LOL. Spent the first 10 mins shouting "there's an edge connector" at the screen :) ... Again I commend you for taking us through the whole process, mistakes-and-all. Respect :)
Are you kidding? It's also got a full working BASIC interpreter in it! If there's any way to communicate with another computer, you could just write a quick program to use PEEK on the whole thing and dump it that way. But that twisty connection to the EPROM reader was funny to see.
@@8bitwiz_ I used to poke and peek all the cartridges for the Coco. I had 64k when the Coco had 4k. Then I uploaded to the cassette tape. Then I had 360 5 1/4 disk drives that I uploaded that turned a 720 5 1/4.
If one has an EPROM dumper that allows adjustment of the supply voltage, I would suggest trying to read out the chip with VDD values starting at 5.25 volts and going downward by 0.25 volts each time until most of the bits no longer read the same value. If one finds that e.g. everything reads identically at 4.5 to 5.25 volts, and from 3.5 volts to 4.25 volts, but one bit reads differently at 4.0 and 5.0 volts, odds are very good that's the erroneous bit. If a bit wasn't programmed quite as strongly as the rest, it might read as blank at higher VDD levels but as programmed at lower VDD levels. If a bit wasn't erased as well as the rest, it might erroneously read as programmed at lower VDD levels but as blank at higher ones (which is why I suggested going up to 5.25 volts).
Great job with this. A common cost-saving measure for simple double-sided boards was to skip the through-hole plating and to "top-solder" instead. Apparently they also decided that supply decoupling capacitors were not necessary but it would be good practice to include these if you're making a new version.
Right. I learned since then that apparently that was pretty common. Do the EPROMs benefit from a decoupling capacitor as well? In the updated design I used a single EPROM with a NOR gate and I used a decoupling capacitor on the NOR gate, but not on the EPROM itself. I guess it may be close enough to help both 😃
@@NoelsRetroLab Not going to do any harm, but there are plenty of boards with no decoupling capacitors so it is not really a worry. Just have wider power and ground traces on the board and you will not need them, as there should be decent enough decoupling on the other end of that connector anyway, plus at lower frequency not really going to be a worry with TTL chips. When you have large S100 boards, with a hundred or more TTL chips on the board, yes you need decoupling, but a lot of avionics boards get by with only one capacitor at power entry, and survive over a way tougher operating regime. One per chip probably not needed, but one per dozen probably will work well enough.
@@NoelsRetroLab , one capacitor per IC is best practice but one shared among the (few) chips on the board is a good compromise. A decoupling cap (effectively) provides a local high-speed reservoir for the IC's instantaneous (switching) current demands so as not to cause supply/ground dips/bounces on the supply rails of the IC as its transistors rattle away. Having any capacitor at all is better than relying on the impedance of the traces (and edge connection) back to the power supply being low enough to avoid troublesome VCC/GND disturbances under these fast transient loads. That said, it's demonstrably OK to omit the capacitor(s) (just not ideal).
@@adfeldhaus A typical EPROM is a purely combinatorial device. On e.g. a 250ns device, if there is a power supply glitch, the outputs may be meaningless until 250ns after the supply stabilizes, but that would only be a problem if the device is operating near the edges of its timing margins (if the system would require that the device produce valid data within 260s, but the supply takes 20ns to stabilize after a glitch, that could result in erroneous operation, but in practice timing margins will seldom be that close.
@@flatfingertuning727 , Good point. Since it's fully static and stateless a (parallel) EEPROM like this should settle down to a correct state pretty quickly after transient supply glitches.
I'm so glad to hear you are cloning this cartridge's circuit board. Something this rare deserves a new lease on life. Thanks also to the fellow who sent you his cartridge to get data from. Seeing you being so careful to honor his wishes was uplifting. Well Done!
Since there were differences in the second ROM, I'm wondering if it wouldn't be a good idea to reprogram that as well. Just to have a consistent version across the two chips.
@@Bisen222 It's not about getting "...copiright [sic] text the same...", it's about making sure that any references in one ROM to the other ROM are actually correct. If there's a jump vector in ROM #1 that points to somewhere in ROM #2, and ROM #2 is different from what ROM #1 expects, that's a problem. You can't replace the last chapter of _Moby Dick_ with the last chapter of _Gone With The Wind_ and expect it to work.
You really make this aspect of hardware and electronics easy to understand and enjoyable to learn. You make me feel inspired to go deeper with my very limited electronics experience. The information you provide - though limited/geared to retro computers - can be transferred over and used on other electronics. Thank you for also taking the time to explain schematics and how to use different tools (like oscilloscopes) in your other videos to troubleshoot. You are a wonderful teacher!
Another excelent forensics video from Noel. It is amazing how well you describe the process and the different approaches to do it. I wish I had had you as a professor at the university. It would be also nice to record the viewers, as I found myself shouting at the screen: "you are getting feedback from the other ROM". Amazing job as always, mate!!.
Interesting. A while ago you mentioned you had a mystery cartridge and I though it was a copy of the Kamil Karimov MSX emulator. I was almost right :-) Anyway, back in 1986 Peter Zevenhoven made a software emulator published by the Dutch Spectravideo CuC users club, basically he was doing the same trick taking a 728 MSX ROM patched with the hardware IO addreses of the 328 hardware turning the 328 in a MSX compatible machine allowing most MSX games to run. MSX emulator is loaded in RAM page 02 of bank 0, next a RAM bank switch is setup and a copy is made to page 21 of bank 2 (disabling build in ROM) and a reset is issued turning the 328 into an MSX. I do not believe CuC sold this emulator (possibly due to MSX copyright infringement), emulator circulated on club events and it was as such distributed. At that time I was an active member of CuC and I still have copy of that emulator on floppy.
I think those unused top pads coming off was due to the fact that the through-holes were not plated. Just not much but a small amount of 30+ year old adhesive to stick them down. Interesting video!
Excellent video as always Noel. Looking forward to seeing what a man of your considerable electronics talents comes up with regarding those new PCB designs!
I wonder why you were not reading out the ROM from the SVI, in basic.. and then save it to floppy. That's usually what people on MSX do to dump ROMs in case they don't want to open the machine.
Hmm, given the choice between soldering a bunch of wires and writing a simple BASIC program to dump the cartridge contents to the screen and doing screen captures to get the contents, I'd choose the second. Unlike with say a game cartridge, you have BASIC at your disposal, a FOR/NEXT loop and a PEEK is all you need to see what's on the cartridge, I would think.
I've always been able to wire up some kind of serial interface for systems. Usually the joystick ports have some kind of output pin. You just need one. Though in BASIC it might be tricky to do async serial timing reliably and fast enough to not take a day to send the data.
Before I retired, I used to do this type of thing a lot. When companies refuse to give you the raw data, you have to improvise and read the PROMs or even ROMs if you are going to reverse engineer the code to update or change. We had the best luck using the clip on spring connectors to read them. Especially if you couldn't read them via edge connectors. And even having the machine/operational code, in order to reverse the code you had to use special programs to decode the readings. But I loved doing such things. An amazing challenge, but also an amazing reward when you "cracked" the code...
I got an old o-scope years ago and nearly first thing I did was dump its two EEPROMs (and posted them to a few forums for others), in case they ever become corrupt. Probably 30 years old at this point, around when EEPROMs start to fail.
Thank you! It's been super fun so far (even though it has been a super, super simple project just to get familiar with PCB layout tools). Looking forward to taking it further.
For tenuous connections like your socket on the bottom of the PCB or the piggy-back, a dielectric grease works wonders. In the states you can get it at any auto parts store sold as "bulb grease" or grease for battery terminals.
Non plated double sided was common in the '80s and thus top soldering was common. They even had all metal sockets that allowed top soldering. If you make a replacement you may want to use a 32K or larger EEPROM or EPROM and a logic chip.
I was wondering on that first EPROM if you tried to erase it via UV light first? I know it shows up as all ones, which means an empty EPROM, but I thought maybe doing a UV erase that would completely reset it and make it work again possibly?
Best not to do anything to alter the original chips. There are a variety of quasi-analog ways of reading out data from a chip that's suffering from bit rot (the simplest of which is using varying VDD levels when reading) but erasing the chip would render any such recovery efforts impossible.
@@djmips If you don't care about preservation, what's the point of all that? Sounds like he basically made a whole new copy of it anyway, reusing I guess just the aluminum panels.
I really appreciate that Noel shows the failures along with the successes! Noel has made me much more comfortable with taking chances when I do repairs (my repairs are far, far simpler I hasten to add).
Great video, Adrian! Thank you! I cut my teeth on 8-bit home computers - I bought an Amstrad CPC-6128 which I still have. I haven’t gone as far as to create this kind of project but I did have fun making ROM socket boards to fit other ROMs. Maybe if I get time, I’d like to go back and explore adding some kind of storage device to my Amstrad. There used to be advertisements for “Winchester Drives” but they were hugely expensive. I’m guessing there are other solutions on CF or SD cards that can be implemented now. All my university notes were transcribed from handwritten scribbles to word processed pages with digitally created images. Neither format is compatible with what I have now. Although, my technical illustrations and technical authoring has evolved hugely since 1990.
Why didn't you just dump it using the SVI? A quick machine language rom dumper would let you save out the image and then you could split it back to the two eproms.
Nice work! I believe you need to erase the old eprom first before it will program correctly again. Since you removed the window cover, you might try exposing it UV light and give it another try. It may very well work again!
@@1971merlin The data read out is binary. On the floating gates it's analog, a certain number of trapped electrons. It may read as a 1 but still have electrons.
There's nothing worse than ploughing on thru a set of well thought out instruction/theories and realising that something you completely forgot about would've saved you time.... loads of time. Been there and done that, too many times!
That chip was toast so it had to be desoldered anyway. But for the other one… cutting the trace would probably be better. And yeah, I think he probably should reprogram the other chip since there could be some differences.
Putting all that heat into the chips, and almost destroying the PCB because he's too retarded to use a scalpel knife? Scary! I realise that he subsequently found that the EPROM was "dead" but he really did not need to remove either chip to test it or try to grab a dump. Also, he tried to "reprogram" without erasing the EPROM first... so that was never gunna work.
Many thanks to you and especially to the owner of the still working cardridge, thanks to all the efforts and trust another piece of homecomputer history was preserved.
Isn't it possible to dump the roms on the computer directly, like its possible on a C64? so u could save all your adapter working, even it's quite amasing.
Yes, once I figured out how they were mapped, that would have been possible. It all started because I thought it was going to be easier to read the EPROM directly. Ha! 😃
I am not sure that I would have touched the faulty cartridge. Just leave it as it is as a piece of historical memorabilia. The PCB is very simple. (Good idea for your first PCB design project.) You could have given the donor his original cartridge back untouched and your first replica cartridge using your reverse engineered PCB. But I suspect it all comes down to what the donor wanted you to do. Either way, your approach to reading the working cartridge was awesome and it is so much more helpful to show your failed attempts and thought process. Makes for a far more interesting and educational video. Until recently, I would have put Adrian Black as my favourite retro computer blogger but, from my perspective, you are just about inching ahead. Keep up the awesome work.
A 27128 is a !6 Kilobyte eprom A0 to A13 = 16K total , but on the edge connector there is no A14. It seems that the eproms are mapped in the same address space.(only 1 eprom selected at a time).
Doh! You're right. I don't know why I said 8KB (and then I went and reinforced that with the text overlay). Each EPROM is enabled separately with a different /CS signal (there are 4 of them in the SVI cartridge connector).
RMA 04-HV - best soldering flux gel I know for repairing all these 30-40 years old devices. It support cleaning solder from contamination (let it boil using soldering iron tip and all mess like rust just float on surface, easily removable) and it support solder flow to both sides of PCB through holes - everyone can do factory quality level soldering using RMA 04. Just be aware that solder can flow to chip legs all they way up on other side using RMA 04.
I initially tried programming it just like that since it was already in an "erased" state (at least the data bits). I tried erasing it and reprogramming it afterwards off camera but it made no difference. It looks like something fundamental in the addressing or control was totally busted.
@@thebyteattic Usually, you can re-program an eprom without blanking it, if you want to change bits from 1 to 0, but not the other way around. This one was all 1s, so looked like a blank one, and should have been reprogramable if it worked.
How very interesting, I look forward to the follow up episodes. Thank you. Not knowing a lot about MSX I'm curious to see how well this 'simple' ROM replacement works. I remember when I designed a 9958 / AY upgrade board my Coleco ADAM, I decided I might as well add some multi-standard compatibility - that required a bunch of jumpers for the differing addressing. It would be funny if all I needed to do was HEX-edit the ROM jump tables! Yep, my take on an ADAM Super Expander will play games for just about any machine containing a Z80, Ti GFX and sound chips, or an AY. Sadly I never got the native 80Col support to work - I'm not a software guy. Such mods are not that hard as much of the work can be found within the RC2014 community... As the ADAM is a bit big my 'daily driver' retro computer is an Amstrad PCW 8512 system board crammed in to a 8256 keyboard with 'pretend' Amber Monitor circuitry, a Gotek, the same GFX and sound upgrades, and a more modern Z80 variant included in a 8256 - RC2014 bus adaptor - got to love how easy Ti made it to integrate their line of GFX chips in to any old system, and that CP/M is so gloriously dumb it will put up with almost anything! :-D PS I have a semi-finished page on Thinyverse about the 8256-keyboard mod if anyone is interested. Just don't be using a 9512 system board, it won't fit.
The PCB was more likely home made so the through holes would not have plated holes. It would also be a good next step to make a NEW pcb with correct plated through holes.
The PCB is not "home made"... there simply was no (easy) technology to plate the through-holes back in those days. It was made from double-sided copper PCB material, which was then etched both sides. That's why they used tube things that looked like rivets if they needed to create a "via".
I wonder how 1 of the 2 EPROM's became FF'ed? It does amaze me vintage EPROM's still work at all. Backups have always been a thing, if only people knew (Not viewers of this channel of course).
Thanks for showing this. It would have been interesting to try to read the EPROM vi athe cartridge connector, I think that would have all the necessary signals on it?
Yes! In retrospect that would have been possible. At the beginning, since i hadn't traced out the connections yet, I wasn't sure if there was something else going on. But it would indeed have been very simple to do that. Maybe I'll mention that in the followup episode. Cheers!
Hi Noel, I just found your video. I happen to be selling an SV-328 with this exact CCG cartridge (I bought it from the Spanish company back in the day!). Unfortunately, my cartridge doesn't work anymore either.
You should tape up EPROM holes as soon as you program them. With some UV-opaque material, of course. There's residual UV everywhere and it slowly erases EPROMs' contents.
@@JCCyC back when I was a kid without a decent UV source, I tried a week with some eproms out in the Aussie sun. Not a single bit flipped.. It takes some high energy UV in a concentrated manner to erase an eprom, and stray light sources is rather unlikely to really cause issues. There are some decent articles on hackaday about using direct sunlight, and for one it was 2 weeks before weak bits started, and 3 weeks before full erasure. That is a lot of light needed. There is little to no UV-B or UV-C in the average house or workshop so you could consider years before bits get erased, if ever.
Would it be possible to use a single 27256 in the modern version? They just seem to be a lot easier to get, and you only need one so no chance of mixing the two ROMs up.
If the ROMs on the borrowed board were slightly different, should you not have reprogrammed the 2nd ROM as well, because they may look like they work until a call from one ROM hits a slightly different location of the other ROM i.e. they are miss matched?
you aparently are not aware of the function of those windows to the die. These are EPROMS , not EEPROMS, which means that they get erased by exposing them to UV light. Since every light source has some UV component in it, exposing those to it may cause the EPROMS to get erased or get data inconsistency. Therefore they are covered. Just take that into account if you didn't knew about it, i am trying to help, I hope my critic is not too harsh. I really enjoyed the video, keep the great work going!
i think they make clip on chip interfaces that can clipped on and give access to the pins without unsoldering. if he is wanting to not want unsoldering because he wants to extract maximum collectors/resale value you could offer to buy it from him. if you have friends or could make friends in the computer forensics department of the police or the fbi you could have them help you as they have to have reliable recovery because criminal cases has to have reliable evidence otherwise the case can be thrown out. you could use exacto knife to cut the trace between the 2 pins and then use solder to re join the pins.
If you are designing a PCB, why not build a MSX to DIP adaptor as a PCB with some dip switches to allow chip selection. It would make restoring other such cartridges a brease in future. Maybe Make a multi system version that does C64 as well?
Great video!! I remember these cartridges from when I was a kid. It would be great to get one so any tips where to search for one of them would be nice. Otherwise, I will try to build one.
Those were actually soldered in place, so they were pretty good. The previous attempts not to much though! (And yes, I still wouldn't trust it more than a few tries 😃)
Cool, and so another rare thing gets saved! I've honestly never seen a pcb that needed to be soldered on both sides, but then again, I don't really solder a lot ;)
A verry long time ago I had to a similar ROM backup, I had used a DPDT switch to switch between the roms as it was literally change 1 pin on each from VCC to Ground and I had a toggle switch handy. The issue I had was not 2 sepearte ROMS as much as bank 1 and bank 0 switching with a reader that could not actually read the full ROM, and using a hex editor to patch both banks into one file and managed to reflash into a single ROM of full capacity ( 2 of 16k into one 32k ROM ).
Interesting little cart there, its awesome you were able to dump the Rom preservation of these things is so important. Really enjoyed the content and breakout of all the pins and the methods you used in those attempts. Really came away with some useful knowege. Liked and Subcribed, looking forward t o seeing more great work!
Would it not have been simpler to just plug the cartridge in and write a small basic program to dump the ROM address space? I've done this on various other systems, though I admit to knowing nothing about the SVI machines.
It should be possible to replace this with a single 16kB EPROM with a few logic gates. With SMD, you could make the whole thing tiny, like 10mm high, basically just the edge connector.
Wouldn't solder flux help with solder flowing? And yes, i also support idea of reading rom via basic. Surely this wouldn't came as pretty as reading individual eproms, yet you'd have only to divide dump into 2 equal parts and reflash single one. surely approach taken by you is much more interesting and educational :)
I have not watched the entire thing yet so maybe you did this already but build a cart slot adapter with a switch to read one or the other eprom...now let's see where you went with this. 🙂 EDIT: Good on you sir! except for the switch between the chips( you did a hardwire for each one when reading) I knew you'd figure out how to do it. You could have also tried to use the good cart and read the data from it using peeks and writing them out to a cassette or some device....that could have worked well too.
It would be pretty neat if you could reach out to another youtuber to help you recover the damaged eprom. If there's a different version it would be a great artifact
I was able to recover dead eproms by heating the up with hot air and reading the multiple times, until two or three consecutive dumps are identical. U sing this method I was able to recover some very rare software from a PDP11 controlled CNC milling machine.
In general, most EPROM technologies are easier to read at high temperatures. So maybe if you would heat it with a hot air gun the FFF would become data.
LOL. Spent the first 10 mins shouting "there's an edge connector" at the screen :) ... Again I commend you for taking us through the whole process, mistakes-and-all. Respect :)
Are you kidding? It's also got a full working BASIC interpreter in it! If there's any way to communicate with another computer, you could just write a quick program to use PEEK on the whole thing and dump it that way.
But that twisty connection to the EPROM reader was funny to see.
I'd have just written some code on the computer itself to dump it over serial or something. Let the original machine do the work.
@@8bitwiz_ I used to poke and peek all the cartridges for the Coco. I had 64k when the Coco had 4k. Then I uploaded to the cassette tape. Then I had 360 5 1/4 disk drives that I uploaded that turned a 720 5 1/4.
If one has an EPROM dumper that allows adjustment of the supply voltage, I would suggest trying to read out the chip with VDD values starting at 5.25 volts and going downward by 0.25 volts each time until most of the bits no longer read the same value. If one finds that e.g. everything reads identically at 4.5 to 5.25 volts, and from 3.5 volts to 4.25 volts, but one bit reads differently at 4.0 and 5.0 volts, odds are very good that's the erroneous bit. If a bit wasn't programmed quite as strongly as the rest, it might read as blank at higher VDD levels but as programmed at lower VDD levels. If a bit wasn't erased as well as the rest, it might erroneously read as programmed at lower VDD levels but as blank at higher ones (which is why I suggested going up to 5.25 volts).
Great job with this.
A common cost-saving measure for simple double-sided boards was to skip the through-hole plating and to "top-solder" instead. Apparently they also decided that supply decoupling capacitors were not necessary but it would be good practice to include these if you're making a new version.
Right. I learned since then that apparently that was pretty common.
Do the EPROMs benefit from a decoupling capacitor as well? In the updated design I used a single EPROM with a NOR gate and I used a decoupling capacitor on the NOR gate, but not on the EPROM itself. I guess it may be close enough to help both 😃
@@NoelsRetroLab Not going to do any harm, but there are plenty of boards with no decoupling capacitors so it is not really a worry. Just have wider power and ground traces on the board and you will not need them, as there should be decent enough decoupling on the other end of that connector anyway, plus at lower frequency not really going to be a worry with TTL chips. When you have large S100 boards, with a hundred or more TTL chips on the board, yes you need decoupling, but a lot of avionics boards get by with only one capacitor at power entry, and survive over a way tougher operating regime. One per chip probably not needed, but one per dozen probably will work well enough.
@@NoelsRetroLab , one capacitor per IC is best practice but one shared among the (few) chips on the board is a good compromise.
A decoupling cap (effectively) provides a local high-speed reservoir for the IC's instantaneous (switching) current demands so as not to cause supply/ground dips/bounces on the supply rails of the IC as its transistors rattle away.
Having any capacitor at all is better than relying on the impedance of the traces (and edge connection) back to the power supply being low enough to avoid troublesome VCC/GND disturbances under these fast transient loads.
That said, it's demonstrably OK to omit the capacitor(s) (just not ideal).
@@adfeldhaus A typical EPROM is a purely combinatorial device. On e.g. a 250ns device, if there is a power supply glitch, the outputs may be meaningless until 250ns after the supply stabilizes, but that would only be a problem if the device is operating near the edges of its timing margins (if the system would require that the device produce valid data within 260s, but the supply takes 20ns to stabilize after a glitch, that could result in erroneous operation, but in practice timing margins will seldom be that close.
@@flatfingertuning727 , Good point. Since it's fully static and stateless a (parallel) EEPROM like this should settle down to a correct state pretty quickly after transient supply glitches.
I'm so glad to hear you are cloning this cartridge's circuit board. Something this rare deserves a new lease on life. Thanks also to the fellow who sent you his cartridge to get data from. Seeing you being so careful to honor his wishes was uplifting. Well Done!
Just a thought, but couldn't you have dumped the ROMs directly from BASIC using PEEK?
Easy enough to write a Basic program that could save the ROMs in Intel format ready for an Eprom programmer.
Since there were differences in the second ROM, I'm wondering if it wouldn't be a good idea to reprogram that as well. Just to have a consistent version across the two chips.
i would not remove a original working chip just to get the copiright text the same.
@@Bisen222 It's not about getting "...copiright [sic] text the same...", it's about making sure that any references in one ROM to the other ROM are actually correct. If there's a jump vector in ROM #1 that points to somewhere in ROM #2, and ROM #2 is different from what ROM #1 expects, that's a problem. You can't replace the last chapter of _Moby Dick_ with the last chapter of _Gone With The Wind_ and expect it to work.
@@mykalimba right you could run into crc mismatch type issues.
@@pedrocx486 I will persevere and somehow survive this.
@@pedrocx486 [sic] burn, dude. 8/8.
You really make this aspect of hardware and electronics easy to understand and enjoyable to learn. You make me feel inspired to go deeper with my very limited electronics experience. The information you provide - though limited/geared to retro computers - can be transferred over and used on other electronics. Thank you for also taking the time to explain schematics and how to use different tools (like oscilloscopes) in your other videos to troubleshoot. You are a wonderful teacher!
Another excelent forensics video from Noel. It is amazing how well you describe the process and the different approaches to do it. I wish I had had you as a professor at the university. It would be also nice to record the viewers, as I found myself shouting at the screen: "you are getting feedback from the other ROM". Amazing job as always, mate!!.
Interesting. A while ago you mentioned you had a mystery cartridge and I though it was a copy of the Kamil Karimov MSX emulator. I was almost right :-)
Anyway, back in 1986 Peter Zevenhoven made a software emulator published by the Dutch Spectravideo CuC users club, basically he was doing the same trick taking a 728 MSX ROM patched with the hardware IO addreses of the 328 hardware turning the 328 in a MSX compatible machine allowing most MSX games to run. MSX emulator is loaded in RAM page 02 of bank 0, next a RAM bank switch is setup and a copy is made to page 21 of bank 2 (disabling build in ROM) and a reset is issued turning the 328 into an MSX. I do not believe CuC sold this emulator (possibly due to MSX copyright infringement), emulator circulated on club events and it was as such distributed. At that time I was an active member of CuC and I still have copy of that emulator on floppy.
I guess the compatibility would be worse with the emulator than the cartridge since the former is eating 16Kb ram that may be needed by the MSX games.
@@manuelastudillo8843 The situations is somewhat worse, SVI 328 can only switch banks in 32K page sizes.
I think those unused top pads coming off was due to the fact that the through-holes were not plated. Just not much but a small amount of 30+ year old adhesive to stick them down. Interesting video!
I had the same thought. Be good to know what Noel thinks about this; could be handy knowledge for similar devices.
Excellent video as always Noel. Looking forward to seeing what a man of your considerable electronics talents comes up with regarding those new PCB designs!
Thank you! So far it's been really fun the initial hardware design that I started doing!
Large solder pads hopefully lol
I wonder why you were not reading out the ROM from the SVI, in basic.. and then save it to floppy. That's usually what people on MSX do to dump ROMs in case they don't want to open the machine.
Hmm, given the choice between soldering a bunch of wires and writing a simple BASIC program to dump the cartridge contents to the screen and doing screen captures to get the contents, I'd choose the second. Unlike with say a game cartridge, you have BASIC at your disposal, a FOR/NEXT loop and a PEEK is all you need to see what's on the cartridge, I would think.
I've always been able to wire up some kind of serial interface for systems. Usually the joystick ports have some kind of output pin. You just need one. Though in BASIC it might be tricky to do async serial timing reliably and fast enough to not take a day to send the data.
Before I retired, I used to do this type of thing a lot. When companies refuse to give you the raw data, you have to improvise and read the PROMs or even ROMs if you are going to reverse engineer the code to update or change. We had the best luck using the clip on spring connectors to read them. Especially if you couldn't read them via edge connectors. And even having the machine/operational code, in order to reverse the code you had to use special programs to decode the readings. But I loved doing such things. An amazing challenge, but also an amazing reward when you "cracked" the code...
I got an old o-scope years ago and nearly first thing I did was dump its two EEPROMs (and posted them to a few forums for others), in case they ever become corrupt. Probably 30 years old at this point, around when EEPROMs start to fail.
Very glad to hear that you are going into board design!!
Thank you! It's been super fun so far (even though it has been a super, super simple project just to get familiar with PCB layout tools). Looking forward to taking it further.
Nice! Anything MSX related that can be preserved should be preserved :D
Very true!
For tenuous connections like your socket on the bottom of the PCB or the piggy-back, a dielectric grease works wonders. In the states you can get it at any auto parts store sold as "bulb grease" or grease for battery terminals.
Very very cool video Noel hopefully some uploads coming soon we miss you
Coming up in an hour! I took a bit of time off over the holidays and then this upcoming video took longer than usual.
Non plated double sided was common in the '80s and thus top soldering was common. They even had all metal sockets that allowed top soldering. If you make a replacement you may want to use a 32K or larger EEPROM or EPROM and a logic chip.
I was wondering on that first EPROM if you tried to erase it via UV light first? I know it shows up as all ones, which means an empty EPROM, but I thought maybe doing a UV erase that would completely reset it and make it work again possibly?
Best not to do anything to alter the original chips. There are a variety of quasi-analog ways of reading out data from a chip that's suffering from bit rot (the simplest of which is using varying VDD levels when reading) but erasing the chip would render any such recovery efforts impossible.
@@flatfingertuning727 No one is going to try and recover the data from that bad EPROM. They already have an image that's good enough.
@@djmips If you don't care about preservation, what's the point of all that? Sounds like he basically made a whole new copy of it anyway, reusing I guess just the aluminum panels.
I really appreciate that Noel shows the failures along with the successes! Noel has made me much more comfortable with taking chances when I do repairs (my repairs are far, far simpler I hasten to add).
Great video, Adrian! Thank you!
I cut my teeth on 8-bit home computers - I bought an Amstrad CPC-6128 which I still have.
I haven’t gone as far as to create this kind of project but I did have fun making ROM socket boards to fit other ROMs.
Maybe if I get time, I’d like to go back and explore adding some kind of storage device to my Amstrad. There used to be advertisements for “Winchester Drives” but they were hugely expensive. I’m guessing there are other solutions on CF or SD cards that can be implemented now.
All my university notes were transcribed from handwritten scribbles to word processed pages with digitally created images. Neither format is compatible with what I have now.
Although, my technical illustrations and technical authoring has evolved hugely since 1990.
This is not Adrian's basement! Hahaa...
Quality vintage preservation, sir. Got a subscription.
Why didn't you just dump it using the SVI? A quick machine language rom dumper would let you save out the image and then you could split it back to the two eproms.
Nice work! I believe you need to erase the old eprom first before it will program correctly again. Since you removed the window cover, you might try exposing it UV light and give it another try. It may very well work again!
A reading of FF is what you get with an erased EPROM. He already read all FF implying in was already erased.
@@1971merlin A necessary, but not sufficient condition.
@@1971merlin The data read out is binary. On the floating gates it's analog, a certain number of trapped electrons. It may read as a 1 but still have electrons.
@@gblargg wrong way round. An uncharged gate reads as a 1. Burning a rom cell adds a charge that flips it to 0.
@@1971merlin What did I write that seemed to differ from that?
There's nothing worse than ploughing on thru a set of well thought out instruction/theories and realising that something you completely forgot about would've saved you time.... loads of time. Been there and done that, too many times!
15:32 Don't de-solder the whole chip, just break the track between the pins and put a solder blob back later
That chip was toast so it had to be desoldered anyway. But for the other one… cutting the trace would probably be better. And yeah, I think he probably should reprogram the other chip since there could be some differences.
Putting all that heat into the chips, and almost destroying the PCB because he's too retarded to use a scalpel knife? Scary!
I realise that he subsequently found that the EPROM was "dead" but he really did not need to remove either chip to test it or try to grab a dump. Also, he tried to "reprogram" without erasing the EPROM first... so that was never gunna work.
Amazing video! A rare piece of 80' spanish engeniering! Thanks for the video and merry xmas!
Thanks, you too!
Many thanks to you and especially to the owner of the still working cardridge, thanks to all the efforts and trust another piece of homecomputer history was preserved.
Isn't it possible to dump the roms on the computer directly, like its possible on a C64? so u could save all your adapter working, even it's quite amasing.
Yes, once I figured out how they were mapped, that would have been possible. It all started because I thought it was going to be easier to read the EPROM directly. Ha! 😃
I am not sure that I would have touched the faulty cartridge. Just leave it as it is as a piece of historical memorabilia. The PCB is very simple. (Good idea for your first PCB design project.) You could have given the donor his original cartridge back untouched and your first replica cartridge using your reverse engineered PCB. But I suspect it all comes down to what the donor wanted you to do. Either way, your approach to reading the working cartridge was awesome and it is so much more helpful to show your failed attempts and thought process. Makes for a far more interesting and educational video. Until recently, I would have put Adrian Black as my favourite retro computer blogger but, from my perspective, you are just about inching ahead. Keep up the awesome work.
A 27128 is a !6 Kilobyte eprom A0 to A13 = 16K total , but on the edge connector there is no A14.
It seems that the eproms are mapped in the same address space.(only 1 eprom selected at a time).
Doh! You're right. I don't know why I said 8KB (and then I went and reinforced that with the text overlay). Each EPROM is enabled separately with a different /CS signal (there are 4 of them in the SVI cartridge connector).
If you made a new board you could use a 16k eprom and some logic to take the cs signals and feed them to A14.
@@ericpaul4575 You mean a 32K eprom? 27256
16K eproms does not have a A14 Line
@@pcuser80 yeah that.
I have never heard of any of the computers you mentioned. I'm immediately intrigued by this channel. Good job!
RMA 04-HV - best soldering flux gel I know for repairing all these 30-40 years old devices. It support cleaning solder from contamination (let it boil using soldering iron tip and all mess like rust just float on surface, easily removable) and it support solder flow to both sides of PCB through holes - everyone can do factory quality level soldering using RMA 04. Just be aware that solder can flow to chip legs all they way up on other side using RMA 04.
Thanks for the links to the ROMs and docs. Looking forward to seeing the videos about cloning of the cartridge.
Wouldn't you have to blank the old chip before trying to program it in the Minipro? Or did you do that off camera?
Yes, I had the same thought. Without blanking it under UV light first, programming won't work.
I initially tried programming it just like that since it was already in an "erased" state (at least the data bits). I tried erasing it and reprogramming it afterwards off camera but it made no difference. It looks like something fundamental in the addressing or control was totally busted.
@@thebyteattic Usually, you can re-program an eprom without blanking it, if you want to change bits from 1 to 0, but not the other way around. This one was all 1s, so looked like a blank one, and should have been reprogramable if it worked.
How very interesting, I look forward to the follow up episodes. Thank you. Not knowing a lot about MSX I'm curious to see how well this 'simple' ROM replacement works. I remember when I designed a 9958 / AY upgrade board my Coleco ADAM, I decided I might as well add some multi-standard compatibility - that required a bunch of jumpers for the differing addressing. It would be funny if all I needed to do was HEX-edit the ROM jump tables!
Yep, my take on an ADAM Super Expander will play games for just about any machine containing a Z80, Ti GFX and sound chips, or an AY. Sadly I never got the native 80Col support to work - I'm not a software guy.
Such mods are not that hard as much of the work can be found within the RC2014 community... As the ADAM is a bit big my 'daily driver' retro computer is an Amstrad PCW 8512 system board crammed in to a 8256 keyboard with 'pretend' Amber Monitor circuitry, a Gotek, the same GFX and sound upgrades, and a more modern Z80 variant included in a 8256 - RC2014 bus adaptor - got to love how easy Ti made it to integrate their line of GFX chips in to any old system, and that CP/M is so gloriously dumb it will put up with almost anything! :-D
PS
I have a semi-finished page on Thinyverse about the 8256-keyboard mod if anyone is interested. Just don't be using a 9512 system board, it won't fit.
Would it have been possible to read out the content of the ROM directly with a small basic program on the machine itself?
Lot of time and effort, thanks for sharing the hard work, especially with a bad quality PCB.
If the cartridges had slightly different versions of the software, the repaired one now has the half 8KB of the other version mixed in.
My spidey sense was like "wait, there are two connected EPROMs! How can you read only one without disabling the other one?" :-)
The PCB was more likely home made so the through holes would not have plated holes.
It would also be a good next step to make a NEW pcb with correct plated through holes.
This was also my thought too
Me three.
The PCB is not "home made"... there simply was no (easy) technology to plate the through-holes back in those days. It was made from double-sided copper PCB material, which was then etched both sides. That's why they used tube things that looked like rivets if they needed to create a "via".
I wonder how 1 of the 2 EPROM's became FF'ed?
It does amaze me vintage EPROM's still work at all.
Backups have always been a thing, if only people knew (Not viewers of this channel of course).
Very exciting, your SVI content is unique and really appreciated, thank you.
Thanks for showing this. It would have been interesting to try to read the EPROM vi athe cartridge connector, I think that would have all the necessary signals on it?
Yes! In retrospect that would have been possible. At the beginning, since i hadn't traced out the connections yet, I wasn't sure if there was something else going on. But it would indeed have been very simple to do that. Maybe I'll mention that in the followup episode. Cheers!
Hi Noel, I just found your video. I happen to be selling an SV-328 with this exact CCG cartridge (I bought it from the Spanish company back in the day!). Unfortunately, my cartridge doesn't work anymore either.
Really enjoyed watching this, it was very interesting to see the different methods you tried to get it working!
You should tape up EPROM holes as soon as you program them. With some UV-opaque material, of course. There's residual UV everywhere and it slowly erases EPROMs' contents.
Now I see you used the original sticker soon after, but for 100% security that should be immediate.
@@JCCyC back when I was a kid without a decent UV source, I tried a week with some eproms out in the Aussie sun. Not a single bit flipped.. It takes some high energy UV in a concentrated manner to erase an eprom, and stray light sources is rather unlikely to really cause issues. There are some decent articles on hackaday about using direct sunlight, and for one it was 2 weeks before weak bits started, and 3 weeks before full erasure. That is a lot of light needed. There is little to no UV-B or UV-C in the average house or workshop so you could consider years before bits get erased, if ever.
You've done a damn good job with this!
Why You don't use any solder flux? I think It would help solder to penetrate to oher side of the bord and facilitate soldering.
Would it be possible to use a single 27256 in the modern version? They just seem to be a lot easier to get, and you only need one so no chance of mixing the two ROMs up.
Yes, but stating the obvious, you would need a custom made PCB.
I salute your dedication to the sticker!
If the ROMs on the borrowed board were slightly different, should you not have reprogrammed the 2nd ROM as well, because they may look like they work until a call from one ROM hits a slightly different location of the other ROM i.e. they are miss matched?
A simple FOR loop and PEEK instruction from BASIC, would solve the problem.
Great video! It's cool which someone takes care of the preservation of old and odd piece of computers history. 👍
I was wondering could you just read the cartridge using a small program on the SVI.
you aparently are not aware of the function of those windows to the die. These are EPROMS , not EEPROMS, which means that they get erased by exposing them to UV light. Since every light source has some UV component in it, exposing those to it may cause the EPROMS to get erased or get data inconsistency. Therefore they are covered. Just take that into account if you didn't knew about it, i am trying to help, I hope my critic is not too harsh. I really enjoyed the video, keep the great work going!
i think they make clip on chip interfaces that can clipped on and give access to the pins without unsoldering.
if he is wanting to not want unsoldering because he wants to extract maximum collectors/resale value you could offer to buy it from him.
if you have friends or could make friends in the computer forensics department of the police or the fbi you could have them help you as they have to have reliable recovery because criminal cases has to have reliable evidence otherwise the case can be thrown out.
you could use exacto knife to cut the trace between the 2 pins and then use solder to re join the pins.
If you are designing a PCB, why not build a MSX to DIP adaptor as a PCB with some dip switches to allow chip selection. It would make restoring other such cartridges a brease in future. Maybe Make a multi system version that does C64 as well?
Cool Video, (Yes we were all thinking the same thing, "what about the other rom?") thanks Noel.
15:15 cut the track, then solder bridge it back when you are done?
Nice video again. How does the defective rom look under the microscope? Are all leads to the die still connected?
Noel. Whats model of desoldering gun do you use?
Great video!! I remember these cartridges from when I was a kid. It would be great to get one so any tips where to search for one of them would be nice. Otherwise, I will try to build one.
Hey-hey, a surprise Noel's Retro Lab Christmas present! Awesome!
There's a "First Noel" joke in there somewhere, I'm sure... 😉
Which version of the rom is newer? Was your cartridge a newer revision?
for future reference I have definitely seen clamps that attach from the top to read roms in situ, this would be a perfect application
nevermind
9:34 i can FEEL any of those wires just randomly de-attaching
Those were actually soldered in place, so they were pretty good. The previous attempts not to much though! (And yes, I still wouldn't trust it more than a few tries 😃)
Cool, and so another rare thing gets saved! I've honestly never seen a pcb that needed to be soldered on both sides, but then again, I don't really solder a lot ;)
A verry long time ago I had to a similar ROM backup, I had used a DPDT switch to switch between the roms as it was literally change 1 pin on each from VCC to Ground and I had a toggle switch handy. The issue I had was not 2 sepearte ROMS as much as bank 1 and bank 0 switching with a reader that could not actually read the full ROM, and using a hex editor to patch both banks into one file and managed to reflash into a single ROM of full capacity ( 2 of 16k into one 32k ROM ).
Yay...MSX content. Thanks !
Interesting little cart there, its awesome you were able to dump the Rom preservation of these things is so important. Really enjoyed the content and breakout of all the pins and the methods you used in those attempts. Really came away with some useful knowege. Liked and Subcribed, looking forward t o seeing more great work!
great job as usual Noel!.. look forward to seeing some of your board designs..
If you use pcbway to make the blanks, can you make it public?
Absolutely! It's already done. Just waiting to finish the video editing 😃
Well done ! This made my day.
16:26 -- your friend was wise to not want this board to be de-soldered.
Would it not have been simpler to just plug the cartridge in and write a small basic program to dump the ROM address space? I've done this on various other systems, though I admit to knowing nothing about the SVI machines.
Very interesting video and approaches to problem solving :)
Hi Noel, I appreciate what you've done here, but is it possible to provide links to both ROM dumps separately so we can write our own ROMs? Cheers.
It should be possible to replace this with a single 16kB EPROM with a few logic gates. With SMD, you could make the whole thing tiny, like 10mm high, basically just the edge connector.
Well done Noel, I did not expect anything else :-D Michael
im looking forward to seeing the cartridge replicated using modern techniques, especially the idea of a 3d printed shell and cool label art :D
I was taught to place a sticker on the EPROM window immediately having written it.
Wouldn't solder flux help with solder flowing?
And yes, i also support idea of reading rom via basic. Surely this wouldn't came as pretty as reading individual eproms, yet you'd have only to divide dump into 2 equal parts and reflash single one.
surely approach taken by you is much more interesting and educational :)
what tool are you using to compare the roms ?
Can you just read ROMs on target computer and put it to floppy/hdd?
Yes, that is precisely what he just did. In this video, the one that you watched before typing your comment.
Excellent work as always!
I have not watched the entire thing yet so maybe you did this already but build a cart slot adapter with a switch to read one or the other eprom...now let's see where you went with this. 🙂
EDIT: Good on you sir! except for the switch between the chips( you did a hardwire for each one when reading) I knew you'd figure out how to do it. You could have also tried to use the good cart and read the data from it using peeks and writing them out to a cassette or some device....that could have worked well too.
Where did you get that Spectravideo T-Shirt Btw? or any of the other T-Shirts, I guess most come from the same place, disclose please! :)
Vote up, nice video clip, thank you for sharing it :)
It would be pretty neat if you could reach out to another youtuber to help you recover the damaged eprom. If there's a different version it would be a great artifact
great work, thank you
I was able to recover dead eproms by heating the up with hot air and reading the multiple times, until two or three consecutive dumps are identical. U sing this method I was able to recover some very rare software from a PDP11 controlled CNC milling machine.
were did you find it
3:00 it's the flux clean the flux melord!
My first idea would be to use an edge connector and an Arduino to work the chip selects, Address, and read out using GPIO.
A very interesting video. Thanks!
If you want to reprogramm it several times, maybe you can use an EEPROM for your board instead of an EPROM.
In general, most EPROM technologies are easier to read at high temperatures. So maybe if you would heat it with a hot air gun the FFF would become data.
3:20 27c128 is 16K EPROM not 8K