Wow! What a nasty issue, absolutely astonishing O_o Thank you very much for this update video, I was really curious to see what it is and I was not disappointed at all. I never expected to see something that sneaky. Also I have to take off my hat for Oli Wright for the great hint. That was really educative. And of course, thank you Adrian for your kind words in my direction :)
RS is Radio Spares, they still exist as RS Electronics and are a massive electronics supplier that started out in the UK and now are international. They used to re brand a lot of parts. Good to know the pencil helped.
When I saw the RS logo, I immediately had RS components in my mind. However, I only knew them as a parts distributor like Mouser, Digikey, etc, and never seen any ICs with their branding on them. I remember buying my very first Raspberry Pi through them, and it had RS branding on the box, but not on the components.
RS still sell a whole load of rebadged stuff; including components. I have found, as a general rule, that the rebadged stuff is usually of a good quality. I don’t think there is anything sketchy with the RS logo being applied over the Motorola logo. It was either done for RS by Motorola or with Motorola’s blessing.
Definitely genuine part, fully meeting datasheet specifications, re-marked by R.S. (many years ago). I have quite a few of these re-marked R.S. devices in my IC stash. Of course totally unlike the modern fake re-marked chips from China which either are totally dead or are marked as higher grade/speed devices. Back in the day when these devices were sold by R.S., R.S. would not touch hobbyists with a bargepole, you had to be a "professional" organisation to buy from them. Nowadays things are different, but as a consequence of their original policy, I do not buy from them.
Came here to say the same. Back in the 80s RS rebranded plenty of semiconductors. I still have some RS bridge rectifiers in my collection of random parts. The fault was clearly the LS373 chip, but either just due to its age, or possibly other damage; certainly not a dodgy rebranded part.
This is why i have been a professional electronics tech for over 20 years. I still love the thrill of the chase and the buzz of success when you find the problem. I still buy faulty parts on eBay just for the fun of it. Its like jigsaw puzzle for me. Congrats on the find. Great job.
And to be such a person and you find something you figure out it gonna be way too much time and effort and nothing else than a living pain to repair and you still do it anyway. that is at least what I end up doing so many times because not only will it reward you when you succeed but I am also a very bad looser which makes me hate giving up on something you start working on so no matter what you just have to make it happen one or another way
The Tester may only be able to perform a more basic test of the IC, not necessarily stress it at a faster speed or heavier bus load. It does a good, general-purpose ID and checkout, no complaints. After all, you don't expect a Swiss Army knife to have a hammer, arc welder and pocket chainsaw, do you? 🙄
Sounds like the chip would have been fine in most applications, like if you swapped it with one of the better chips on the same board things would have been fine. Which makes me wonder if a binary pass/fail is even the best way for the tester to report its results. Maybe there should be a way to report that it passes general purpose tests but has a defect known to cause problems in a few specific use cases, with a code that can be looked up for more details.
it is! and yes they used to, i have many of their repainted and remarked ones! strange a non uk made device has basically uk supplied parts in them! they did and do supply overseas but again weird they didnt use chips direct from a manufacturer!
@@enojelly9452 I have recollection of counterfeit transistors being reported in the electronics press in the late 80's so its not a stretch to imagine they were doing it to 74 logic too. Its odd because then, as now, RS were considered a "premium" supplier that supplied mainly government, education and design departments of electronics companies, they had a comprehensive stock and fast delivery but were not cheap, so not the place that a PC clone manufacturer would choose as a supplier. Perhaps the counterfeit operation just chose RS for the part to give it value. I know RS did sell their own branded parts, but I would have thought they would have got them marked by the OEM rather than repainting already marked chips.
I've had some 74LS chip makes become glitchy due to noise on the VCC supply (especially prevalent on flip flops and latches), while another brand would work perfectly. It may be that the RS Components rebadged parts were less immune to noise and were thus rejected by Motorola and were then bought and rebadged by RS Components. The clue here may be that they once worked on this motherboard, but due to aging caps and power supply components, now fail with the higher power supply noise that is (probably) now present. This would also explain why the chip was passed as "good" by the part tester. The spec I used to build with back in the late 70's was to target less than 200mV of supply noise peak to peak, but some 74LS series parts needed less than 150mV even back then. Nowadays, rather than replacing aging ceramics, I just tack a new 0.1uF cap across the power pins of the misbehaving chip to see if that fixes the glitch prior to going to the trouble of replacing it. So, this was really a long winded way of asking if you did a supply noise measurement at this particular component?
Any chip that has latching can fail, and this looks like a classic case. The output tries to go high, but the internal latch has failed, so it goes high for the internal setup time, then fails to get latched high. The DMA transfer takes place on the trailing edge of the clock, the rising edge setting up the transfer, then the far end latching state on the trailing end, here the setup is on the enable going low, with the other side latching as the enable pin makes the outputs tristate again. You can see the runt pulses of the internal latches on the good chip as well, just they get properly latched and held high with internal feedback. If the chip was decapped you probably will find either metallisation failure or purple plague on the chip surface, best to take the other 2 Motorola parts and replace them as well, as they are off the same production line, and Motorola did have issues with purple plague on chips in this era. Best here to snip them out and solder in new ones.
Hi Adrian. The chip tester just sets up some logic levels on the inputs of the DUT, then latches them over to the IC in the ZIF socket, while using an edge-triggered flip-flop to ‘capture’ the values at the output (pins). So, when it pulls the CE-bar (chip enable, negative-logic) low, it captures the initial high pulse you saw on the ‘scope - it never looks at the logic value after that, until it’s ready to run another test. It also doesn’t test for proper rise/fall times, either; that would make the chip tester WAY more complicated and expensive! As Dave Jones on EEVBlog would say, this is a trap for young players! 😏
Awesome work! So satisfying to see it work! Louis Rossmann says often that the satisfaction of repairs is in solving a fault on a board, and then the absolute joy of a customers' reaction to them getting machine fixed with data intact.
Well done! This is exhibit A of why I never rely on a chip tester to tell me that a chip is "Good". It will only tell you that it's not definitively bad, not that it's good. And sometimes "good" is contextual - plenty of 80s era designs where some chips will test just fine but not work in the target (old Atari computers and the TRS-80 line are notorious for that).
Great work. So rewarding seeing these old machines coming back to life while learning of different logic problems associated with these chips and boards. Thanks so much for sharing.
Told you those address lines would be interesting : ) Glad you found it. Probably a failed junction on the output transistor for AL12. Good thing it's a common enough part you can easily swap. Corrosion probably caused excessive current on that line and blew the junction or a bond wire.
I saw a lot of 74LS-chip failing in several instruments (voltmeters and oscilloscopes) from the 70s and 80s. So I tend to replace first most caps but also many 74LS (with HC types), like LS00, LS04 and also LS373.
Adrian I want you to know that it is fun to see you go through the process of fixing a machine. I have done Computer Interface Technology as a study but now work in administration. Seeing you work makes me want to poke hardware again.
"RS" is Radio Spares over here in the UK, these days called RS Components, rather unusual to think they could have re-marked that particular IC though... :\
Yeah, I was going to mention it. RS Components. (Also known as Electrocomponents elsewhere in the world) They are known for loads of rebranded quality components, from ICs and caps to soldering irons and wellies. They also sell original branded stuff too. Been using them since the '90s. I also worked on one of their corporate conferences. I think that was back in '99. Maybe. They are still going. I don't think they do the old paper catalogue anymore, its just online these days.
It would be interesting to see the latch signal input as well as the associated data input. That might make it clearer why the output is behaving that way. Thanks for sharing this.
That RS logo on the LS373 is from Radio Spares, which is a British company that (unsurprisingly) sells electronic components. I assume they took OEM parts and rebranded them for sale under their own brand. They still exist today.
It's such a satisfaction for us viewers to see the solution of a difficult problem. It's probably not worth the time for you in the first place but please continue to support us viewers with the final solution of a difficult issue ! Thank your for all the videos 2021. Please stay healthy and safe and I hope you will enjoy us with nice videos in 2022 ! Greetings from Germany !
It just goes to show that sometimes there's more to the story than a chip simply being "good" or "bad". Based on the number of times you've brought up this particular class of logic chip, I'm guessing it's not particularly rare and might even still be in production to a limited degree, but there are bound to be cases like this with less common chips as well. Actually, now I wonder if all these rebadged chips have the same defect. You should consider trying them all out in that socket to find out. It could be useful to have some documentation online somewhere showing that these RS-rebadged chips will test OK in the chip tester as well as working in most applications, but are not suited for this particular use case.
A full capability semiconductor test system would provide reference loading on all pins and verify drive current, output voltages and timing across a range of input and output test vectors. My first job was as a digital designer at GenRad working on board test which is similar but less specific as to tolerances. Those devices were very, very complex and used custom analog driver/sensor designs to look at the analog levels of the inputs and outputs in great detail and had very precise timing systems (high speed ECL drivers and such) to drive a range of input timing to verify performance of the chips and measure the input to output timing of same. I expect the old test systems from back then are no longer around as they were large, expensive and complex to use. A failure like the one you were seeing with that LS373 would likely not show up in a simple pattern tester like the one you have.
Still, this is the weirdest failure mode I've seen for a simple logic chip. It's as if the enable line is capacitavely coupled, so that it activates for some time and then falls off.
@@ovalteen4404 It might very well be capacitively coupled or the drive transistor may have deteriorated to the point where it cannot drive any significant load. The load from a mosfet input (which most modern devices would have is likely far lower than the load from an old style TTL input where if needs to drive a BJT (and thus a decent amount of current) rather than the gate on a FET. Hard to say as I haven't looked at semiconductor failure analysis in over 30 years.
I can think of a few reasons the chip shows good in the tester but fails on the MB. Speed first comes to mind. The tester might not be testing in nanosecond pulses like the MB would be doing. Loading might be another reason, where the chip can't source enough current for the MB but the tester didn't load it as much. Also how many outputs are at the same level could be a factor, though the tester probably would have ran through all possible output combinations. Lastly, could have a bad/broken VCC connection internally where it is getting VCC sourced from the other pins.
ttl chips usually sink more current than they can source , just had a look at a datasheet for a TI one, 2.6ma source, 24ma sink, faulty one seems like it can pulse high enough but not hold for long so maybe bad latching part of the internal circuit , not the output driver part ??
@@Walczyk i dont know, i dont know how this chip tester tests, presumably sets an input byte checks output isnt the same , pulses latch input, then check it is and then goes to next byte, 0 to 255 in turn? problem is if it doesnt 'wait' long enough when testing to miss a failed 'latch' it may make a false 'pass'
[RS] is RS Electronics. A big component seller here in the UK. In fact it was always joked that you could get most of the parts for a BBC Micro from RS. So either those chips came from RS or that Motorola had rebranded them for RS but then needed to provide some to the MB manufacturer because of stock shortages or similar.
As a designer in the 80s, I was always advised - "if you're going to design a new part in to a board, make sure you can also get it from RS, that way you can be sure it will likely have multiple suppliers and will be around for a long time". Of course, it was never quite that simple, but it was a good approach to ensuring component availability for a design/
Ha! You tricked me and snuck a video out on new year's eve! Feel free to ignore my comment on your other video, since it's completely redundant now. Glad you could fix the board, although those re-marked chips are very strange. I'd be tempted to replace the others just in case. But fixing a faulty board is a great way to end the year. Anyway, happy new year to you too! Hope you have a great 2022!
There’s not a lot of power supply decoupling capacitors on that board. I wonder if adding a couple more 100nF capacitors nearby across the supply rails would make the bad chip behave better?
Great video. The branding on the chip is radio spares Aka RS component's. They used to brand stock with there own logo and sometimes even there own part numbers!!! Though majority of the parts they supplied were from good quality suppliers.
Nice find Adrian! Thanks, as always, for the great troubleshooting content that you share with all of us. I hope that you have a very Happy and Healthy New Year! Fred
I had similar issue with my Atari STe, and ACSI/DMA port buffers (LS244, LS245). They were also testing perfectly fine, but data on hard disk was being corrupted. Replacing these chips helped.
I'm glad you took the scope out. It would have been quite unsatisfying otherwise as a closure, since the chip tester reported it good. What an interesting failure miode.
RS is Radio Spares. They rebadge stuff all the time, and have done for years. It’s not usually junk. I wonder if the manufacturer used RS because they couldn’t get stock elsewhere. Usually RS is expensive but good at holding stock.
@@adriansdigitalbasement2 thats what seems strange, them using parts from a UK based 3rd party supplier!! i doubt it was a 'bad' chip, from new, just failed, looks to me more like a latching problem than the output drivr section as it does briefly pulse high, but not hold, i have loads of RS remarked chips in a box of old but unused TTL stuff
@@adriansdigitalbasement2 given that Hong Kong would have been a British colony when this board was made, and Taiwan and Hong Kong very deeply work with one another in electronics, it makes total sense that there would be a RS office that could have been a supplier.
@@adriansdigitalbasement2 oh I'm fully aware lol traced everything, found a bad trace..patched it and no change at all. Makes board #2 that's kicking my butt...2nd only to my rev3 5150..
Another problem that can happen with jellybean ICs is that some circuit designs are stricter in their timings that the jellybean standard allows for. It's rare, but happens, and when it does you usually have to go with a specific manufacturer.
That’s an RS Components logo on the ‘373. A UK component supplier (presumably elsewhere too) they’ve been around for decades. Didn’t know they rebadged chips, I’ve not seen it recently on anything I’ve had from them.
You don't need all address lines to work for refresh. RAM access pattern is different during refresh, and IIRC only A0-A7, maybe up to A8 or A9 are used to select the row to be refreshed. All banks are refreshed at the same time. There is a special REFRESH signal on the ISA bus to put ISA RAM boards into refresh mode. ISA RAM boards don't need an address counter, because the row number to be refreshed is on the *low* address bits.
Likely the chips were "rebadged" due to region restrictions in place at the time. Chip shortage caused by the IBM PC success left European and Asian manufacturers with no chips. The chip manufacturers set up regional allocation schemes and thus began the original "outsourcing" where companies set up factories in Europe and Asia. 74LS373 failures are actually quite common, quite likely a timing issue. The 8237 being an 8 bit chip, the CPU actually programs the upper address lines for transfers.
The RS symbol on the 74373 is a logo for Radio Spares (RS Components) and electronic spares supplier. "In 1937, J H Waring and P M Sebestyen established a company called Radiospares Limited, supplying radio repair shops with spare parts. The business began selling electronic components in 1954 and rebranded as RS Components in 1971." Taken from RS Components website. They would have re-branded the chip for their stock under license from Motorola I guess. I'm surprised if the chip was a reject. RS Components are known as a quality supplier. The problem could be caused by the battery acid corrosion, but then again somebody has had this very chip fail before. I wonder if this was a common failure? I have no memory of this particular fault from back in the day.
RS (Radio Spares) are a well known UK distributor of electronic components that still exists today. I guess the rebranding was a similar as the major supermarkets have their own house brands - possibly they can set a lower price and move a bigger volume.
Memory refresh only needs to use chip select along with RAS, presenting only 8 bits of the lower address on the line. Then the chips do a RAS only refresh, not needing to have any other selection. Saves time, in that all columns in the chip are refreshed at the same time, as the columns contain the read amplifiers, so they read the bit, and immediately write it back, doing 128 columns at a time. Larger chips implement this still, but as they have more address lines the number increases, but still you only use RAS to do refresh, and for most chips you can actually do refresh by simply having enough reads, so long as your reads go through the RAS only address space fast enough to hit every location.
Adrian, i think, that stickers are a good idea, but a bracket (metal one or 3d printed) can be even more helpful to avoid confusion of incorrect xt-ide card installation. Thanks for this interesting story about pc-xt mobo - never know, that such little chip can be treacherous marginal and make this trouble appear.
congrats, you have found a needle in a haystack) Also I would consider checking if something else drives that line low and newer chip simply wins this race condition (check line resistance against ground, also something sharing that line might become hot). I wonder how this part would act in tl866 chip check.
DRAM refresh doesn't actually need all of the address lines - depending on age it either need just the row address lines (RAS) but not the column address lines (CAS) of the memory chip (RAS-Only Refresh Mode), or it just needs the RAS/CAS lines to be pulsed properly and doesn't need any address lines (CAS before RAS Refresh Mode). IIRC an XT is old enough that which of these two are used depends on which memory chip they used - the 4164/41256 datasheets I found via Google only mention the first but the really ancient 4464 datasheet I found mention both. But the upshot is that either some or all address lines are irrelevant during memory refresh so these can be broken on the DMA circuitry and the machine will work unless DMA is necessary for other purposes like your floppy drive. Not going to claim I figured this out during the previous video - I triggered on the "the DMA works because DRAM refresh needs all memory lines" which made me go "No it doesn't - wonder if...", you had already tried a different DMA controller but then the schematic came up showing three different buffer drivers which could be involved... The oscilloscope shot of the failure was REALLY nice!
100% correct! For RAS only refresh cycles the row address is supplied externally (9 bits for a 256k x something chip. For CAS before RAS refresh which came in at around the 256K generation (if memory serves) the DRAM chips would each maintain their own internal refresh row address counter. I designed non-PC computer memory boards using both methods in the 1980s.
Adrian, often you have specific schematics pulled up in your videos, could you list for us the web sites, or online resources you use to find them? I'm interested in trying to fix my Tandy 286 clone, but I'd feel much more confident if I had pinouts and schematics for it. Thank you.
At work, from time to time, our customers would have us build a prototype, whether a new product or a ‘respin’ of an existing product, and sometimes we or the customer would find that we might have odd issues with one brand of IC, yet it would work perfectly with another brand…
i've just found pdfs of the datasheets of both the TI and motorola versions and state identical source and sink current capability, the dud one looks like it can pulse high correctly, but not hold it so i'd say its more a latching problem, not its output driver..? however back in the 80s they may have been specified differently??
In the tester the IC only has to drive a very light load, but in the PC more inputs are attached and are switched faster. During production, the ICs were already marginal. But were seen as a cheap alternative if the system works with it.
Seems to me that the retro chip tester pro doesn't simulate a realistic load when tasting these logic chips. Either that or the load on address line 15 is unusually high. A partially failed bypass capacitor could cause this.
Two possible reasons I can think of for the rebadging - The chips were marginal and did not pass Motorola QC and were sold of to another company, under the strict requirements that they would not enter the market as Motorola parts - Motorola had an over supply of chips, not wanting to drop the price on Motorola branded parts, they were sold to another company, again with the strict requirement that they would not enter the market as Motorola parts.
Could be also that RS simpy wanted their own brand, so they could purchase chips at the cheapest supplier without having to convince the customer to switch to another brand.
@@danielmantione RS did rebrand a lot of parts to their house numbering scheme, as it made internal stock control easier, handling stuff with only a RS part number, and a generic TTL part number, with a date or batch code from RS to trace them. Still do, especially with test equipment as that allows them to have special tweaks made just for them in the firmware.
SeanBZA is right. RS components used to buy parts from other manufacturers, rebrand them and sell them. It was their core business model during the 80s. The parts were legit, although they were selling them more expensive than the usual retail outlets.
Radio Shack / Tandy also sold chips with their own part numbers. (Inside their computers usually) But they worked with the manufacturers of the chips to make custom markings on the chips. They weren’t painted on LOL.
@@adriansdigitalbasement2 All IC manufacturers will custom mark product for a customer order, you just have to order 10k parts and they will do that for you. Most common on mask ROM, when you wanted to get past the prototype stage, and was before the evolution of OTP EPROM becoming popular, which killed the mask ROM business for all the small volume, and even made big inroads in large volume, as they could do upgrades much easier, no need to wait for the chip bulk to be finished, just the units that have been programmed.
Nice catch. I have several IC testers, including an expensive commercial one, and I have found several chips test OK on some and bad on others, since they each use different methods, levels, etc for testing (and some cheap ones could even have poorly designed tests). I would love to try that chip in all of my testers, it sounds like a good benchmark. Let me know if interested, I have your E-mail from prior correspondence (TI-99 CPU purchases and I sent some cartridges to you).
Relabeling is not a new thing, in the good old days with vacuum tubes for audio and tv, it was quite normal that the mfg's bought tubes from each other you could buy a tube from Tungsol; but it was a RCA tube labeled (printet) and packaged at RCA.
I know that two of these re-badged chips haven't stopped working (yet), but if it was me I would swap them out for something I was more confident in. Once bitten, twice shy if you know what I mean. Yes, I know that these are legitimate re-brands for corporate reasons rather than anything dodgy, but it is the kind of thing that just leaves me with suspicions at the back of my mind.
I sense a pattern... it seems most of the standard components in old computers that die seem to be some sort of buffers or bus drivers. Maybe the three-state logic wasn't all that good in the early days? Not to mention the usual ESD suspect 😉
Bus driver chips got abused a lot in the 80's. The BBC Micro is notorious for overdriving the 74LS245 on the databus, by having too many devices connected to it. They actually had to add a resistor array to the databus to fix issues created by using the chip beyond it's specs. in the datasheets this is the value for "fanout" or how many standard 74 logic chips an output can drive.
I think these chips do remarkably well, considering they have lasted nearly 40+ years years in some cases. Modern day chips would do well to last as long.
@@station240 I recall Steve Furber talking about that. They had no idea that they were abusing the 245, and the resistors were basically a voodoo change that seemed to do the trick, but they'd no idea why it worked.
@@a4000t Indeed, most of these chips are hanging in remarkebly well; it's just that the likelyhood of something else failing seems to be (within my field of view) quite a bit lower than the bus drivers/transcievers. The overdriving that station240 mentioned might be a reason, but I wouldn't expect that on an IBM design. Just an observation based on quite a few replaced ones of these here :)
How reliable is that chiptester pro then? Could you do a test of a confirmed failed chip to see if it actually comes up as faulty? Great stuff as always Adrian.
seems like it takes time for the chip to start acting up, maybe if you tested it for a couple of cycles then it would test bad eventually in the tester thingy, but maybe it needs continuous use to show the signs of fault happy new year!
I'm surprised the retro chip tester tested that chip as good. It must be missing some tests to fully evaluate it. It may be worth contacting the creator to suggest a better test that will properly test the chip.
Stephen, the guy who made the Chip Tester says that it can't accurately test output drivers as it puts basically no load or capacitance on the outputs ... he says it often won't detect such problems and there is just no way to accurately test that kind of thing without making it super expensive and complicated.
@@adriansdigitalbasement2 i'd say its not a drive issue as the outpu driver part of the chip is separate from the latch part, according to the datasheet i've seen, more likely the chip tester tests it too quickly, not waiting to see if the output changes after a short delay after removing a latch trigger/enable
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
Wow! What a nasty issue, absolutely astonishing O_o Thank you very much for this update video, I was really curious to see what it is and I was not disappointed at all. I never expected to see something that sneaky. Also I have to take off my hat for Oli Wright for the great hint. That was really educative. And of course, thank you Adrian for your kind words in my direction :)
RS is Radio Spares, they still exist as RS Electronics and are a massive electronics supplier that started out in the UK and now are international. They used to re brand a lot of parts. Good to know the pencil helped.
When I saw the RS logo, I immediately had RS components in my mind. However, I only knew them as a parts distributor like Mouser, Digikey, etc, and never seen any ICs with their branding on them. I remember buying my very first Raspberry Pi through them, and it had RS branding on the box, but not on the components.
@@Colaholiker they did an awful lot of branding
RS still sell a whole load of rebadged stuff; including components. I have found, as a general rule, that the rebadged stuff is usually of a good quality. I don’t think there is anything sketchy with the RS logo being applied over the Motorola logo. It was either done for RS by Motorola or with Motorola’s blessing.
Definitely genuine part, fully meeting datasheet specifications, re-marked by R.S. (many years ago). I have quite a few of these re-marked R.S. devices in my IC stash. Of course totally unlike the modern fake re-marked chips from China which either are totally dead or are marked as higher grade/speed devices. Back in the day when these devices were sold by R.S., R.S. would not touch hobbyists with a bargepole, you had to be a "professional" organisation to buy from them. Nowadays things are different, but as a consequence of their original policy, I do not buy from them.
Came here to say the same. Back in the 80s RS rebranded plenty of semiconductors. I still have some RS bridge rectifiers in my collection of random parts. The fault was clearly the LS373 chip, but either just due to its age, or possibly other damage; certainly not a dodgy rebranded part.
This is why i have been a professional electronics tech for over 20 years. I still love the thrill of the chase and the buzz of success when you find the problem. I still buy faulty parts on eBay just for the fun of it. Its like jigsaw puzzle for me. Congrats on the find. Great job.
And to be such a person and you find something you figure out it gonna be way too much time and effort and nothing else than a living pain to repair and you still do it anyway. that is at least what I end up doing so many times because not only will it reward you when you succeed but I am also a very bad looser which makes me hate giving up on something you start working on so no matter what you just have to make it happen one or another way
You should post some videos.
Same here the skill is now getting rarer as no one repairs or trains to this level anymore as it is to expensive professionally
Nice job! You should see if the dev of the Retro Chip Tester Pro wants that chip to analyze in order to improve the firmware.
I was thinking the same. Maybe it is just matter of trimming the timings or add some delays
The Tester may only be able to perform a more basic test of the IC, not necessarily stress it at a faster speed or heavier bus load. It does a good, general-purpose ID and checkout, no complaints. After all, you don't expect a Swiss Army knife to have a hammer, arc welder and pocket chainsaw, do you? 🙄
Sounds like the chip would have been fine in most applications, like if you swapped it with one of the better chips on the same board things would have been fine. Which makes me wonder if a binary pass/fail is even the best way for the tester to report its results. Maybe there should be a way to report that it passes general purpose tests but has a defect known to cause problems in a few specific use cases, with a code that can be looked up for more details.
The RS on the 74LS373 latch chip looks like the RS Components Logo. They may have relabeled it as one of the own branded products.
it is! and yes they used to, i have many of their repainted and remarked ones! strange a non uk made device has basically uk supplied parts in them! they did and do supply overseas but again weird they didnt use chips direct from a manufacturer!
@@YoureUsingWordsIncorrectly In the 80s?
@@enojelly9452 I have recollection of counterfeit transistors being reported in the electronics press in the late 80's so its not a stretch to imagine they were doing it to 74 logic too. Its odd because then, as now, RS were considered a "premium" supplier that supplied mainly government, education and design departments of electronics companies, they had a comprehensive stock and fast delivery but were not cheap, so not the place that a PC clone manufacturer would choose as a supplier. Perhaps the counterfeit operation just chose RS for the part to give it value. I know RS did sell their own branded parts, but I would have thought they would have got them marked by the OEM rather than repainting already marked chips.
Yep. They did this a lot back in the 70s and 80s. Totally legit parts.
@@IanSlothieRolfe Yes, but my answer was to the comment about China… did they have such a semiconductor industry in the 80s already?
I've had some 74LS chip makes become glitchy due to noise on the VCC supply (especially prevalent on flip flops and latches), while another brand would work perfectly. It may be that the RS Components rebadged parts were less immune to noise and were thus rejected by Motorola and were then bought and rebadged by RS Components. The clue here may be that they once worked on this motherboard, but due to aging caps and power supply components, now fail with the higher power supply noise that is (probably) now present. This would also explain why the chip was passed as "good" by the part tester. The spec I used to build with back in the late 70's was to target less than 200mV of supply noise peak to peak, but some 74LS series parts needed less than 150mV even back then.
Nowadays, rather than replacing aging ceramics, I just tack a new 0.1uF cap across the power pins of the misbehaving chip to see if that fixes the glitch prior to going to the trouble of replacing it. So, this was really a long winded way of asking if you did a supply noise measurement at this particular component?
Very interesting point. Would love to see if putting additional bypass caps fixes the issue.
Any chip that has latching can fail, and this looks like a classic case. The output tries to go high, but the internal latch has failed, so it goes high for the internal setup time, then fails to get latched high. The DMA transfer takes place on the trailing edge of the clock, the rising edge setting up the transfer, then the far end latching state on the trailing end, here the setup is on the enable going low, with the other side latching as the enable pin makes the outputs tristate again. You can see the runt pulses of the internal latches on the good chip as well, just they get properly latched and held high with internal feedback.
If the chip was decapped you probably will find either metallisation failure or purple plague on the chip surface, best to take the other 2 Motorola parts and replace them as well, as they are off the same production line, and Motorola did have issues with purple plague on chips in this era. Best here to snip them out and solder in new ones.
@@SeanBZA if these had known issues, yes, probably best to replace, like with certain 'bad' capacitor plague affected capacitor brands
I am amazed at how there is always someone who knows the solution. Whatever your problem is, someone has the answer. The trick is finding them.
For the ISA to IDE, better to add a metal plate at the external side.
Excellent work.
Your chip tester runs at a slower speed and not the 4.77mhz the PC mobo runs at so it would be fine on a 1mhz system like an Apple.
Hi Adrian. The chip tester just sets up some logic levels on the inputs of the DUT, then latches them over to the IC in the ZIF socket, while using an edge-triggered flip-flop to ‘capture’ the values at the output (pins). So, when it pulls the CE-bar (chip enable, negative-logic) low, it captures the initial high pulse you saw on the ‘scope - it never looks at the logic value after that, until it’s ready to run another test. It also doesn’t test for proper rise/fall times, either; that would make the chip tester WAY more complicated and expensive! As Dave Jones on EEVBlog would say, this is a trap for young players! 😏
Awesome work! So satisfying to see it work! Louis Rossmann says often that the satisfaction of repairs is in solving a fault on a board, and then the absolute joy of a customers' reaction to them getting machine fixed with data intact.
RS Components is a UK firm still in existence. Originally the RS stood for Radiospares. Their logo is very recognisable.
Well done, Adrian, but you forgot to show us putting the bad chip in the dead parts bin. Have a happy new year.
send it to me and i'll test it on my 'Leap' IC tester and see what that says 😉
He needs a new bin labeled "evil". Any part that can pretend to be good and fail like this is beyond "bad".
@@scottlarson1548 hehe
The videos where you actually bust out the real tools and figure things out completely are just excellent. HNYAB!
Excellent troubleshooting! An oscilloscope is definitely a must-have for knowing exactly what's going on in most digital logic circuits.
Well done! This is exhibit A of why I never rely on a chip tester to tell me that a chip is "Good". It will only tell you that it's not definitively bad, not that it's good. And sometimes "good" is contextual - plenty of 80s era designs where some chips will test just fine but not work in the target (old Atari computers and the TRS-80 line are notorious for that).
Great work.
So rewarding seeing these old machines coming back to life while learning of different logic problems associated with these chips and boards.
Thanks so much for sharing.
Told you those address lines would be interesting : )
Glad you found it. Probably a failed junction on the output transistor for AL12. Good thing it's a common enough part you can easily swap. Corrosion probably caused excessive current on that line and blew the junction or a bond wire.
I saw a lot of 74LS-chip failing in several instruments (voltmeters and oscilloscopes) from the 70s and 80s. So I tend to replace first most caps but also many 74LS (with HC types), like LS00, LS04 and also LS373.
Me it was always 7474 dual flip flops, they were the first chip to change, in any place with wonky logic operation.
I've fixed lots of arcade machines that didn't like putting HC components in places of LS ones. 98% of the time you're fine with HC though.
Look at that! It freaking works!
Adrian I want you to know that it is fun to see you go through the process of fixing a machine. I have done Computer Interface Technology as a study but now work in administration. Seeing you work makes me want to poke hardware again.
"RS" is Radio Spares over here in the UK, these days called RS Components, rather unusual to think they could have re-marked that particular IC though... :\
Ah, that's where I 've seen that logo before.
They used to re label a lot. I have RS branded 78k12s somewhere from the same era
Yeah, I was going to mention it. RS Components. (Also known as Electrocomponents elsewhere in the world) They are known for loads of rebranded quality components, from ICs and caps to soldering irons and wellies. They also sell original branded stuff too. Been using them since the '90s. I also worked on one of their corporate conferences. I think that was back in '99. Maybe. They are still going. I don't think they do the old paper catalogue anymore, its just online these days.
@@Thebrainymonkey yep i think online only now, not even the CD catalogue, i have the 50th year paper catalogue somewhere, from 1987 !
It would be interesting to see the latch signal input as well as the associated data input. That might make it clearer why the output is behaving that way. Thanks for sharing this.
Wow, that would've been difficulty to find without any clues. I'd swap out the other two re-branded chips also.
That RS logo on the LS373 is from Radio Spares, which is a British company that (unsurprisingly) sells electronic components. I assume they took OEM parts and rebranded them for sale under their own brand. They still exist today.
It's such a satisfaction for us viewers to see the solution of a difficult problem. It's probably not worth the time for you in the first place but please continue to support us viewers with the final solution of a difficult issue ! Thank your for all the videos 2021. Please stay healthy and safe and I hope you will enjoy us with nice videos in 2022 ! Greetings from Germany !
It just goes to show that sometimes there's more to the story than a chip simply being "good" or "bad". Based on the number of times you've brought up this particular class of logic chip, I'm guessing it's not particularly rare and might even still be in production to a limited degree, but there are bound to be cases like this with less common chips as well. Actually, now I wonder if all these rebadged chips have the same defect. You should consider trying them all out in that socket to find out. It could be useful to have some documentation online somewhere showing that these RS-rebadged chips will test OK in the chip tester as well as working in most applications, but are not suited for this particular use case.
Troubleshooting this board was lengthy but fascinating.
A full capability semiconductor test system would provide reference loading on all pins and verify drive current, output voltages and timing across a range of input and output test vectors. My first job was as a digital designer at GenRad working on board test which is similar but less specific as to tolerances. Those devices were very, very complex and used custom analog driver/sensor designs to look at the analog levels of the inputs and outputs in great detail and had very precise timing systems (high speed ECL drivers and such) to drive a range of input timing to verify performance of the chips and measure the input to output timing of same. I expect the old test systems from back then are no longer around as they were large, expensive and complex to use. A failure like the one you were seeing with that LS373 would likely not show up in a simple pattern tester like the one you have.
Still, this is the weirdest failure mode I've seen for a simple logic chip. It's as if the enable line is capacitavely coupled, so that it activates for some time and then falls off.
@@ovalteen4404 It might very well be capacitively coupled or the drive transistor may have deteriorated to the point where it cannot drive any significant load. The load from a mosfet input (which most modern devices would have is likely far lower than the load from an old style TTL input where if needs to drive a BJT (and thus a decent amount of current) rather than the gate on a FET. Hard to say as I haven't looked at semiconductor failure analysis in over 30 years.
13:20 Wow! This is like unmasking the villain at the end of a Scooby-Doo episode! 👺
I can think of a few reasons the chip shows good in the tester but fails on the MB. Speed first comes to mind. The tester might not be testing in nanosecond pulses like the MB would be doing.
Loading might be another reason, where the chip can't source enough current for the MB but the tester didn't load it as much.
Also how many outputs are at the same level could be a factor, though the tester probably would have ran through all possible output combinations.
Lastly, could have a bad/broken VCC connection internally where it is getting VCC sourced from the other pins.
Could it also just not be properly testing that address high?
ttl chips usually sink more current than they can source , just had a look at a datasheet for a TI one, 2.6ma source, 24ma sink, faulty one seems like it can pulse high enough but not hold for long so maybe bad latching part of the internal circuit , not the output driver part ??
found motorola data sheet, says the same
@@andygozzo72 does that mean the test cannot detect it, or it wasn't designed to test it?
@@Walczyk i dont know, i dont know how this chip tester tests, presumably sets an input byte checks output isnt the same , pulses latch input, then check it is and then goes to next byte, 0 to 255 in turn? problem is if it doesnt 'wait' long enough when testing to miss a failed 'latch' it may make a false 'pass'
[RS] is RS Electronics. A big component seller here in the UK. In fact it was always joked that you could get most of the parts for a BBC Micro from RS. So either those chips came from RS or that Motorola had rebranded them for RS but then needed to provide some to the MB manufacturer because of stock shortages or similar.
the prototype bbc very likely did use all rs parts, maybe someone could ask Sophie Wilson 😉
As a designer in the 80s, I was always advised - "if you're going to design a new part in to a board, make sure you can also get it from RS, that way you can be sure it will likely have multiple suppliers and will be around for a long time". Of course, it was never quite that simple, but it was a good approach to ensuring component availability for a design/
@@GodmanchesterGoblin from your user name, you're probably not far from me 😉
@@andygozzo72 In case it makes a difference, I am in Godmanchester in the UK, not Godmanchester in Canada. 😉
@@GodmanchesterGoblin thats what i meant, i'm in a cambridgeshire village so not far from there 😉
Ha! You tricked me and snuck a video out on new year's eve! Feel free to ignore my comment on your other video, since it's completely redundant now.
Glad you could fix the board, although those re-marked chips are very strange. I'd be tempted to replace the others just in case. But fixing a faulty board is a great way to end the year.
Anyway, happy new year to you too! Hope you have a great 2022!
Crazy you Could find that.
Nice done 👍👍👍, always a great experience to watch one of your movies.
Happy New year as Well from Denmark.
amazing job adrian !! also huge thanks to the commentor for finding the issue too :DDDD
There’s not a lot of power supply decoupling capacitors on that board. I wonder if adding a couple more 100nF capacitors nearby across the supply rails would make the bad chip behave better?
I wish a wonderful old years day and a happy new year.
Holy freaking ghost of an XT! What a great debugging session.
Great video. The branding on the chip is radio spares Aka RS component's. They used to brand stock with there own logo and sometimes even there own part numbers!!! Though majority of the parts they supplied were from good quality suppliers.
Oh that noise, scraping off the paint with that Q-tip...! I love these videos but I'm like, Adrian, noooooo!
Yes I agree that's the RS components (Large UK electronics distributor) logo.
definitely, i have many of those RS remarked chips in a box of old but unused chips 😉
Nice find Adrian! Thanks, as always, for the great troubleshooting content that you share with all of us. I hope that you have a very Happy and Healthy New Year! Fred
I had similar issue with my Atari STe, and ACSI/DMA port buffers (LS244, LS245). They were also testing perfectly fine, but data on hard disk was being corrupted. Replacing these chips helped.
I'm glad you took the scope out. It would have been quite unsatisfying otherwise as a closure, since the chip tester reported it good. What an interesting failure miode.
Nice job. Funny how you allude to swapping parts out earlier as not being proper procedure. And in the end, that's what fixed it : )
Heh it's true. I should have looked at these signals on the scope first to figure out exactly what was wrong...
Blindly swapping parts is not good practice, but once you have your finger on a suspect, I see nothing wrong with giving it a shot.
Really enjoyed all your videos in 2021, you are such an inspiration.
Just finished watching the other video and this was uploaded 1 hour ago 😂❤️.
RS is Radio Spares. They rebadge stuff all the time, and have done for years. It’s not usually junk. I wonder if the manufacturer used RS because they couldn’t get stock elsewhere. Usually RS is expensive but good at holding stock.
Would it be normal to have them installed from the factory in a Taiwan clone motherboard?
RS as in RS Components? The UK based company?
@@adriansdigitalbasement2 yep.
@@adriansdigitalbasement2 thats what seems strange, them using parts from a UK based 3rd party supplier!! i doubt it was a 'bad' chip, from new, just failed, looks to me more like a latching problem than the output drivr section as it does briefly pulse high, but not hold, i have loads of RS remarked chips in a box of old but unused TTL stuff
@@adriansdigitalbasement2 given that Hong Kong would have been a British colony when this board was made, and Taiwan and Hong Kong very deeply work with one another in electronics, it makes total sense that there would be a RS office that could have been a supplier.
Very nice sleuthing, and great breakdown of the actual issue! Happy New Year!
Thanks for the quick update. Seriously, the suspense was killing me.
Great video's leading to the fix. Very educational. Thank you Adrian!
Would be nice to test that broken chip on the tester while the scope is hooked up and see how it behaves there regarding that faulty output pin.
Nice! I'm still fighting a 386 (later model) that's stuck in boot loop, show bios screen and resets
So much harder on those types of boards with large VLSI chips that contain most of the logic. Much harder to troubleshoot things.
@@adriansdigitalbasement2 oh I'm fully aware lol traced everything, found a bad trace..patched it and no change at all. Makes board #2 that's kicking my butt...2nd only to my rev3 5150..
As others have said the RS is RS Components, use the company all the time, but didn't realise they had repainted ICS.
yep, they rebranded most of their parts,at times
Another problem that can happen with jellybean ICs is that some circuit designs are stricter in their timings that the jellybean standard allows for. It's rare, but happens, and when it does you usually have to go with a specific manufacturer.
That’s an RS Components logo on the ‘373. A UK component supplier (presumably elsewhere too) they’ve been around for decades. Didn’t know they rebadged chips, I’ve not seen it recently on anything I’ve had from them.
they did it a lot back then, i have loads of rs branded logic in a shoebox!
You don't need all address lines to work for refresh. RAM access pattern is different during refresh, and IIRC only A0-A7, maybe up to A8 or A9 are used to select the row to be refreshed. All banks are refreshed at the same time. There is a special REFRESH signal on the ISA bus to put ISA RAM boards into refresh mode. ISA RAM boards don't need an address counter, because the row number to be refreshed is on the *low* address bits.
happy new years to you i a big fan to iv learned alot wacthing your channel both of them iv been a fan for a long time now
Likely the chips were "rebadged" due to region restrictions in place at the time. Chip shortage caused by the IBM PC success left European and Asian manufacturers with no chips. The chip manufacturers set up regional allocation schemes and thus began the original "outsourcing" where companies set up factories in Europe and Asia. 74LS373 failures are actually quite common, quite likely a timing issue. The 8237 being an 8 bit chip, the CPU actually programs the upper address lines for transfers.
The RS symbol on the 74373 is a logo for Radio Spares (RS Components) and electronic spares supplier. "In 1937, J H Waring and P M Sebestyen established a company called Radiospares Limited, supplying radio repair shops with spare parts. The business began selling electronic components in 1954 and rebranded as
RS Components in 1971." Taken from RS Components website.
They would have re-branded the chip for their stock under license from Motorola I guess. I'm surprised if the chip was a reject. RS Components are known as a quality supplier. The problem could be caused by the battery acid corrosion, but then again somebody has had this very chip fail before. I wonder if this was a common failure? I have no memory of this particular fault from back in the day.
RS (Radio Spares) are a well known UK distributor of electronic components that still exists today. I guess the rebranding was a similar as the major supermarkets have their own house brands - possibly they can set a lower price and move a bigger volume.
Great work Adrian! Always nice to see a happy ending to those!
BTW is that GZDoom on your taskbar? When will we see you playing some crazy wads?
Hahaha yeah good catch -- it is GZDoom which I will run sometimes for some "Stress relief" :-)
Memory refresh only needs to use chip select along with RAS, presenting only 8 bits of the lower address on the line. Then the chips do a RAS only refresh, not needing to have any other selection. Saves time, in that all columns in the chip are refreshed at the same time, as the columns contain the read amplifiers, so they read the bit, and immediately write it back, doing 128 columns at a time. Larger chips implement this still, but as they have more address lines the number increases, but still you only use RAS to do refresh, and for most chips you can actually do refresh by simply having enough reads, so long as your reads go through the RAS only address space fast enough to hit every location.
Great deep-dive into this problem Adrian!
ah my guess was wrong, nice video and repair though! Lucky Ollie had the answer, that would have been a pain to diagnose entirely by yourself.
Adrian, i think, that stickers are a good idea, but a bracket (metal one or 3d printed) can be even more helpful to avoid confusion of incorrect xt-ide card installation.
Thanks for this interesting story about pc-xt mobo - never know, that such little chip can be treacherous marginal and make this trouble appear.
congrats, you have found a needle in a haystack) Also I would consider checking if something else drives that line low and newer chip simply wins this race condition (check line resistance against ground, also something sharing that line might become hot). I wonder how this part would act in tl866 chip check.
Happy new year! Glad to see you like the band.
DRAM refresh doesn't actually need all of the address lines - depending on age it either need just the row address lines (RAS) but not the column address lines (CAS) of the memory chip (RAS-Only Refresh Mode), or it just needs the RAS/CAS lines to be pulsed properly and doesn't need any address lines (CAS before RAS Refresh Mode).
IIRC an XT is old enough that which of these two are used depends on which memory chip they used - the 4164/41256 datasheets I found via Google only mention the first but the really ancient 4464 datasheet I found mention both. But the upshot is that either some or all address lines are irrelevant during memory refresh so these can be broken on the DMA circuitry and the machine will work unless DMA is necessary for other purposes like your floppy drive.
Not going to claim I figured this out during the previous video - I triggered on the "the DMA works because DRAM refresh needs all memory lines" which made me go "No it doesn't - wonder if...", you had already tried a different DMA controller but then the schematic came up showing three different buffer drivers which could be involved...
The oscilloscope shot of the failure was REALLY nice!
100% correct! For RAS only refresh cycles the row address is supplied externally (9 bits for a 256k x something chip. For CAS before RAS refresh which came in at around the 256K generation (if memory serves) the DRAM chips would each maintain their own internal refresh row address counter. I designed non-PC computer memory boards using both methods in the 1980s.
Just here to drop my Happy New Year to you Adrian! Best wishes! 😊🎆
Adrian, often you have specific schematics pulled up in your videos, could you list for us the web sites, or online resources you use to find them? I'm interested in trying to fix my Tandy 286 clone, but I'd feel much more confident if I had pinouts and schematics for it. Thank you.
> sees bitwarden in your extension bar
a man of culture, I see :3c
Huzzah! Hurray! Now, let's actually watch this.
I think it would be great to see you installing the turbo components!
At work, from time to time, our customers would have us build a prototype, whether a new product or a ‘respin’ of an existing product, and sometimes we or the customer would find that we might have odd issues with one brand of IC, yet it would work perfectly with another brand…
i've just found pdfs of the datasheets of both the TI and motorola versions and state identical source and sink current capability, the dud one looks like it can pulse high correctly, but not hold it so i'd say its more a latching problem, not its output driver..? however back in the 80s they may have been specified differently??
REALLY NICE work!!!! Enjoyed that! Thankyou!
You should go through some of the bags of chips that it marked bad to see if you're getting false positives and false negatives.
In the tester the IC only has to drive a very light load, but in the PC more inputs are attached and are switched faster. During production, the ICs were already marginal. But were seen as a cheap alternative if the system works with it.
looks more latching problem, rather than output 'drive' capability,
Seems to me that the retro chip tester pro doesn't simulate a realistic load when tasting these logic chips. Either that or the load on address line 15 is unusually high. A partially failed bypass capacitor could cause this.
Two possible reasons I can think of for the rebadging
- The chips were marginal and did not pass Motorola QC and were sold of to another company, under the strict requirements that they would not enter the market as Motorola parts
- Motorola had an over supply of chips, not wanting to drop the price on Motorola branded parts, they were sold to another company, again with the strict requirement that they would not enter the market as Motorola parts.
Could be also that RS simpy wanted their own brand, so they could purchase chips at the cheapest supplier without having to convince the customer to switch to another brand.
@@danielmantione RS did rebrand a lot of parts to their house numbering scheme, as it made internal stock control easier, handling stuff with only a RS part number, and a generic TTL part number, with a date or batch code from RS to trace them. Still do, especially with test equipment as that allows them to have special tweaks made just for them in the firmware.
SeanBZA is right. RS components used to buy parts from other manufacturers, rebrand them and sell them. It was their core business model during the 80s. The parts were legit, although they were selling them more expensive than the usual retail outlets.
Radio Shack / Tandy also sold chips with their own part numbers. (Inside their computers usually) But they worked with the manufacturers of the chips to make custom markings on the chips. They weren’t painted on LOL.
@@adriansdigitalbasement2 All IC manufacturers will custom mark product for a customer order, you just have to order 10k parts and they will do that for you. Most common on mask ROM, when you wanted to get past the prototype stage, and was before the evolution of OTP EPROM becoming popular, which killed the mask ROM business for all the small volume, and even made big inroads in large volume, as they could do upgrades much easier, no need to wait for the chip bulk to be finished, just the units that have been programmed.
Nice catch. I have several IC testers, including an expensive commercial one, and I have found several chips test OK on some and bad on others, since they each use different methods, levels, etc for testing (and some cheap ones could even have poorly designed tests). I would love to try that chip in all of my testers, it sounds like a good benchmark. Let me know if interested, I have your E-mail from prior correspondence (TI-99 CPU purchases and I sent some cartridges to you).
Happy new year.
Good fix Adrian! Thank you for sharing it with us and Happy New Year! to you as well💖👍😎JP
Awesome troubleshooting 😊
Relabeling is not a new thing, in the good old days with vacuum tubes for audio and tv, it was quite normal that the mfg's bought tubes from each other you could buy a tube from Tungsol; but it was a RCA tube labeled (printet) and packaged at RCA.
oh, yeah, here in the uk i've seen that a lot , such as mullard/philips made ones remarked as mazda, etc.
The logo on the re-badge chip looks like RS Components (rs-oline). I wonder if RS did their "own line" of ICs back then?
I know that two of these re-badged chips haven't stopped working (yet), but if it was me I would swap them out for something I was more confident in. Once bitten, twice shy if you know what I mean. Yes, I know that these are legitimate re-brands for corporate reasons rather than anything dodgy, but it is the kind of thing that just leaves me with suspicions at the back of my mind.
I sense a pattern... it seems most of the standard components in old computers that die seem to be some sort of buffers or bus drivers. Maybe the three-state logic wasn't all that good in the early days? Not to mention the usual ESD suspect 😉
usually chips that get hot excluding the CPU etc which is usually reliable
Bus driver chips got abused a lot in the 80's.
The BBC Micro is notorious for overdriving the 74LS245 on the databus, by having too many devices connected to it.
They actually had to add a resistor array to the databus to fix issues created by using the chip beyond it's specs.
in the datasheets this is the value for "fanout" or how many standard 74 logic chips an output can drive.
I think these chips do remarkably well, considering they have lasted nearly 40+ years years in some cases. Modern day chips would do well to last as long.
@@station240 I recall Steve Furber talking about that. They had no idea that they were abusing the 245, and the resistors were basically a voodoo change that seemed to do the trick, but they'd no idea why it worked.
@@a4000t Indeed, most of these chips are hanging in remarkebly well; it's just that the likelyhood of something else failing seems to be (within my field of view) quite a bit lower than the bus drivers/transcievers. The overdriving that station240 mentioned might be a reason, but I wouldn't expect that on an IBM design. Just an observation based on quite a few replaced ones of these here :)
I love that you typed su instead of dir :)
How reliable is that chiptester pro then? Could you do a test of a confirmed failed chip to see if it actually comes up as faulty? Great stuff as always Adrian.
seems like it takes time for the chip to start acting up, maybe if you tested it for a couple of cycles then it would test bad eventually in the tester thingy, but maybe it needs continuous use to show the signs of fault
happy new year!
Happy 2022!
That's the logo of RS components. RS stands for RadioSpares.
I'm surprised the retro chip tester tested that chip as good. It must be missing some tests to fully evaluate it. It may be worth contacting the creator to suggest a better test that will properly test the chip.
Stephen, the guy who made the Chip Tester says that it can't accurately test output drivers as it puts basically no load or capacitance on the outputs ... he says it often won't detect such problems and there is just no way to accurately test that kind of thing without making it super expensive and complicated.
@@adriansdigitalbasement2 i'd say its not a drive issue as the outpu driver part of the chip is separate from the latch part, according to the datasheet i've seen, more likely the chip tester tests it too quickly, not waiting to see if the output changes after a short delay after removing a latch trigger/enable
Ending 2021 on a high. Hopefully we can hold the high far into 2022 unlike that chip!
A clear blow to the reputation of the Retro Chip tester Pro.
For future testing I'd recommend one of those ISA POST code cards, they've helped me quite a lot.
The original PC didn't generate post codes. He tried that earlier.
@@Rorschach1024 Ah my bad
That RS at 9:18 looks like the logo of Radiospares, now known as RS components
RS is RS Components Ltd, the logo matched.
great video again! Thank you & Happy new year to you as well! 🙂