The address lines could be ok, since the starting address on a 32-bit machine is 0xFFFFFFF0h (on a 16.bit 8086 it is 0xFFFF0h). This is where the CPU is searching for the BIOS initialization, (usually a jump). So it would be fine to see all bits high except the first 4. I'm not an expert as well, but I would check out if some address bus activity after the north/south bridges happens. On older systems I usually measure the address lines in the ISA slot, but this one is much more modern and the ISA bus is abstracted over multiple layers. If you can check out the address lines after the north bridge, that may be would give you some hints. Could happen, that the north bridge is already dead, what is a shame with BGA. It is a pain to solder, so far I was not once successful with that :) Good luck!
ahhhh so those address lines being high might actually be the "initial state" I was looking for. Very interesting thanks. I can check if there is something on the PCI bus as the southbridge seems to get data from it? Thanks for that, it's an important piece in the puzzle!
necroware and tony my favourite retro computer youtubers together brainstorming! this PCB doesn't have a chance ! It will start working just out of fear :)
Replacing a BGA seems to be very difficult and a lot of work. There is this crazy solder paste that forms those balls by itself with hot air. Would love to be able to do it one day.
As @necro_ware told the adresslines should be in a special state after reset goes high. Then the CPU sends Out a "read" signal, then the cpu reads the Data what is written to the datapins. The sequence after reset is: - Put start adress on the adress lines and hold them - enable the "Data read" Line - recieve Operation Code (instruction) at the data Pins - do the instruction - go ahead with the next thing Do: Check the read pin at the CPU. It should go active after reset Write down every signal on data0 to data31 after reset (the first instruction the CPU recievs) and check the internet for the instruction. If the first instruction is something valid (a jump, a clear-register x or similar) the Problem is the *second* instruction. If the first instruction is an invalid opcode the cpu may freeze (since the 286 invalid opcodes forces a freeze).
Almost all high state on address lines are probably normal at start. First address used by processor is 0xFFFFFFF0, and it may stuck somewhere further along the buses. First you can check if A0-A3 lines are low at the start. If not the processor doesn't try to start at all. If are they did. The you can check if this addresses appear on PCI bus - this can mean North bride is OK. An then on XD bus.
Tony my humble contribution... I'm pretty sure you already tried this but just in case you haven't... 1. Many boards I have dont like the post analyzer... specially on the ISA slot... some show the same behavior as yours Try starting the board without the post analyzer and just plug the pc speaker to see if you hear an error code.. or plug the post analyzer on the PCI slots instead 2. Many times I need to press the rest button several times to make the cpu start 3. With the post analyzer pluged tap on the power components with your fingers .. capacitors, inductors, transistors to see if you see some fluctuations on the voltage leds on the post analyser
Interesting, I never had that issue but worth checking for sure. It would be silly if the board was just working and just not starting because of... me :D Thanks for that!
This is from back in the days where you had to check EVERYTHING for it to work. The jumpers set the cpu voltage,whether it is single or dual voltage , the clock, the multiplier, cache config, BIOS voltage. And it all had to be right for it to fire up in most cases. The retroweb has a manual on this board with the jumper settings for the V58. /motherboard/manual/v58-60ec698ab4d08855318212.pdf Now from what I can can see, the Pentium 200 is a P54C chip, with a 3.3V Vcore. JP8 sets whether the CPU is single or dual voltage, this should be set to single for the P54C which is 3-5 and 4-6 closed. Dual voltage is 1-3 2-4 So check this jumper block. JP11 is the Vcore, it should be on 1-2 by default which is 3.3V. 3-4 is 2.8V, 5-6 is 2.9V, 7-8 is 3.2V, 9-10 is 3.5V. Make sure you are on 1-2. THe pentium 200 is Min/Recommended/Max V core 3.135V / 3.3V / 3.6V Bus speed I think is set correct at 66Mhz. this is set with jumper JP2 and JP3 both should be set to 3-5 4-6 for 66Mhz. The there's the switches, which set the multiplier - SW1 S4 S5 and S6. your multiplier should be 3 which is S4 off, S5 on, S6 off. double check these. Might even be worth checking the switch block with a multimeter to confirm the switches are set, and not in need of a bit of switch cleaner. if that all checks out. You might be on to confirming the clocks to the north and south bridge, and supply voltages. These will need to be working too. You seen in the block diagram how these all need to be working in the chain for the system to read the bios and boot. There will be a clock generator chip kicking about supplying clocks to all the board chips, including the CPU, so the fact it has a clock is a good sign, but it's worth confirming the clocks and supplies on the the rest.
Thank you for taking the time to comment! I have that manual, I know it by heart now :) I have 3.3V at the CPU (both sides) and 66MHz so those should be ok. Are you suggesting that the chipset has its own clock and power? I didn't consider that...
@@tony359 Yes, if you check the M1541 datasheet, you will see it's very involved with clocks, as it is the interface between the CPU, cache, memory, PCI bus etc, and has inputs from those clocks, and output refresh clocks for the RAM. IF it's in the address/data path, chances are there's a clock somewhere, to keep it all syncronised. ;-) The south bridge M1543 has inputs for the RTC 32k clock, and the PCI bus clock. and maybe a 14Mhz clock input, probably for running the show. There's a whole block in the the block diagram for clock and reset. So yes, both North and South bridge, have more than one clock, some in, some out to other parts of the system, and are worth checking. In fact I was thrown off a bit, I just checked the chip numbers - you have the M1531/M1543 which is the ALADDiN IV+. retroweb /chipsets/179 If you look under documentation, then 2 chip docs available, you'll get tthe 2 datasheets, complete with block diagram, and pinout data. The idea still stands though, check the clocks and voltages to the chips, as much as you can, which won't be easy since they are BGA. But you may be able to trace some.
Faced a similar issues a while back. It ended up being the Super I/O that was dead. On the platform I was troubleshooting, I was missing basic clocks generated by that IC, and that's how I figured it out. So, finding the your Super I/O datasheet and making certain everything is being generated could be a good next step (Would also prevent the CPU from initializing, which seems to be your case). Good Luck!
Hi Tony, you are doing well mate, don't give up. I'm no expert on these but I have been working with electronics for over 50 years and you just have to have confidence in yourself and what you have done so far. The one thin about you I am impressed with is your logical step by step diagnosis and your general approach to the different problems. On this you have done everything right thus far, so don't give up, the only criticism I have if any is don't jump about, all you will do is confuse yourself. Decide a course of action and follow it logically from one end to the other, so have you got activity on the Bios chips data and address lines, if so follow it through that strange little but to the next chip, is there data on the data and address lines coming out the south bridge? follow it. Is there data going in the North bridge and so on, don't jump from one end to the other, it will probably turn out to be some small logic chip failed or a broken track somewhere, Even getting basic and actually checking continuity of address and data lines with your meter in continuity mode (Beeping) and check the actual tracks from chip to chip is all I can offer.
Criticisms are always very welcome of course! I see your point and yes, I tend to jump around! But I just thought that if nothing's coming out of the CPU, well nothing will reach the BIOS, or the Southbridge or the Northbridge? I've inspected for broken traces a few times - so far I haven't seen anything. And thanks for your kind words :)
@@tony359maybe it’s not the cpu being on hold, but the cpu is waiting for data or activity from the chipset. So the data from the bios might be held back by somewhere on its way. The bios chip should also output the data that is saved at the program start address on its data lines. You could compare the output on the data lines with the contents of the bios file. I am wondering if the correct data from the bios chip is coming out and if it is stuck at the first byte or if it outputs multiple bits and bytes. I believe the bios chip outputs 8 bits in parallel on the 8 data pins, byte after byte for every change of the 8 address bits on the address pins. Someone on yt built an open source bios programmer and recently, he showed the status of the address bits with leds on a bread board with many wires ;)
There are already some good comments that point to the same what I already thought about, which is the reset vector. As this points to the beginning of the last 16 bytes of addressable space, it is normal that mostly all address pins go high. The reset vector is the first code the CPU looks for, and as there are only 15 bytes available (the 16th is the checksum), it is a Jump command that enters the main BIOS POST routine. There must be some logic to translate the CPU address to the limited amount of address pins on the BIOS, which I'm afraid is likely the northbridge.
Uhm interesting. So as Necroware mentioned, you're suggesting that the "pulses" I was looking for are in fact the "address lines being high". That is the first activity from the CPU which gets translated by the Northbridge into something that can work for the BIOS? So I guess I should focus my attention to the PCI bus? Maybe I see something there which is not reaching the BIOS? Thanks for your help!
@@tony359 Exactly. So I think the behavior of the CPU is pretty much expected. The address request should pass through to address lines all around the busses (ISA, PCI and also the BIOS chip which also should have most or all address pins high.On a (whatever)C1024, I would e.g. expect A15 down to A5 to be high and the lower pins to start low, plus /CE being low (Chip Enable Active Low) and the BIOS ROM to output some data. If there are not address strobes on the address lines, no Chip Enable Low or no data flowing out of the ROM chip, this is already strange. If you have the proper address, chip enable and data coming out but obviously not getting back to the CPU, then you have again another problem. I assume a 16 bit ROM and a 32 Bit CPU meaning that the data will be read in two steps and assembled in the chipset, so again likely a task for the northbridge. However, follow the traces and look what happens where.
I need to double check the BIOS, I think I could see the address lines stuck to high but Data lines were likely floating as they read 1.5V. I guess I should check /CE and make sure it's low?
@@tony359 Addition: I found another datasheet with another label description. Check Pin 2 (/E or /CE) and Pin 20 (/G or /OE - Output Enable!) if it is a 27C1024. For other types, refer to their data sheet, but I think, /CE and /OE have both relevance, and /OE even more, which I first overlooked because it wa labelled as /G (which I didn't understand) in the first datasheet I looked at
Amazing, in-depth video, as always. Thanks for bringing great content, Tony. As I read the comments, there's a lot of knowledgeable people, hoping they can give you the right hints. What I think may need to be checked, although I might be wrong, apart from re-balling the North and South bridges, is the CPU socket itself - inspect it for corrosion or dirt.
I suspect the problem is with one of the chipset BGAs. There might be a bad connection due to physical stress or oxidization. Try to push the BGA towards PCB moderately and hold during reset. Sometimes, it helps enough for the board to POST.
A week ago I had the same issue with S7 MB, zero activity. The problem was in the NB, maybe more than 10 legs were unsoldered, after full reflow the board works just fine.
I'd suspect the north bridge also. Check if it is getting too hot to the touch, or try pressing it, in case it's a bad solder ball underneath the chip.
I have pressed everything on the board for quite some time as that's exactly how I diagnosed bad BGA on the Amptron! But no luck. Nothing is getting hot either, I did check with the thermal camera... Thanks for your help!
Unless someone has a better idea, and if one has too much time on their hands and is determined to make it work, one could map out the back of the PCB socket pinout on a paper, and then go through all non-address/data non-voltage input pins (since you have presumably checked voltages and you have checked that nothing interesting happens to address and data) and map out their startup behaviour and see if something looks suspicious after cross checking with the p5 datasheet. You can just use colour markers and designate colour for each of the common behaviours, or designate a letter per behaviour. It's "only" a couple dozen pins.
If you check on my little piece of paper at some point in the video, at the bottom is the behaviour of the Amptron board! :) It's maybe a bit too much, it's easy to lose track of things. Regardless, the CPU on this board doesn't do anything on any of its pins (that I tested).
BGA chipsets. I have a couple of boards that are doing similar things. Re-ball those & it should start working again. I need to save up so I can get the equipment to do the same for my boards.
I have several MB with the same behaviour and I failed to continue the repair because of the BGA chipset and the lack of the knowledge how the startup sequence is happening in these systems. I am very curious how this repair will go and I wish you patience and luck to drill down to the root of the problem. Suscribed!
@@tony359 What I found a bit strange: I was not able to find any document how the 486, 586 processor should work on "pin" level. What is the power up sequence, what levels should be present at a given pins with timing regarding each other in the power up sequence. There are timing diagrams for the buses, but no detailed explanation on these in the datasheet. The pin functional descriptions most of the time very condensed, left a lot of guessing about functionality and interoperability of other pins. Standard documentation is a set of puzzle pieces and you have to put the puzzle together to see and understand the full picture necessary to be able to diagnose an MB.
@@tony359 Thank you the tip for the test eprom to write the post code. It is an extremelly usefull debugg tool to check the CPU, SB.NB part of the MB. Always learning something.
Very interesting information in this video! Broken chipsets are a problem it seems. I once killed a mainboard by placing it on some metal. I guess I shorted some pins of the chipset and it’s dead. Not much one can do I guess.
Fingers crossed you find it, I have a socket 7 which has been a pain. Original AMD k6-2 shows a code but no beep or boot, a AMDk6-2 500 AFX does nothing. Hopefully there is something I can learn from this as well.
Hi Tony, thanks for all your great videos. Have you tried this with another type of cpu, like a Cyrix M2, or Amd K6 ? or even perhaps one of the slower and older intel cpu's?
some (but not all) pure pci based main chipset boards (the isa is handled by the southbridge on this one, if i recall) don't output post codes to the isa bus, so if you haven't stuffed the post card in a pci slot, thats worth a shot. i'd be tracing down the clock signals at all points to ensure its getting what it needs where it needs, confirming jumpers for all elements since some of these older ones have them for almost everything it seems. beyond that, if you have a plain jane p100/133 or whatever in the lower end laying around, set the board to run that and try it that way. could be the 3.3v side of the equation.
I cannot remember if I tried the post analyser on the PCI slots, I shall do that! I do have a Pentium 90 - what would be the difference with the P200 I have? Thanks for your help!
@@tony359 well, in theory the p54cs (the 200) should work in any board a p54c (the 90) would also work in, even if at reduced clock speeds, but i've found some boards of mine don't seem to like anything over 133 or 150 for some strange reason i haven't quite tracked down yet, despite saying they have the capability. if you haven't already stumbled across it, the rcollins site has a lovely set of early pentium datasheets if you search for 'pentium p54c datasheet', then click on the rcollins link. some of which i couldn't find on theretroweb for whatever reason. i'm not aware of any physical socket differences between the two generations, and its mostly just microcode. if you're feeling adventurous i suppose you could try the bios patcher necroware uses at times to update the microcode and whatever, see if that yields results. and then there's the reality some some specific cpus just don't play well with some boards at all. or, like you theorized, that cpu may not actually work. i can't recall if they ever made any mmx variants in that black and silver package type that were mmx, but didn't have the laser etching on the heat spreader. i'd recommend double checking the cpu code bottom against the cpu world db to confirm its a p54cs variant, and not a p55c (mmx). if the 90 works in that board, either there's something that doesn't like the 200 in that board, or the 200 is doa. (they're pretty hardy, so i 'm skeptical of it being dead without a confirmation) there's also an embedded variant of the 200, but i can't imagine there's anything different between the two beyond microcode as well. i suppose beyond those, its a matter of checking the IC pins for loose ones where possible. if the bga chips have solder joint or pad issues, thats going to be hard to look for without removing them, sadly. i'm also vaguely wondering if all those rail caps around the socket are still good. gotta love those pesky retro motherboards.
summary of post youtube seems to have deleted entirely; microcode in the bios. but sometimes some boards hated on mid to later chips. you might also try the bios modding tool necroware has shown for older systems. fair warning on that though, it does corrupt some few bioses. so if it does weird stuff, don't be surprised. but if it does, at least it would be booting. might also test the caps around the cpu socket. and look at all non-bga chips for dodgy solder joints. can't for the life of me recall all the info i had dumped in that vanished post.
Thanks Tony for all your efforts on my motherboard. Just a point. The cpu fan header is not powering the fan and when I first got the board the jumper in that white jumper socket ( I think it is used for an external voltage regulator) was in the wrong position, I’m wondering if that has done something to the motherboard
The socket with the 5 pins? That's for the "hardware monitor" - it's so unclear on the manual. How do you know the jumper was in the wrong position? The manual really doesn't say anything about it. If I remove it, the "error" LED stays on all the time. And the Fan socket doesn't have 12V on it? It might just be that there is a fuse and it's blown - I suspect the fan is directly wired to the PSU 12V line - but worth checking, thank you! And no worries, it's fun to get stuck! (well... :D )
I think I tried all the options and that seems to be the "correct" one, though it'd be nice if the MANUAL said something about that! :D I'll check that fan terminal, thanks.
I'm wondering if the CLR CMOS circuit is stuck on, the board would not start if that's the case. Or you simply have the CLR CMOS jumper set on.. This is just a hunch. :)
I would also say check all states of all bios chip pins and check exactly at what state it is stuck. What address, data, and so on do we see on each pin? Check the data sheet of the bios chip for the pins for that.
@@tony359 try moving it to the other position just in case to rule out mislabeling. Not likely but known to happen... I would start by checking continuity on the traces that's going through the jumper, see if continuity changes when you move the jumper. If it doesn't, do a visual inspection for shorts or faulty components. I haven't had coffee yet so not thinking clearly lol, but the chipset is north + south bridge right? Check for activity on the north bridge. Another good practice, wiggle components on the board and look at the back of the board to see if anything moves eveen a tiny bit. If so, reflow the solder.
You could check all the traces of the cpu, pci and xD bus for continuity. Maybe a short has burnt one of the traces? Also it might be worth checking the chipset pins for activity, they should do something as well. Also the pci bus. I guess the chipset is broken. But if there is activity on the pci and xD bus you could narrow it down maybe.
Amptron PM-9600 has the same chipset that Acer V58 does. You can compare signal lines of chipset: their resistance to the ground (on a bare board) and their behavior (with a CPU running)
oh, interesting! I can give that a go though I'm not sure all the pads are accessible but I might have seen something at the back of the board. Thanks!
@@tony359 , I would try to run both models without a BIOS chip and watch CPU and BIOS address/data signals on an oscilloscope. Once I repaired a much more modern board with the same symptoms (No post code , inactive reset). The problem was found in a zero ohm resistor in SPI bus circuit (the resistor was between chipset and SPI BIOS chip). By the time I found out it I had already replaced chipset. I hope you will manage to resurect the mainboard and I wish you good luck. But from the other side not every mainboard can be repaired for resonable amount of time and effort
Hi Tony, Sometimes is not everything repairable. I had a moterboard to. What i do, nothing helps. The board is now hanging on the walll in my room. Greetings from Steven from the Netherlands
Looking at the Pentium board, go back to the ISA and PCI slots, between the back of the first two PCI slots from the ISA slots side, there appears to be a black electrolytic cap with holes in the top. Are these Sharpie marks or a blown cap? If that description is unclear, go to your diag pcb in a ISA slot and move down to the PCI slots and the cap I am referencing is near the side of the pcb between the first 2 PCI slots.
It does sound like there's an issue happening between the BIOS and the CPU as the two are OK individually. You may have to check everything in between. Although, it would be quite funny if it was just a jumper this whole time :). Best of luck to you
Check the ADS# pin as well. The CPU should drive this low once it is in a bus cycle, so if it never starts one to even try and read from the BIOS ROM then it should remain negated (high). The CPU also isn't necessarily going to execute byte by byte like older CPUs, there's a lot of parallelism going on (64 bits at a time). The BRDY# pin will be asserted by the chipset towards the CPU once it has presented some data in response to a read, or taken some data in response to a write, so that might be a good one to look at too. Also rather than trying random address bus pins I would be a bit more targeted. The Pentium doesn't have address 0-2 pins, instead it has 8 "byte enable" pins. So I'd also take a look at these to see if there is any activity. For each byte out of an 8 byte chunk that the CPU wants to read/write it will assert the corresponding BEx# pin (maybe multiple at a time too). And of course, know your enemy! Look at the behaviour of the pins on a known working board so that you know what they should be doing out of reset.😊
I think also checking NA# might be worthwhile. This seems to be a signal from the chipset to the CPU that it can accept a new bus cycle. If the chipset doesn't assert this signal the CPU may never start trying to access the BIOS ROM.
Thank you! I am learning more and more about the Pentium and I understand the first thing it would do is call for FFFF which is... high. So maybe what I observed was expected and as you said I should look at more specific pins. Thank you!
Check the south bridge because the south bridge loads the BIOS into the RAM. Processor without a code can not doing anything. That is the case according to my knowledge.
something I have forgotten to check, absolutely! I can check the socket with the microscope, I should be able to see if the catches are there. And then maybe check continuity with the multimeter! Thanks for that!
Hey Tony359 maybe you could try to gentlty push, hold down the SMD chips on the board and then power on? I had two old boards this procedure helped. There were bad solder joints due to e.g. mechanical stress. I could not see that from visual inspection. However resoldering the identified chips then solved the problems permanently.
Yes, I found the issue on the Amptron this way and then I re-flowed the ICs for a permanent fix. On the V58 I pushed and pushed but I never saw a glimpse of activity unfortunately. But good idea, thanks!
Could you try pushing down with a finger firmly on the Northbridge and/or Southbridge and while at it power on the board and see if there is activity on the bus? Sometimes the balls underneath are cracked and not making good contact...
Thank you! I have already, for quite some time :) It didn't make it on the video. Though, I didn't check if there was activity, I was only checking on the Post Card. I can try that, thanks. The Amptron was fixed that way :)
There's two interrupt pins you could check, INTR, NMI i would assume if they were shorted to active that would prevent execution also looking at the datasheet it appears theres an INIT pin does that change at all upon boot?
I have checked with thermal camera, there are no obvious shorts on the board. Comparing the lines against the Amptron - same or similar chipset - is a great idea! Thanks!
Well, it's an old board, all the settings are manual and they directly affect the voltage/clock at the CPU. It doesn't matter if it's an AMD/Cyrix/Intel, as long as it works with 3.3V and 66MHz, it should work. One thing to consider is the multiplier: that I guess is sent to the CPU for internal multiplication. If that comes wrong, then the CPU might try to run at the incorrect frequency internally so maybe worth to check! Thank you!
Have you checked the signals on the CPU from the top? Could simply be a broken connector in the socket. CPU is behaving like there is no clock at all. But the CPU looks like the pins are available for contacting from the top.
I checked with the thermal camera and I didn't see anything suspicious. That said, I didn't realise the Amptron is the same chipset so I can try and compare a bit. Thanks for that!
Do you see any address lines activity at the BIOS chip? If not, are there any 74-series buffer chips near it? The reason I ask is because I believe the BIOS is located at the correct address on Socket 7 boards, and is needed to initialize chipset registers. (Granted, this is based on observations from a Cyrix MediaGX damaged VIA MVP3 board, where the CPU does start to execute BIOS code but fails after the first POST code, because the chipset was fried by the incompatible CPU…)
No, no activity and I mentioned the LS logic chip at some point in the video - that's also good unfortunately. Interesting about the BIOS being needed to initialise the Chipset, thanks for that!
i was wondering if this board is one of those that will not power up if the CR2020 battery is flat, I've watched The Electronics Repair on UA-cam and he found a few boards like that.
You mentioned moving jumper JP1 at the beginning of the video but not being sure what it is for, this is used so the system can be configured for either an AT or ATX power supply, ATX supports wake from standby, AT does not, there is a special connector marked JP6 near that jumper where the board expects the 5v standby power (since it lacks a ATX connector) I guess this used some special Acer power supply, never seen this before, anyway there is also a second jumper, if you look at The Retro Web there is a manual with only jumpers, this shows when you move JP1 you also need to move JP15, did you do that? Other than that some boards actually measure the power consumption of the fan and will not start if a load from the fan is not sensed... Finally there is a green LED on the board, is it called HW Monitor? I can't see in the video but this might be related to why it will not start, refer to the same jumper diagram on the retro web: JP7, JP9 & JP10
Yes, I did move JP15 when I discovered the option. I didn't think of adding a fan on the connector - all I know is the connector seems to be dead, as it's not getting power. But that might mean it's going through some logic which, as you suggests, checks for a load before providing voltage AND turning on the board. Yes, the green LED is off (well, it looks barely lit) and it goes full brightness if I remove the blue jumper from the "HW monitor" socket, for which there is NO info whatsoever on the manual. JP7, 9 and 10, I tried several combinations but I don't understand what they do (JP10 says factory set do not alter). I'll try with a fan, thanks!
@@tony359 I can't provide a direct link but the jumper only manual is on the Retro Web that I mentioned, those jumpers are listed in that jumper manual...
If the chipset is broken, last option would be to replace it with a known working chipset. It might not be worth it for the hardware, but it would be nice to know if that is the case here.
there may be something in between the cpu and the northbridge and the cpu or the northbridge is bad or even the southbridge .... you have to check every component between the cpu and the northbridge en between the northbridge and the southbridge ...
Hi Tony! A you checked the clock crystal near the SB? Also, near the bios chip preset the crystal, i dont know for what it. Try to check for bad or broken any elements, try to cheking the CPU slot pins, check pins on any other chips, try to press on the NB, SB chips and then start the MB.
Yes, I had forgotten about the CPU pins. I intended to check them and then I forgot. I can inspect with the microscope thanks. The clock near the RAM should be the reference one for the clock IC nearby - if that doesn't work, you shouldn't have 66MHz at the CPU. The one close to the BIOS: interesting. I wonder what that is for. I'll check it thanks! I've inspected the board - front and back - multiple times for a broken trace or broken "something"... :( Thanks for your help!
The little crystal... i would think that looks like a 32Khz RTC crystal. There should be an RTC logic and NVRAM chip nearby, upheld by the battery which is also nearby, since that and the BIOS share a bus, called XD bus off the south bridge. And yet it's a little surprising that there is no activity from the CPU at all, not attempted BIOS read, no address-data bus activity? Seems like there should be a more immediate reason that should be more directly measurable.
what about the /BOFF signal? The backoff input is used to abort all outstanding bus cycles that have not yet completed. In response to BOFF#, the Pentium processor will float all pins normally floated during bus hold in the next clock
Sir please reply... this question is very very confusing... I have 2 identical 12V DC 7.5Ah batteries say X and Y. One side of a normal bulb is connected to postive terminal of X battery, and another side of the bulb is connected to the negative terminal of the Y battery. The negative terminal of X battery and the positive terminal of Y battery are not connected to anything. So, in this case, I want to know whether there will be a current flow to the bulb, and whether the bulb will glow??? Give reasons if bulb will glow, and reasons why bulb will not glow. Thanks in advance.
I'm not sure why you're asking me to explain you why a light bulb would turn on or not (and consider both scenarios and also provide a explanations for both scenarios) on a motherboard repair video. Your question reminds me my Physics school tests but that was a long time ago. That said Physics was my favourite subject and I usually got high markings on those tests :)
I see the 2 videos and no see if you check the clock in the bios (bios send data have cpu or not if they have power and clock) so the clock is missing, the bios is broken or the chipset is killing it. Have a small cristal near the bios.
I haven't checked that - how does the clock work with the BIOS? The BIOS IC does not have a clock input, I thought it would just follow the ISA clock? Anyways, I do not see any activity at all at the BIOS chip. Thanks for now!
Sorry It took a while to look for the dataset, in this case it says that it is internal (I see most outside) you have the 5 v on the pin 32 vcc? This chip is a pain have to many variations to read depending on pins 32 /24 /22 are high or low (this pins need to change low or high to allow read if all stack high not work) to send data pin 24 and 22 low and pin 31 high . (Data says logic for this pins to be high or low but who make the logic the bios chip, the chipset or other chip)
You can read in the programmer with the oscilloscope see that you should have it on the pins (it seems that it reads and records it so you should see the necessary signals) but I'm afraid it's going to be the chipset, hopefully you just have to Resolder it
@@tony359 If you go to that, it's better to do the reball. Many cracks is not ball itself, but inter-metallic joint on the pad (copper-nickel-gold-tin). It separates in the layer and reflow not fix it. You can see it on the pad as grey layer that is hard to solder (but not impossible, if you clean it, it will work).
I'm sorry for suggesting the obvious, but what about CMOS battery? Some boards won't start with flat battery and I once almost discarded one of good boards for that reason.
How much memory has that ram stick got? According to the manual that motherboard supports a maximum of 192MB, one 128 and a 64MB or 128MB plus 2x 32MB. So one 256MB stick should not work. Aside from obvious missing things like voltages, clock, reset, dead chipset ics, it's usually cpu or ram issues that give a dead post card. I recently have 1/2 repaired a faulty Jetway socket 7 motherboard with dead cache, now I found out the pins in the cpu socket are badly oxidised which causes intermittent no post or hangups. Your board looks really clean so likely no corrosion problems. Only other thing is have a real close look for damaged traces.
Only 32MB I'm afraid. But even with incompatible or bad RAM, I'd expect the CPU to execute the first few instructions and then complain about bad RAM. If you remove the RAM from your system, you still get something at the POST card, then obviously it stops. Unless, i guess, the RAM is shorting or compromising the CPU bus? But now I'm thinking about that, the RAM is wired to the Northbridge. The RAM works anyways, you can see in the video when I test the CPU and RAM on a different board. But thanks so much for the suggestions!
@@tony359 Maybe You have already checked but I would check voltages to the cache and tag ic's as well as any other You haven't checked. The cache chips have 2.5v and 3.3v rails possibly. A bit of a long shot though.
EDIT My bad here, JP7 is ok, Vcore is correct on JP11. This is a bit of an oddball socket 7 motherboard, have never seen that HWM Vcore on any other S7 mobo.
Ah, that's the "hardware monitor", I don't know what that does. but I tried all possible variations... Thanks for the hints, all very valid. I haven't checked the cache voltage so something to try for sure. Thanks.!
I think this is relevant here,tony has bypassed all bios checks,ie switches,jumpers, ram,and so on. So,because he has programmed a prom to simply get the cpu to write a value out on a bus,he is checking to see if the MOST basic function is working, if so,he can move from there.The glue logic,(ie northbridge,and or southbridge)and the cpu are being tested to see if they can do the most basic thing ie read an opcode and operand from the prom and then write a byte or word to an address. His concept works because he tested it,on another mb,and the data was written to the test card.This is a sensible step. The only caveat,there maybe slight differences between the addressing or bus arrangements between the two motherboards,he has to check this.if they're the same,then he has a glue logic to prom bus issue,a glue logi to pci bus issue,ie no seven segment activity,or a cpu to glue logic issue,and he can proceed from there. Jumpers,bios,cmos ram,are irrelevant guys,please lets not confuse him,there are other people on the forum,who are pointing him on the right direction here. Lets us try to give as helpful suggestions as possible.
Really strange... The only jumper I see that does not fit me is JP5, 2-3 pins sais for BIOS flashing 5V in manual try 3-4 which sais normal in manual and has default asterisk and NC on the board
@@tony359 may be just before quit the board you may try to "cook" the north bridge ( heat the chip with no more than 160° i think for a few seconds) and try
It lacks hyper specialised knowledge like that and will run in circles. But you're welcome to try it yourself and see whether you get any useful output out of it :D It is after all a text prediction machine, a lossy highly compressed representation of text used for training, and there is very little in depth CPU troubleshooting content that the training process could ingest. A PC mainboard is after all a field replacement unit, where if something is wrong with that isn't BIOS or battery, jumpers or installed components, it would simply be junked and replaced by a working one. Component level repair except rework at manufacturer where they have secret manuals and specialised tools, is effectively a new phenomenon.
The address lines could be ok, since the starting address on a 32-bit machine is 0xFFFFFFF0h (on a 16.bit 8086 it is 0xFFFF0h). This is where the CPU is searching for the BIOS initialization, (usually a jump). So it would be fine to see all bits high except the first 4. I'm not an expert as well, but I would check out if some address bus activity after the north/south bridges happens. On older systems I usually measure the address lines in the ISA slot, but this one is much more modern and the ISA bus is abstracted over multiple layers. If you can check out the address lines after the north bridge, that may be would give you some hints. Could happen, that the north bridge is already dead, what is a shame with BGA. It is a pain to solder, so far I was not once successful with that :) Good luck!
ahhhh so those address lines being high might actually be the "initial state" I was looking for. Very interesting thanks. I can check if there is something on the PCI bus as the southbridge seems to get data from it? Thanks for that, it's an important piece in the puzzle!
necroware and tony my favourite retro computer youtubers together brainstorming! this PCB doesn't have a chance ! It will start working just out of fear :)
@@fft2020 LOL! Necroware is the Master, I am just the Padawan! :)
Replacing a BGA seems to be very difficult and a lot of work. There is this crazy solder paste that forms those balls by itself with hot air. Would love to be able to do it one day.
As @necro_ware told the adresslines should be in a special state after reset goes high. Then the CPU sends Out a "read" signal, then the cpu reads the Data what is written to the datapins.
The sequence after reset is:
- Put start adress on the adress lines and hold them
- enable the "Data read" Line
- recieve Operation Code (instruction) at the data Pins
- do the instruction
- go ahead with the next thing
Do:
Check the read pin at the CPU. It should go active after reset
Write down every signal on data0 to data31 after reset (the first instruction the CPU recievs) and check the internet for the instruction.
If the first instruction is something valid (a jump, a clear-register x or similar) the Problem is the *second* instruction.
If the first instruction is an invalid opcode the cpu may freeze (since the 286 invalid opcodes forces a freeze).
Almost all high state on address lines are probably normal at start. First address used by processor is 0xFFFFFFF0, and it may stuck somewhere further along the buses. First you can check if A0-A3 lines are low at the start. If not the processor doesn't try to start at all. If are they did. The you can check if this addresses appear on PCI bus - this can mean North bride is OK. An then on XD bus.
Amazing, I shall do that!
Tony my humble contribution...
I'm pretty sure you already tried this but just in case you haven't...
1. Many boards I have dont like the post analyzer... specially on the ISA slot... some show the same behavior as yours
Try starting the board without the post analyzer and just plug the pc speaker to see if you hear an error code.. or plug the post analyzer on the PCI slots instead
2. Many times I need to press the rest button several times to make the cpu start
3. With the post analyzer pluged tap on the power components with your fingers .. capacitors, inductors, transistors to see if you see some fluctuations on the voltage leds on the post analyser
Interesting, I never had that issue but worth checking for sure. It would be silly if the board was just working and just not starting because of... me :D Thanks for that!
@@tony359 Yeah, he's telling you right, man. Give it a go and see what we know... 😄
This is from back in the days where you had to check EVERYTHING for it to work. The jumpers set the cpu voltage,whether it is single or dual voltage , the clock, the multiplier, cache config, BIOS voltage. And it all had to be right for it to fire up in most cases. The retroweb has a manual on this board with the jumper settings for the V58. /motherboard/manual/v58-60ec698ab4d08855318212.pdf
Now from what I can can see, the Pentium 200 is a P54C chip, with a 3.3V Vcore. JP8 sets whether the CPU is single or dual voltage, this should be set to single for the P54C which is 3-5 and 4-6 closed. Dual voltage is 1-3 2-4 So check this jumper block.
JP11 is the Vcore, it should be on 1-2 by default which is 3.3V. 3-4 is 2.8V, 5-6 is 2.9V, 7-8 is 3.2V, 9-10 is 3.5V. Make sure you are on 1-2. THe pentium 200 is Min/Recommended/Max V core 3.135V / 3.3V / 3.6V
Bus speed I think is set correct at 66Mhz. this is set with jumper JP2 and JP3 both should be set to 3-5 4-6 for 66Mhz.
The there's the switches, which set the multiplier - SW1 S4 S5 and S6. your multiplier should be 3 which is S4 off, S5 on, S6 off. double check these. Might even be worth checking the switch block with a multimeter to confirm the switches are set, and not in need of a bit of switch cleaner.
if that all checks out. You might be on to confirming the clocks to the north and south bridge, and supply voltages. These will need to be working too. You seen in the block diagram how these all need to be working in the chain for the system to read the bios and boot. There will be a clock generator chip kicking about supplying clocks to all the board chips, including the CPU, so the fact it has a clock is a good sign, but it's worth confirming the clocks and supplies on the the rest.
Thank you for taking the time to comment! I have that manual, I know it by heart now :)
I have 3.3V at the CPU (both sides) and 66MHz so those should be ok.
Are you suggesting that the chipset has its own clock and power? I didn't consider that...
@@tony359 Yes, if you check the M1541 datasheet, you will see it's very involved with clocks, as it is the interface between the CPU, cache, memory, PCI bus etc, and has inputs from those clocks, and output refresh clocks for the RAM. IF it's in the address/data path, chances are there's a clock somewhere, to keep it all syncronised. ;-)
The south bridge M1543 has inputs for the RTC 32k clock, and the PCI bus clock. and maybe a 14Mhz clock input, probably for running the show. There's a whole block in the the block diagram for clock and reset. So yes, both North and South bridge, have more than one clock, some in, some out to other parts of the system, and are worth checking.
In fact I was thrown off a bit, I just checked the chip numbers - you have the M1531/M1543 which is the ALADDiN IV+. retroweb /chipsets/179
If you look under documentation, then 2 chip docs available, you'll get tthe 2 datasheets, complete with block diagram, and pinout data. The idea still stands though, check the clocks and voltages to the chips, as much as you can, which won't be easy since they are BGA. But you may be able to trace some.
I've taken a look and yes, I should have continued reading :) I'll check the clocks and also see what I can probe on the chipset. Thanks for now!
Faced a similar issues a while back. It ended up being the Super I/O that was dead. On the platform I was troubleshooting, I was missing basic clocks generated by that IC, and that's how I figured it out. So, finding the your Super I/O datasheet and making certain everything is being generated could be a good next step (Would also prevent the CPU from initializing, which seems to be your case). Good Luck!
I should indeed check whether the rest of the clocks work - PCI, ISA and the other reference clocks others have mentioned around the board. Thanks!
Hmm interesting. I hope you can come to a conclusion to the issue! Congrats on the 20k!
Thank you!
Hi Tony, you are doing well mate, don't give up. I'm no expert on these but I have been working with electronics for over 50 years and you just have to have confidence in yourself and what you have done so far. The one thin about you I am impressed with is your logical step by step diagnosis and your general approach to the different problems. On this you have done everything right thus far, so don't give up, the only criticism I have if any is don't jump about, all you will do is confuse yourself.
Decide a course of action and follow it logically from one end to the other, so have you got activity on the Bios chips data and address lines, if so follow it through that strange little but to the next chip, is there data on the data and address lines coming out the south bridge? follow it. Is there data going in the North bridge and so on, don't jump from one end to the other, it will probably turn out to be some small logic chip failed or a broken track somewhere, Even getting basic and actually checking continuity of address and data lines with your meter in continuity mode (Beeping) and check the actual tracks from chip to chip is all I can offer.
Criticisms are always very welcome of course! I see your point and yes, I tend to jump around! But I just thought that if nothing's coming out of the CPU, well nothing will reach the BIOS, or the Southbridge or the Northbridge? I've inspected for broken traces a few times - so far I haven't seen anything.
And thanks for your kind words :)
@@tony359maybe it’s not the cpu being on hold, but the cpu is waiting for data or activity from the chipset. So the data from the bios might be held back by somewhere on its way. The bios chip should also output the data that is saved at the program start address on its data lines. You could compare the output on the data lines with the contents of the bios file. I am wondering if the correct data from the bios chip is coming out and if it is stuck at the first byte or if it outputs multiple bits and bytes. I believe the bios chip outputs 8 bits in parallel on the 8 data pins, byte after byte for every change of the 8 address bits on the address pins. Someone on yt built an open source bios programmer and recently, he showed the status of the address bits with leds on a bread board with many wires ;)
Thank you. Yes, I guess those "stuck lines" might actually be intentionally and something just stops at some point.
Good effort Tony, I admire your methodical thinking and get the impression that it is not the last video in the 'series'. Good luck!
eheh, fingers crossed! And thanks!
There are already some good comments that point to the same what I already thought about, which is the reset vector. As this points to the beginning of the last 16 bytes of addressable space, it is normal that mostly all address pins go high. The reset vector is the first code the CPU looks for, and as there are only 15 bytes available (the 16th is the checksum), it is a Jump command that enters the main BIOS POST routine. There must be some logic to translate the CPU address to the limited amount of address pins on the BIOS, which I'm afraid is likely the northbridge.
Uhm interesting. So as Necroware mentioned, you're suggesting that the "pulses" I was looking for are in fact the "address lines being high". That is the first activity from the CPU which gets translated by the Northbridge into something that can work for the BIOS?
So I guess I should focus my attention to the PCI bus? Maybe I see something there which is not reaching the BIOS?
Thanks for your help!
@@tony359 Exactly. So I think the behavior of the CPU is pretty much expected. The address request should pass through to address lines all around the busses (ISA, PCI and also the BIOS chip which also should have most or all address pins high.On a (whatever)C1024, I would e.g. expect A15 down to A5 to be high and the lower pins to start low, plus /CE being low (Chip Enable Active Low) and the BIOS ROM to output some data.
If there are not address strobes on the address lines, no Chip Enable Low or no data flowing out of the ROM chip, this is already strange. If you have the proper address, chip enable and data coming out but obviously not getting back to the CPU, then you have again another problem. I assume a 16 bit ROM and a 32 Bit CPU meaning that the data will be read in two steps and assembled in the chipset, so again likely a task for the northbridge. However, follow the traces and look what happens where.
I need to double check the BIOS, I think I could see the address lines stuck to high but Data lines were likely floating as they read 1.5V. I guess I should check /CE and make sure it's low?
@@tony359 Yes guess so
@@tony359 Addition: I found another datasheet with another label description. Check Pin 2 (/E or /CE) and Pin 20 (/G or /OE - Output Enable!) if it is a 27C1024. For other types, refer to their data sheet, but I think, /CE and /OE have both relevance, and /OE even more, which I first overlooked because it wa labelled as /G (which I didn't understand) in the first datasheet I looked at
Amazing, in-depth video, as always. Thanks for bringing great content, Tony. As I read the comments, there's a lot of knowledgeable people, hoping they can give you the right hints. What I think may need to be checked, although I might be wrong, apart from re-balling the North and South bridges, is the CPU socket itself - inspect it for corrosion or dirt.
Indeed, I had forgotten to check that. It's on my list! Thanks!
I suspect the problem is with one of the chipset BGAs. There might be a bad connection due to physical stress or oxidization. Try to push the BGA towards PCB moderately and hold during reset. Sometimes, it helps enough for the board to POST.
I tried that - that's how I diagnosed the Amptron! - but I couldn't get any POST code. It could still be the BGA of course. Thank you!
A week ago I had the same issue with S7 MB, zero activity. The problem was in the NB, maybe more than 10 legs were unsoldered, after full reflow the board works just fine.
That happens very often - this is a BGA unfortunately. Difficult to re-solder.
I'd suspect the north bridge also. Check if it is getting too hot to the touch, or try pressing it, in case it's a bad solder ball underneath the chip.
I have pressed everything on the board for quite some time as that's exactly how I diagnosed bad BGA on the Amptron! But no luck. Nothing is getting hot either, I did check with the thermal camera... Thanks for your help!
Unless someone has a better idea, and if one has too much time on their hands and is determined to make it work, one could map out the back of the PCB socket pinout on a paper, and then go through all non-address/data non-voltage input pins (since you have presumably checked voltages and you have checked that nothing interesting happens to address and data) and map out their startup behaviour and see if something looks suspicious after cross checking with the p5 datasheet. You can just use colour markers and designate colour for each of the common behaviours, or designate a letter per behaviour. It's "only" a couple dozen pins.
If you check on my little piece of paper at some point in the video, at the bottom is the behaviour of the Amptron board! :) It's maybe a bit too much, it's easy to lose track of things. Regardless, the CPU on this board doesn't do anything on any of its pins (that I tested).
BGA chipsets. I have a couple of boards that are doing similar things. Re-ball those & it should start working again.
I need to save up so I can get the equipment to do the same for my boards.
Yeah, BGA is a pain!
I have several MB with the same behaviour and I failed to continue the repair because of the BGA chipset and the lack of the knowledge how the startup sequence is happening in these systems. I am very curious how this repair will go and I wish you patience and luck to drill down to the root of the problem. Suscribed!
Thank you and welcome! Fingers crossed, watch this space :)
@@tony359 What I found a bit strange: I was not able to find any document how the 486, 586 processor should work on "pin" level. What is the power up sequence, what levels should be present at a given pins with timing regarding each other in the power up sequence. There are timing diagrams for the buses, but no detailed explanation on these in the datasheet. The pin functional descriptions most of the time very condensed, left a lot of guessing about functionality and interoperability of other pins. Standard documentation is a set of puzzle pieces and you have to put the puzzle together to see and understand the full picture necessary to be able to diagnose an MB.
@@tony359 Thank you the tip for the test eprom to write the post code. It is an extremelly usefull debugg tool to check the CPU, SB.NB part of the MB. Always learning something.
Very interesting information in this video! Broken chipsets are a problem it seems. I once killed a mainboard by placing it on some metal. I guess I shorted some pins of the chipset and it’s dead. Not much one can do I guess.
yes that is a possibility and a difficult one to diagnose...
Fingers crossed you find it, I have a socket 7 which has been a pain. Original AMD k6-2 shows a code but no beep or boot, a AMDk6-2 500 AFX does nothing. Hopefully there is something I can learn from this as well.
Fingers crossed!
Hi Tony, thanks for all your great videos. Have you tried this with another type of cpu, like a Cyrix M2, or Amd K6 ? or even perhaps one of the slower and older intel cpu's?
Not yet but you're not the only one suggesting, thanks!
some (but not all) pure pci based main chipset boards (the isa is handled by the southbridge on this one, if i recall) don't output post codes to the isa bus, so if you haven't stuffed the post card in a pci slot, thats worth a shot.
i'd be tracing down the clock signals at all points to ensure its getting what it needs where it needs, confirming jumpers for all elements since some of these older ones have them for almost everything it seems.
beyond that, if you have a plain jane p100/133 or whatever in the lower end laying around, set the board to run that and try it that way. could be the 3.3v side of the equation.
I cannot remember if I tried the post analyser on the PCI slots, I shall do that! I do have a Pentium 90 - what would be the difference with the P200 I have? Thanks for your help!
@@tony359 well, in theory the p54cs (the 200) should work in any board a p54c (the 90) would also work in, even if at reduced clock speeds, but i've found some boards of mine don't seem to like anything over 133 or 150 for some strange reason i haven't quite tracked down yet, despite saying they have the capability.
if you haven't already stumbled across it, the rcollins site has a lovely set of early pentium datasheets if you search for 'pentium p54c datasheet', then click on the rcollins link.
some of which i couldn't find on theretroweb for whatever reason.
i'm not aware of any physical socket differences between the two generations, and its mostly just microcode. if you're feeling adventurous i suppose you could try the bios patcher necroware uses at times to update the microcode and whatever, see if that yields results.
and then there's the reality some some specific cpus just don't play well with some boards at all.
or, like you theorized, that cpu may not actually work.
i can't recall if they ever made any mmx variants in that black and silver package type that were mmx, but didn't have the laser etching on the heat spreader. i'd recommend double checking the cpu code bottom against the cpu world db to confirm its a p54cs variant, and not a p55c (mmx).
if the 90 works in that board, either there's something that doesn't like the 200 in that board, or the 200 is doa. (they're pretty hardy, so i 'm skeptical of it being dead without a confirmation)
there's also an embedded variant of the 200, but i can't imagine there's anything different between the two beyond microcode as well.
i suppose beyond those, its a matter of checking the IC pins for loose ones where possible. if the bga chips have solder joint or pad issues, thats going to be hard to look for without removing them, sadly.
i'm also vaguely wondering if all those rail caps around the socket are still good.
gotta love those pesky retro motherboards.
summary of post youtube seems to have deleted entirely; microcode in the bios. but sometimes some boards hated on mid to later chips. you might also try the bios modding tool necroware has shown for older systems. fair warning on that though, it does corrupt some few bioses. so if it does weird stuff, don't be surprised. but if it does, at least it would be booting.
might also test the caps around the cpu socket. and look at all non-bga chips for dodgy solder joints.
can't for the life of me recall all the info i had dumped in that vanished post.
sorry, I know the feeling when you write a long post and the Universe says "no" :)
Thanks Tony for all your efforts on my motherboard. Just a point. The cpu fan header is not powering the fan and when I first got the board the jumper in that white jumper socket ( I think it is used for an external voltage regulator) was in the wrong position, I’m wondering if that has done something to the motherboard
The socket with the 5 pins? That's for the "hardware monitor" - it's so unclear on the manual. How do you know the jumper was in the wrong position? The manual really doesn't say anything about it. If I remove it, the "error" LED stays on all the time.
And the Fan socket doesn't have 12V on it? It might just be that there is a fuse and it's blown - I suspect the fan is directly wired to the PSU 12V line - but worth checking, thank you!
And no worries, it's fun to get stuck! (well... :D )
@@tony359 Yes that is the socket. I don’t know for sure to be honest but I looked at other photos online to establish the correct position
I think I tried all the options and that seems to be the "correct" one, though it'd be nice if the MANUAL said something about that! :D I'll check that fan terminal, thanks.
i recently worked on a slot1 board, also dead with no activity on the bios pins, turned out to be a dirty processor socket, give deoxit a try i guess
Yes, I haven't checked the socket, it's a great idea!
I'm wondering if the CLR CMOS circuit is stuck on, the board would not start if that's the case. Or you simply have the CLR CMOS jumper set on..
This is just a hunch. :)
I would also say check all states of all bios chip pins and check exactly at what state it is stuck. What address, data, and so on do we see on each pin? Check the data sheet of the bios chip for the pins for that.
Jumper is in the normal position. How would I find out if the CLR CMOS circuit is stuck on? Thank you!
@@tony359 try moving it to the other position just in case to rule out mislabeling. Not likely but known to happen...
I would start by checking continuity on the traces that's going through the jumper, see if continuity changes when you move the jumper. If it doesn't, do a visual inspection for shorts or faulty components.
I haven't had coffee yet so not thinking clearly lol, but the chipset is north + south bridge right? Check for activity on the north bridge.
Another good practice, wiggle components on the board and look at the back of the board to see if anything moves eveen a tiny bit. If so, reflow the solder.
Next thing can be checking the caches - it's connected on the cpu bus and any fault block cpu completely.
Nothing is burning on the board. What would you suggest I checked on the cache? Thank you!
@@tony359 Power, solder joints, bent pins. You can also desolder chips, board should be able to start without it.
I'll add it to the list, thanks! :)
You could check all the traces of the cpu, pci and xD bus for continuity. Maybe a short has burnt one of the traces? Also it might be worth checking the chipset pins for activity, they should do something as well. Also the pci bus. I guess the chipset is broken. But if there is activity on the pci and xD bus you could narrow it down maybe.
that's difficult as the chipset is BGA unfortunately. But I need to check the PCI bus for sure, thank you!
Amptron PM-9600 has the same chipset that Acer V58 does. You can compare signal lines of chipset: their resistance to the ground (on a bare board) and their behavior (with a CPU running)
oh, interesting! I can give that a go though I'm not sure all the pads are accessible but I might have seen something at the back of the board. Thanks!
It is not the same chipset? That is an Ali M153x and this is an M154x.
@@SianaGearz , according to theretroweb resource both mainboards are based on ALi M1531/M1543 (ALADDiN IV+). I trust the resource
@@tony359 , I would try to run both models without a BIOS chip and watch CPU and BIOS address/data signals on an oscilloscope. Once I repaired a much more modern board with the same symptoms (No post code , inactive reset). The problem was found in a zero ohm resistor in SPI bus circuit (the resistor was between chipset and SPI BIOS chip). By the time I found out it I had already replaced chipset.
I hope you will manage to resurect the mainboard and I wish you good luck. But from the other side not every mainboard can be repaired for resonable amount of time and effort
Reflowing those chips??
Hi Tony, Sometimes is not everything repairable. I had a moterboard to. What i do, nothing helps. The board is now hanging on the walll in my room. Greetings from Steven from the Netherlands
Hi Steven - of course. I just have this guts feeling that I can make this work! :) Failing that, a spot on the wall is always available! Thank you!
Have you checked that cpu clock is present ? i couldn't see it on post card btw Great video !
yes, I checked it with the scope, it's in the video at some point. Thanks for that!
Looking at the Pentium board, go back to the ISA and PCI slots, between the back of the first two PCI slots from the ISA slots side, there appears to be a black electrolytic cap with holes in the top. Are these Sharpie marks or a blown cap? If that description is unclear, go to your diag pcb in a ISA slot and move down to the PCI slots and the cap I am referencing is near the side of the pcb between the first 2 PCI slots.
it's just some marker I'm afraid! But thanks!
For not being an "expert" you do make an honest effort in your troubleshooting!
ahah thanks, my "expert" benchmark is pretty high, I know :)
Thanks for watching!
It does sound like there's an issue happening between the BIOS and the CPU as the two are OK individually. You may have to check everything in between. Although, it would be quite funny if it was just a jumper this whole time :). Best of luck to you
That's my guts feeling as the manual is missing some or doesn't have much description for some...
Check the ADS# pin as well. The CPU should drive this low once it is in a bus cycle, so if it never starts one to even try and read from the BIOS ROM then it should remain negated (high).
The CPU also isn't necessarily going to execute byte by byte like older CPUs, there's a lot of parallelism going on (64 bits at a time). The BRDY# pin will be asserted by the chipset towards the CPU once it has presented some data in response to a read, or taken some data in response to a write, so that might be a good one to look at too.
Also rather than trying random address bus pins I would be a bit more targeted. The Pentium doesn't have address 0-2 pins, instead it has 8 "byte enable" pins. So I'd also take a look at these to see if there is any activity. For each byte out of an 8 byte chunk that the CPU wants to read/write it will assert the corresponding BEx# pin (maybe multiple at a time too).
And of course, know your enemy! Look at the behaviour of the pins on a known working board so that you know what they should be doing out of reset.😊
I think also checking NA# might be worthwhile. This seems to be a signal from the chipset to the CPU that it can accept a new bus cycle. If the chipset doesn't assert this signal the CPU may never start trying to access the BIOS ROM.
Thank you! I am learning more and more about the Pentium and I understand the first thing it would do is call for FFFF which is... high. So maybe what I observed was expected and as you said I should look at more specific pins. Thank you!
Check the south bridge because the south bridge loads the BIOS into the RAM. Processor without a code can not doing anything. That is the case according to my knowledge.
they're not easy to check as they're BGA... But thanks!
Is there connection between socket and the cpu? I had one slot1 board which seemed totally dead. And all I have to do was to use the contact cleaner.
something I have forgotten to check, absolutely! I can check the socket with the microscope, I should be able to see if the catches are there. And then maybe check continuity with the multimeter! Thanks for that!
Hey Tony359 maybe you could try to gentlty push, hold down the SMD chips on the board and then power on? I had two old boards this procedure helped. There were bad solder joints due to e.g. mechanical stress. I could not see that from visual inspection. However resoldering the identified chips then solved the problems permanently.
Yes, I found the issue on the Amptron this way and then I re-flowed the ICs for a permanent fix. On the V58 I pushed and pushed but I never saw a glimpse of activity unfortunately. But good idea, thanks!
Great video!
Thank you!
Could you try pushing down with a finger firmly on the Northbridge and/or Southbridge and while at it power on the board and see if there is activity on the bus? Sometimes the balls underneath are cracked and not making good contact...
Thank you! I have already, for quite some time :) It didn't make it on the video. Though, I didn't check if there was activity, I was only checking on the Post Card. I can try that, thanks. The Amptron was fixed that way :)
There's two interrupt pins you could check, INTR, NMI i would assume if they were shorted to active that would prevent execution also looking at the datasheet it appears theres an INIT pin does that change at all upon boot?
Thank you, I'll add that to my list!
Maybe you could measure and compare the resistance values of the address and data lines. And/or check capacitors for shorts.
I have checked with thermal camera, there are no obvious shorts on the board. Comparing the lines against the Amptron - same or similar chipset - is a great idea! Thanks!
Try different CPU's perhaps, it might be configured for a totally unexpected one.
Well, it's an old board, all the settings are manual and they directly affect the voltage/clock at the CPU. It doesn't matter if it's an AMD/Cyrix/Intel, as long as it works with 3.3V and 66MHz, it should work. One thing to consider is the multiplier: that I guess is sent to the CPU for internal multiplication. If that comes wrong, then the CPU might try to run at the incorrect frequency internally so maybe worth to check! Thank you!
Have you checked the signals on the CPU from the top? Could simply be a broken connector in the socket. CPU is behaving like there is no clock at all. But the CPU looks like the pins are available for contacting from the top.
It's on my list, thanks! I plan to inspect the socket with the microscope and maybe check for continuity. Thank you!
I would check for short on chipset and compare resistances with working board
I checked with the thermal camera and I didn't see anything suspicious. That said, I didn't realise the Amptron is the same chipset so I can try and compare a bit. Thanks for that!
Do you see any address lines activity at the BIOS chip? If not, are there any 74-series buffer chips near it?
The reason I ask is because I believe the BIOS is located at the correct address on Socket 7 boards, and is needed to initialize chipset registers.
(Granted, this is based on observations from a Cyrix MediaGX damaged VIA MVP3 board, where the CPU does start to execute BIOS code but fails after the first POST code, because the chipset was fried by the incompatible CPU…)
No, no activity and I mentioned the LS logic chip at some point in the video - that's also good unfortunately.
Interesting about the BIOS being needed to initialise the Chipset, thanks for that!
i was wondering if this board is one of those that will not power up if the CR2020 battery is flat, I've watched The Electronics Repair on UA-cam and he found a few boards like that.
It's a good idea - but unfortunately the battery is ok... Thanks anyways!
Any scratched traces under the board?
None I can see unfortunately...
You mentioned moving jumper JP1 at the beginning of the video but not being sure what it is for, this is used so the system can be configured for either an AT or ATX power supply, ATX supports wake from standby, AT does not, there is a special connector marked JP6 near that jumper where the board expects the 5v standby power (since it lacks a ATX connector) I guess this used some special Acer power supply, never seen this before, anyway there is also a second jumper, if you look at The Retro Web there is a manual with only jumpers, this shows when you move JP1 you also need to move JP15, did you do that? Other than that some boards actually measure the power consumption of the fan and will not start if a load from the fan is not sensed... Finally there is a green LED on the board, is it called HW Monitor? I can't see in the video but this might be related to why it will not start, refer to the same jumper diagram on the retro web: JP7, JP9 & JP10
Yes, I did move JP15 when I discovered the option. I didn't think of adding a fan on the connector - all I know is the connector seems to be dead, as it's not getting power. But that might mean it's going through some logic which, as you suggests, checks for a load before providing voltage AND turning on the board.
Yes, the green LED is off (well, it looks barely lit) and it goes full brightness if I remove the blue jumper from the "HW monitor" socket, for which there is NO info whatsoever on the manual. JP7, 9 and 10, I tried several combinations but I don't understand what they do (JP10 says factory set do not alter).
I'll try with a fan, thanks!
@@tony359 I can't provide a direct link but the jumper only manual is on the Retro Web that I mentioned, those jumpers are listed in that jumper manual...
I'm very familiar with those two manuals. But my board has jumpers the manual doesn't have.
If the chipset is broken, last option would be to replace it with a known working chipset. It might not be worth it for the hardware, but it would be nice to know if that is the case here.
Oh yes, this has nothing to do with "financial reasons", this is for fun :) If I could find out what is broken, it would be great!
Is there a scorch mark on the underside of the board near the CPU?
I'll check again but I checked for damage so many times! :) I'll check around the CPU thanks!
@@tony359 I wish I could screenshot it and save it to here so you could see it.
can you give me a time so I can check?
Is there a "reset BIOS" jumper and is it in the correct state?
Yes, that's in the correct position - but thanks!
there may be something in between the cpu and the northbridge and the cpu or the northbridge is bad or even the southbridge .... you have to check every component between the cpu and the northbridge en between the northbridge and the southbridge ...
Yes but how do I know what's in between the CPU and the Northbridge? In theory there shouldn't be much, just the CPU bus though. Thanks!
@@tony359 there are always decoupling caps in that path
Thank you, I'll take a look!
Hi Tony!
A you checked the clock crystal near the SB?
Also, near the bios chip preset the crystal, i dont know for what it.
Try to check for bad or broken any elements, try to cheking the CPU slot pins, check pins on any other chips, try to press on the NB, SB chips and then start the MB.
Yes, I had forgotten about the CPU pins. I intended to check them and then I forgot. I can inspect with the microscope thanks. The clock near the RAM should be the reference one for the clock IC nearby - if that doesn't work, you shouldn't have 66MHz at the CPU. The one close to the BIOS: interesting. I wonder what that is for. I'll check it thanks!
I've inspected the board - front and back - multiple times for a broken trace or broken "something"... :(
Thanks for your help!
The little crystal... i would think that looks like a 32Khz RTC crystal. There should be an RTC logic and NVRAM chip nearby, upheld by the battery which is also nearby, since that and the BIOS share a bus, called XD bus off the south bridge.
And yet it's a little surprising that there is no activity from the CPU at all, not attempted BIOS read, no address-data bus activity? Seems like there should be a more immediate reason that should be more directly measurable.
@@SianaGearz Yes, i mean the RTC crystal marked on board as X3
If it's got a dallas chip replace that because you won't get beep codes if it's dead
No Dallas chip unfortunately and battery is healthy. But thanks!
what about the /BOFF signal?
The backoff input is used to abort all outstanding bus cycles
that have not yet completed. In response to BOFF#, the Pentium
processor will float all pins normally floated during bus hold in
the next clock
Amazing, I missed that! Thanks, I shall check that!
@@tony359 /FRCMC also sounds like it would be able to spoil the fun
@@tony359 and R/S signal might also be able to let the CPU appear dead (?) .. datasheet talks about putting the CPU into idle for debugging via JTAG
Thanks - I checked R/S in the video and it's not active. But I'll check the other two thanks!
Sir please reply...
this question is very very confusing...
I have 2 identical 12V DC 7.5Ah batteries say X and Y.
One side of a normal bulb is connected to postive terminal of X battery, and another side of the bulb is connected to the negative terminal of the Y battery.
The negative terminal of X battery and the positive terminal of Y battery are not connected to anything.
So, in this case, I want to know whether there will be a current flow to the bulb, and whether the bulb will glow??? Give reasons if bulb will glow, and reasons why bulb will not glow.
Thanks in advance.
I'm not sure why you're asking me to explain you why a light bulb would turn on or not (and consider both scenarios and also provide a explanations for both scenarios) on a motherboard repair video. Your question reminds me my Physics school tests but that was a long time ago. That said Physics was my favourite subject and I usually got high markings on those tests :)
I see the 2 videos and no see if you check the clock in the bios (bios send data have cpu or not if they have power and clock) so the clock is missing, the bios is broken or the chipset is killing it.
Have a small cristal near the bios.
I haven't checked that - how does the clock work with the BIOS? The BIOS IC does not have a clock input, I thought it would just follow the ISA clock?
Anyways, I do not see any activity at all at the BIOS chip.
Thanks for now!
Sorry It took a while to look for the dataset, in this case it says that it is internal (I see most outside)
you have the 5 v on the pin 32 vcc?
This chip is a pain have to many variations to read depending on pins 32 /24 /22 are high or low (this pins need to change low or high to allow read if all stack high not work) to send data pin 24 and 22 low and pin 31 high . (Data says logic for this pins to be high or low but who make the logic the bios chip, the chipset or other chip)
@@yoyomismo2052 Yes, I have VCC and GND on the BIOS chip. And I don't see any activity anywhere on that IC.
You can read in the programmer with the oscilloscope see that you should have it on the pins (it seems that it reads and records it so you should see the necessary signals) but I'm afraid it's going to be the chipset, hopefully you just have to Resolder it
Try reflowing the northbridge BGA chip.
Yes, that's always an option! Thank you!
@@tony359 If you go to that, it's better to do the reball. Many cracks is not ball itself, but inter-metallic joint on the pad (copper-nickel-gold-tin). It separates in the layer and reflow not fix it. You can see it on the pad as grey layer that is hard to solder (but not impossible, if you clean it, it will work).
@@petr56321313 Does that happen with that era's chips, with leaded solder?
@@petr56321313 Yes thanks, I am familiar with that problem. I've encountered when I reballed my first (old) GPU.
I'm sorry for suggesting the obvious, but what about CMOS battery? Some boards won't start with flat battery and I once almost discarded one of good boards for that reason.
... excellent (obvious) idea! I'm not sure I checked that. I'll put that in my list! Thanks.
The battery was replaced
I can't even remember my own videos? :D
@@Roadkill7878 I don't think so - it is the same battery that was shown in previous video. Why do you think it was replaced?
@@86smoke because it’s my motherboard and I replaced it!
How much memory has that ram stick got? According to the manual that motherboard supports a maximum of 192MB, one 128 and a 64MB or 128MB plus 2x 32MB. So one 256MB stick should not work. Aside from obvious missing things like voltages, clock, reset, dead chipset ics, it's usually cpu or ram issues that give a dead post card. I recently have 1/2 repaired a faulty Jetway socket 7 motherboard with dead cache, now I found out the pins in the cpu socket are badly oxidised which causes intermittent no post or hangups. Your board looks really clean so likely no corrosion problems. Only other thing is have a real close look for damaged traces.
Only 32MB I'm afraid. But even with incompatible or bad RAM, I'd expect the CPU to execute the first few instructions and then complain about bad RAM. If you remove the RAM from your system, you still get something at the POST card, then obviously it stops. Unless, i guess, the RAM is shorting or compromising the CPU bus? But now I'm thinking about that, the RAM is wired to the Northbridge. The RAM works anyways, you can see in the video when I test the CPU and RAM on a different board.
But thanks so much for the suggestions!
@@tony359 Maybe You have already checked but I would check voltages to the cache and tag ic's as well as any other You haven't checked. The cache chips have 2.5v and 3.3v rails possibly. A bit of a long shot though.
EDIT My bad here, JP7 is ok, Vcore is correct on JP11. This is a bit of an oddball socket 7 motherboard, have never seen that HWM Vcore on any other S7 mobo.
Ah, that's the "hardware monitor", I don't know what that does. but I tried all possible variations...
Thanks for the hints, all very valid. I haven't checked the cache voltage so something to try for sure. Thanks.!
I think this is relevant here,tony has bypassed all bios checks,ie switches,jumpers, ram,and so on.
So,because he has programmed a prom to simply get the cpu to write a value out on a bus,he is checking to see if the MOST basic function is working, if so,he can move from there.The glue logic,(ie northbridge,and or southbridge)and the cpu are being tested to see if they can do the most basic thing ie read an opcode and operand from the prom and then write a byte or word to an address.
His concept works because he tested it,on another mb,and the data was written to the test card.This is a sensible step.
The only caveat,there maybe slight differences between the addressing or bus arrangements between the two motherboards,he has to check this.if they're the same,then he has a glue logic to prom bus issue,a glue logi to pci bus issue,ie no seven segment activity,or a cpu to glue logic issue,and he can proceed from there.
Jumpers,bios,cmos ram,are irrelevant guys,please lets not confuse him,there are other people on the forum,who are pointing him on the right direction here.
Lets us try to give as helpful suggestions as possible.
Bios OK? Test Jumper Frequencys
I used the same BIOS chip to test the 80h on the Amptron.
The jumper frequency is ok as I measure 66Mhz at the CPU socket. Thank you!
Really strange... The only jumper I see that does not fit me is JP5, 2-3 pins sais for BIOS flashing 5V in manual try 3-4 which sais normal in manual and has default asterisk and NC on the board
I can try - though that should only affect the behaviour when flashing the BIOS I guess? Thanks for that!
change the motherboard battery...
It’s good unfortunately but thanks
@@tony359 may be just before quit the board you may try to "cook" the north bridge ( heat the chip with no more than 160° i think for a few seconds) and try
@@pyromiko Take a look here: ua-cam.com/video/Q8inNVAGj4A/v-deo.html
@@tony359 excellent! Very happy ending!! Regards!
I know nothing about motherboards however if I were in your situation I'd try "AI" and see what it says. 😉
It lacks hyper specialised knowledge like that and will run in circles. But you're welcome to try it yourself and see whether you get any useful output out of it :D
It is after all a text prediction machine, a lossy highly compressed representation of text used for training, and there is very little in depth CPU troubleshooting content that the training process could ingest. A PC mainboard is after all a field replacement unit, where if something is wrong with that isn't BIOS or battery, jumpers or installed components, it would simply be junked and replaced by a working one. Component level repair except rework at manufacturer where they have secret manuals and specialised tools, is effectively a new phenomenon.
I tried and it said "did you try turning it off and on again?" 😉
No beeps means there's no cpu function so try changing it
Did you actually watch the video? Tony put the CPU in another board and it worked fine!
unfortunately it's not so easy...
@@stoptheirlies don't judge too harshly .. I watched the whole video .. twice .. and still suggested to check the R/S pin *facepalm* .. it happens 😁
Exactly! :) It's fine and I appreciate every suggestions :)