I just can't escape computers with RAM problems!
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
Вставка
- Опубліковано 7 тра 2024
- This is turning into some kind of meme at this point at how many things I work on with RAM problems. This time we have a PC XT clone motherboard that used to work, but is not exhibiting random RAM failures. Let's see if I can fix this motherboard.
Part 1: This part!
Part 2: • I might have to give u...
-- Video Links
Broken Book8088 unboxing:
• 0078 A broken Book 808...
Book8088 Repair:
• Finding an intermitten...
Turbo XT / Super XT BIOS:
github.com/virtualxt/pcxtbios
SuperSoft Diagnostic ROM:
www.minuszerodegrees.net/supe...
Adrian's Digital Basement Merch store:
my-store-c82bd2-2.creator-spr...
Support the channel on Patreon:
/ adriansdigitalbasement
Adrian's Digital Basement (Main Channel)
/ @adriansdigitalbasement
My GitHub repository:
github.com/misterblack1?tab=r...
-- Tools
Deoxit D5:
amzn.to/2VvOKy1
store.caig.com/s.nl/it.A/id.16...
O-Ring Pick Set: (I use these to lift chips off boards)
amzn.to/3a9x54J
Elenco Electronics LP-560 Logic Probe:
amzn.to/2VrT5lW
Hakko FR301 Desoldering Iron:
amzn.to/2ye6xC0
Rigol DS1054Z Four Channel Oscilloscope:
www.rigolna.com/products/digi...
Head Worn Magnifying Goggles / Dual Lens Flip-In Head Magnifier:
amzn.to/3adRbuy
TL866II Plus Chip Tester and EPROM programmer: (The MiniPro)
amzn.to/2wG4tlP
www.aliexpress.com/item/33000...
TS100 Soldering Iron:
amzn.to/2K36dJ5
www.ebay.com/itm/TS100-65W-MI...
EEVBlog 121GW Multimeter:
www.eevblog.com/product/121gw/
DSLogic Basic Logic Analyzer:
amzn.to/2RDSDQw
www.ebay.com/itm/USB-Logic-DS...
Magnetic Screw Holder:
amzn.to/3b8LOhG
www.harborfreight.com/4-inch-...
Universal ZIP sockets: (clones, used on my ZIF-64 test machine)
www.ebay.com/itm/14-16-18-20-...
RetroTink 2X Upconverter: (to hook up something like a C64 to HDMI)
www.retrotink.com/
Plato (Clone) Side Cutters: (order five)
www.ebay.com/itm/1-2-5-10PCS-...
Heat Sinks:
www.aliexpress.com/item/32537...
Little squeezy bottles: (available elsewhere too)
amzn.to/3b8LOOI
--- Instructional videos
My video on damage-free chip removal:
• How to remove chips wi...
--- Music
Intro music and other tracks by:
Nathan Divino
@itsnathandivino - Наука та технологія
I had a thought about what the problem could be, then immediately forgot it. Darn memory error!
Hahah. Good one,
I love the way the Supersoft diagnostic fails to detect a valid floppy controller, but then still thinks it should carry out a floppy read test!
I've not heard it mentioned but static ram particularly early 80s is EXTREMELY static sensitive. Static damage isn't always immediately obvious which is why people often think they've had no trouble. The metal oxide layer is extremely thin and esd blows pinholes in it that grow with and without use and will be the reason they fail unless theres a mechanical failure of the package. Where I used to work we implemented extensive esd protection and field failure returns went from 5% which was average to almost nothing 0.3% for serial numbers after the changes were made. If it wasn't for that changing my mind I wouldn't mention it. A static strap is ideal.
I agree, RAM is still among the most ESD sensitive parts these days but back then they had practically zero protections.
Static discharges small enough that you won't even feel it is enough to kill this old stuff a lot of the time.
If run within it's specs and without being exposed to any static events RAM should have among the lowest failures of any component in a system.
Yeah, back in the 80s it was still a bit hit and miss with ESD, some manufacturers of chips had quite big issues for quite a few years and it took quite many years before the industry figured out a working standard how to protect from ESD damage. But still, even today the ESD damage is estimated in billions of dollars per year.
And in my professional work I have to do mandatory training every year and measure workbenches, soldering stations, shoes etc pretty much all the time - it's boring but needed.
At home I just use some cheap wrist strap and an ESD-mat and keep plastics away from my working area, with this I know I'm probably 90% safe and since it will only affect me if I break something it's alright.
I'd wager that he's grounded somehow - chains around his ankles, metal shoes, tinfoil underwear? There'll be something.
Yes, it actually became a huge problem even for populated boards. They ended up having to add some protection, but cost was always an issue, and RAM was always pushed to extreme limits because nobody ever had enough fast RAM. I'd hope he is protected he's knowledgeable to the extreme, but I still no people who think esd is a myth but I wager its the no.1 failure mode of these machines.
I’ve noticed that lately. I rarely used an ESD strap anymore when working on new stuff, just hang onto the chassis when moving stuff around. But if I forget when working on anything older than XP era, I’m replacing chips or a CPU.
I'm so glad you mentioned the glitching on the video, I was about to troubleshoot my own system.
It could use screen ram.
With all RAM in sockets, why not pull all the RAM chips, install 1st bank, run test, if it passes, add 2nd bank or RAM chips, and test until all RAM chips are identified and bad / good.
I was wondering if he would do that and test it with his 512k static ram card to just check system stability. Can that thing even map to $000000?
As he said the error is random.😁
I really expected the ram to get pulled and a more targeted testing of it needed to be done. Today he looked/pondered of almost anything else except the ram. I know most people don't have a ram chip on a card but that would be a cheap and simple way of ruling out or pointing at the ram...
@@josedias5514 I believe also that his TL866 eprom programmer can check the ram chips. This is fairly quick to do, despite the number of ram chips involved.
Seems rather as bus/refresh problem.
My hardware is 46 years old and I reckon theres at least a couple of faulty RAM banks. The RTC's also buggered because I cannot remember the date everytime I resume from sleep. Not just the date, I often get the year wrong. Soundcard is a bit dodgy too and I can only see lower resolutions without bulky adapters. Input output ports still work but I'm sure one of them will develop a fault soon, I know that I have a low-level interrupt that wakes me from sleep before the sun comes up - and it takes a lt longer to boot than it used to but we get there in the end once overclocked with a bit of coffee. But just like those old ST506 HDD's I remember a complete set of song lyrics from 1988. But pretty much everything else I know and do is incompatible with the 2020's and I'm not able to run modern apps like social media because I spent my life trying to avoid being social. My wife is from a different land and so we have incompatible character sets some line up but not others. The good news is that I have no idea what my mother-in-law is saying and just direct it to /dev/nul anyway. The kids are unicode so thats ok and gives them plenty of space for a whole bunch of characters I have no idea what they mean. But we get by, as long as I can still run the software they need me to run like paying for stuff and driving them around then I'm sure the'll keep me going. I thought about retro-brightening the wife but that didn't go down too well, we just have a carrier bag to keep all the bits in when they fall off - that way someone else can restore it later. Hopefully there's a few more years yet, but there comes a day when we will no longer boot and those memories become unreadable forever. But someone will still be running a bit of my code.
looks like marginal timing, that is consistent with changing the logic family of a part that is involved in the - in this case - memory access. The timing on these things was petty much always close to the edge, and putting in a latch, buffer or in this case decoder that is a different logic family can push it over the edge. If a "c" part fixes it that would be proof. Very hard to diagnose accurately without a logic analyzer.
Hello Adrian,
As you're troubleshooting the crusty 245 chips, I can't help notice what appears to be multiple breaks in the traces that run past the 'ends' of the 245 chips -- right where you were saying that there had been previous corrosion. Have you tested those traces? Even if you tested them previously as good, could they have continued to degrade? Is it a visual bug from the camera angle and lights?
Love your channel as always! Craig Zody
Yeah that's unfortunately a visual artifact from the camera and lighting. They are actually OK
That extra ISA footprint is likely to be for a riser card socket. This thing is leading you on a merry chase! It may degenerate to an intermittent data buss line? Man, I hate a problem like this - you really have to dig into the guts to find it, and when you do, then comes the "oh, it's so obvious" head-smack of realization. Well, let's see how it goes in Part Two.
Presumably this diagnostic tool cannot store a history of errors due to not utilising the RAM - ?
Nothing more frustrating than intermittent problems.
Children. Children are more frustrating than intermittent problems. Definitely.
@TheJonBrawn
Children are intermittent problems, one could say 😂
That's the first time I've seen a "Taiwanese Special" XT clone with a ripped off Microsoft Cassette BASIC!
I think your ram issue is a red herring. I would focus on why is it saying there is a clock and a game port. How does it detect them when they aren't present?
I agree why is it saying it detects something, that is not there.
A Varta, A Rifa, and MT RAM walk into a bar *...
* - the bar is called Dead Parts Bin
Maybe the problem isn't the RAM chips. Maybe its power quality. When I used to do this stuff I found a lot of machines with bad PSUs that would mostly work. Also, this era might have a linear power supply which are more susceptible to power fluctuations. The grid is not controlled as tightly as it used to be also last 2 dwelling voltage approaches 127V during peak load and grid would have been 115V when these were made.
the extra "isa slot" can that be for risercard when used in some cases
Yes, because no standard ISA card would go in there without the components on the board fouling whatever might be in the neighboring socket.
I'm pretty sure about that. Once i found a similar board with only that "middle slot" populated. I assume it was from a Triumph Adler POS system.😁
What is with the game port and clock? its obvious that neither are installed, but something sure is detecting them. i wonder if that is related in any way, shape or form.
ABC = Adrian Being Confused 😃
Those two 245s are definitely corroded, I thought for sure you had it with the socket and swap. Oh the joys of battery damage. I haven't worked on one of these in 30 years, and thank goodness that wasn't yet an issue back then, or I'd given up as a fledgling.
Hi Adrian. Great channel. Just an observation, you never seem to use any Anti-Static precautions. Many moons ago I was an electronics engineer working on mainframes and also early PC's. The training we had inisisted on anti static bags, mats and wrist straps. We were shown that after a static discharge many chips including memory would take months to years to degrade to the point of failure.
Hi Adrian, thankx for this video. I like your content and videos very much. Greetings from Steven from the Netherlands
All of the memory issues detected with the diagnostic EEPROM had a C in the same place in the address where the problem was. I.E. 7C945, 1CE45, etc. That may or may not mean anything, but it stuck me as a pattern, and a pattern quite often indicates where a problem might be. Just a thought.
My question while I'm watching this is why don't you take it down to a single bank of ram? It takes so little time to remove the chips and go down to one bank.
Then swapping the listed bits in the diagrom. Bit 6 seems to be constant with one , then 2.
Well, you know what they say about old age.
First you lose the ability to hold your liquids (batteries), then you lose your memory (RAM/ROM) and finally you lose your shape (rust/corrosion).
And you tend to work on really old computers :)
I'm getting the feeling of something that works once it has "warmed up." It seemed like the longer the board had power running through it, the "better" it did. The problem with that being, finding where such a devil of an issue is can be a serious headache. I've seen the opposite as well, of course.
CMOS parts also have better logic levels- they can pull up much harder than NMOS parts can. That might be a reason why it would make a difference.
You are completely ignoring the other issue and concentrating entirely on RAM , which is wrong. By this I refer to “detected” gameport and clock. I would assume this is related to the RAM problem you have. What are the chances that two different problems appear at same time on perfectly working before motherboard ?
So it can’t be RAM or RAM related transceiver to start with.
been working on a couple of 5150's lately, this was very informative. Thanks!
The unpopulated extra slot is for a riser card with 90 degree angle.
If you don't know the circuit design, swapping CMOS for TTL and vice versa is generally a bad idea. Different voltage levels and timings.
Not to mention fan out. I don't know as CMOS has the "umph" to power a bunch of non-CMOS.
Thanks Adrian great eposide cheers
Back in the early days of the IBM PC I found myself helping to fix some of the computers at the company I was working at in addition to my regular job. I do remember one machine that was showing RAM related problems. I eventually tracked down the problem to a bad chip that was used for parity for one of the RAM rows.
You are having bad luck because Rammy isn't on the scene! LOL! These computer often remind me of how old and crusty I feel some days and my memory seems to be failing too. Par for the course I suppose.
Where's Rammy when you need him? :)
Sockets lead to anarchy.
with the ram 245 removed, would your memory board work?
Come for the RAM problems, stay for the "It freakin' works!" 😉
When comes the Restauration of the Commodore 1084 monitor that was with the field found Amiga 500?
I tried a set of fuses under 5 amps in size and blasted 20Amps through them. My testing setup was getting hot but the fuse held for over 10 minutes... if only they could make their conductors out of the same material as their fuses!
And the plastic components as durable as the stickers on them.😁
adrian's self repairing machines basements a sci-fi story taking place in far future of artificially robotic controlled universe .
28:55 If the test program writes known bit pattern to memory and it can read the same pattern back even when the hardware reports parity error, then you know the chip containing the parity bit is failing. That doesn't crash the system because it's only parity bit.
However, the error that causes actual data to get corrupted is going to crash or hang the system if that part of the memory is used for any logic.
Someone should do a superset and call it Adrian Black: Attack of the Bad Ram.
I looked for common bits between all those parity error addresses throughout the video and there were none.
I’m surprised that Rammy wasn’t out and rampaging around the work area. 😮😮😮
sometimes it is just the solder joint.. sometimes it is the ripple on the supply voltage and sometimes it is the decoupling capacitors. Few knows their purpose but they are actually important.
its amasing you reviel the whole thought prosses of everrything on going
Parity errors are detected by dedicated RAM-facing circuitry. Since you are sometimes getting parity errors and sometimes other RAM errors, it may appear that the parity logic is actually working and truly detecting real issues.
You don't appear to have BIG issues accessiny video memory - no garbage on your screen. By this I am not referring to the actual video RAM but the bus lines to it (and the slots), which are likely shared with the onboard RAM to some extent, to support RAM expansion cards.
Then you see spurious devices like that game port, sometimes showing up, not at other times.
You may have some garbage/noise coming from something... perhaps due to bad addressing or CS/EN lines, activating (good) chips when they shouldn't.
It may be the PCB itself or some of the control logic...
Record the addresses in which the errors appear, in binary. See which address bits are always the same, if any, check those.
I advise against swapping in a 74HCT245 for a 74LS245. Most driver control circuits lead to open inputs while switching and as little as 15 ns open suffices for bad ringing that propagates throughout the board. Ask me how I know. Pullup resistors can avoid that issue, but you don't always have those.
IIRC HCT parts are CMOS parts and LS parts are TTL parts
@@commentidelloziopera HC uses CMOS levels, HCT uses TTL levels.
I see scratches around a suspected chip, I check the trace it crosses. Visual inspection on the back, microscope.
You usually pull out the oscilloscope instead of chip swapping, wish you'd done that here as well but hey. :)
I dunno if it's related, as it's a much later machine, but I had a RAM stick for my A1200(it has an expansion card with SIMM sockets) that I guess was borderline. It would boot and work most of the time, but sometimes it would just green screen on boot or send me to see the guru after a random amount of time. After switching to a different RAM stick it has worked fine ever since.
Might just be borderline "sometimes works" RAM chips, hence the inconsistency? Might be worth a try.
Things always seem to fix themselves on video. heh. Adrian, do we get a dump of the BASIC ROM? I haven't seen that one before.
Adrian, I'm only 20 minutes in, but I'm curious - and I honestly don't recall, it's been SO long since I've messed with XTs.. is there a chip that's in control of the onboard RAM? To me, that sounds like the culprit.. but I don't know if there's a separate chip for controlling the RAM.
the only thing I can recall is maybe the DRAM refresh circuitry, but I guess it's embedded into the CPU or something
Adrian, I've heard of the "capacitor plague." As of late, I believe your motherboards are suffering from a RAM plague. Break out the oscilloscope!🐏🐏
I was 4th !! I'd like to thank all the little people that made this possible...
It would be handy if that RAM test had an ability to write the results to a text file on a floppy drive of somesort, that way you could see the errors after testing and figure out where they were without having to be a fast writer to scribble down all the codes that pop up and change quickly...
BD 100 or DeoxIT is good for cleaning dust and corrosion on any connectors, potentiometers or switches.
WD-40 have contact cleaner and rust remover in their product range also.
Send this brilliant man more ram
LOL that 88 chip looks like somebody took a chisel to it
i remember back in the day using a signal injector to test the gates and flipflops on these chips to see if the output was what was expected from the inputs..lol not a fun job but it could pinpoint failed chips..
I would start with the bad PCB trace possibility. Check the tracks around the area you had the corrosion in. Also check the ISA socket to make sure all the crustiness is completely gone. I suspect you have a cracked, corroded or damaged track or two.
I had simular glitching on a camer in OBS, swapping camera USB ports, sorted it for me, some ports have limited capability, as you know, anyway thought I put my 2 pence worth in.
Adrian is an Interpol Fan 💯
On other projects like the Tandy I've seen how other friends of the channel have written utility ROMs to help diagnose memory issues. The PC hardware standard dedicated address space to each card slot specifically to allow ROMs to add drivers or utility support code. Theoretically you should be able to make a ISA board with the memory diagnostics ROM on it (or subset). Someone familiar with a reverse assembler should be able to make it from a dump of the ROM.
IIRC not all XTs (and clones) support option ROMs
256k chips are newer than 64k chips. I have to wonder if 256k MT ram will become the new 64k MT ram. lol
By chance, if you don't have one, purchase an i.c. chip tester, remove the ram chips one at a time, and test to see if any failure shows on the tester. It's a long slow tedious process but it might help to find the cause of the problem.
Wasn't clear - was the IC you removed to repair the laptop a C or non C version? The 82C88 variant is lower power / CMOS. It also has a higher rail voltage rating than the non C.
"From the makers of 'Is It Cake?' comes the next Netflix hit series 'Is It RAM's Fault?'"
The Great British RAM-Off.
RAM-Master
Prepare the battering Ram board.
Is it MT Ram?
@@michaelstoliker971 That's sort of the inverse of "It's not Lupus!" on House.
82c88 pops up in the most random devices.
i wonder if it was just because of being widely available
i mean it's MT Ram that is installed on this Mainboard...I wouldn't be surprised if the Ram Randomly failed.
256kb are fair, 64kb was crap.
Did you notice the black marks on the traces nearby the replaced transceiver?
Is t there a delay line in address strobe circuit?
Yeah, I at the point that I think Adrian is going to need an Exorcism to get this board to work.
And praise the Omnissia!😁
🎵There's a bad chip on the rise 🎵
🎵MT RAMs, where you learn to compute without RAM🎵
(sung to empty rooms)
Assumption is the mother of all f ups.
You could say computers are like people. As they get older they lose their memory.
You think you have problems - have you seen the slightly odd IBM5150 clonish Centurion/EDS system / expansion boards with some anomalies that David at Usagi Electric has recently looked at
You removed the 245 and only 4 bits failed the critical memory region test (and they were the same 4 bits both times) - is there another 245 serving the other 4 bits?
Right or not right. There is no quite.
I'm thinking address latches and perhaps delay line...
The 6502 is more efficient because of not latching the adress-lines ... but didn't the VIC-Chip halts the CPU everytime it refreshes the display and renders a lot of computing time useless?
IIRC the 6502 triggers a new cycle both at rising and falling edge of the clock; this allows the VIC to halt the CPU half of the time (IIRC on the falling edge of the clock) to access (V)RAM and to refresh it periodically without slowing the computer down too much
Is that a picture of a gnu on your shirt? :)
Decoupling capacitors?
@43.54 seems there are a couple of traces broken. No?
Hey Adrian could any of those capacitors on the RAM be marginal?
I have a 486 motherboard that as soon as anything touches the -12V rail on the ISA bus the PSU goes into shutdown. I have no idea what is going on, I've tried several different PSUs and ATX-AT converters and it is all the same :( There are no shorts on the board, I also removed the tantalums just to be sure and nothing changed :( Does anyone have any ideas, even a (friendly!) forum where I could ask for some help!
That doesn't really make sense. How are you sure it's the -12v rail on the ISA bus that's causing problems?
What about system error 01 at 16:31 ?
There was more than 1 system one error. I’m curious to know which error that one is
At least it's not an Apple II clone 😅
It's funny that the CPU is the smallest chip on the motherboard on these old boards.
It's not, the CPU is tied for largest chip.
I think Adrian is an Interpol Fan. 😅
😊😊😊good
What you also can do is remove al chips and deoxit d5 the whole board.
No offense intended whatsoever, but I just adore the way Adrian's X86 videos sometimes get so close to being outright hamfisted with the crossed fingers while posting after partswapping. Sure it's happened in some of his Commodore videos, but those don't usually lead to a visual blow to his theatrical lack of confidence in territory that's a bit alien from his usual area of expertise.
Sure, I'm really playing it up, but it makes me think of the scenes that have the guy in my avatar, Corvax, rapidly losing confidence that result in him resorting to more desperate measures with sloppy execution that only open more problems until there's nothing left to but to drop to the floor and whimper in the fetal position until the King barges in and accidentally digitizes himself into the matrix, where the next step is to turn to treason and "run away"!
Lots of Ramdom. 😂
My money is on a cold solder joint(s)......
Those traces look terrible, really corroded with the typical dark soldermask. I would suspect they are (because very thin) or will be soon disconnected and broken. To not check those is kinda strange, obvious damage, thin trace, typical discoloration. A problem waiting.
If in doubt yank it out.😁
it should be normal to wish stuff to be working lol, well repair channels. sometimes stuff has to not work to have rest. stuff fixing itself would make repairers out of job wouldnt it.
The extra slot is likely for a desktop riser as an option. But it's odd to see it in a full xt form factor motherboard
that is what I thought too a rizer daughter board?
How many ways can we spell "riser"? Any other bids? :P
@@Nukle0nha yeah good catch
@@Nukle0n I can work on electronics -spelling takes second place - LOL
@@Nukle0n Reiser
Reizer
Ryser
Ryzer
Ryeser
Ryezer
Reizor
Reisor
Rysor
Ryzor
Ryesor
Ryezor
Or they can't escape you!!!!!
You can't mix and match ram sticks and not expect to have some problems.