I'm very late to the party but, a lot of Z80 based computers used to put their ROM into higher memory addresses and then use hardware to block the CPU from reading anything but zeros (NOP) after a reset until the address bus reached the first address of the ROM. Or putting it another way, after reset the CPU would execute nothing but NOP (no operation) instructions until it reached the memory address where the ROM is located. Once it hit that address, the data bus would "open for business" until the next reset. This was a work-around for the fact that the Z80 CPU would always start running instructions from memory location zero after a reset but CP/M programs were required to run in low memory with some program data starting at 0 and the program itself starting at location 256 (0x100H). While it certainly was possible to put a ROM at location 0, the expectation that programs would need to be able to run starting at location 0 was so common that some way of making the low memory available was almost always available, whether that was by the above technique or by memory "paging" to replace the low memory ROM with RAM.
When you have repaired a ZX Spectrum you made a child somewhere very happy again! Good work! Carry on with it. Oh, wait a minute! That CHILD is in his fifties by now.
Sinclair got stung by using the fault upper RAM chips, once the supply ran out he had to use fully working upper RAM chips! I built a circuit that allowed me to page in the other 32k RAM, was a lot of fun. :)
That was fascinating seeing a replacement be made for the ZX8401. I've never considered recreating simpler chips like that and was worried they'd be lost to time. Really awesome video!
I've seen the same kind of behaviour with the blanking pin on other systems - the Archimedes for example can give a very strange picture unless that pin is pulled high enough. It's going to relate to the sync signal - my guess is the sync is slightly different between those two ULAs. Since the blanking pin (at the TV end) supports "punching a hole" in the composite signal, with the RGB output (from what I understand), a difference in composite signal used for the sync, could affect how the output looks when using RGB. I guess what's happening is one ULA has cleaner sync than the other, and the one with slighly worse sync needs a more definitive "punch" using that blanking pin. On the archimedes for example you can rarely get anything other than a white box (with black border), until you use pin 16 to punch away the composite signal with the RGB. On other systems you just get the TV switching entirely into composite - I've never understood why, but it has to be related to the sync in each case I think. Subbed btw! EDIT: Thinking about the Archimedes example - it makes sense to me now (the white screen) - The Arc outputs only H or V sync (which cables combine), but that sync signal obviously lacks video (hence the white when we aren't "punching" through the sync source with the RGB picture info). The RGB has to punch in there or there is nothing for the TV to display. Cant believe how long that took me to work out lol
Great explanation! That makes a lot of sense! I did look at the sync signal but I forgot it wasn't really used for RGB out. It uses synch on composite instead, and that's what I should have looked at. Very interesting to hear about the Archimedes results. That's one computer I haven't touched yet but it's on my want list. Someday... 😀
@@NoelsRetroLab If you have a LOT of time and patience, I could send you a severely battery damaged Archimedes A3000. They are very very time consuming though.
@@GadgetUK164 Oh man, so tempted! But for my first Archimedes I better wait to get one in working condition, even if cosmetically it needs some work. It can be pretty tough to repair something without ever having seen one working in the first place 😀 If you ever have a working one for sale that is not super expensive (wishful thinking, I know), let me know!
@@NoelsRetroLab Is it really? I may have misinterpreted what you aimed at, but I tend to see it the other way. You could still build "anytning" using the TTL standard that was developed in the early 1960s. Sure, these are surface mounted CMOS versions. But surface mount was used even when Sinclair produced computers (especially in Japan) and CMOS has actually been around since the 1960s (although being much slower than today).
I've seen that same kind of image issue with different ULA's via RGB creating the artifact at the end of words. I went through several revision of RGB cables before I eventually figured out it was something that's different with 2 seperate ULAs, just three weeks apart. Someone mentioned you need voltage on the signal. Then I tried different TV's and again saw different issues with the same ULA that had artifacts. One one TV (LCD) it had no issues, also no issues on a CRT. Two other LCD's had the issue. So it looks like it's a combination of things. Great to see that you solved it on the same TV with the added voltage.
Hi Noel! You rock my friend! Such good quality, time and care put in each video released. I cant imagine the next step when some patreon funds arrive to you allowing more content, more time spent and other things! This is easily the best channel on YT for retro happiness. Regards from Portugal, and stay safe, the numbers in Spain about covid are not good
Many thanks for this new video Noel, that was excellent as usual! I'm not into Sinclair machines at all, yet I was really captivated by the part explaining the memory addressing and I was taken aback by learning how Sinclair solved the "48K / 64 K" equation by using cheap defective RAM chips that actually helped reduce the cost further, that was just astounding!! Lastly I really enjoyed the 'hands-on' part on SMD soldering, very useful demonstration and practical tips, thank you very much! :-)
You're welcome. Glad you liked the video! Yeah, sometimes the strangest (and interesting) decisions go into the architecture of these computers. It's all part of their unique quirks 😀
Glad that helped. To be honest, I usually transfer them over to the Mac because it's my main computer, and there I use HexFiend, which does a binary diff, but has a nice GUI and shows the differences very clearly.
Yeah, it's great! There are a couple EPROM models that I wanted to use that it doesn't support, but it's been great for the most part (in spite of the horrible, horrible UI design).
Your SMD soldering technique is pretty good. Liquid flux is fine ( we used no clean flux pens at work which are liquid ). Use more flux. Once you've secured the corners try drag soldering - You apply solder and the tip to the SMD component and swipe down the legs in a smooth movement.
Thanks! I've actually started trying the paste flux and I like it much better! Probably because I can have more of it. There's going to be a fair amount of SMD soldering in the next TI video in a couple of weeks!
Thanks! They hold it nicely in place AND you're guaranteed to solder the straight. Not like I would have ever soldered pin headers accidentally at a angle.... not me, nooo... 😃
"aligning the SMD chip may take a while" (he drops the chip on the PCB and it magically aligns perfectly on it by itself) LOL Great video as usual! (How do you erase those eeproms?)
The ones we see in this video are the UV-erasable type - you can recognise those by the quartz window; If one is _VERY_ patient they can be erased by exposing them to direct sunlight over a long period, but normally one would use a dedicated device with an UV light source. You can see examples of such erasers here: duckduckgo.com/?t=ffnt&q=uv+eprom+eraser&atb=v175-1&iax=images&ia=images
Hahaha, yes, that was the ONE time I dropped the chip and it was in perfect alignment! I had recorded the comment before selecting the video, so that wasn't he best example. You know how fiddly it can be to get them just right. Edit: And Berty is right. I have one of those UV boxes and in 15-20 minutes they're fully erased.
I had to rewind the video a bit as I was wondering "is that chip magnetic or what"! I have a bucket of those UV-erasable eeproms but I do not have an UV light to erase them. I tried sunlight but did not work - not long enough I suppose. Also, we do not get much sunshine here in the UK... :(
True. I still have it around. Maybe I'll try it just for fun. But then again, the EPROM was screwing up the memory tests somehow, so there has to be something slightly off about it.
Have you heard of the ZX Harlequin? It’s a reverse engineered spectrum kit that doesn’t need a ULA. The issue is they can’t sell the rom with it. I was thinking of just getting an eprom and trying to use an image off the internet to program it with. Not sure it’s the same one but my local electronics shop has EPROMs and the mini pro for sale.
Yes, I have a couple kits of those still to assemble. I'll probably cover them one day and get in the details of what's different about them. I suspect it'll be a bit like the Inves Spectrum+, since they had to reimplement all the ULA logic in a different way (in the case of the Harlequin all with discrete logic!).
Great repair! It would have been nice to see whether the databus cleared up after changing the ROM. I was working on a board with a similar problem, dirty address and databus and it didn't occur to me the ROM could be corrupted, I did flash a ROM with the data read off the old one and ofcourse it showed no improvement.
I suspect those ULA's came from different batches, different tolerances, etc. In other words, they aren't 100% exact duplicates. The "faulty" one probably just crosses (or gets close to) some threshold that causes it to be slightly inferior to the first one.
That makes a lot of sense. Now I'm very curious whether before this accident it behaved like the other one, it it's something that happened after the expansion was plugged in while it was on.
Good point. I'm going to guess that it's exactly the same and that the difference comes elsewhere in the signals. I mean, that pin is JUST tied to the battery. Although now that I think about it, if the battery isn't hooked up, then it's not just ground or high, it's an open circuit. I wonder if that causes the TV to interpret it in unexpected ways.
From what little reading I’ve done: On some 128k toastrack cables, with no battery, they use the composite signal to provide the switching voltage. Somehow relying on the tv electronics impedance to take it into the right range (or something) but this puts a nasty draw on the ULA, it can make it hot and increase likelihood of failure. Maybe you’re seeing that effect with your cable???
Great video! Interesting things to find out. Regarding the bad image that the original ULA provided, it may be related to the blanking signal. Those RGB cables use the composite signal as a blanking signal. But since that signal does not have a constant voltage, it is helped with a capacitor. Maybe the original ULA provides a slightly lower voltage or more unstable voltage for that, therefore the TV switches from composite to RGB randomly. I've experienced a close situation with two Toastracks: one of them had an annoying image flicker every 2-3 seconds the other behaved the same but this happened a few times each minute. All dissapeared when I have provided the proper blanking signal to the DIN connnector.
That type of RGB video uses the composite video signal to sync the image. Is that what you mean? I think you're right and it's something related to that (see coment below by GadgetUK164).
@@biv-8515 The person I bought it from is out of stock at the moment but he's getting more. You can see his Ebay listing here: www.ebay.es/itm/224111543392 Send him a message and he'll reserve one for you.
some people do a similar thing with the Amstrad cpc using the lum as sync and the S sync as blanking which causes ghosting, adding a capacitor will remedy the ghosting but causes the the intermittent loosing of signal (in my experience anyway) .. with the +2 & +2A you can use the 12v feed with a suitable resistor fitted on the motherboard to bring the voltage down when combined with the tv's 75 ohm impendence gives the suitable voltage (somewhere over 1 volt if memory serves) +2 has some different configuration options via the links ie RGB levels, brightness signal ,audio sync, composite ... not sure about the toast rack as i have not managed to buy one yet as the prices are high 😒
Did you get to open that SCART cable to look at how exactly the battery is being used? I never saw one like this, but it is usual practice in high-quality cables to boost the signals through repeaters, so to compensate for the signal loss of the cable itself. This is often done in long-length HDMI cables, for instance. Usually it is done by tapping the 5V line of the output connector, but then again, the same technique could be used to fill out your missing pin without a battery hanging from the cable. That they chose to use a battery suggests they thought the power rail (and perhaps the ULA itself) couldn't drive the signal hard enough across the length of the cable. If you open it and find a transceiver or repeater of some sort inside, there's your answer.
I haven't, but usually there's just a resistor and then sends the voltage straight over. I've never seen anything fancier than that in these kind of SCART cables.
That battery SCART cable is just weird. I can only imagine they use a fairly high ohm resistor dropper to get the 9V to the low voltage required as a signal. But why not just use diode(s) from the TTL signals and capacitor as a charge pump to provide the missing composite/RGB signal switching voltage instead?
@@Drew-Dastardly Yeah, they need to drop it down to 3V, so I'm sure they do exactly that. They could have used an RS232 as well. I never thought of trying to use some of the TTL signals to charge a capacitor. I wonder if that's just too bulky to be contained in the cable. For my own 128K, I'm thinking of hooking up 5V to the unused pin in the monitor connector and using that. The only downside is that I can't use the same cable with other 128Ks (and if I forget, then things like this happen!).
Another great video. Great to see how modern components can be used to replace parts that are no longer made and also the enthusiasm and love people have for these old computers. Are these kits for the ZX8401 available to purchase by the public or are there any Gerber files for the SMD pcb .
if you need one, no problem, I will make another bunch of them. The last fews I had were sold out to Zxrenew. Whatever, you also could do one by yourself with the schematic that Noel's shared. Or contact me by PM, still I keep one.
You know, I'm not sure. If you look at the UK board of the 128K there's some kind of coil there. I'm not sure if it's a ferrite core or what it is exactly. 2.bp.blogspot.com/-So4FZHf9CN0/WBd34G1cCqI/AAAAAAAAKR8/fmnMf4BCs7shLLI-0ZrteH2DE3nV4RWTgCLcB/s1600/UK%2Bvs%2BSpanish%2B128K%2Bv2.jpg
@Mr Guru Sorry for posting here, but I don't find your post about how to solder it with a hot air gun. I was looking for those videos that you commented about it, but all what I watched was with solder paste. I have not solder paste. Is it obligatory for this technique? thanks again
@Mr Guru I have tried it... the result was not very nice, I will try it one more time, but... it seems to me that better i have not enough skills for this way. I operated it as follow: 1 add flux, 2 add iron to the pads, 3 more flux, 4 hot air gun... not nice how it had finished. Maybe if I fix only one pad to the chip... but the problem is that left too much solder iron on the pads. Need to learn more from Noel :)
14:12 I have a feeling that possibly the VCC pin (14) of the NAND chip wasn't quite soldered. When you removed the chip, you could see the solder on almost all of the pins get disturbed, but the solder for pin 14 barely changed. I think maybe the pin was floating just above the solder, and not quite making proper contact. Really neat vid, though, and great explanation of the circuit. Subbed! ;)
Good eyes! That's certainly a possibility. I did do a continuity check between each pin and the pad, but I suppose that if it was barely soldered, just putting the test lead on it would push it down and record continuity just while I was testing.
The issue with the blurry video output from the ULA could perhaps be due to slight timing differences between the colour clock and the RGB outputs. On some CRTs with RGB SCART input, and no voltage on SCART pin 16, they can tend to "overlap" the RGB signal over the Composite signal. I would have thought the SCART cable is using Composite Sync from the Speccy, though, so maybe not? (this can sometimes happen even without any voltage on pin 16. Some TVs just don't process the RGB input properly, as they either expect *only* the Composite Video pin to be hooked up, or RGB with voltage on pin 16.) If the SCART cable is using Composite Video for sync, then that could explain the difference. The timing difference could be causing a delay after the RGB is processed by the chroma encoder chip. I don't think it's a problem with the RGB output directly. As you say, it looks fine once you hooked up the battery. What confuses me is how you got a decent image before on that TV without a voltage on pin 16 in the first place. lol :p
@@NoelsRetroLab I still have problems with SMD soldering at times, even after quite a few years. Yep, I've definitely had that before when continuity tested fine, and it turned out the tiny amount of pressure from the probe was enough to make the connection. hehe I repaired a board for somebody about six months ago which I suspected had solder joints on a QFP that hadn't reflowed properly. I used a cheap USB "microscope" camera, and did the careful "poke test" on the chip pins... drive.google.com/file/d/1hbmWG1EPzY6ZHh2CwrFV3onNRpzhAR3U/view?usp=sharing I didn't add all of that flux paste either, btw. lol I used to use flux paste for many years, as it does tend to stick around for longer without boiling off. But now I use liquid flux for pretty much everything, and got a bit quicker with the dwell times of the iron. That clip is a good example of what often happens when a larger QFP chip is removed using hot air, and the pins get bent very slightly. The pins then no longer sit flat on the pads after replacement. I won't tell you who's board it was, as he'd probably kill me. lol
Thanks for the upload Noel. You could try and see if the bad 27256 will fully erase but I doubt it. I also have the same type of programmer which I converted to the 'A' version from the 'CS' version.
I doubt it too since it had some things that were interfering with the data or address bus. How did you upgrade it? That might be really relevant to me right now because that allows it to use higher voltage right? I'm trying to burn some 2732As for the TI99 right now, but those require 21V and they fail (in spite of being listed as being supported in the EPROM programmer). Edit: Never mind, I see the CS also supports 21V. But my more recent II Plus only goes to 18V. How annoying!
@@NoelsRetroLab at 20:27 the good EPROM was installed in the programmer offset (pin 1 of the EPROM is in socket position 2), and dumped (offset not detected as Check ID disabled, guessing this was done because the read of the good chip failed due to the offset). Bad EPROM is installed with no offset and dumped - imagine comparing the two read dumps will be different due to this (unless programmer software was made at Hogwarts?). Also, the new EPROM was programmed with no offset using the dump of the good EPROM, so don't see how this worked (will put this down to the magic of editing, the curse of continuity, and the desire to tell good story). Anyway... where I'm going with this is, as Andy says, you might be able to get away with erazing and reprograming that EPROM). Thanks for the vids - have been watching them all day.
just thinking out loud .. when the battery fixed the ghosting , and you got different results back then and now , could it be as simple as corrosion on the pins of the IC ? .. or if one is weaker than the other , are there any pull up/down resistors for that part of the ciruit ?
Could be. I don't recall seeing any obvious corrosion though. Rather puzzling, especially how the behavior changed (although that's consistent with your corrosion theory after pulling it in and out a couple of times).
@@IDPhotoMan Argh, it's a crime to have a CPC 6128 and no way to use it, don't you know? 😀If you can get some sort of RGB monitor, you can easily make a video cable (and you can feed it with standard 5V DC and 12V DC power supplies). If you need any help ping me on Discord.
@@IDPhotoMan you can use most old analogue TVs.. check out 'the 8-bit guy' channel he has a vid on converting the TV to RGB, and its quite simple.. you can then wire it directly into the 6128s monitor port... or if ya just want to test its working..you can get a black and white composite compatible signal by using one of the color pins and one of the sync pins on the 6128s monitor port. i did that back in the late 80s/early 90s and flipped 2 screens in ram, overlayed some audio and then recorded it to VHS! there was an article about it in one of the old Amstrad Action magazines..
I think the way addressing works on DRAM is more to do with the internal construction than price, size etc. From what I vaguely remember SRAM/ROMs with multiplexed buses did it by having the data and address bus share pins. Anyhow nice video as always.. interesting how much the 8 bits used gate arrays really. You would have thought in the numbers they were making they would have reduced them right down to a single chip pretty quickly.
Oh interesting, I assumed it was just a cost-saving issue and not some other reason. Now I'm going to have to look into it 😀 Eventually they did reduce them to just about a single chip like in the case of the Spectrum +2 or the Amstrad GX4000 (but I really dislike working with those because... there's not much to fix beyond blown RAM!).
Awesome to see that discrete SMD are small enough now that one can use them to replace such custom chips in almost the same size package. Would love to see what options are available if the video ULA was bad. Any way you can make a video on that? That would be very cool to get into.
Yes, isn't that cool? That's a very clear way of how technology has improved since the 80s. As for the ULA replacements, I'll make a note about it. In the meanwhile, have you read The ZX Spectrum ULA by Chris Smith? Excellent book. www.zxdesign.info/book/
@@NoelsRetroLab Thank you sir. I am an electronic tech/hobbiest and I am learning HDL and to make a new ULA that can replace the old one with possible enhancements would be great...I'm sure it's been done already though since the Next Zx does not than it's original. I have a TS2068 that needs repair and would love info on that ULA so I could possibly fix that if there is a problem with it. It does have some issues. I've just not had time to look at it or the many retro computers and consoles I have sitting in my garage and shed.
LOL i didnt know Sir Clive was 'that' much of a scrooge to use defective memory!! thats hillarious! Nice video, and nice replacements and fixes!.. i bet with the ULA issue, that they would also be different on different TVs too.. maybe not showing the issue (without the battery) on one TV, but does on another... some TVs have real trouble when the signals are not perfect (eg the blanking and syncs) others just display artifact, and some are so tolerant they can miss complete signals (eg blanking) do you know of any CPC chip replacement kits like this...i really dont wanna go the FPGA or CPLID (and learn how to program them) route if possible
Yes, you're right. Especially because I'm leaving the blanking signal totally open. It's not like I was grounding it or applying some voltage. I won't make that mistake again! :-)
That was pretty ingenious of Sir Clive. No wonder the speccy was so much cheaper than its rivals at the time! RAM was quite expensive back then. To cut a deal and get chips destined for landfill was great thinking. I am also minded of the fake cache RAM chips of the mid1990's used in 486's. That however was criminal and not ingenious.
@@Drew-Dastardly Exactly. This was ingenious but totally legit. Did you see the video I did on the Amstrad CPC 472? That one was definitely not very legit: th-cam.com/video/_ErfWfjN9iU/w-d-xo.html
@@2010Thex I suspect from those computers that you're in the US or Canada. There you could get hold of a Timex Sinclari. It's the closest thing you can get in America without importing a real Sinclair ZX Spectrum. But even importing them won't be too bad.
very interesting video and I just subscribed. Due to the fact that this board had a catastrophic failure, it must've been a voltage spike going down the power rails to destroy so many ICs. Maybe a good time to recap the board and power supply to stop this from happening again.
how to unsolder a faulty SMD-chip without a heatgun: use a sharp knife to cut the legs on both sides and take out the body - take out leg by leg with just your soldering iron
Honestly, it's not being smart, it's just about slowly learning what's going on and having some experience (even if it's just by watching videos). I bet you could fix it or at least do some initial analysis without too much problem. Give it a try!
Funny, I didn't know that Windows 10 shell had a COMP command, when I needed it, I believe it was with W98 or ME I used : FC /b Rom1.bin Rom2.bin FC stands for "File Compare" and /b forces a binary comparison, otherwise it will compare as text files, disregarding extra spaces and line feed characters... Sorry for saying this but I'm Portuguese and I've never been too fond of Spanish ROMs - Usually I flash them with the standard version. Some programs use ROM's content to generate random data, so there might be some compatibility issues (at least in one program I wrote, that's basically a tape copier for the 128k it uses the ROM error messages, like "Tape loading error" and these have a specific length, so when I call the error report routine I have to specify the exact length... One program I've wrote is not 100% compatible with the +3 - The sounds sometimes sound ok, others too low. I've tried everything, even disable memory paging, but nothing seems to work, it's completely random but a cold boot and selection of "48K Mode" in the main menu solves the problem... I believe it has something to do with a short sound produced by the +3 Menu when we move the cursor up or down and press enter... If anyone knows a solution to this problem, please leave a comment below...
I had no idea about comp either, I had to look it up. I use Mac on my day-to-day work, but I use that crappy Windows laptop for EPROM burning, Arduino programming, etc. On the Mac I much prefer Hex Fiend or just diff from the command line. Yeah, the UK 128K ROM had some bugs fixed too, so that makes sense. I believe there's even a project that has patched a few more bugs, but I can't remember the name of that project now.
As a kid my OG 128k had a strange issue when you do that built in maintenance feature (never knew exactly how to get into it, but I just mashed all the keys when powering on) it got near the end and always crashed but games work, I think it was a ram issue, I last played it around 1998 and in the early or mid 00's tried turning it on to just have loads of rainbow lines, its still at my parents but their house is damp so it may be beyond repair. Always wondered if it was as I thought a ram issue.
It sounds like a subtle RAM issue, although the fact that games worked fine is weird. Games usually push the machine to the limits and use up every bit of available RAM. But it would be possible. The machine might be recoverable. You should rescue it and see if it can be salvaged 😃
Hello Noel, I'm myself also fixing one spanish toastrack, and your videos are being very helpful, thank you. Please let me ask you this question: Could a toastrack prompt correctly... pass all diagnostics in green... but still not work properly? This is exactly what is happening to me after the repair. My machine cannot run the most of the 128k games properly although 48k work (the most of them although not all). Weird isnt?
It's unusual, but it's certainly possible. Try both with ZX Diagnostics and DiagROM. Also, to make sure it's not an overheating issue, try ZX Diagnostics on "soak" mode and leave it running for several hours. I've seen ONE instance where a ZX Spectrum +2 was faulty but ZX Diagnostics didn't catch it. It ended up being some weird RAM chip with different timings. Is the problem you're having 100% consistent, or does it change from run to run? It sounds like an interesting problem!
@@NoelsRetroLab Definitely consistent. It happens in the same titles, in the same moments. Basically... 128k games load... with a blue frame... after 3 seconds... black frame. Then you press any key and the title crashes. This happens in almost every 128k title. The machine was badly damaged, many Z80 data lines were broken and I had to reconnect them using the data path documentation. I spent many hours checking the connections with my tester. I even found a couple of mistakes in the schematics rendered by Ian Worsley that you can find on spectrumforeveryone. Full recap and a few RAM chips replaced as well. I tried "soak" mode for about 2 hours (40-50 iterations) and nothing bad happened so I am kind of puzzled. I think RAM could be the cause, so I double checked the connections to no avail. Since everything is socketed, I will play with the RAM chips and I will try DiagROM. PCF1306 could also be the cause? I guess I can manage to build my own replacement following the schematics shared on speccy.org At this point, this little fella it's starting to be a real challenge.Thank you for your advice!
Doesn't your TV/Monitor allow manually choosing between composite and RGB? Many (most?) TVs with SCART had the possibility. I had never seen the 9V battery connector in a SCART cable.
Not that I know. I don't think most TVs do, especially not CRT ones. It's the same input, just different interpretation of the signals. Maybe new LCDs have that but I haven't really looked closely since this across all of them.
Yes, that makes sense. It's weird that there was no clear pattern either (it wasn't every end of the line with a certain color that was triggering it).
I usually print them myself from the logos I got (or re-mastered myself) that I used for the boxes where I store the computers :-) I put them up here, although I don't know how their international shipping is: www.latostadora.com/retrolab/
Good question. Having a test cartridge (or even an EPROM) is the easiest way. If not, things get a bit trickier. Maybe I'll do a video about this sometime. Basically I would first look for hot chips, then look for data bits stuck on high or low, then look at the address bus, and finally maybe measure resistance between each data line and ground and look for some that are unusually low or high.
Yep Clive was a scavenger for parts for his machines. Using those defective 4164s was quite brilliant really. In a way, free market ruined the home computers in the 80's as they all had horrible compromises in terms of memory management and video capabilities. The prices just hadn't dropped enough yet to put decent machines on the market for an affordable price for consumers.
Yes, that absolutely gave him an edge over competitors if you manage to use parts like that. I wouldn't go as far as to say it ruined the home computers, but it certainly put a lot of pressure on them (and because of that made them very unique and quirky 😃).
That was left as an excercise to the reader 😃 Nah, seriously. One day I'll do a video on the memory addressing of the 128K Spectrum, but bank 5 corresponds to the lower RAM, which was already tested during the crazy screen color phase. Once that passes, we know we can display stuff on screen and it tests all the other banks.
Not sure. That happened a lot in older PCBs: They would have some parts that would be left empty. Look at the huge circle area in the top right corner of the board even! (I think the UK board had some kind of coil there or something).
The upper RAM must be 0x8000 - 0xFFFF to be the Upper 32K. The ROM is in the first 16K I think.
Yes! You're totally right. Brain-fade on my part! Otherwise it cuoldn't be using A15 either. Thanks for pointing that out!
I'm very late to the party but, a lot of Z80 based computers used to put their ROM into higher memory addresses and then use hardware to block the CPU from reading anything but zeros (NOP) after a reset until the address bus reached the first address of the ROM. Or putting it another way, after reset the CPU would execute nothing but NOP (no operation) instructions until it reached the memory address where the ROM is located. Once it hit that address, the data bus would "open for business" until the next reset.
This was a work-around for the fact that the Z80 CPU would always start running instructions from memory location zero after a reset but CP/M programs were required to run in low memory with some program data starting at 0 and the program itself starting at location 256 (0x100H).
While it certainly was possible to put a ROM at location 0, the expectation that programs would need to be able to run starting at location 0 was so common that some way of making the low memory available was almost always available, whether that was by the above technique or by memory "paging" to replace the low memory ROM with RAM.
When you have repaired a ZX Spectrum you made a child somewhere very happy again! Good work! Carry on with it. Oh, wait a minute! That CHILD is in his fifties by now.
Sinclair got stung by using the fault upper RAM chips, once the supply ran out he had to use fully working upper RAM chips! I built a circuit that allowed me to page in the other 32k RAM, was a lot of fun. :)
Ooh, that looks like fun. You could probably use that extra RAM for hacking stuff like we did with the extra 64KB on an Amstrad CPC 😃
That was fascinating seeing a replacement be made for the ZX8401. I've never considered recreating simpler chips like that and was worried they'd be lost to time. Really awesome video!
Glad you liked it!
I've seen the same kind of behaviour with the blanking pin on other systems - the Archimedes for example can give a very strange picture unless that pin is pulled high enough. It's going to relate to the sync signal - my guess is the sync is slightly different between those two ULAs. Since the blanking pin (at the TV end) supports "punching a hole" in the composite signal, with the RGB output (from what I understand), a difference in composite signal used for the sync, could affect how the output looks when using RGB. I guess what's happening is one ULA has cleaner sync than the other, and the one with slighly worse sync needs a more definitive "punch" using that blanking pin. On the archimedes for example you can rarely get anything other than a white box (with black border), until you use pin 16 to punch away the composite signal with the RGB. On other systems you just get the TV switching entirely into composite - I've never understood why, but it has to be related to the sync in each case I think. Subbed btw! EDIT: Thinking about the Archimedes example - it makes sense to me now (the white screen) - The Arc outputs only H or V sync (which cables combine), but that sync signal obviously lacks video (hence the white when we aren't "punching" through the sync source with the RGB picture info). The RGB has to punch in there or there is nothing for the TV to display. Cant believe how long that took me to work out lol
Great explanation! That makes a lot of sense! I did look at the sync signal but I forgot it wasn't really used for RGB out. It uses synch on composite instead, and that's what I should have looked at. Very interesting to hear about the Archimedes results. That's one computer I haven't touched yet but it's on my want list. Someday... 😀
@@NoelsRetroLab If you have a LOT of time and patience, I could send you a severely battery damaged Archimedes A3000. They are very very time consuming though.
@@GadgetUK164 Oh man, so tempted! But for my first Archimedes I better wait to get one in working condition, even if cosmetically it needs some work. It can be pretty tough to repair something without ever having seen one working in the first place 😀 If you ever have a working one for sale that is not super expensive (wishful thinking, I know), let me know!
So cool to replace a 40 pin chip with some SMD chips. I haven't had a ZX Spectrum 128K but in school they have the ZX81.
Yeah, it's pretty neat. Is a very visual remind of how technology has advanced since then.
And don't worry, I have a couple ZX81s waiting their turn.
@@NoelsRetroLab Is it really? I may have misinterpreted what you aimed at, but I tend to see it the other way. You could still build "anytning" using the TTL standard that was developed in the early 1960s. Sure, these are surface mounted CMOS versions. But surface mount was used even when Sinclair produced computers (especially in Japan) and CMOS has actually been around since the 1960s (although being much slower than today).
Well done Noel. Another superb repair. I swear I could smell the glue when you stuck that derby label onto the new EpROM
Hahaha Thank you!
I've seen that same kind of image issue with different ULA's via RGB creating the artifact at the end of words. I went through several revision of RGB cables before I eventually figured out it was something that's different with 2 seperate ULAs, just three weeks apart. Someone mentioned you need voltage on the signal. Then I tried different TV's and again saw different issues with the same ULA that had artifacts. One one TV (LCD) it had no issues, also no issues on a CRT. Two other LCD's had the issue. So it looks like it's a combination of things. Great to see that you solved it on the same TV with the added voltage.
Hi Noel! You rock my friend! Such good quality, time and care put in each video released. I cant imagine the next step when some patreon funds arrive to you allowing more content, more time spent and other things! This is easily the best channel on YT for retro happiness. Regards from Portugal, and stay safe, the numbers in Spain about covid are not good
Thank you so much! Glad you're enjoying it. We're all safe over here so far. Fortunately things are better in northern Spain. Cheers.
Nice reapir. Love the schamtic on he corner indicating the probe points. Atention to detail, love it.
Thanks! 👍
Many thanks for this new video Noel, that was excellent as usual! I'm not into Sinclair machines at all, yet I was really captivated by the part explaining the memory addressing and I was taken aback by learning how Sinclair solved the "48K / 64 K" equation by using cheap defective RAM chips that actually helped reduce the cost further, that was just astounding!!
Lastly I really enjoyed the 'hands-on' part on SMD soldering, very useful demonstration and practical tips, thank you very much! :-)
You're welcome. Glad you liked the video! Yeah, sometimes the strangest (and interesting) decisions go into the architecture of these computers. It's all part of their unique quirks 😀
Nice tip with the comp good.bin bad.bin, i've never seen that before but will use it in the future, thanks Noel !
Glad that helped. To be honest, I usually transfer them over to the Mac because it's my main computer, and there I use HexFiend, which does a binary diff, but has a nice GUI and shows the differences very clearly.
Nice to see it working properly again. I love those EPROM Read/Writers.
Yeah, it's great! There are a couple EPROM models that I wanted to use that it doesn't support, but it's been great for the most part (in spite of the horrible, horrible UI design).
Another superb video, plug in replacements are availiable at ZXrenew!
Nice fixes Noel!, and good to see that there are people actually making replacement chips for these iconic machines …
I really appreciate the explanation of the multiplexers. I've wondered why so many computers have them. I get it now!
For some reason YT unsubbed me from your channel, glad it popped you back into my recommended feed.
Yay! Welcome back! 😀
That's some cool back to the future stuff. Doc Brown would have replaced that chip with a set of valves though.
Great Scott! Fantastic idea, Marty! 😃
Hey Noel great to see you again.
I hope you’re well.
What a interesting idea to make the chip.
Your SMD soldering technique is pretty good. Liquid flux is fine ( we used no clean flux pens at work which are liquid ). Use more flux. Once you've secured the corners try drag soldering - You apply solder and the tip to the SMD component and swipe down the legs in a smooth movement.
Thanks! I've actually started trying the paste flux and I like it much better! Probably because I can have more of it. There's going to be a fair amount of SMD soldering in the next TI video in a couple of weeks!
Good idea putting the pin headers on breadboard
Thanks! They hold it nicely in place AND you're guaranteed to solder the straight. Not like I would have ever soldered pin headers accidentally at a angle.... not me, nooo... 😃
always we learn something new here
That's the idea! :-) Cheers.
Thank you for this video Noel, as always this excellent and hard work is a really magister lesson for me!!!!!
Thank you and glad you enjoyed it! Hopefully we can get together after the virus madness is over.
"aligning the SMD chip may take a while"
(he drops the chip on the PCB and it magically aligns perfectly on it by itself) LOL
Great video as usual! (How do you erase those eeproms?)
The ones we see in this video are the UV-erasable type - you can recognise those by the quartz window; If one is _VERY_ patient they can be erased by exposing them to direct sunlight over a long period, but normally one would use a dedicated device with an UV light source. You can see examples of such erasers here: duckduckgo.com/?t=ffnt&q=uv+eprom+eraser&atb=v175-1&iax=images&ia=images
Hahaha, yes, that was the ONE time I dropped the chip and it was in perfect alignment! I had recorded the comment before selecting the video, so that wasn't he best example. You know how fiddly it can be to get them just right.
Edit: And Berty is right. I have one of those UV boxes and in 15-20 minutes they're fully erased.
I had to rewind the video a bit as I was wondering "is that chip magnetic or what"!
I have a bucket of those UV-erasable eeproms but I do not have an UV light to erase them. I tried sunlight but did not work - not long enough I suppose. Also, we do not get much sunshine here in the UK... :(
@Mr Guru thanks for the tip, I will check it out tomorrow morning! thanks
Great video. Your videos are always really interesting, I can't miss any.
Thank you! Soon you're going to get some really interesting ones to you personally :-)
@@NoelsRetroLab That will be so cool ;)
Very interesting. Great vid Noel.
Glad you enjoyed it
One thing you could have done for fun, would have been to erase the bad EPROM and see whether you could reprogram it.
True. I still have it around. Maybe I'll try it just for fun. But then again, the EPROM was screwing up the memory tests somehow, so there has to be something slightly off about it.
Have you heard of the ZX Harlequin? It’s a reverse engineered spectrum kit that doesn’t need a ULA. The issue is they can’t sell the rom with it. I was thinking of just getting an eprom and trying to use an image off the internet to program it with. Not sure it’s the same one but my local electronics shop has EPROMs and the mini pro for sale.
Yes, I have a couple kits of those still to assemble. I'll probably cover them one day and get in the details of what's different about them. I suspect it'll be a bit like the Inves Spectrum+, since they had to reimplement all the ULA logic in a different way (in the case of the Harlequin all with discrete logic!).
Great explanation Noel, thanks :) and happy you fix it at the end. :)
Thank you!
These videos are really fun.
Thank you!
Great repair! It would have been nice to see whether the databus cleared up after changing the ROM. I was working on a board with a similar problem, dirty address and databus and it didn't occur to me the ROM could be corrupted, I did flash a ROM with the data read off the old one and ofcourse it showed no improvement.
Very well done video! Enjoyed it a lot. Thanks!
Glad you enjoyed it!
I suspect those ULA's came from different batches, different tolerances, etc. In other words, they aren't 100% exact duplicates. The "faulty" one probably just crosses (or gets close to) some threshold that causes it to be slightly inferior to the first one.
That makes a lot of sense. Now I'm very curious whether before this accident it behaved like the other one, it it's something that happened after the expansion was plugged in while it was on.
Excelente video. Muchas gracias!
Open the SCART connector and compare the pin in question at that end of the cable between the two ULAs
Good point. I'm going to guess that it's exactly the same and that the difference comes elsewhere in the signals. I mean, that pin is JUST tied to the battery. Although now that I think about it, if the battery isn't hooked up, then it's not just ground or high, it's an open circuit. I wonder if that causes the TV to interpret it in unexpected ways.
From what little reading I’ve done: On some 128k toastrack cables, with no battery, they use the composite signal to provide the switching voltage. Somehow relying on the tv electronics impedance to take it into the right range (or something) but this puts a nasty draw on the ULA, it can make it hot and increase likelihood of failure. Maybe you’re seeing that effect with your cable???
Great video! Interesting things to find out.
Regarding the bad image that the original ULA provided, it may be related to the blanking signal. Those RGB cables use the composite signal as a blanking signal. But since that signal does not have a constant voltage, it is helped with a capacitor. Maybe the original ULA provides a slightly lower voltage or more unstable voltage for that, therefore the TV switches from composite to RGB randomly. I've experienced a close situation with two Toastracks: one of them had an annoying image flicker every 2-3 seconds the other behaved the same but this happened a few times each minute. All dissapeared when I have provided the proper blanking signal to the DIN connnector.
That type of RGB video uses the composite video signal to sync the image. Is that what you mean? I think you're right and it's something related to that (see coment below by GadgetUK164).
@@NoelsRetroLab Where can I buy a kit for the PCF1306 IC replacement?
@@biv-8515 The person I bought it from is out of stock at the moment but he's getting more. You can see his Ebay listing here: www.ebay.es/itm/224111543392 Send him a message and he'll reserve one for you.
some people do a similar thing with the Amstrad cpc using the lum as sync and the S sync as blanking which causes ghosting, adding a capacitor will remedy the ghosting but causes the the intermittent loosing of signal (in my experience anyway) .. with the +2 & +2A you can use the 12v feed with a suitable resistor fitted on the motherboard to bring the voltage down when combined with the tv's 75 ohm impendence gives the suitable voltage (somewhere over 1 volt if memory serves) +2 has some different configuration options via the links ie RGB levels, brightness signal ,audio sync, composite ... not sure about the toast rack as i have not managed to buy one yet as the prices are high 😒
Did you get to open that SCART cable to look at how exactly the battery is being used? I never saw one like this, but it is usual practice in high-quality cables to boost the signals through repeaters, so to compensate for the signal loss of the cable itself. This is often done in long-length HDMI cables, for instance. Usually it is done by tapping the 5V line of the output connector, but then again, the same technique could be used to fill out your missing pin without a battery hanging from the cable. That they chose to use a battery suggests they thought the power rail (and perhaps the ULA itself) couldn't drive the signal hard enough across the length of the cable. If you open it and find a transceiver or repeater of some sort inside, there's your answer.
I haven't, but usually there's just a resistor and then sends the voltage straight over. I've never seen anything fancier than that in these kind of SCART cables.
That battery SCART cable is just weird. I can only imagine they use a fairly high ohm resistor dropper to get the 9V to the low voltage required as a signal. But why not just use diode(s) from the TTL signals and capacitor as a charge pump to provide the missing composite/RGB signal switching voltage instead?
@@Drew-Dastardly Yeah, they need to drop it down to 3V, so I'm sure they do exactly that. They could have used an RS232 as well. I never thought of trying to use some of the TTL signals to charge a capacitor. I wonder if that's just too bulky to be contained in the cable. For my own 128K, I'm thinking of hooking up 5V to the unused pin in the monitor connector and using that. The only downside is that I can't use the same cable with other 128Ks (and if I forget, then things like this happen!).
Another great video. Great to see how modern components can be used to replace parts that are no longer made and also the enthusiasm and love people have for these old computers. Are these kits for the ZX8401 available to purchase by the public or are there any Gerber files for the SMD pcb .
I believe you can buy already-made ones on Ebay, or get the Gerber files from the link I put on the description (the original ones by José Leandro).
if you need one, no problem, I will make another bunch of them. The last fews I had were sold out to Zxrenew. Whatever, you also could do one by yourself with the schematic that Noel's shared. Or contact me by PM, still I keep one.
@@ElectronGordo Feel free to share your Ebay link here since that's totally relevant. Or let me know and I'll put it in the description.
@@NoelsRetroLab Would be great. That way people could have access to it when they are not enough skilled to DIY.
Thanks for the video Noel. I'm curious what the white circle with 4 PCB pads is for. I guess something wasn't used.
You know, I'm not sure. If you look at the UK board of the 128K there's some kind of coil there. I'm not sure if it's a ferrite core or what it is exactly. 2.bp.blogspot.com/-So4FZHf9CN0/WBd34G1cCqI/AAAAAAAAKR8/fmnMf4BCs7shLLI-0ZrteH2DE3nV4RWTgCLcB/s1600/UK%2Bvs%2BSpanish%2B128K%2Bv2.jpg
@Mr Guru Sorry for posting here, but I don't find your post about how to solder it with a hot air gun. I was looking for those videos that you commented about it, but all what I watched was with solder paste. I have not solder paste. Is it obligatory for this technique? thanks again
@Mr Guru I have tried it... the result was not very nice, I will try it one more time, but... it seems to me that better i have not enough skills for this way. I operated it as follow: 1 add flux, 2 add iron to the pads, 3 more flux, 4 hot air gun... not nice how it had finished. Maybe if I fix only one pad to the chip... but the problem is that left too much solder iron on the pads. Need to learn more from Noel :)
@Mr Guru Thank you, I will do it, but maybe some PCFs later, cause first one was already spoilt (bended) by the heat. And I get many errors now.
Mate, this is SUPERB! Right up my street! thank you! GOD bless you
Thank you so much! Glad you enjoyed it!
Good practice is to put some lubricant to the IC socket making it easier to remove and install. Deoxit or even WD-40 will do.
Another great Vid . Thanks Noel :-)
My pleasure!
Good video. You explain well.
Thank you!
14:12
I have a feeling that possibly the VCC pin (14) of the NAND chip wasn't quite soldered.
When you removed the chip, you could see the solder on almost all of the pins get disturbed, but the solder for pin 14 barely changed.
I think maybe the pin was floating just above the solder, and not quite making proper contact.
Really neat vid, though, and great explanation of the circuit.
Subbed! ;)
Good eyes! That's certainly a possibility. I did do a continuity check between each pin and the pad, but I suppose that if it was barely soldered, just putting the test lead on it would push it down and record continuity just while I was testing.
The issue with the blurry video output from the ULA could perhaps be due to slight timing differences between the colour clock and the RGB outputs.
On some CRTs with RGB SCART input, and no voltage on SCART pin 16, they can tend to "overlap" the RGB signal over the Composite signal.
I would have thought the SCART cable is using Composite Sync from the Speccy, though, so maybe not?
(this can sometimes happen even without any voltage on pin 16. Some TVs just don't process the RGB input properly, as they either expect *only* the Composite Video pin to be hooked up, or RGB with voltage on pin 16.)
If the SCART cable is using Composite Video for sync, then that could explain the difference.
The timing difference could be causing a delay after the RGB is processed by the chroma encoder chip.
I don't think it's a problem with the RGB output directly. As you say, it looks fine once you hooked up the battery.
What confuses me is how you got a decent image before on that TV without a voltage on pin 16 in the first place. lol :p
@@NoelsRetroLab
I still have problems with SMD soldering at times, even after quite a few years.
Yep, I've definitely had that before when continuity tested fine, and it turned out the tiny amount of pressure from the probe was enough to make the connection. hehe
I repaired a board for somebody about six months ago which I suspected had solder joints on a QFP that hadn't reflowed properly. I used a cheap USB "microscope" camera, and did the careful "poke test" on the chip pins...
drive.google.com/file/d/1hbmWG1EPzY6ZHh2CwrFV3onNRpzhAR3U/view?usp=sharing
I didn't add all of that flux paste either, btw. lol
I used to use flux paste for many years, as it does tend to stick around for longer without boiling off.
But now I use liquid flux for pretty much everything, and got a bit quicker with the dwell times of the iron.
That clip is a good example of what often happens when a larger QFP chip is removed using hot air, and the pins get bent very slightly.
The pins then no longer sit flat on the pads after replacement. I won't tell you who's board it was, as he'd probably kill me. lol
*waves* GDay Ash fancy seein you here! LOL
@@WacKEDmaN I get about, mate. lol
Thanks for the upload Noel. You could try and see if the bad 27256 will fully erase but I doubt it. I also have the same type of programmer which I converted to the 'A' version from the 'CS' version.
I doubt it too since it had some things that were interfering with the data or address bus.
How did you upgrade it? That might be really relevant to me right now because that allows it to use higher voltage right? I'm trying to burn some 2732As for the TI99 right now, but those require 21V and they fail (in spite of being listed as being supported in the EPROM programmer).
Edit: Never mind, I see the CS also supports 21V. But my more recent II Plus only goes to 18V. How annoying!
@@NoelsRetroLab at 20:27 the good EPROM was installed in the programmer offset (pin 1 of the EPROM is in socket position 2), and dumped (offset not detected as Check ID disabled, guessing this was done because the read of the good chip failed due to the offset). Bad EPROM is installed with no offset and dumped - imagine comparing the two read dumps will be different due to this (unless programmer software was made at Hogwarts?). Also, the new EPROM was programmed with no offset using the dump of the good EPROM, so don't see how this worked (will put this down to the magic of editing, the curse of continuity, and the desire to tell good story). Anyway... where I'm going with this is, as Andy says, you might be able to get away with erazing and reprograming that EPROM). Thanks for the vids - have been watching them all day.
just thinking out loud .. when the battery fixed the ghosting , and you got different results back then and now , could it be as simple as corrosion on the pins of the IC ? .. or if one is weaker than the other , are there any pull up/down resistors for that part of the ciruit ?
Could be. I don't recall seeing any obvious corrosion though. Rather puzzling, especially how the behavior changed (although that's consistent with your corrosion theory after pulling it in and out a couple of times).
I don't even have a Spectrum, but love your videos - new patreon supporter :-)
Thank you so much! Much appreciated!
@@NoelsRetroLab I DO however have a working Amstrad CPC6128... but no monitor... and i live in the U.S. lol. Some day i'll get it fired up.
@@IDPhotoMan Argh, it's a crime to have a CPC 6128 and no way to use it, don't you know? 😀If you can get some sort of RGB monitor, you can easily make a video cable (and you can feed it with standard 5V DC and 12V DC power supplies). If you need any help ping me on Discord.
@@NoelsRetroLab thank you, it would be fun to get it going.
@@IDPhotoMan you can use most old analogue TVs.. check out 'the 8-bit guy' channel he has a vid on converting the TV to RGB, and its quite simple.. you can then wire it directly into the 6128s monitor port...
or if ya just want to test its working..you can get a black and white composite compatible signal by using one of the color pins and one of the sync pins on the 6128s monitor port. i did that back in the late 80s/early 90s and flipped 2 screens in ram, overlayed some audio and then recorded it to VHS! there was an article about it in one of the old Amstrad Action magazines..
I think the way addressing works on DRAM is more to do with the internal construction than price, size etc. From what I vaguely remember SRAM/ROMs with multiplexed buses did it by having the data and address bus share pins.
Anyhow nice video as always.. interesting how much the 8 bits used gate arrays really. You would have thought in the numbers they were making they would have reduced them right down to a single chip pretty quickly.
Oh interesting, I assumed it was just a cost-saving issue and not some other reason. Now I'm going to have to look into it 😀
Eventually they did reduce them to just about a single chip like in the case of the Spectrum +2 or the Amstrad GX4000 (but I really dislike working with those because... there's not much to fix beyond blown RAM!).
Finally, Did you got any information about why with one ULA appears the ghosting efect and with the other one not? cheers
Good job on that SMD soldering Noel, looks easy, but we all know it isn't :)
Haha, thanks! It's actually not bad at all. Easier than it seems I think.
It's another project, but I would like to see if you can fix your TI99
Working on it right now!
Awesome to see that discrete SMD are small enough now that one can use them to replace such custom chips in almost the same size package. Would love to see what options are available if the video ULA was bad. Any way you can make a video on that? That would be very cool to get into.
Yes, isn't that cool? That's a very clear way of how technology has improved since the 80s. As for the ULA replacements, I'll make a note about it. In the meanwhile, have you read The ZX Spectrum ULA by Chris Smith? Excellent book. www.zxdesign.info/book/
@@NoelsRetroLab Thank you sir. I am an electronic tech/hobbiest and I am learning HDL and to make a new ULA that can replace the old one with possible enhancements would be great...I'm sure it's been done already though since the Next Zx does not than it's original. I have a TS2068 that needs repair and would love info on that ULA so I could possibly fix that if there is a problem with it. It does have some issues. I've just not had time to look at it or the many retro computers and consoles I have sitting in my garage and shed.
LOL i didnt know Sir Clive was 'that' much of a scrooge to use defective memory!! thats hillarious!
Nice video, and nice replacements and fixes!.. i bet with the ULA issue, that they would also be different on different TVs too.. maybe not showing the issue (without the battery) on one TV, but does on another... some TVs have real trouble when the signals are not perfect (eg the blanking and syncs) others just display artifact, and some are so tolerant they can miss complete signals (eg blanking)
do you know of any CPC chip replacement kits like this...i really dont wanna go the FPGA or CPLID (and learn how to program them) route if possible
Yes, you're right. Especially because I'm leaving the blanking signal totally open. It's not like I was grounding it or applying some voltage. I won't make that mistake again! :-)
That was pretty ingenious of Sir Clive. No wonder the speccy was so much cheaper than its rivals at the time! RAM was quite expensive back then. To cut a deal and get chips destined for landfill was great thinking.
I am also minded of the fake cache RAM chips of the mid1990's used in 486's. That however was criminal and not ingenious.
@@Drew-Dastardly Exactly. This was ingenious but totally legit. Did you see the video I did on the Amstrad CPC 472? That one was definitely not very legit: th-cam.com/video/_ErfWfjN9iU/w-d-xo.html
I enjoy your trouble shooting videos
Thanks! Really glad to hear that.
Excellent video!!!
Glad you liked it!
I have apple IIs and Commodore 64s... now o feel like I need to get a ZX Spectrum!:) Again excellent work and videos
@@2010Thex I suspect from those computers that you're in the US or Canada. There you could get hold of a Timex Sinclari. It's the closest thing you can get in America without importing a real Sinclair ZX Spectrum. But even importing them won't be too bad.
@@NoelsRetroLab Yes, I live in Canada but will have to see of Time Sinclair is just as good as the Spectrum series. Thanks
very interesting video and I just subscribed.
Due to the fact that this board had a catastrophic failure, it must've been a voltage spike going down the power rails to destroy so many ICs. Maybe a good time to recap the board and power supply to stop this from happening again.
Hola Noel!!. Conozco varios LLopis por aqui. Saludos desde Alicante.
Hi Noel, any chance of a build/review of the "zx dandanator"?
I hadn't thought about it, but I'll add it to the list. Especially now that they're coming up with a new version/add on.
how to unsolder a faulty SMD-chip without a heatgun: use a sharp knife to cut the legs on both sides and take out the body - take out leg by leg with just your soldering iron
I wish I was this smart so I could fix my Colecovision.
Honestly, it's not being smart, it's just about slowly learning what's going on and having some experience (even if it's just by watching videos). I bet you could fix it or at least do some initial analysis without too much problem. Give it a try!
Funny, I didn't know that Windows 10 shell had a COMP command, when I needed it, I believe it was with W98 or ME I used :
FC /b Rom1.bin Rom2.bin
FC stands for "File Compare" and /b forces a binary comparison, otherwise it will compare as text files, disregarding extra spaces and line feed characters...
Sorry for saying this but I'm Portuguese and I've never been too fond of Spanish ROMs - Usually I flash them with the standard version. Some programs use ROM's content to generate random data, so there might be some compatibility issues (at least in one program I wrote, that's basically a tape copier for the 128k it uses the ROM error messages, like "Tape loading error" and these have a specific length, so when I call the error report routine I have to specify the exact length...
One program I've wrote is not 100% compatible with the +3 - The sounds sometimes sound ok, others too low. I've tried everything, even disable memory paging, but nothing seems to work, it's completely random but a cold boot and selection of "48K Mode" in the main menu solves the problem... I believe it has something to do with a short sound produced by the +3 Menu when we move the cursor up or down and press enter... If anyone knows a solution to this problem, please leave a comment below...
I had no idea about comp either, I had to look it up. I use Mac on my day-to-day work, but I use that crappy Windows laptop for EPROM burning, Arduino programming, etc. On the Mac I much prefer Hex Fiend or just diff from the command line.
Yeah, the UK 128K ROM had some bugs fixed too, so that makes sense. I believe there's even a project that has patched a few more bugs, but I can't remember the name of that project now.
Nice. This is better than school :-)
As a kid my OG 128k had a strange issue when you do that built in maintenance feature (never knew exactly how to get into it, but I just mashed all the keys when powering on) it got near the end and always crashed but games work, I think it was a ram issue, I last played it around 1998 and in the early or mid 00's tried turning it on to just have loads of rainbow lines, its still at my parents but their house is damp so it may be beyond repair.
Always wondered if it was as I thought a ram issue.
It sounds like a subtle RAM issue, although the fact that games worked fine is weird. Games usually push the machine to the limits and use up every bit of available RAM. But it would be possible.
The machine might be recoverable. You should rescue it and see if it can be salvaged 😃
Hello Noel, I'm myself also fixing one spanish toastrack, and your videos are being very helpful, thank you. Please let me ask you this question: Could a toastrack prompt correctly... pass all diagnostics in green... but still not work properly? This is exactly what is happening to me after the repair. My machine cannot run the most of the 128k games properly although 48k work (the most of them although not all). Weird isnt?
It's unusual, but it's certainly possible. Try both with ZX Diagnostics and DiagROM. Also, to make sure it's not an overheating issue, try ZX Diagnostics on "soak" mode and leave it running for several hours.
I've seen ONE instance where a ZX Spectrum +2 was faulty but ZX Diagnostics didn't catch it. It ended up being some weird RAM chip with different timings.
Is the problem you're having 100% consistent, or does it change from run to run? It sounds like an interesting problem!
@@NoelsRetroLab Definitely consistent. It happens in the same titles, in the same moments. Basically... 128k games load... with a blue frame... after 3 seconds... black frame. Then you press any key and the title crashes. This happens in almost every 128k title.
The machine was badly damaged, many Z80 data lines were broken and I had to reconnect them using the data path documentation. I spent many hours checking the connections with my tester. I even found a couple of mistakes in the schematics rendered by Ian Worsley that you can find on spectrumforeveryone.
Full recap and a few RAM chips replaced as well.
I tried "soak" mode for about 2 hours (40-50 iterations) and nothing bad happened so I am kind of puzzled. I think RAM could be the cause, so I double checked the connections to no avail. Since everything is socketed, I will play with the RAM chips and I will try DiagROM.
PCF1306 could also be the cause? I guess I can manage to build my own replacement following the schematics shared on speccy.org
At this point, this little fella it's starting to be a real challenge.Thank you for your advice!
Doesn't your TV/Monitor allow manually choosing between composite and RGB? Many (most?) TVs with SCART had the possibility. I had never seen the 9V battery connector in a SCART cable.
Not that I know. I don't think most TVs do, especially not CRT ones. It's the same input, just different interpretation of the signals. Maybe new LCDs have that but I haven't really looked closely since this across all of them.
how did ram survived? (mine was toasted lol) nice equipment for testing zx spectrum.....
Probably the other chips shorted quickly and spared the RAM (especially since it's further away from the power supply).
The extra pixels are not really pixels, they are display artefacts. Dunno much about how and why but I've seen them on some setups since the '80s.
Yes, that makes sense. It's weird that there was no clear pattern either (it wasn't every end of the line with a certain color that was triggering it).
Where do you get your cool ZX tee shirts?
I usually print them myself from the logos I got (or re-mastered myself) that I used for the boxes where I store the computers :-) I put them up here, although I don't know how their international shipping is: www.latostadora.com/retrolab/
How to manually test for bad DRAM chips if you do not have test cartridge?
Good question. Having a test cartridge (or even an EPROM) is the easiest way. If not, things get a bit trickier. Maybe I'll do a video about this sometime. Basically I would first look for hot chips, then look for data bits stuck on high or low, then look at the address bus, and finally maybe measure resistance between each data line and ground and look for some that are unusually low or high.
Could you have just erased the bad EPROM and rewritten it with the data from the good one?
Yep Clive was a scavenger for parts for his machines. Using those defective 4164s was quite brilliant really.
In a way, free market ruined the home computers in the 80's as they all had horrible compromises in terms of memory management and video capabilities. The prices just hadn't dropped enough yet to put decent machines on the market for an affordable price for consumers.
Yes, that absolutely gave him an edge over competitors if you manage to use parts like that.
I wouldn't go as far as to say it ruined the home computers, but it certainly put a lot of pressure on them (and because of that made them very unique and quirky 😃).
I see some magic there... at 20:25 you insert one ROM and ... it reads "good"???
I guess no one noticed that the diagnostics didn’t test RAM bank 5?
That was left as an excercise to the reader 😃 Nah, seriously. One day I'll do a video on the memory addressing of the 128K Spectrum, but bank 5 corresponds to the lower RAM, which was already tested during the crazy screen color phase. Once that passes, we know we can display stuff on screen and it tests all the other banks.
Why is resistor R127 Missing??
Not sure. That happened a lot in older PCBs: They would have some parts that would be left empty. Look at the huge circle area in the top right corner of the board even! (I think the UK board had some kind of coil there or something).
Hi Noel.. I dropped you an email.. Would appreciate your response.. Thanks.. - Andy
Oops, it was caught in the spam filter. I'll reply now! 👍
99.99% of new chips are good. 99% of problems is bad soldering. This is at least, in mine case. ULA is not very complex chip compared with Atari ones.
ZED, it's ZED 80, ZED-X SPECTRUM etc... AAAARRRRGGGHHHH!!!!!!!!
Otherwise, great vid!
I totally get "zed"-X Spectrum since it's a British name, but why "zed" 80? Wasn't it created in the US?
@@NoelsRetroLab Yeah, I'm actually inclined to agree with you re: Z80, regardless of how wrong it sounds to me!
Keep the great content coming!
Cringe on using superglue to attach a label.