-
I am in strong agreement with what the guest named 3DG4 has been proclaiming. It reminds me a lot of mechanics who just want to replace parts and not diagnosis the real problem. I have my Master's in EE and nowhere did academia teach me how to troubleshoot something like a motherboard with an oscilloscope. I do think that this may be where the more specialised workforce comes in, with motherboard repair techs, etc.
What I find troubling about spending 20 minutes reading this thread is that I came out of it without any additional information as to how to begin troubleshooting a motherboard with an oscilloscope. Could someone perhaps link a few good self-help tutorials with vivid detail on how to troubleshoot a motherboard with an oscilloscope and DMM? My particular area of interest is 386/486. I do have a 386 motherboard I'm working on which shows no signs of life and I have tried all the usual diagnosis steps that I know of, e.g. check for CLK2 signal reaching the CPU, coming from the two crystal oscillators, confirm jumpers set properly, confirm jumpers making contact, confirm PSU working properly, check all SMD solder points for loose contacts, check bottom of board for visible cold solder joints, run the multi-meter across all resistors and caps checking for shorts, ensured the BIOS is working in an identical board, tried new CMOS battery, tested the SRAM in an SRAM tester to confirm it is good, and replaced all socketed chips with know good ones. The board has no electrolytic caps, and no MOSFETS. No beep codes are emitted. I have confirmed the board is setup correctly because I have an identical board which works to compare with. The board basically has a QFP 206 DMA controller, and what looks like a QFP north bridge (UMC 481), a QFP south bridge (UMC 482), and various other TTL 74F DIP and QFP ICs.
I'd like to start a new thread about how to troubleshoot this, but first I'd like to get a few recommendations from you experts about some good walk-me-through-it guides to early motherboard troubleshooting with an oscilloscope. I do have a cheap 100 MHz Ebay-special type of ATTEN oscilloscope which has worked well enough for what designs I needed it for in the past.
Seems like I would need to know a typical 386, early 486-era start-up routine to even know where to begin probing. Although UMC datasheets are often very hard to come by, the datasheets for the 206 and 481/482 have been procured! -
I have no repair experience with 386 boards. Only with older machines. I would focus on the CPU first. Has it ground, power, clock and reset signal? So it should start running then, trying to get data (code from ROM) via the bus.
-
Well, the CPU is soldered to the motherboard. it is an AMD 386DX in the QFP package. According to the datasheet, revision C1 or later of this chip can be floated by soldering the FLT# to GND. At present, the FLT# is connected to Vcc. I was considering cutting the trace which connects the CPU to Vcc, and connecting FLT# to GND. Once I do that, presumably I can insert any PGA-132 CPU into the adjacent PGA-132 socket (or I can just cut off the whole AMD 386DX).
However, I did put into the PGA-132 socket a PGA AMD 386DX CPU of the same "D" revision (while the QFP AMD 386DX is still soldered and not floated) and I still did not get any signs of life. While this test is not entirely conclusive in that the CPU is not the issue, it makes me want to take another direction FIRST before going back to the CPU.
I have 4 of these motherboards. They all came in a "dead" state. One one of the boards, one of the 206 QFP leads had separated from the pad. Soldering it back fixed the board. On the next board, the jumpers were too worn out and not making solid contact. Fixing the jumper corrected the board. Unfortunately, the solution for these remaining two boards isn't so straight forward and was hoping that someone with experience troubleshooting 386/486/socket 7 motherboards could guide me through the process.
I tried to attach a high res photo of board, but unfortunately, the 2 MB size limit of the forum won't allow for detailed viewing of the components.
-
Whatever Scope you decide on, it should have X/Y mode, so that you can get or build a component tester attachment (sometimes called 'Octopus'), preferably one where you have multiple voltage and current ranges and frequencies. It is the biggest repair aid for boards, especially if you know the circuit only partially.
-
Take a look at the Good Will Instek website. They have an India only sales promotion. The UI may not be very good, but the hardware is quite nice. Not familiar with these particular models.
http://www.gwinstek.com/en-global/Page/DGS_D_Promotion -
Also, when looking at computer (and anything else microprocessor or -controller based) repair and diagnosis, it might be a good idea to look for a so called MSO (mixed signal oscilloscope), which has some logic channels as well.
They can be used (on those instruments I know) not only for display, but also for complex triggering. -
I don't think there is a special procedure to diagnose a socket7 board.
A 386 is a CPU like any other CPU. It needs power,gnd,clock. After a reset signal it comes into a defined state and starts running. It tries to find some code to process by putting addresses on the bus and reading code from there. You can see those signals with a scope or logic analyzer.
-
I, personally, am not looking for a scope and I have started another thread more pertinent to my current goals.
https://www.eevblog.com/forum/repair/how-to-troubleshoot-your-defective-386486-motherboard-with-an-oscilloscope/ -
Something to make for your efforts is a "free run harness" for the different CPUs. This a device that can be inserted between the CPU and the MB and forces a NOP instruction on the databus. This causes the CPU to cycle through all the addresses and will facilitate locating address line problems and bad ROM devices.
Long ago I made one for a 6502 using a socket and was able to locate a blown PROM using a 5 MHz recurrent sweep scope to probe the address lines. Unfortunately, the replacement part cost so much I wound up buying a new computer instead. -
Apologies everyone but i might be going slightly off topic but if you are bored and have 5 minutes...
As a long served career repair tech specializing in Broadcasting equipment, i find it virtually amazing folks set about to repair mother boards from a PC. Agreed there have been times when i become involved in application-specific equipment which had been designed around a small generic 486 MoBo of the sort that would have been intended for a laptop. Here it was serving as a Master Camera Setup Controller Unit (the actual mother board and LCD served as part of the piece of equipment) which was at the time a very expensive piece of gear.
This is when the rules change i guess and it is WORTH repairing what is a mass-produced item that could so easily be completely replaced for benefit of reliability. That's where it's worth working on an aging MacBook too i guess, simply because they are currently of a value that says it is worthwhile to at least have a look...
I recall a 386CPU laptop with a terrible blue colored LCD display (no manufacture's name) that had been an extremely expensive purchase. To service it needed an equally expensive repair kit consisting of a batch of external programs (that ran on another PC and connected via a com serial port using some kind of inhouse serial protocol) and a special ROM chipset swapout and a bunch of floppy disks so i could begin to find the trouble! Using an oscilloscope would have been pointless.
In a similar case i had an audio-level/phasing indication metering unit from a studio sound control desk on my bench that was 486 CPU based that ran some kind of high level system ontop of fundamental DOS! The unit had been dropped and hit the floor hard. It now randomly crashed id program or would power itself down. Being extremely expensive to replace to the point it was unlikely the company would invest in such an item under current financial conditions i tried everything to make the unit become reliable. It was one thing to actually work through a bus looking at various blocks of data being passed using an oscilloscope but it wasn't until i plumbed in a my 16bit capable Logic Analyzer that i could capture blocks of activity that represented what the unit was calculating or processing i.e. actually doing. There was absolutely no possibility of a service manual so i was working blind... After the 5th day of doing so much it came down to the obvious fact that somewhere there was a crack or cracks internally of some tracks on the nine multi-layer board because of the fall so that meant it was pretty much the end.... I could have spent hours back engineering all the ASIC data sheets, the bus layouts for the 486 and it's ancillary chipsets find the breaks and try jumper even though they were probably internal vias but it simply wasn't worth it. I could not ensure reliably.
Another job was a PC built back when the 486CPU was the latest beast possible. For some reason the computer would randomly crash or simply not do something then lockup needing a reboot. It was used like that for years probably because it wasn't in a high use place did it ever get looked at.
Then the day came and it was rebuilt with a new and faster motherboard, larger RAM and something better than Win95... Out of interest i picked the old motherboard of the workshop trash a popped the old 486 from its socket after bashing off the heatsink.... I turned the chip over to see its pins and HELLO!... one pin in the central area folded over flat at 90 degrees.... It HAD to have been installed like that all those years ago and why it randomly caused the computer to do what it did...
So all i am saying is that chasing a fault on a PC motherboard could really be in actual fact a waste of time (unless you are doing it as a hobby) because today the things are so heavily loaded with such highly stressed components the biggest killer of all being heat and that starts the minute the dust gets in there and airflow cooling is compromised. Where i work a PC might be classed reliable for 2 yrs if that unless its a server crate or blade that's designed differently for even harsher work. The equipment has to be reliable so the minute there's a failure and brandnew unit usually ends up in its place with the benefit of fresh latest software... And if anyone showed me a 486 or 386 PC today i would probably laugh because the world has moved on i am sure unless it's in some very high value application equipment.
-
@peteb2
Did you read the start of this thread? The OP is in India and is a retired automotive engineer. In India he stated that it is commonplace to make repairs to component level. That's pretty much a lost art in the developed world, though TheSignalPath is trying hard to correct that.
But beyond that, if you have no money and no prospect of getting any other education, learning to repair things is a good start even if the repairs themselves are not very profitable. When I was in graduate school, I repaired audio gear scavenged from the trash using a VOM and a soldering iron. I didn't make a lot selling the stuff, but I *did* make money.
The fact that some things can't be repaired doesn't make determining the cause of the failure worthless. Many times the failures are trivial to fix. And troubleshooting is the same no matter what the technology is. It's a *very* valuable skill which can only be acquired by practice. And the more you know, the fewer tools you need.
Arguing that it is worthless to try reminds me of people who can't do math. Why? Because they won't try. The people who are good at anything are the ones who don't give up. It's taken me 40 years to learn to play guitar. But I finally reached the point I can put a CD on and play along. Some of my favorite stuff is from Clapton's "Unplugged" album. I don't know what I'm going to do. I just listen and my fingers do the talking. I only screw up when I think about it. -
I'm in the component level repair business, no computers or consumer stuff, at first only measurement and calibration gear but now also more and more industrial stuff (often pcb's that are part of something and do not work without the rest. The only way there is to test components.
But most gear has a lot of digital stuff inside. 90% of what I do has no documentation of schematics and also gear that is declared non-repairable by the official dealer/manufacturer. So a bit like what is common in India but rvery are in my country.
My most used gear are in order of most use:
-Very good soldering/desoldering gear
- several DMM's,
- a good real LCR meter,
- several scopes (350 MHz MSO, isolated 60 MHz scopemeter, 100MHz analog scope) I bought the scopemeter after some testing and for some speacial use but it now use it even more as my other scopes ( a Siglent see my review: http://www.pa4tim.nl/?p=5896 )
- special scope probes, I could not do without current probes so I have them in many flavours from a Tek 6042 upto a TTi iprober.
- component testers : curve tracer (tek 576), IC testers, a number of selfmade component testers.
- function generator
- and a lot more that I need for very specialized things but not often.
A thing I almost never use and can live without is an ESR meter (however I do have several of them) But everyone has its own way of doing things. I could not do without a scope and my Tek curve tracer.
I also play guitar (since 1982 or so) and agree with you on that -
Very odd that you mentioned this because for some reason I have had Eric Clapton's Unplugged "Layla" song going in my head at random times throughout the week, although these days, I don't actively listen to music.
Back on track - rhb and peteb2 - I really appreciate your comments. I was kind of hoping to cease using this particular thread because I started a thread more specific to my needs, https://www.eevblog.com/forum/repair/how-to-troubleshoot-your-defective-386486-motherboard-with-an-oscilloscope/
but it looks like I'll be reading on two fronts. I was actually hoping that the OP would have updated us on his/her progress, but I guess he/she has moved on.
In response to some comments, I am doing this strictly as a hobby. I used to to work as an engineer by profession and now have taken up collection of specific periods of computing history. I also setup systems around such hardware as a hobby. There is actually a fairly substantial following in this field of interest as of recent. e.g. you can check out the vintage computer forum, http://www.vcfed.org/forum/forum.php , or Vogons at http://www.vogons.org/ . The interest is mostly centred around vintage gaming on period-correct hardware, although there are others, like me, who don't game but are more into the hobby aspect of restoration and tinkering. 15-20 years ago, I could buy very high-end 1990's-era motherboards, CPUs, graphics cards, SCSI adapters, etc. on eBay for next to nothing. These same items today, if you can even find them, now sell on eBay in the hundreds of dollars. So aside from the hobby aspect of it, one may be able to reap substantial capital gains when selling. By way of example, back in 2003, I could buy a rare Cyrix 5x86-133/4x for $25. There was an eBay posting this year for this CPU, which only happens about every 5-10 years, and the bids were running north of $500 (although the reserve was never met). Just 6 years ago, you could buy a Voodoo 5 graphics card on eBay for $30. People are now paying $200+.
Another aspect of the hobby for me is that it brings me back to an age which I consider a golden era in computer history. I am personally fond of the 386-486 era, but my interest does extend into Tualatins, though the passion for me diminishes substantially.
The two dead boards I'm trying to get working are not for me. They are for a younger chap who seems to be developing a vintage computer fever. They are actually his boards. I have always enjoyed trying to repair old motherboards or computer hardware. Sometimes I give up, and stick the board in a box for a decade or more; then at a later point, when I have slowly accumulated more knowledge, I am often able to get them running again, though not always. This is a hobby and I enjoy learning and solving problems in the EE arena. My main limitation now is time and sleep. With 3 restless youngsters at home and a wife, my only time to tinker is the midnight hour and for a short duration. Tonight, looks like I'll only have time to read posts.
peteb2 - very interesting stories. Thank you for sharing. Sounds like that 486 pin was just barely making contact in its 90 degree bent position. Perhaps some mild corrosion or oxidation over time made the contact intermittent.
peteb2 - it is very likely that the boards of interest have been dropped. They were sitting around some community college waiting to be scrapped. Who knows how many static bodies rubbed all over them. The fact that the other two boards (the ones I got working) were contact issues makes me lean in the direction that the 2 dead boards may have similar issues, and as per your story, could be inner-layer contact issues.