Oric Repair: The "Impossible" RAM Failure
ฝัง
- เผยแพร่เมื่อ 16 พ.ค. 2024
- This episode is sponsored by PCBWay www.pcbway.com
What should have been a relatively straightforward repair of an Oric Atmos, turned into a wild chase for the true source of the problem. Tests kept pointing me in one direction, but my gut told me it was something else.
Support Noel's Retro Lab on Patreon: / noelsretrolab
You can also support Noel's Retro Lab on TH-cam by joining this channel:
/ @noelsretrolab
Chapters:
00:00 Intro
01:22 Look inside
03:43 Voltages
05:21 Clocks
08:16 RAM
09:17 Diagnostics ROM
12:50 RAM circuit
15:02 ULA
16:08 Power
18:51 Last try
20:38 Testing
21:13 Making sense of things
23:46 Conclusion
Links:
Oric diagnostics ROM oric.signal11.org.uk/html/dia...
Blake's 7 www.defence-force.org/index.p...
Music tracks:
Funky Stars by McKlain mcklain.bandcamp.com/track/fu...
Battro OST by McKlain mcklain.bandcamp.com/track/ba...
More awesome music by McKlain: www.mcklain.com
🛠 Tools I use ➤ noelsretrolab.com/tools.html
Connect with Noel's Retro Lab:
Discord ➤ / discord
Facebook ➤ / noelsretrolab
Twitter ➤ / noelsretrolab
Instagram ➤ / noelsretrolab
Mailing list ➤ noelsretrolab.com - วิทยาศาสตร์และเทคโนโลยี
The massive speaker made this the perfect machine for sneaking into the computer shop, and writing a large for-next loop, followed by the EXPLODE keyboard and running away. Not that I ever did that.
Those were some of the best BASIC commands ever implemented in a computer too! 🤣
OMG I remember that! That was a nostalgic trip down memory lane!
10 EXPLODE
20 WAIT 200
30 GOTO 10
RUN
o.k, that's what i used to do on the amstrad back in the days...
Yeah did a good impression of a metal guitar given the right params.
Awesome video, bad luck indeed! You might be helped a lot by the Romulator when working on 6592 systems like this, it’s very much worth the $35 USD price.
Thanks, Adrian! I'll definitely have to check the Romulator. It seems like a very different way to go about diagnostic systems. I like it!
HAHA Adrian, I was thinking of that while watching this, and am glad you chimed in. Because the 6502 is so difficult to diagnose a system for (due to it being register anemic), the Romulator makes it so much easier.
The biggest problem with most RAM testers, and even chip testers, is that they tend to be forgiving about the timing values. For RAM testing especially, though, you need a tester that knows the minimum required RAS time, minimum required CAS time, and minimum data in/out times, and it needs to test according to those timings. TTL chips should also be tested against their datasheet timing specs.
The Oric 1 was my first personal computer. I was a poor student, and I loved the opportunity to finally be able to have my own programable computer in BASIC.
I really like the way you fault find. logical and concise! I run a small workshop in Derbyshire UK repairing
vintage studio and musical equipment. I hope you don't mind but I have used a few of your videos to show to my apprentices.
It helps to have a second voice of reason when trying to get the youngsters to do things in a logical way and not jump in all guns blazing,
assuming they've seen the fault before, and not just the ramblings of some crazy 60 year old ex BBC engineer just doing it the old BBC way.
Anyway Thank you.
I love you finally got your hands on an Oric and made an episode about it. Amazing repair! And thanks for showing a game of mine :)
DRAM can be very touchy about timing delay, especially with the timing of RAS and CAS, and how much time you must have them both selected, and how much time the bus must be held on the trailing edge of the pulse for the chip to latch the address in, and then the amount of time it needs to be inactive before the other pulse. Likely the timing here is just on the edge of acceptable on that chip, and the driving from the ULA is a little weak, so the edge is slow, and thus the latches do not get the correct address latched before it starts to change to the other, or the ULA is writing or reading the data lines before the chip has finished decoding the address, so the written data or read data is not always correct.
Seeing as screen data, also from the ULA, is mostly correct, with no bad single bit errors to show consistent failure, it must be only the CPU ram read and write timing that is off spec, at least for that chip, though likely the others are close to it as well. Likely the original design is very picky about the manufacturer of the DRAM, as they relied on the timing of a specific chip being better than the specification, and this finally has drifted with age.
Would be interesting to know who made that DRAM, and if the others you tried were the same manufacturer, and probably the same fab as well, from the number imprinted under the chip in the ejector pin marks.
Crystal failing is common enough, though you could simply have turned it around in circuit and it likely would have worked again, TTL crystal oscillators do that. Yes I have also experienced the IC working, just not in this unit, but it works in another, because the timings are just different enough, and also it works cold, but not hot, or vice versa.
Correct. According Noel's problem: a memory cycle ends when both RAS and CAS go high again. Usually this doesn't happen exactly at the same time, sometimes CAS is triggered slightly before RAS (e.g. shown in Am9064 datasheet), sometimes RAS is triggered slightly before CAS (e.g. shown in HM4864 datasheet). Noel's defect DRAM still works when CAS is raised before RAS but not vice versa, so there is most likely a defect in the column/row drivers. Hard to find this problem since a tester usually provides only one method.
I replaced all the ram in my spectrum with dual port SRAM. I can now hack and edit programs on the fly.
While I have no experience on these boards, I had my fair share of work adjusting the clock skew on DDR2
@@HappyBeezerStudios
I am assuming you were not changing anything on the DDR2 module itself?
In which case, isn't likely to be a clock skew problem.
If you were dealing with a problem affecting the timing of multiple signals going to the DDR2 module then it is likely to be a setup/hold time issue.
Clock skew is to do with only the clock line going to multiple registers/flipflops.
I've played the Blake's 7 game! It's really good, and the characters are even really good likenesses of the actual actors!
I used to watch the TV series when I was a small boy, so this made the game extra-special. 😄
I really enjoyed this quest for the truth. The tension and resolution was great! Congrats Noel! You really aced it.
Glad you enjoyed it!
Wow, you just transported me back about 35 years when one of the members of my local computer club had an Atmos and I remember thinking it was really cool! The aesthetic was quite unique in it’s day. Thanks.
I remember a guy in our town had an Oric Club at his house. I only went once because he charged everyone a £1 each for using his electricity at what was then about £0.06 per KW/hour that was quite a markup!
It's surprising, because the Oric 1 is a horrendously ugly thing, and the Atmos looks so much better.
An Oric Atmos AND Blake's 7? 80s-gasm! Noel, you rock!
Oric reminds me of ORAC, the insuperable supercomputer from Blake's 7. That's a fond memory from my youth. This computer looks interesting, and the designers made some interesting choices in the product's design phases. Thanks!
It influenced the name of the computer.
I've seen other youtubers with RAM chips that pass tests but still fail in the machine. I think it's more common than people think as most haven't had access to a decent chip tester until recently.
It just goes to prove - the best tester is the circuit itself. (I've seen that from guys who work on tube-type guitar amps.)
To test proper, you need to be able to monitor timings with precision, and more precision requires better components and design. Just like oscilliscopes.
@@rockytom5889 Really, what you'd need to test the "worst case" timing, or perhaps a range of timings.
It's always great to see an Oric get some love, I had an Oric-1 48k back in the early 80's, it was a solid little machine - I loved mine.
I even ended up reviewing a few Oric games for a magazine.
Still have my Oric-1. I liked how you could type "beep, "explode" and others and it would play those sounds out the speaker.
It was a nice little touch. Although it made a lot of games sound alike.
The Oric Atmos was my first computer in 84. I sorely miss it, having had to sell it to finance my next computer. I've been thinking of buying one for a long time now, and this video only makes the temptation greater. Nice work on finding that elusive ram issue, I bet a number of people would have been stumped as well.
Always loved the look of the Atmos, very cute little machine. Brilliant video as always Noel, glad you got it working and nice job tracking down that really bizarre memory timing issue.
My first experience of a computer was my friend's dad's Oric-1, I was absolutely amazed by it. My friend then got an Atmos himself, so fond memories, especially of Zorgons Revenge, an awesome game. More Oric videos please!!!
One note about the use of 4164: the Oric allows the ROM to be disabled, and then you have a full 64K of accessible RAM. The bizarre decision is that there's no way to do this without external hardware! Disk drive controller interfaces did this to put the OS in the upper 16K of RAM and complete the bank switching circuitry. This trick was at the centre of many Oric Atmos ads claiming you didn't lose RAM if you used additional peripherals. It was because you lost RAM if you *didn't* have any.
A minor quibble: you keep calling them "demultiplexers", but those chips are actually _multiplexers_ (they take the full range of address lines and multiplex them onto a smaller set). A demultiplexer would go the other direction (input the column and row addresses and output the full 16-bit address).
Great work with finding the issue. The Oric architecture is seemingly simple, but it’s definitely on the temperamental side.
For sure! I didn't even talk about how there's a random character that appears on screen half the time you turn it on. That could be a fun deep dive to figure out why that's happening 😃
FYI, RAS and CAS are row address strobe and column address strobe, not signal. Nice bit of sleuthing and a good result, though.
Love your video's, but it's Not a Laser thermometer. You can rip out that laser and it'll still work. It's an IR thermometer or IR gun thermometer to be more exact.
Ha! I had no idea. I assumed it was measuring some variation of the laser reflected back or something. Goes to show it's outside my area of expertise! 😃
I'm actually amazed that there is still so much interest in this machine. It had a fantastic keyboard and I loved it, and wrote several bits of software myself
Super interesting and very well presented. Thanks, Noel!
Wow, an Oric! I started my micro journey with a home-build Microtan 65, which got expanded over time, I even made a DIY RAM expansion for it. Of course, as soon as the Oric appeared my Tangerine loyalty won and I got one. Never had any problems with it and got a lot of use from it for a number of years. Moved on to the Amiga family eventually.
I had the Oric 1, then upgraded to the Atmos and then to the full suit - printer, disc drive and their power supply etc. I built an assembler and EEPROM programmer to make a moving message (another 6502) LED display - I sold a few :) Finally, I sold the whole (dust-gathering, but working) package about 5 years ago for a healthy price. Great memories and a smashing video, thank you - Beamer, from the UK.
Fantastic episode! I love your deep dives and not sticking to simpliest solutions and simply displaying "the end" ;)
Thanks for documenting and sharing the highs and lows of this diagnostic. Always good to learn from testing issues.
Brilliant stuff as usual noel. enjoy your deep dives into these retro computers.
Best Ram fault diagnostic vid I've ever watched Noel Thanks a bunch.
Amazing video Noel. Explanation and supporting graphics are awesome as always. Congrats!.
Extremely educational, detailed and enjoyable video 🙂👌🏻
Great video, thoroughly enjoyed the diagnosis process and dogginess to keep on going.
Love you videos, very informative for the architecture of boards and how the signals go.
This was really interesting. Looking forward to more videos on the Oric.
Really nice video. It give me some tips to keep in mind about memory signals.
Nice! You learn most if something unexpectedly fails. And you take the time and track it down to the source. Well done!
I really enjoy your diagnostic videos. They have taught me a lot about looking for the unlikely when diagnosing a board that should work but doesn't.
Great video -- a very methodical approach to figuring out the actual failure. Good work :-)
The Oric screen video system, oh boy.
The screen memory was also used to store control codes for any character that was displayed.
Control codes for character colour, size, flash etc had to be placed on the screen anywhere to the left of the character you wanted to display on that row.
This meant you could not have a character on the screen if a control code was in that memory address, else it would overwrite the control code.
The first two columns were reserved for overall background and foreground colours but could be overwritten if so desired.
I believe Ceefax and Oracle teletext pages used the same system.
Coming from an ZX81 to the Oric 1 took me a fair bit of getting used to.
Hi, I was happy to see ORIC-1 & ATMOS again. I bought my first ORIC-1 (48k) in 1983 and two years later one with 16k RAM. Still original packt in my storage. Learned 6502 assembler and a lot About Hardware. A Real Flash back. Thanks for the Video!! Regards - Henry
Excellent bit of diagnosis. Definitely emphasises the need not to put absolute trust in new components. Found this very interesting.
I enjoyed this video a lot. Thanks for sharing it.
Excellent vvideo. I'm learning a lot (I guess) with your detailed explanations and research of the issue. Thanks!
Blake's Seven can't have been a coincidence to have - Orac on an Oric!
A faulty crystal?! I’ve never seen that before. It would’ve been something I’d come back to when everything else has failed 🤣 Well spotted!
its not a common problem but they can and do go bad, i had a colour decoder crystal fail in a 1970s colour tv in the early 90s , it was still oscillating but must've gone enough off frequency to stop the decoder working, new crystal cured it
@@andygozzo72 I literally in 35 years of electronics have never had had that happen. I would really find it as a last resort. It’s such a passive robust component.
I built a tester for the SRAM modules on the TRS-80 Model 100 which have four 2K SRAM chips on a ceramic carrier. On one module my tester said it was fine, but it would fail in the machine. Fortunately, my friend made an awesome test harness for the M100/T102 machines, and it was able to report the failure as in A10.
Looking at A10 and nearby address lines on the scope, when the module was in the M100, there was a very obvious RC curve to A10 when it changed states. In my tester this curve was not apparent. My friend said, "The bus drivers on the M100 are right at their maximum rating for driving the capacitance on the address bus." My Arduino based tester had much stronger outputs, so a little extra capacitance did not make a difference.
You can have other issues where a gate draws a bit too much current and the chip will test OK on a tester but fail in circuit as the chip driving it and possibly other chips does not have the fanout to drive that load.
Interestingly I was able to soak that module in alcohol for hours and fix it. I suspect it was contaminated with the awful flux they used on the M100 which can turn slightly conductive.
Great analysis! I consider your videos to be required viewing in my list of retro channels.
Great troubleshooting tour video! It's wonderful to hear your reasoning for the different steps. Although I don't have an Oric, I'm pretty sure, I can bring some bits (sic!) of knowledge learned from this video to good use on other computers.
I really love the look of the outsides and the insides.
Noel - thanks so much for this video. I have exactly the same symptoms on my Oric Atmos. I started remove the ram a year or so ago and ended up lifting a load of traces, so I put it off for another day. But this comprehensive look at the various areas it can fail has given me a lot more confidence to try it again! Excellent video.
Well done!
I enjoy your videos so much, I wish you were able to upload more often, but I understand it's hard work and you have other responsibilities as well... Looking forward to the next one!
Good video. Thank you
Extremely good video. Thanks! Didn't know about that Oric's upside down PCB quirk :)
Another fantastic deep dive into a problem. My own mental checklist of “maybe it’s this” or “hmm, maybe it’s that” as you went along and I was getting increasingly confused along with you. I love the follow through and not simply leaving it as “it’s fixed somehow”.
Love the link between Blake's 7 and the ORIC Computer. The ORIC was Avon's little toy in the series, though a "slightly" different computer.
Orac was Avon's toy.
Great prononciation of the name of the Bulgarian clone Pravetz 8D.
As always, it was very good!
I would like to see more details the RAS and CAS signals in the oscilloscope.
Thank you.
The coolest thing about the Oric is that the ROM is copied into RAM on power on (that's where the other 16k 'goes').
You can then *overwrite the copy in RAM*, thus patching the ROM! For example, want to change the font in text mode? Overwrite the sprites!
Oh interesting! I didn't realize that. I assumed the ROM was simply mapped in that space. But wait, how does that work? The copying must be done by the ROM code itself, so there has to be a way in software to select either reading from ROM or from RAM for that addressing space, right?
@@NoelsRetroLab I don't remember the exact details (this was about 40 years ago!) but I got this from a book that I ordered from the local library.
I guess the test is to see if you can POKE above C000!
I'd like to know where you read that :)
My understanding is that no matter what you do, on a stock machine you always access the ROM part.
When you plug a floppy disk drive (Microdisc, Cumana or Jasmin), you get additional electronic circuitry which allows to switch between the ROM, the overlay RAM, and the onboard eprom on the disk drive, which allows the DOS to be in memory (in overlay ram) without using any additional memory - with the exception of the $400-$4FF which is used to swap between ROM and RAM depending if you are executing BASIC commands or doing disk accesses.
So yes, with a disk drive plugged, you can indeed do what you mentioned (copy the ROM to RAM and patch it), it's actually exactly what I did for my "The Hobbit improved" which was in video ("Oriclopedia: Let's improve the Hobbit") just a few days ago.
This is exactly what a PC does when you have the 'Shadow Video ROM' selected in the BIOS - it copies the contents of the ROM to RAM for slightly faster execution. As you say it allows the user to POKE new values into the character definitions now held in RAM to create your own custom fonts. :)
@@aCivilServant The Oric ULA always fetch the font from B400/B800 in TEXT mode and 9800/9C00 in HIRES mode, there's no need to patch the ROM font to have custom fonts.
Fantastic. Thx
Love your channel keep up the good work!
When in doubt, blame the RAM, and when it's not the RAM, it's probably the RAM... :P
Great detective story! A friend had both the Oric and Atmos, and it was a fun system.
Nice dive into a machine I've never heard of. I enjoyed this.
Glad you enjoyed it!
Great vid !!!
As you've experienced, sticking a DRAM into a simple tester or POST is often not sufficient. In the good ole S100 Bus days, the CP/M MEMR Rasmussen RAM test (the vintage equivalent of Memtest86) was the best assurance for DRAM to be operating properly. This program runs through a series of comprehensive tests such as walking bit tests to make sure the DRAM will operate as expected. These older machines often had no parity RAM error checking so when things went awry, they went totally awry!
Great video and chiptune music (Quazar - Funky Stars) One of my favorites. Cannot get beter than this 🙂
I have fond memories of Oric 1. It was the second machine I used after C64. We typed a few programs from the manual and also played some games. The Atmos was also well known here and on display at local shops. I almost bought one! (But in the end I got a CPC464)
That's what is lacking on many today's electronic repair persons: Perseverance to find the right problem, and you did it. Congratulation!
I upgraded from a second hand ZX81 to a brand new Oric Atmos. I loved it! I spent hours writing BASIC programs and I think I had all the games ever produced for it.
Grew up in France and can confirm. I had an Atmos and it was popular (albeit quite briefly and less so than the original Oric 1). I remember drooling over the Stratos announcement...
This is AMAZING!!!
I had 2 Oric-1's back in the day of which both died and I got an Oric Atmos as a replacement for them. Wonderful machine and the keyboard is one of the best.
Back in the late 1980's, when i was out shopping to buy my first home computer. I was in my high street "Dixon's" store in Liverpool city center and looking at all the makes and brands of computers out on display and was so near to buying an Oric Atmos that day but instead bought a Commodore 16 computer instead, only because there was more games out (on cassette tape) at that time for the C16 than the Atmos computer. Love watching your video uploads and highly jealous of all the different t-shirts (with the computer logos on the fronts) that you have in your collection. Great work again, Noel :)
The Oric was awesome. I have owned one since a teen and still have it in it's boxed. Shortly afterwards I got an Atmos, love the keyboard, that too is still boxed. Since then I have four more from various places and even managed to get hold of the official printer, okay it was a rebadged MCP40.
I loved this video! The Atmos is a nice computer and It was nice with a twist in the repair
Its called the murphy law. Bad luck that can happen to the best IT geek. Congratulations for the resilient enfort to solve the problem. Best regards from Portugal.
Great job! I had my friends Korg EX8000 synthesiser on and off the bench for almost a year. Ended up being the crystal!!!! I took one out of an old STM Nucleo board and the thing fired back up to life. I think it is getting 'that' time for old crystals.
Great episode Noel! I greatly enjoy these videos which resemble mystery stories, Sherlock Holmes for detective mysteries, House M.D for medical mysteries, Noel's Retro Lab for retrocomputing mysteries ;-)
Interesting video... Oric looks pretty cool!
Wow! I don't remember ever hearing about the Oric! Man, I would have practically cut off my left foot for a system like that in 1983. It's so small and compact, yet so much packed onto that tiny little MB! Hell, I want one now! Can you get them on ebay? Imma go look, but I bet you can't. ;)
Thank you very much for the video - it was like reading a detective novel, but about electronics. =) Great diagnostics work indeed.
Nice video, Still have an Oric Atmos computer in the loft along with a ZX81, the keyboard was great on this computer compared to most from that era.
You've got me interested to see if I can get my Oric to work, I've never had it running.
The Oric had two significant hurdles to overcome. Te Oric 1 shipped with a faulty ROM as far as I can remember. A friend of mine bought one and was given a free upgrade to the Atmos by the shop he bought it from. By the time it came out the ZX Spectrum was king of the hill in the UK and the C64 in the USA due to the massive amount of software released for each machine. In my opinion, it was usually the software which could make or break a machine in those days, hence Alan Sugar commissioning software houses to create a large selection of games, and some serious software when he put the CPC464 onto a very crowded 8-bit market and make it a success.
haha, because the main board was turn around, they use the 7905. ;-)
As a teen I used to visit a place called Hempstead Valley shopping centre which had a John Menzies shop in it. John Menzies was a defunct store, a bit like WH Smith, in that it was a news agent that dabbled in selling other stuff including home computers. There was a boxed Oric Atmos that sat on a high shelf which was unsold for the entire 3 years I visited. Not that I ever wanted to own an Oric Atmos, but I bet if I had the money at the time, an ability to haggle and the patience to wait, I could have made a considerable profit by taking it off their hands.
good vid thanks
I'm old enough to remember the Oric - never had one though. I had a ZX81 and a ZX Spectrum.
In a previous video here, I mentioned doing the replacements and doing them one at a time into another machine for people who don't have the chip tester, but was poo-poo'ed saying mass chip replacements isn't the way to do it. I like the fact you used a known good working Oric to test the RAM and discovered that _was_ the issue, even though the other test with the chip tester said they were all good. It's be interesting if you posted a follow up (pinned) message that you tested the other batch of RAM with the new firmware, and whether or not it found a bad chip.
very impressive!
I used to sell the Oric-1 in Iceland, working for Bókabúð Braga back in the time (of course along with all the other 8-bit computers available at the time... )
There were suprisingly many buyers...
As long as we learn something than it's always fun when tech throws us for a loop. I have a ram stick that works only in specific slots. It's an 8 GB DDR 3 and it won't work in DIMM slot 1 on any motherboard that I currently have for DDR 3, but it does run in DIMM 2 and has no issues that I can find. I'm not using it not sure I should use it and it just doesn't make any sense to me right now.
Great video. I had exactly the same issue with mine. It was faulty with multiple failed ram chips so I did a complete swap with some 4164’s from China. Wouldn’t work. I swapped the multiplexers, inverter, tried a different ULA. I fell short of swapping the CPU. Turned out that although the ram worked, the timing was not as advertised. Finally found a set that worked!
Another grest video, Noel. Thanks!
Nasty ol' RAM chip! Well done! That would have drove me nuts.
Wow! That was bad luck! But great to see you not giving up and finally deducing the course.
Really enjoyed the troubleshooting process, having it last this long is like a huge mystery with a twist ending where you're always guessing who the culprit will be.