's more useful as parts. I ended up giving it to a friends 5 year old son.
Don't think he played on it much either. LOL.
Yeah, and it's too bad. It's one of those "what could have been" cases had Amstrad decided to put some effort into it.
@@NoelsRetroLab It's Amstrad cashing in and cutting corners, again.
@@NoelsRetroLab I think you are spot on with your "too little, too late" remark in the video: What is the point of bringing a Z80 based console on the market in 1990? If I wanted a Z80 console at that time, I would have bought the established Sega Master System (Japan: 1985, US: 1986, EU: 1987), just because I knew there were a lot of cool games available, and I could trade games with friends. Does this console have a unique selling point over the Master System?
I don't think "some effort" would have been enough to save that device, except if for you "some effort" means being a whopping 3 years earlier to the market.
@@tw11tube Seriously! I don't know how it ever got past the planning stage. Or maybe it was initially pitched and designed as a 16-bit console, and then, somewhere along the way, someone cut costs and stuck a Z80 on it. 🤪
so basicaly with 2000s stuff we are doomed... really i would trade a maybe bulkier design if it ment more repairable =(
nice try 👍
Ben Hack is the guy to know how to fix this :)
This is one of the worst consoles. It is just what the Amstrad should have been from the get go. I am a big Amstrad fan, but it just was never made for gaming. Only a handful of programmers could actually get anything out of it. That people can do it today does not really count. Anything related to moving graphics was painfully slow. On top of that it seems the Spectrum gets more out of it's AY chip than the Amstrad.