First thing I'd do is figure out if the EEPROM data is still good, by comparing to another unit, or checksum if printed on the labels. Also try verifying at a range of supply voltages. EEPROMs are the one part that has an inherent decay mechanism, so high on the list of suspects.
I know right , had 10 year old dvd players with blown caps :( although , that might be cause these days everything has switching power supplies instead of a transformer.. ("always on ") versus only on with main switch.
capacitors failing is a thing starting in the mid 90s up to mid 00s. in the mid 90s it was due to SMD electrolytics leaking, and in the mid 00s it was just poor quality. i've worked on a few very old TVs and computers and, on a C64 i found just two bad electrolytics. every other cap was within spec (OVER rated). on TVs, the only bad cap i ever found was the +B cap but that one is under a lot of stress. on modern devices, there are no bad caps, just bad supply design that puts OK caps under unnecessary stress.
@@drgenio2006 I know some worker stole an electrolite formula for caps , and sold it t competitors , but it was a bad formula , planted for him. it was the cause of a lot of problems with caps. after 1999.
Pretty nuts that this thing required a full on 10MHz 68k CPU. At the time it was released, this thing still had all the computing power of an entry level Unix workstation. Just to pass frames between two ports at 10Mbit
This was an absolutely beautiful video!!! Those cinematic shots, the phenomenal music composition, the cut to black; I could go on and on! Not one bit went unnoticed and I appreciated all of it! Bravo!
Haha! I don't usually get sentimental about caps and fans, and will go ahead and replace both while I'm cleaning up an older piece of gear. Saves me from having to do it later when they eventually fail. (Although, I can't agree strongly enough that Noctua is not the first and last word on fans. They have their place. That place is not "everywhere.") But, for an outfit that is directed more toward preservation and demonstration in a museum context, I'm actually glad for this too.
Their industrialPPC fans would be a better fit than their consumer PC fans, they have a lot better static pressure than their normal fans while being quieter than equivalent industrial fans.
Why would they be underrated? Fan technology has improved significantly since this bridge was made, and (the correct) modern fans would perform better both in airflow and static pressure, while making less noise.
@@game-tea For one ... have they? I dunno. We had jet engines in the late 80s. I don't think moving air through a rack chassis was exotic stuff at the time. ;-) Some things have gotten optimized -- bearings, blade shape, beveling the bezels -- but the real difference in noise comes down to three factors, in this order: 1) Speed (RPM). 2) Turbulence caused by obstructions. 3) Mechanical coupling of vibration. The first is directly correlated to CFM and static pressure. You can make any fan silent by not spinning the blades, but that would have a bit of an impact on its efficacy. Noctua fans runs slow. That's why they're so quiet. And that's why they're often underrated -- because so many people conclude that Noctua fans are so quiet because .. brown .. and never consider that an engineer calculated things like junction temperature and heatsink-to-air resistance to arrive at a CFM rating that would allow the device to run in the temperature range it's specified to run in. Granted, there's often significant margin there, and as long as you don't ask it do the things the manual says it can do (fully loaded at 40C ambient), it'll probably be OK. At least for a while. Noctua fans won't do anything about item 2 above. When you push air right through a grille with some routed circular slits, or a bunch of drilled holes, you're going to get buffeting noise. 3D print a spacer to move the fan back by an inch or two and that'll improve. The final point -- well, Noctua fans DO improve that, on account of the rubber mounts. You can also retrofit something similar onto the stock fan without compromising the device's cooling. If it's even a significant source of noise in any given case... So yes. "Underrated" is usually accurate. They're PC fans meant to be used in groups to form a slight breeze through a big open case, and they do a reasonably good job at that. They are not cure-alls, and they do not possess some exclusive mysterious voodoo that breaks the laws of physics.
Remember it is just a 10 megabit/s 2-port switch! Unbelievable how much hardware went into this. Look what is inside the minimal switch (e.g. 5-port) today... and that one is 10-100 times faster too.
well. a bit more than that. it's an optical converter as well. and even for most of 90s most people just used hubs. this one does routing of the ethernet packages, like a switch, having a bespoke computer in there to deal with the bandwidth.
@@Rob2 sure but it doesn't just dump the ethernet on that module, I think. that's just fiber-to-pin. they did have remote ethernet repeaters at the time too though using the same optical cabling, the manuals for this mention those on the sections where they speak of how to measure and certify the cabling(and how to build the fiber cable). of course you can get the fiber stuff in a wifi/ethernet switch box but for the time they did pretty well with a board that's not that much more complicated or bigger than an amiga 1000, they didn't make a complicated chassis and separate everything into a lot of different boards as was the style in 80s, instead it's almost like an appliance switch in design just with bigger chips and less integration
@@lasskinn474 This box has two general-purpose ethernet controllers, a powerful general-purpose CPU with program EPROM and scratch RAM plus the required glue logic, an associative memory implemented with programmable logic and more RAM, and then it has only AIU ports so the media interface is external or on that fiber optic board. All of that is present in a $20 switch you buy at the local store, except it has all of it in a single chip, and it is 100 times faster. So there has been quite an advance in technology...
Many props for dumping the ROMs. If you end up opening the second unit up at some point, would you mind checking if the EPROM labels are different and dumping them as well if so? Alternate firmware revisions are always neat to have.
I sure wasn't expecting a Hand Tool Rescue cameo lol. That gave me quite the spittake. Also, what a truly beautiful video. Really. It should be on the first channel imho. Thoroughly enjoyed it
Fantastic clip, and to top it off - there is even a cat! This video reminded me I have the early 1992 made Ethernet bridge by D-Link. 1U factor with AUI/BNC for both ports. I feel the urge to get it from the garage and test it ;)
I used a few of these back in the 80’s. So happy to see these restorations as this was a revolutionary box! Good luck with the next steps in debugging this board!
21:06 Looking at the holes in the case, your board looks significantly bent in comparison, since you also seem to have narrowed it down to that area, id check all the solder joints in the area, possibly just reflow them all.
I have seen so many caps leak of that age and destroy the boards. I worked in a small plant that hand assembled devices along the size of that. I had a station where I populated the boards with devices and hand soldered them on.
Fantastic video! Just a heads-up - I'm sure you know this, but I've heard that a lot of the switching supplies from this era can fry themselves if you don't supply dummy load.
The 2 EEPROMs my have different firmware on them to signify the Tests from Each side (AOI and Fiber) before it gets to the Motorola controller. Also, On the Fiber there may be a burnt out connection on the FIber Board. you may want to test the diode to see if that will even flash.
7:50 safety tip: do high voltage stuff with one hand, and keep your other hand somewhere safe like behind your back (not touching anything grounded obv). either attach the clips one by one, or use some probes u can hold like chopsticks. your arms are a pair of wires made of meat, connected roughly in series with your heart. if you can ensure one of those two wires is left floating, then at least in the case of DC or low-frequency AC, you *theoretically* can't form a circuit that includes the heart. not perfect, if the shock forces your muscles to contract, good luck controlling where either of your hands end up, not to mention your arm not actually being a perfectly homogenous conductor that'll form a perfectly predictable current path, or the fact that your other hand is never truly floating due to capacitive coupling. but always seemed like a smart way to get a small bonus to survival probability should all the other precautions fail.
We had one of those at work between the 10base5 and 10base2 network segments we used. Not sure if it was the model with 2 AUI connectors (and a 10base2 interface) or if it had a 10base2 BNC right on the box. But it was the same big box from Digital...
I wasn't born at this time and to me 10M is painfully slow (yet I started my life browsing internet on a DSL [and i still use 56k modem for my job]). Idk why but seeing people repair stuff is for me so beautiful.
You should be able to in-circuit test the RAM chips without having to de-solder them. I would use an ABI boardmaster, but you can often use an EPROM programmer that has RAM test functionality by making a dip socket to dip test clip cable
Cool piece of history. Just think, these days all of that integrated logic is done on a single chip in a device that runs off 3.3v at magnitudes more speed - or often just done simply by software running on the host CPU.
Great start to the restoration. Looking forward to the next instalment. Just wondering if the diagnostic ROM outputs to a serial with some sort of result code you might be able to use for further investigation?
Yeah sorta-- the manual says the self test records the step it failed on in the NVRAM (which is soldered to the board). After editing the video I went ahead and desoldered it and popped it in the reader. There was barely anything on it, and nothing in plain text that I could make sense of. I just put a copy of it on files.serialport.org if anyone wants to take a peek.
@@TheParallelPort Maybe you can put nvram to the socket to be able to dump it more than once? I think about running the bridge again to see which part of the dump is changing (if changes at all) because now there is too much to hook on something.
Might also be interesting to look at the Chip Enable lines of the RAM, since you've identified that there are different banks for different purposes. That would at least narrow it down to banks that have passed, banks that have not yet been tested, and (ideally) the bank that is under test when it fails.
Those PAL chips... Ouch, PALs were burned during the manafacture process, sadly they seem to last only about 25-30years before they start to "forget", GAL chips were better as you could reprogram them after they've been made.... if you have the logic array image thats supposed to be on them that is.
Hey, I might need some help with restoring an old ericson round 1989-95 era. I'd love some help with it the only existing problem with it is a part in the powersuply that has exploded. Hope you dee this!
You said the -12v rail was low. Some old memory chips use that voltage, they might not function with it low. You should make sure the power supply works completely. Next you would think such a device would have a serial port for diagnostic output, but i couldn't see anything like that in the video.
these memories only need 5V if it would be a modern highly integrated stuff, then the 3 pins next to the dip switch could be a serial port (still could be interesting) --- but this one would need a dedicated UART IC to do it, except there is none to be seen --- welp, seems like time for some good-old logic-analyzer fun....
The oscilloscope at the end, shown probably 2 things: -using 1:1 probe, its common mistake. 1:10 have lower capacitance -3-level signals, so few outputs short circuited, or some output broken. In this case good to watch by IR camera. 30 yeaes ago, I've used own fingers and drop of alcohol to see the evaporation speed. Thanks to remembering about AUI. SGI Indy have it, together with RJ45. I was very glad than have permission to change local network from coax to twisted pairs, and forgot about everydays network falling and electrick shocks.
I admire people who even bother with old power supplies. I probably would try to replace all of them with a USB supplied PPS or some micro ATX adaptation
Honestly if you know what your doing then fixing these old psus are not that difficult. There more reliable then anything modern. And for restoration purposes (I.e keeping things oem) this is the only way
A suggestion re the EPROMS - try combining them into a single 16 bit wide file and see if a disassembler can make sense of the data. Maybe there is a Ghidra addict out there who can turn them into something sensible and rule out bit rot. Bonus points for repairing the fans !
If your transceivers have an SQE/heartbeat switch, have you tried changing its setting? I don't know if it should be on or off for this bridge, I only know it should be off when used on repeaters and on when used on NICs. It's quite possible that the bridge is either waiting for the transceiver's heartbeat signal to know the collision detection of the transceiver is working, but doesn't see it; or it doesn't need it and is confused by it being present.
Wow. Would not have suspected the aux transformer as the source of failure. Also really odd to see that as its own fully bespoke part vs. a winding on the main transformer. This is definitely from an era before these designs were optimized.
One of the problems was screwing the case together before fully testing it - NEVER do that! By Murphy's Law, you are then guaranteed to find another problem that means un-casing whatever it is again. (And if it does not work, you have not wasted time). Try a thermal camera to see if anything is noticeably warmer or cooler than other similar parts on the board? You can sometimes test RAM by piggy backing another similar IC on top, just with the pins set in slightly to give a good spring fit contact. You could also try copying the EPROMs for your working one, to test the board with two wired transceivers, in case the firmware is different to work with the fibre optic unit?
You should put a heat resistsnt very long lasting sticker on the inside of it somewhere with a note about this video, when it was cleaned and a summery of whatever you want. For the next person to get it. Hopefully humans will pass things down like this more into the future and stories dont get so easily lost. RUclips is not forever. Nothing is. Also not sure what type of screwdrivers you use but Im hoping they are soft tip. It always sucks to strip/scratch a screw that is just as much part of the relic as any capacitor.
Something you can do before desoldering all of those DRAM chips is checking that all logic chips are functional (check that signals aren't between levels or pulsing erratically), checking for broken connections may also be of interest, especially of clock, chip enable and read/write control wires. You may also want to inspect any chips next to the power supply, because while it didn't burn degradation may have happened on the chips (or the board) that are physically closest to it. This advice is not related to this specific machine, so please keep that in mind.
I had to fix a few stray wires on transformers before. Strange how it happens. Old Crusty fans. Then I love how fiber just blows everything away for distance / speed.
My college was DEC everything that wasn't a PC. I'm certain they used several of these to spread the college LAN out to the residential computer labs. Somewhere in my junk, I have several of those plastic case shells, which the college threw out, but I grabbed them for future projects. Apparently, they were relocating the equipment into something rackmounted.
Be sure to thoroughly check the PSU and install a crowbar cicuit if it doesnt already have one. An overvoltage situation on this ginourmous motherboard would be horrible, just think about replacing every second chip.
Probe the address lines of the memory with a logic analyser then you can step through what fails without taking all the chips out (although you might as well get them all socketed if one needs replacing). Check out some of the videos from Curious Marc about how they debug stuff
So what would companies have used before this item was available. I presume this was the pioneer from which a whole industry followed? Excuse my ignorance but this seems like a breakthrough product judging by your description. Thank you fir youe time and efforts.
not very familiar with history, but probably either one large ethernet stretched to its limits (I think it was coax at the time given the AUI ports, not yet twisted-pair), or separate ethernets connected via routers when possible (assuming the network ran only routeable protocols like IP or DECnet) -- but if I understand correctly, routers were quite a bit larger though, and often the size of a whole mini PC (e.g. a whole Unix machine to route IP, or a whole VAX running OpenVMS to route DECnet) so this bridge would've been very lightweight -- or other local network types that had different limitations (FDDI maybe? Token Ring? I have no idea which ones were the most common alternatives)
@@leggysoft TLDR: it's a switch, not a hub. You need to remember Ethernet used shared media back in the day. If one device is sending a packet nothing else on the entire network can transmit. A hub/media converter changes the physical interface but does not address this limitation. If a device connected to port A is transmitting it will be retransmitted on port B so both sides of the network will be blocked. A bridge/switch is much more complicated. If a device on port A is transmitting the line on port B will be left free. The switch will only transmit on port B if the packet is destined for a device on that side of the network. This can greatly cut down on the network utilization and improve performance. This requires enough smarts to learn and remember which side of the network each device is on.
Did anybody else see the upside-down PCB number in the corner of the PSU board at 6:37 and panic? It's "5015085D" upside-down, but at first glance it was close enough to a possibly-damaged "asbestos" to set off alarm bells.
Those aren't "snap rings" on the fans, those are "Jesus clips". As when one goes flying across the room the first words out of your mouth are "Jesus H. Christ!" 🤣🤣🤣🤣🤣
The RUclips channel Cathode Ray Dude recently mentioned he has one of these I think. If you want to get hold of another there might be a coop opportunity. Remove this commenting you want to limit advertising, but I just thought you should know.
Well, 10base5 was better than 10base2, but the latter was much easier to work with. (even easier than UTP! you could always add another device in the same room, never out of switchports or wall outlets)
I think you've confused this for one of those scripted dramas where everything works out in the end. This is how it goes in the real world. Sometimes the thing works. Sometimes it doesn't. At any rate, my man here will no doubt be spending some time nursing this back to health. So I guess everything _will_ work out in the end. If it hasn't worked out, it's not the end.
First thing I'd do is figure out if the EEPROM data is still good, by comparing to another unit, or checksum if printed on the labels. Also try verifying at a range of supply voltages. EEPROMs are the one part that has an inherent decay mechanism, so high on the list of suspects.
Easier to do in this case, since you have a working unit!
Wouldn't the EEPROM be where the MAC address is stored?
@@soundspark 22:50 Mac address is stored on the PROM
36 years old caps still good - unbelievable
I know right , had 10 year old dvd players with blown caps :(
although , that might be cause these days everything has switching power supplies instead of a transformer.. ("always on ") versus only on with main switch.
These are pre-capacitor plague.
capacitors failing is a thing starting in the mid 90s up to mid 00s. in the mid 90s it was due to SMD electrolytics leaking, and in the mid 00s it was just poor quality. i've worked on a few very old TVs and computers and, on a C64 i found just two bad electrolytics. every other cap was within spec (OVER rated). on TVs, the only bad cap i ever found was the +B cap but that one is under a lot of stress. on modern devices, there are no bad caps, just bad supply design that puts OK caps under unnecessary stress.
@@drgenio2006 I know some worker stole an electrolite formula for caps , and sold it t competitors , but it was a bad formula , planted for him. it was the cause of a lot of problems with caps. after 1999.
I remember DVD players was failing in mass due to bad caps. Caps was put to the hottest place. Planed.
Pretty nuts that this thing required a full on 10MHz 68k CPU. At the time it was released, this thing still had all the computing power of an entry level Unix workstation. Just to pass frames between two ports at 10Mbit
This was an absolutely beautiful video!!! Those cinematic shots, the phenomenal music composition, the cut to black; I could go on and on! Not one bit went unnoticed and I appreciated all of it! Bravo!
5:16 is real art 😍
+1 for the Hand Tool Rescue tools haha
Great wrench... I got the bigger one! ++1 for the HTR plug too!
Nice cameo by Hand Tool Rescue.
Thanks for fixing those fans and not just slapping an under rated Noctua fan in there.
Haha! I don't usually get sentimental about caps and fans, and will go ahead and replace both while I'm cleaning up an older piece of gear. Saves me from having to do it later when they eventually fail. (Although, I can't agree strongly enough that Noctua is not the first and last word on fans. They have their place. That place is not "everywhere.")
But, for an outfit that is directed more toward preservation and demonstration in a museum context, I'm actually glad for this too.
Their industrialPPC fans would be a better fit than their consumer PC fans, they have a lot better static pressure than their normal fans while being quieter than equivalent industrial fans.
Why would they be underrated? Fan technology has improved significantly since this bridge was made, and (the correct) modern fans would perform better both in airflow and static pressure, while making less noise.
@@game-tea For one ... have they? I dunno. We had jet engines in the late 80s. I don't think moving air through a rack chassis was exotic stuff at the time. ;-) Some things have gotten optimized -- bearings, blade shape, beveling the bezels -- but the real difference in noise comes down to three factors, in this order:
1) Speed (RPM).
2) Turbulence caused by obstructions.
3) Mechanical coupling of vibration.
The first is directly correlated to CFM and static pressure. You can make any fan silent by not spinning the blades, but that would have a bit of an impact on its efficacy.
Noctua fans runs slow. That's why they're so quiet. And that's why they're often underrated -- because so many people conclude that Noctua fans are so quiet because .. brown .. and never consider that an engineer calculated things like junction temperature and heatsink-to-air resistance to arrive at a CFM rating that would allow the device to run in the temperature range it's specified to run in. Granted, there's often significant margin there, and as long as you don't ask it do the things the manual says it can do (fully loaded at 40C ambient), it'll probably be OK. At least for a while.
Noctua fans won't do anything about item 2 above. When you push air right through a grille with some routed circular slits, or a bunch of drilled holes, you're going to get buffeting noise. 3D print a spacer to move the fan back by an inch or two and that'll improve.
The final point -- well, Noctua fans DO improve that, on account of the rubber mounts. You can also retrofit something similar onto the stock fan without compromising the device's cooling. If it's even a significant source of noise in any given case...
So yes. "Underrated" is usually accurate. They're PC fans meant to be used in groups to form a slight breeze through a big open case, and they do a reasonably good job at that. They are not cure-alls, and they do not possess some exclusive mysterious voodoo that breaks the laws of physics.
Honestly, I am replacing everything I have with Noctuas, because I am figuratively allergic to any kind of fan noise.
Remember it is just a 10 megabit/s 2-port switch!
Unbelievable how much hardware went into this. Look what is inside the minimal switch (e.g. 5-port) today... and that one is 10-100 times faster too.
well. a bit more than that. it's an optical converter as well.
and even for most of 90s most people just used hubs. this one does routing of the ethernet packages, like a switch, having a bespoke computer in there to deal with the bandwidth.
@@lasskinn474 The optical converter is a seperate board in this bridge that is tiny when compared to the "motherboard" with all the functions on it...
@@Rob2 sure but it doesn't just dump the ethernet on that module, I think. that's just fiber-to-pin.
they did have remote ethernet repeaters at the time too though using the same optical cabling, the manuals for this mention those on the sections where they speak of how to measure and certify the cabling(and how to build the fiber cable).
of course you can get the fiber stuff in a wifi/ethernet switch box but for the time they did pretty well with a board that's not that much more complicated or bigger than an amiga 1000, they didn't make a complicated chassis and separate everything into a lot of different boards as was the style in 80s, instead it's almost like an appliance switch in design just with bigger chips and less integration
@@lasskinn474 This box has two general-purpose ethernet controllers, a powerful general-purpose CPU with program EPROM and scratch RAM plus the required glue logic, an associative memory implemented with programmable logic and more RAM, and then it has only AIU ports so the media interface is external or on that fiber optic board.
All of that is present in a $20 switch you buy at the local store, except it has all of it in a single chip, and it is 100 times faster.
So there has been quite an advance in technology...
Many props for dumping the ROMs. If you end up opening the second unit up at some point, would you mind checking if the EPROM labels are different and dumping them as well if so? Alternate firmware revisions are always neat to have.
I sure wasn't expecting a Hand Tool Rescue cameo lol. That gave me quite the spittake.
Also, what a truly beautiful video. Really. It should be on the first channel imho. Thoroughly enjoyed it
What an amazing musical taste at 5:20...
Fantastic clip, and to top it off - there is even a cat! This video reminded me I have the early 1992 made Ethernet bridge by D-Link. 1U factor with AUI/BNC for both ports. I feel the urge to get it from the garage and test it ;)
I used a few of these back in the 80’s. So happy to see these restorations as this was a revolutionary box! Good luck with the next steps in debugging this board!
I hope there's more in the series, from the dialogue at the end it sounded like it was being shelved. 😢
Your video editing is top notch. I love the epic sounding background music and you have a nice narrator's voice.
all but except getting the name of his channel right lol
Trust Papst to make a servicable computer fan.
Some of the best parts out there.
I have a couple of Panaflos (still made in Japan) I bought in 2006 and they're still working like the first day.
21:06 Looking at the holes in the case, your board looks significantly bent in comparison, since you also seem to have narrowed it down to that area, id check all the solder joints in the area, possibly just reflow them all.
I have seen so many caps leak of that age and destroy the boards. I worked in a small plant that hand assembled devices along the size of that. I had a station where I populated the boards with devices and hand soldered them on.
Fantastic video! Just a heads-up - I'm sure you know this, but I've heard that a lot of the switching supplies from this era can fry themselves if you don't supply dummy load.
The 2 EEPROMs my have different firmware on them to signify the Tests from Each side (AOI and Fiber) before it gets to the Motorola controller. Also, On the Fiber there may be a burnt out connection on the FIber Board. you may want to test the diode to see if that will even flash.
The quality of components used were amazing even when these are from 1988 it's unbelivable
7:50 safety tip: do high voltage stuff with one hand, and keep your other hand somewhere safe like behind your back (not touching anything grounded obv). either attach the clips one by one, or use some probes u can hold like chopsticks. your arms are a pair of wires made of meat, connected roughly in series with your heart. if you can ensure one of those two wires is left floating, then at least in the case of DC or low-frequency AC, you *theoretically* can't form a circuit that includes the heart. not perfect, if the shock forces your muscles to contract, good luck controlling where either of your hands end up, not to mention your arm not actually being a perfectly homogenous conductor that'll form a perfectly predictable current path, or the fact that your other hand is never truly floating due to capacitive coupling. but always seemed like a smart way to get a small bonus to survival probability should all the other precautions fail.
Thank you for your service Sir
This isn't the military?
@@lonely.toaster What sort of question is that ???
@@TonyLingyour wording sounds like this channel served the military
You should collaborate with others on this, and build a Vintage Internet. Old hardware, old software. You could totally put Archie on it too.
We had one of those at work between the 10base5 and 10base2 network segments we used.
Not sure if it was the model with 2 AUI connectors (and a 10base2 interface) or if it had a 10base2 BNC right on the box.
But it was the same big box from Digital...
Wow, fans with real bearings
Nice work on this. Don't fret, I know you'll get her running.
* Great choice on the troubleshooting music too!
Check all your control voltages to spec , at a bunch of different components?
Just one of many ideas.
Although you may have already done
I wasn't born at this time and to me 10M is painfully slow (yet I started my life browsing internet on a DSL [and i still use 56k modem for my job]). Idk why but seeing people repair stuff is for me so beautiful.
20:00 that's a whole lot of unobtamium parts it sounds like 😍
The Digital Ethernet Bridge 100...
When you absolutely must have your bridge built like a freakin tank.
^.^
not me flinching while disconnecting the PWR supply
god damn look at the size of those capacitors
You should be able to in-circuit test the RAM chips without having to de-solder them. I would use an ABI boardmaster, but you can often use an EPROM programmer that has RAM test functionality by making a dip socket to dip test clip cable
Cool piece of history. Just think, these days all of that integrated logic is done on a single chip in a device that runs off 3.3v at magnitudes more speed - or often just done simply by software running on the host CPU.
wait, is it parallel or serial port?
Great start to the restoration. Looking forward to the next instalment. Just wondering if the diagnostic ROM outputs to a serial with some sort of result code you might be able to use for further investigation?
Yeah sorta-- the manual says the self test records the step it failed on in the NVRAM (which is soldered to the board). After editing the video I went ahead and desoldered it and popped it in the reader. There was barely anything on it, and nothing in plain text that I could make sense of. I just put a copy of it on files.serialport.org if anyone wants to take a peek.
@@TheParallelPort Maybe you can put nvram to the socket to be able to dump it more than once? I think about running the bridge again to see which part of the dump is changing (if changes at all) because now there is too much to hook on something.
Might also be interesting to look at the Chip Enable lines of the RAM, since you've identified that there are different banks for different purposes. That would at least narrow it down to banks that have passed, banks that have not yet been tested, and (ideally) the bank that is under test when it fails.
What about the -12V which was out of spec by 2-3V?
great music selection and a fun exploration of hardware from my youth.
Did you get the -12v nearer spec?
Yeah, thankfully! Forgot to add that bit in, but with the board & fans connected, it was something like -12.05V.
10/10, would bridge again.
25:35 - Wikipedia says it is DA15.
Is it possible that there is a diagnostics UART on there somewhere spitting out data about the self test?
Those PAL chips... Ouch, PALs were burned during the manafacture process, sadly they seem to last only about 25-30years before they start to "forget", GAL chips were better as you could reprogram them after they've been made.... if you have the logic array image thats supposed to be on them that is.
I think it may be a psu output issue, try feeding to voltage from another power supply
Hey, I might need some help with restoring an old ericson round 1989-95 era. I'd love some help with it the only existing problem with it is a part in the powersuply that has exploded. Hope you dee this!
You say in the intro serial port but the channel is parallel port?
You said the -12v rail was low. Some old memory chips use that voltage, they might not function with it low. You should make sure the power supply works completely.
Next you would think such a device would have a serial port for diagnostic output, but i couldn't see anything like that in the video.
these memories only need 5V
if it would be a modern highly integrated stuff, then the 3 pins next to the dip switch could be a serial port (still could be interesting) --- but this one would need a dedicated UART IC to do it, except there is none to be seen --- welp, seems like time for some good-old logic-analyzer fun....
@@5mf1nc I cannot read the part number in the video, so I cannot be sure of what it needs.
that hurt a bit to watch you use a #1 screw driver on #2 screws and strip them slightly
The oscilloscope at the end, shown probably 2 things:
-using 1:1 probe, its common mistake. 1:10 have lower capacitance
-3-level signals, so few outputs short circuited, or some output broken. In this case good to watch by IR camera. 30 yeaes ago, I've used own fingers and drop of alcohol to see the evaporation speed.
Thanks to remembering about AUI. SGI Indy have it, together with RJ45. I was very glad than have permission to change local network from coax to twisted pairs, and forgot about everydays network falling and electrick shocks.
I admire people who even bother with old power supplies. I probably would try to replace all of them with a USB supplied PPS or some micro ATX adaptation
You should likely not poke into instrumentation like this, if you have ideas like that.
Honestly if you know what your doing then fixing these old psus are not that difficult. There more reliable then anything modern. And for restoration purposes (I.e keeping things oem) this is the only way
A suggestion re the EPROMS - try combining them into a single 16 bit wide file and see if a disassembler can make sense of the data. Maybe there is a Ghidra addict out there who can turn them into something sensible and rule out bit rot.
Bonus points for repairing the fans !
If your transceivers have an SQE/heartbeat switch, have you tried changing its setting? I don't know if it should be on or off for this bridge, I only know it should be off when used on repeaters and on when used on NICs.
It's quite possible that the bridge is either waiting for the transceiver's heartbeat signal to know the collision detection of the transceiver is working, but doesn't see it; or it doesn't need it and is confused by it being present.
Wow. Would not have suspected the aux transformer as the source of failure. Also really odd to see that as its own fully bespoke part vs. a winding on the main transformer. This is definitely from an era before these designs were optimized.
Just beware that there werent any burnt down playgrounds underneath that bridge 😏😂
why didnt you reflow the boards and clean them?
One of the problems was screwing the case together before fully testing it - NEVER do that!
By Murphy's Law, you are then guaranteed to find another problem that means un-casing whatever it is again.
(And if it does not work, you have not wasted time).
Try a thermal camera to see if anything is noticeably warmer or cooler than other similar parts on the board?
You can sometimes test RAM by piggy backing another similar IC on top, just with the pins set in slightly to give a good spring fit contact.
You could also try copying the EPROMs for your working one, to test the board with two wired transceivers, in case the firmware is different to work with the fibre optic unit?
This soundtrack is intense…I’m on tender hooks wondering what is coming up next.
tenterhooks*
"DRAM" I see you channeling Asianometry. It was seen. I seent it.
there must be a serial console on that thing
I think part of the problem is that your -12v rail is really off an its upsetting it
Interesting insight, but this old DEC technology also deserves to be preserved
Wish things were still made in America
The repairable fans are epic. Normaly you just throw away fans. But no those ones have screws and replaceable bearings!
Let's all get this guy to 5K subs this year.
Very retro tech
You should put a heat resistsnt very long lasting sticker on the inside of it somewhere with a note about this video, when it was cleaned and a summery of whatever you want. For the next person to get it. Hopefully humans will pass things down like this more into the future and stories dont get so easily lost. RUclips is not forever. Nothing is.
Also not sure what type of screwdrivers you use but Im hoping they are soft tip. It always sucks to strip/scratch a screw that is just as much part of the relic as any capacitor.
Something you can do before desoldering all of those DRAM chips is checking that all logic chips are functional (check that signals aren't between levels or pulsing erratically), checking for broken connections may also be of interest, especially of clock, chip enable and read/write control wires. You may also want to inspect any chips next to the power supply, because while it didn't burn degradation may have happened on the chips (or the board) that are physically closest to it.
This advice is not related to this specific machine, so please keep that in mind.
I had to fix a few stray wires on transformers before. Strange how it happens. Old Crusty fans. Then I love how fiber just blows everything away for distance / speed.
My college was DEC everything that wasn't a PC. I'm certain they used several of these to spread the college LAN out to the residential computer labs.
Somewhere in my junk, I have several of those plastic case shells, which the college threw out, but I grabbed them for future projects. Apparently, they were relocating the equipment into something rackmounted.
Interesting pattern around the back edge of the main PCB. I wonder if it's more than just ornamental
What I want to find is whatever bridge Comic Book Guy was using for his Token-Ring-Ethernet LAN configuration.
Adrian's Digital Basement will get it going lol
Lethal? It's hte volts that jolts but the mills (milliamps) that kills right?
Be sure to thoroughly check the PSU and install a crowbar cicuit if it doesnt already have one.
An overvoltage situation on this ginourmous motherboard would be horrible, just think about replacing every second chip.
That would be a CuriousMarc - level project!
@@Rob2 yep, I had that one in my mind when writing the comment :D
4:34 WOW look at that marvel
So the first Ethernet bridge is created to combine two Ethernet bridges?
Why?
Probe the address lines of the memory with a logic analyser then you can step through what fails without taking all the chips out (although you might as well get them all socketed if one needs replacing). Check out some of the videos from Curious Marc about how they debug stuff
Today on the Serial Port lol but the channels name is the Parallel Port.
I keep seeing a blue axial cap with corrosion on one of it's leads. Might be worth checking out.
Get yourself some smooth jaws for that vice! You will thank yourself. :)
M.
So what would companies have used before this item was available.
I presume this was the pioneer from which a whole industry followed?
Excuse my ignorance but this seems like a breakthrough product judging by your description.
Thank you fir youe time and efforts.
not very familiar with history, but probably either one large ethernet stretched to its limits (I think it was coax at the time given the AUI ports, not yet twisted-pair), or separate ethernets connected via routers when possible (assuming the network ran only routeable protocols like IP or DECnet) -- but if I understand correctly, routers were quite a bit larger though, and often the size of a whole mini PC (e.g. a whole Unix machine to route IP, or a whole VAX running OpenVMS to route DECnet) so this bridge would've been very lightweight -- or other local network types that had different limitations (FDDI maybe? Token Ring? I have no idea which ones were the most common alternatives)
The AUI-connector is called DA-15 (a because of the width).
MC68000 rules!
36 years ago fiber was being used...till now its every where for networking why did it take 36 years
Is this dinosaur basically just a 2 port managed switch?
Managed? I thought they were usually unmanaged.
@@eDoc2020 If it was a 2 port dumb switch this giant box is nothing more than a media converter which does not need all those electronics.
@@leggysoft TLDR: it's a switch, not a hub.
You need to remember Ethernet used shared media back in the day. If one device is sending a packet nothing else on the entire network can transmit. A hub/media converter changes the physical interface but does not address this limitation. If a device connected to port A is transmitting it will be retransmitted on port B so both sides of the network will be blocked. A bridge/switch is much more complicated. If a device on port A is transmitting the line on port B will be left free. The switch will only transmit on port B if the packet is destined for a device on that side of the network. This can greatly cut down on the network utilization and improve performance. This requires enough smarts to learn and remember which side of the network each device is on.
Did anybody else see the upside-down PCB number in the corner of the PSU board at 6:37 and panic? It's "5015085D" upside-down, but at first glance it was close enough to a possibly-damaged "asbestos" to set off alarm bells.
I don't understand this, I see the 501... but what possibly damage asbestos
Those aren't "snap rings" on the fans, those are "Jesus clips". As when one goes flying across the room the first words out of your mouth are "Jesus H. Christ!" 🤣🤣🤣🤣🤣
The RUclips channel Cathode Ray Dude recently mentioned he has one of these I think. If you want to get hold of another there might be a coop opportunity.
Remove this commenting you want to limit advertising, but I just thought you should know.
Moist
10 base 5 is lame... 10 base 2 is better :D
Well, 10base5 was better than 10base2, but the latter was much easier to work with.
(even easier than UTP! you could always add another device in the same room, never out of switchports or wall outlets)
How underwhelming end. Sorry, loved to see this but having a memory issue killing a project is underwhelming.
I think you've confused this for one of those scripted dramas where everything works out in the end. This is how it goes in the real world. Sometimes the thing works. Sometimes it doesn't. At any rate, my man here will no doubt be spending some time nursing this back to health. So I guess everything _will_ work out in the end. If it hasn't worked out, it's not the end.
the boards werent' dirty at all and all the caps looked fine. just power it up first
Well that was a waste of my time
I'm not sure what you were expecting from a channel that does video diaries of old networking equipment ....