For the past 20 years my first step is ALWAYS “remove, check/replace CMOS battery” for this exact reason. You’d be amazed how often this first step fixes the issue.
I should do the same. As I have few boards that don't POST, but I haven't tried with a new battery.
2 года назад+6
I did guess its was the 2032, its and odd issue BUT there is a Crystal between the chipset (small round cap locking thing glued down) and the socket, you can see it in the video. That crystal runs the Real Time clock inside the chipset. I debugged a Super Socket 7 board myself that keept erasing the CMOS despite the battery being new, and I had the data sheet for the VIA MVP3 chipset. I checked the pinout and the crystal is for Real Time clock, i also checked the VBAT pin and it had 3V from 2032. In my case it was the Sound card being plugged in to the line in on my "lab" computer where I record YT videos, yea for some reason that created a potential difference it seems dropping VBAT over time effectively erasing CMOS and stopping the Real Time Clock. Anyways I did NOTE from my testing that if the "crystal", the one you have between the chipset and the socket is not working, aka no power, the system wont post or beep just like your so I did call the cmos battery as the culprit. Also your "power on and off" in short succession keeps the caps in the PSU and on the motherboard charged and when ever there is enough voltage from the PSU the Real Time clock will run of that so the RTC is running and thats why your computer can post the second time. It just cant post with no RTC running initially. So my guess is that the way this board was designed its a timing issue, power dont get to the RTC to start "ticking" fast enough and then the system gets stuck and wont post because its expect the RTC to be working. Maybe a fresh PSU and a recap could help but it could just be the way this board works and a new 2032 obviously solved the problem. But yea the Crystral must be oscillating for the RTC in the chipset for the system to post. Also 4.95V is well with in AT specs, the reason it drops low is probably because the HDD is putting no laod on the 12V rail, the spindle is on that rail and on of the few 12V loads on a old PC as you know. And the HDD not spinning up is most likely doe to bios not initializing, I have seen this before, its funny how its all tied together. I'm pretty sure it would run just fine with 4.95V, the fact that it posted with 5.13V was because the RTC could run on what was left in the cap from the post attempt before it and why it went to 5.13V when posting is because the bios kicked off as it should and the HDD was happy whit what it saw and spooled up. Most old PSU's do require some minimum load on the rails to give the right voltages. But I know you already know this but for anyone else this could be good to know. I hope this explains the problem. But yea RTC problems can be annoying at times doing unexpected things. Anyways love your videos and I hope the explanation did not come off as "I know best", just happened to have similar problem on my last SS7 build and did dig pretty deep in to the CMOS rabbit hole XD
I would just double check any capacitors that are in circuit with that battery to be safe. since it works on subsequent boots, it's obviously looking for a charge on an associated capacitor, probably to safely switch the CMOS ram safely over to main power, and the capacitor isn't able to fully charge on a weaker battery. It might be by design, but it could also be an indication that capacitor is old.
It’s more than likely the other way around. The coin cell and main 5V rail are combined after their own respective diodes, and residual charge in a cap somewhere on the CMOS V+ rail allows for a successful warm boot, but isn’t quite up to charge on a cold boot. It is quite likely that the 5V system rail is fed through a current limiting resistor to support safely charging rechargeable cells, since most motherboards had provisions for both. That would form a low pass filter, meaning the downstream cap would take some 10s or 100s of milliseconds to charge to a suitable running voltage from zero. Ergo, cold boots fail.
@@BenState An inductor _in series_ would limit changes in current, yes. But so would a capacitor to ground after a current-limiting resistor. Usually whatever capacitor is in-circuit to a CMOS battery wouldn't be an electrolytic. It's likely either a ceramic or tantalum, so typical "old capacitor" failure modes aren't as likely here.
Not too odd, the Board probably holds timing configuration in the bios memory and fails to start reading the values, on the second attempt these values are refreshed from main power and the system starts.
@@SomeAngryGuy1997 The voltage drop probably was just a red herring. Some part of the system drawing a bit more power during initialization. 4.95v is totally fine.
@@thetaleteller4692 Yeah, for ATX the specification is 5V+/-5%, IE anything between 5.25V and 4.75V is defined *fine*! I assume this is an older AT style PSU, I can't as easily find the limits for those but I very much doubt it's tighter specs than modern ATX powersupplies, so 4.95V isn't an issue. The "blinking keyboard LED" kind of hints the BIOS goes into some kind of crash loop, so yeah, definitely possible it's drawing a bit more power than it will once it gets things set up correctly.
@@thetaleteller4692 Yes. I kinda sorta guess the HDD not starting doesn't load down the 12V line enough and maybe this PSU uses the 12V for regulation, who knows. Maybe it uses a resistor network to average all the rails to do some analog math, so loading the 12V will cause the 5V to go up. I have seen PSUs do that.
Changing the battery is the very first thing I change out when I get an old computer in my shop. Some of these old computers are not all the same and it never hurts to replace them right away.
Audio quality aside (really not a problem), I like the fact that the voiceover took a backstep. It feels like you are more "connected" with the video. Maybe a mix of it in the future as voiceovers do have a place and function. Great job and keep it up.
There are motherboard series that only work with a cmos battery. I've seen quite a few in my career. You don't think about that in the beginning, but an empty cmos battery can cause a lot of problems. A point to keep in mind when working with old computers.
I forget that some systems are quite quirky without a good CR2032. I had this "cheap" board which had a (dead) CR2032 externally wired to a clock chip (not DALLAS). It would POST but not much more than that. I like the boards which actually have these coin battery holders as standard, makes swapping out much easier.
What an odd issue. Thanks for sharing. The recording works great by the way. The live voice-over sounds nice, in fact it has (at least using headphones) clear directionality, coming the voice from behind, which is unusual in an interesting way.
I've had a very similar problem with a same era gigabyte board. The voltage on the cmos battery was enough to maintain some if the cmos data, but corrupt the contents. I had to force reset the cmos using a jumper to clear for a few minutes, put a new battery in and all was good again.
Great video and I'm glad you found the problem. I'm dealing with an IBM PC 300GL that boots, but after 3 to 5 minutes quits receiving keystrokes and emitts a high frequency noise that comes and goes. Once you start hearing that noise the computer basically locks up. This does it in DOS or Windows. Swapped ram, cpu, power supply, and video card and still locks up. Anyway, thank you for the cool video.
Hi Sir, regarding to this, I have a mobo in my stach (dont remember the brand/model) that NEVER boots if CR2032 is empty/low power/missing I took some time to trace the problem some years ago, and you now present a very similar problem
An older series of Dell computers that I worked on had a similar issue. A dead CMOS battery would result in a seemingly dead computer. Ever since that repair, I've always replaced the batteries in computers more than a year or two old, just to be safe.
There is a pin on the IDE cable that can hold the drive in initialization, before the spindle starts up. I have seen controllers and motherboards keep the drive from starting up, and unplugging the cable allows them to start normally. It’s not “normal” because the controller should be initializing the drives, but if the system isn’t booting, that would be one potentially expected outcome. Understanding that, there’s a good chance the computer doesn’t have much, if any, load on the 12V rail, and the PSU may be having trouble regulating the unloaded 12V rail when the drive isn’t spinning. The PSU doesn’t have independent control of every rail, so that could be why the 5V rail was a little low when the drive was shut down. That’s all very characteristic of an older switching supply. Nothing to worry about there. Having the IDE controller stuck in a reset state would cause that chain of events and perfectly explain all the symptoms you see at the beginning.
A few minutes after watching this, the new voltage regulators in my Abit AB-SM5 started to work, the Pentium MMX 166 MHz is finally getting hot again ! No POST codes for a while but after I probed the regulators I started to see POST codes and finally a boot screen.. so thanks for the video .. I think it gave me some luck !
when i was in high school back in 2007 i worked as a school based trainee working along side the schools computer tech and the most common thing id see on computers from the socket 7 to socket 370 era is that if the battery was low or dead then the computer would provide almost the exact same issue to what you demonstrated in this video and removing or replacing the battery would resolve the issue every time
I was sure it was the battery as soon as I saw the error on screen. I have seen this before, I have a few motherboards that either refuse to boot at all, or act weirdly at boot if they have a discharged battery.
Like the transistor and Capacitor could be faulty between the CPU and Bridge.... It's a very rare issue that the CMOS battery caused this. Could be corrupting the CMOS chip.
had a packard bell 166 mmx back in the day did weird things like that if the 2032 was almost dead, wouldn't initialize HDD, false start like this one , or it would boot without video, or boot normally if the energy was right, noticed my settings were wonky, not even a low battery warning, changed the backup battery out and problems went away
I have never seen a motherboard that was a that picky about it's coin cell battery. What a strange error/problem indeed. Glad to see you got it back up and running again. Have a good one.
I ran into machines that wouldn't boot due to a bad coin cell. If it was borderline, the bios may have been corrupt. Removing it and reinserting it would sometime help. Removing the cell completely would also verify it was the cell
I once had an old Shuttle Nforce2 motherboard whose bios battery always drained after a few days when not plugged into the wall. It was a real pain to replace so many batteries and reconfigure the bios. I eventually took it out of the computer case, and noticed there was a sort of sticky substance on the motherboard near the cr2032 battery. I washed the substance off with alcohol and water, and after it dried a few days, it worked perfectly. No more battery drain. There must have been a very minor short to ground from the sticky stuff. Might've been flux or soda.
This is a very common phenomenon with various laptops, even much more modern ones. Some appear dead when the CMOS battery is low. Off the top of my head, various Sony models were known for doing that. I've seen it happen with desktop motherboards before, but it is much more uncommon. Some are especially weird and boot fine without a battery installed, but refuse to boot with a low battery like 1V.
Wistron-produced Acers (7551 ,5738,5536, 5740, 7736, 7738, 7740 and there might be a lot more than I know) will definitely NOT boot with no CMOS or low voltage batteries. I've been given a lot of 5738 and a few 5536s, and most of them weren't POST-ing due to that CMOS thingy.
My DOS machine is a P133 MMX. 384 megs of ram. 2 gb compact flash, dual channel storage. It's so overkill, I love it and getting so much ram into it probably slows it down a little, but it makes me cackle. I have a really pimped out Mac iici too. I just love retro computers and games. I love how massive your tower is there! 😍 Thanks for the video!
It's fun to completely pimp out old retro systems. I have a PowerMac G4 400 with 256 MB RAM, running MacOS 9.2. Originally it had 768 MB. I doubt the OS uses more than 8 MB.
Could it be that the old coin cell battery is shorted? Maybe the problem was not due to the battery being totally discharged but to it being actually faulty, it's a quite rare occurrance but it could happen that those coin cell batteries develop a short circuit internally.
I had a couple of old XP machines that would post, but not boot into the o/s due to a dead CMOS battery. Apparently Windows didn't like the discrepancy between the original BIOS date and the o/s date. Once I replaced the battery and updated the BIOS time and date, everything was fine. The CPU cooler on that MMX looks so tiny and dainty lol
One possible temporary workaround may have been to boot to another OS via external device (DOS is ideal), change the date, then warm reboot until you had the battery replaced.
i'm pretty sure my lga-2011 pc had this issue, it would get stuck in a boot loop, i thought it was a ram issue because i reseated them and it turned on... then i thought it was a psu issue and i replaced the psu... still did it, thought it was caps so i bought a new mobo but then i read a forum post about a guy having the same issue and he replaced the battery and guess what happened? it boots normally... i did the same and guess what happened? it boots normally... now i have an extra computer that i will turn into a NAS or HTPC.
I have some PCI+ISA 233MHz Pentium and AMD P75 boards that once they warm up they won't re-boot without a full power down. Good find on the battery, I would not have thought it would cause such an issue.
I would have never expected the battery to be the culprit. I had some issues with a pentium 2 system and it required replacing 3 caps on the motherboard to fix it. It would only boot at the third attempt, the first two times it would beep with a sort of alarm. That is a chaintech board with an Award bios. I don’t mind the off the cuff style, I tend to work without scripts for most of my video’s.
Interesting, I haven't seen a PC before that won't start without the CMOS battery. Although I haven't really had a reason to remove the battery on a "modern" (e.g. post pentium-era) motherboard. I have seen it on a Mac, my Powerbook 1400c won't boot without the PRAM battery connected. In fact, it won't boot if the PRAM battery is flat. It took me a while to work that out when I first got it home and it wouldn't boot. Powered up, nothing. Tear the thing apart to get to the PRAM battery and it was fully discharged. Wouldn't power on without it either. Put it back together and still won't boot. But I left it plugged in and a while later I'm startled by a loud "BONG" sound, which was the classic Mac boot chime. Turns out you just had to let it charge the battery for a while before it would boot. Anyway, I don't mind the off the cuff style. Not quite as polished, but I'd say it's definitely still entertaining. About a million times better than I'd do, considering how bad it was when I tried to record a narrated screen capture how-to video for work.
I was thinking that someone put a jumper on the reset button header, and for some reason, when executing the power-off and power-on sequence fast enough, the PC would just start. But that's a good finding. I've seen some other motherboards that do the same. Really annoying
This may sound crazy, but I've seen coin cell batteries get fried after a brown out. Scorch marks and everything, replaced with new. It was a dell optiplex with a core i7 8700. Optiplex is a real workhorse! I've seen many a computer with bios battery messages at boot and/or you have to power on twice...nothing I ever bothered trying to fix beyond trying a battery...in those cases new coin cell didn't fix. It's always old stuff for the most part.
I have the "Baby AT" version of this case. just built a Retro system in it. weird thing is that only the power LED works, likely might have to grab my spare leds and replace them, but still a good solid case, even if a bit yellowed. EDIT: I actually have a variant of this board as well. though my setup runs a k6-2 400.. wow
I have an 98 machine that is doing the exact same thing but it does not give me a low battery warning. That being said, I will change out the battery to see what happens.. Thanks for sharing .
Would be nice if you could merge the stereo audio streams. The balance between the left and right channels varies per shot and it can be a little disorienting to the ear.
A nice little feature I learned about Windows 10 (and I'm sure 11 also) a while back was the "Turn mono audio on or off" setting. Go to search and type mono and it will come up. It makes listening to videos with headphones much more tolerable when they have uneven audio or sound on only one channel. Just remember to change it back to stereo afterward.
I'm guessing the IDE cable itself is broken inside and moving it just creates a connection or possibly the ID channel it's plugged into is somehow faulty itself it's only two things I can think of
I would think the BIOS settings got corrupted and when you removed the battery the actually defaulted. I would have tried booting with no battery to make sure.
Would it start with no battery? Not sure how scrambled CMOS would dip the 5V though - I have had one case of scrambled CMOS no boot when I changed a low battery fast enough that the CMOS only partially cleared. I thought the fitted Intel HSF only applied to the overdrive model
It wouldn’t be caused by scrambled data particularly. Most motherboards will only reply with a pwr_good signal (to the PSU) if every voltage is in fact within margins. With some motherboards, this includes the CMOS battery. I’ve had the exact same issue with either the: - Acer (MicroStar) MS-7326, or the; - MSI 945GCM7-F MS-7507. It’s a 775 motherboard, but the issue remained identical. Despite the difference in era. Great vod @Davy!
I'd have thought that a modern CMOS would have a simple checksum to ensure the data wasn't scrambled, and just reset to default values if the checksum didn't match. Maybe I'm assuming too much though.
It cannot be stressed enough, weird behavior in POST (in notebooks as well as desktop pcs) is very often associated with a weak or empty cmos battery. Always check for it before taking further measures like disassembling the whole thing.
My Pentium 133 has an issue with detecting a keyboard, whatever I do, it wont detect my keyboard, tried a different one, same issue. I’m kinda at a loss as it worked perfectly before.
I do a simple test that almost work on all motherboards back in the day... When no image i remove the battery and test if no little tickle on my tongue between + and - then i realize that battery is run out, i replace and always work. Is a ultra quick test when you tired to use tester
I came across to some Acer laptop few years ago, it came to me for diagnostics because it was only blinking it's lights and nothing else. All it needed was also brand new CMOS battery and it came back to life after that. Seems like some machines need working battery to work properly, some machines do not care at all.
I had quite a few mother boards even in the xp or win7 era which did not boot with a dying CR2032, since then I always swap the battery on old motherboards I get.
I had this issue back in the days of P2/3 (I just remember no boot) with some Tomato motherboards (and what made me think of them as low quality...yeah, they were bought because they were cheap). So when I saw the 1st symptom...I was "how's the battery", then when I saw the warning on boot, I skipped the video to find if the battery was indeed the problem. Sorry for the "low watch time", but I would have changed the battery 1st, before removing from the case.
I had a similar fault on a more recent system. Would not boot. Had a power up shut down cycle. Checked the voltage of that battery and it was 3v. A fresh one is 3.3v. System booted when I replaced the battery. I dont remember ever seeing this before In the 20 years I've built/repaired/used computers 🤷🏻♂️
Maybe some quartz generator on the motherboard need full working battery during POST? Long time ago i experienced similar issue with my Asus laptop : dead battery = dead laptop (only fan noise) Sorry for my english :/
Great video, nothing wrong with going unscripted. I had a Win98 PC that started doing this. It also turned itself off and eventually wouldn’t turn on at all, even with a new PSU. I should have tried changing the battery before I sold it for spares.
The audio......... Maybe those dedicated chest mics... Yes, it's very strange the mobo needing a new full coin cell. Do you have ebay or place where you sell your things?
I have a Desktop version with that Case Design. In my PC is a Pentium 166 Mhz with MMX 16MB RAM, 1,6 GB HDD a 2 MB ATI Mach64 and also a Sound Blaster 16 I found that PC on the street and it works with Windows 95 and Office for Windows 95
My guess is, to show the people that are new into retro computing to *always* check the battery first. I would bet the battery was the first thing he tested, then he thought it would make a good educational video that could potentially save a lot of time for some people.
Because it's RUclips and he needs to pad videos out for the views and extra algorithm sightings 🤣 Kidding, I like the diagnosis procedures EDIT: next thing retrospecter needs to do is go all this video is sponsored by or put a shit ton of ad triggers 🤣🤣🤣🤣
Years ago, a tech at PC Doctors in Huntington WV told me that he made more money from bad batteries than any other hardware problem. I have no reason to doubt him. Over the years, I have had several 486 and Pentium motherboards refuse to post without a battery or with a failing battery, all of them had CPU voltage auto-detection or some other "new tech" that was cutting edge at the time. Enough of them that for a very long time, I refused to buy any motherboard until it was at least revision 2 or higher.
To me, it sounds like the 5v rail might have bad caps. The fact that it takes quite some work to get it to provide stable 5v tells me there is a cap that has trouble charging up, but once it does, it starts working fine.
Had the same motherboard, had to do similar troubleshooting. Took me awhile to realize that the battery is the cause of the problems. It's kind of amusing seeing someone else struggling as I have been.
Very weird... The only thing I can imagine is that it is somehow connected to the auto voltage selection that goes bonkers when the battery is low and might, due to some overvoltage protection, short the 5V (which was used for the VRMs of old computers up to the Athlon and P4) partially to ground. By the way, the 430TX chipset has a cacheable area limitation: 64MB. Anything larger than this will become awfully slow, when accessing these RAM areas. This is why I chose the 430HX back in these days, despite of this still using the PIIX3 southbridge without UDMA (my HDDs were PIO back in the days anyway, but I was aiming for large RAM), whereas the 430TX already had an UDMA enabled PIIX4.
A word of caution about 430HX motherboards - some cache the lowest 64MB, others cache the entire 512MB, it depends on the size of the tag ram on the motherboard. IIRC many motherboards support either config, just swap tag ram chip and move a jumper. So likely solvable but still a "check this" item if you plan to have more than 64MB on an 430HX board. IIRC my recollection is that there were some motherboards that used the 430HX system controller (but paired it with the 430TX controller chip (PIIX4/)!, I assume Intel didn't officially support this but it provided the best of both worlds, 512MB cacheable memory and UltraDMA (DMA mode 3). I've found a few forum posts from people who claim they have motherboards like this which suggests my memory might be real.
It would take a pretty large short to drive it down to 4.9 volts. It would think it be enough that the heat would be noticeable if you were looking for it. But the AutoVRM on those boards is absolutly the culprit non-the-less. The ones with the Dallas clock boxes on them don't post when the battery runs down. As for the 64mb limit. Its not really _that_ bad when you have cas2 ram timings. Its a deficit of about 10-15% performance for things exclusively ran from those blocks of ram. Windows itself will reside in cachable range and most applications will be small enough to be mostly resident there too. On computers with about 128mb of ram the vcache will be taking up the bulk of the upper 64mb anyhow. And this is even less an issue if you are running 96 or 80mb of ram. Now if you pop 256mb ram on them... and then start to do a bit of more intense stuff, then its going to have its effects. Memory specific benchmarks will show a much larger difference than is actually realized in practice.
I have a Gateway Pentium D system (GX7022E) that will do exactly the same thing if the CMOS battery is low. Over 15 years old and still going strong, but it likes to eat power supplies.
I have a 486 motherboard that only boots after pressing the reset button on the case. Seems to clear a power good signal or something. But after seeing this, I will double check my CMOS Battery!
This wouldn't be my first encounter with such issues. I had this more than once - a dead CMOS battery should usually be fixed as a first try, since it's a matter of seconds... BTW, as feedback, I watch YT on a TV with a 5.1 receiver switched to Dolby Pro Logic II, so I get the voice centered. In this, your voice moves and sometimes even comes from the back (which is kinda weird), but is almost never directly from the center. Maybe that's something you want to tackle in your recording setup...
I don't think that is uncommon. I've ran across multiple PCs and laptops that refuse to POST or on on occasion it would fail to even power on-with a dead CMOS battery.
I've had this issue with Socket 478 and 775 motherboards over the years. Failing to detect or boot from IDE drives when the CMOS battery is marginal. You may not get a battery warning from the BIOS but it's still problematic for some reason. Something to do with IDE detection or reading the stored settings from the CMOS to boot correctly. Replace the battery with a new one is standard now with any board I come across that's more than a few years old 😃
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
For the past 20 years my first step is ALWAYS “remove, check/replace CMOS battery” for this exact reason.
You’d be amazed how often this first step fixes the issue.
I keep a bulk bad of 2032 battery's for every board I buy off of eBay. Unless it looks like new it is replaced... 🙂
LLAP
I should do the same. As I have few boards that don't POST, but I haven't tried with a new battery.
I did guess its was the 2032, its and odd issue BUT there is a Crystal between the chipset (small round cap locking thing glued down) and the socket, you can see it in the video.
That crystal runs the Real Time clock inside the chipset.
I debugged a Super Socket 7 board myself that keept erasing the CMOS despite the battery being new, and I had the data sheet for the VIA MVP3 chipset.
I checked the pinout and the crystal is for Real Time clock, i also checked the VBAT pin and it had 3V from 2032.
In my case it was the Sound card being plugged in to the line in on my "lab" computer where I record YT videos, yea for some reason that created a potential difference it seems dropping VBAT over time effectively erasing CMOS and stopping the Real Time Clock.
Anyways I did NOTE from my testing that if the "crystal", the one you have between the chipset and the socket is not working, aka no power, the system wont post or beep just like your so I did call the cmos battery as the culprit.
Also your "power on and off" in short succession keeps the caps in the PSU and on the motherboard charged and when ever there is enough voltage from the PSU the Real Time clock will run of that so the RTC is running and thats why your computer can post the second time.
It just cant post with no RTC running initially.
So my guess is that the way this board was designed its a timing issue, power dont get to the RTC to start "ticking" fast enough and then the system gets stuck and wont post because its expect the RTC to be working.
Maybe a fresh PSU and a recap could help but it could just be the way this board works and a new 2032 obviously solved the problem.
But yea the Crystral must be oscillating for the RTC in the chipset for the system to post.
Also 4.95V is well with in AT specs, the reason it drops low is probably because the HDD is putting no laod on the 12V rail, the spindle is on that rail and on of the few 12V loads on a old PC as you know.
And the HDD not spinning up is most likely doe to bios not initializing, I have seen this before, its funny how its all tied together.
I'm pretty sure it would run just fine with 4.95V, the fact that it posted with 5.13V was because the RTC could run on what was left in the cap from the post attempt before it and why it went to 5.13V when posting is because the bios kicked off as it should and the HDD was happy whit what it saw and spooled up.
Most old PSU's do require some minimum load on the rails to give the right voltages.
But I know you already know this but for anyone else this could be good to know.
I hope this explains the problem.
But yea RTC problems can be annoying at times doing unexpected things.
Anyways love your videos and I hope the explanation did not come off as "I know best", just happened to have similar problem on my last SS7 build and did dig pretty deep in to the CMOS rabbit hole XD
Absolutely love this response ! Thx a lot.
If I still have the PC in the coming days I might hook up my oscilloscope.
I would just double check any capacitors that are in circuit with that battery to be safe. since it works on subsequent boots, it's obviously looking for a charge on an associated capacitor, probably to safely switch the CMOS ram safely over to main power, and the capacitor isn't able to fully charge on a weaker battery. It might be by design, but it could also be an indication that capacitor is old.
It’s more than likely the other way around. The coin cell and main 5V rail are combined after their own respective diodes, and residual charge in a cap somewhere on the CMOS V+ rail allows for a successful warm boot, but isn’t quite up to charge on a cold boot.
It is quite likely that the 5V system rail is fed through a current limiting resistor to support safely charging rechargeable cells, since most motherboards had provisions for both. That would form a low pass filter, meaning the downstream cap would take some 10s or 100s of milliseconds to charge to a suitable running voltage from zero. Ergo, cold boots fail.
@@nickwallette6201 chokes are for that. I would suspect capacitors with a high ESR.
@@BenState An inductor _in series_ would limit changes in current, yes. But so would a capacitor to ground after a current-limiting resistor.
Usually whatever capacitor is in-circuit to a CMOS battery wouldn't be an electrolytic. It's likely either a ceramic or tantalum, so typical "old capacitor" failure modes aren't as likely here.
As a fan, I'll watch any of your videos in any shape or form. I just find this so interesting.
Not too odd, the Board probably holds timing configuration in the bios memory and fails to start reading the values, on the second attempt these values are refreshed from main power and the system starts.
That doesn't explain the voltage drop.
@@SomeAngryGuy1997 The voltage drop probably was just a red herring. Some part of the system drawing a bit more power during initialization. 4.95v is totally fine.
@@thetaleteller4692 Yeah, for ATX the specification is 5V+/-5%, IE anything between 5.25V and 4.75V is defined *fine*!
I assume this is an older AT style PSU, I can't as easily find the limits for those but I very much doubt it's tighter specs than modern ATX powersupplies, so 4.95V isn't an issue.
The "blinking keyboard LED" kind of hints the BIOS goes into some kind of crash loop, so yeah, definitely possible it's drawing a bit more power than it will once it gets things set up correctly.
@@thetaleteller4692 Yes. I kinda sorta guess the HDD not starting doesn't load down the 12V line enough and maybe this PSU uses the 12V for regulation, who knows. Maybe it uses a resistor network to average all the rails to do some analog math, so loading the 12V will cause the 5V to go up. I have seen PSUs do that.
The "off the cuff" style is great. 👍
yeah it sounds more authentic than this voice-over thing...
Changing the battery is the very first thing I change out when I get an old computer in my shop. Some of these old computers are not all the same and it never hurts to replace them right away.
Audio quality aside (really not a problem), I like the fact that the voiceover took a backstep. It feels like you are more "connected" with the video. Maybe a mix of it in the future as voiceovers do have a place and function. Great job and keep it up.
Goes back to the basic rule of debugging "fix the most obvious problem first" even if it seems stupid to do so..
There are motherboard series that only work with a cmos battery. I've seen quite a few in my career. You don't think about that in the beginning, but an empty cmos battery can cause a lot of problems. A point to keep in mind when working with old computers.
It's still interesting though that the board would post if you power cycled it quick enough.
I forget that some systems are quite quirky without a good CR2032. I had this "cheap" board which had a (dead) CR2032 externally wired to a clock chip (not DALLAS). It would POST but not much more than that. I like the boards which actually have these coin battery holders as standard, makes swapping out much easier.
What an odd issue. Thanks for sharing. The recording works great by the way. The live voice-over sounds nice, in fact it has (at least using headphones) clear directionality, coming the voice from behind, which is unusual in an interesting way.
I've had a very similar problem with a same era gigabyte board. The voltage on the cmos battery was enough to maintain some if the cmos data, but corrupt the contents. I had to force reset the cmos using a jumper to clear for a few minutes, put a new battery in and all was good again.
Great video and I'm glad you found the problem. I'm dealing with an IBM PC 300GL that boots, but after 3 to 5 minutes quits receiving keystrokes and emitts a high frequency noise that comes and goes. Once you start hearing that noise the computer basically locks up. This does it in DOS or Windows. Swapped ram, cpu, power supply, and video card and still locks up. Anyway, thank you for the cool video.
Hi Sir, regarding to this, I have a mobo in my stach (dont remember the brand/model) that NEVER boots if CR2032 is empty/low power/missing
I took some time to trace the problem some years ago, and you now present a very similar problem
An older series of Dell computers that I worked on had a similar issue. A dead CMOS battery would result in a seemingly dead computer. Ever since that repair, I've always replaced the batteries in computers more than a year or two old, just to be safe.
There is a pin on the IDE cable that can hold the drive in initialization, before the spindle starts up. I have seen controllers and motherboards keep the drive from starting up, and unplugging the cable allows them to start normally. It’s not “normal” because the controller should be initializing the drives, but if the system isn’t booting, that would be one potentially expected outcome.
Understanding that, there’s a good chance the computer doesn’t have much, if any, load on the 12V rail, and the PSU may be having trouble regulating the unloaded 12V rail when the drive isn’t spinning. The PSU doesn’t have independent control of every rail, so that could be why the 5V rail was a little low when the drive was shut down.
That’s all very characteristic of an older switching supply. Nothing to worry about there.
Having the IDE controller stuck in a reset state would cause that chain of events and perfectly explain all the symptoms you see at the beginning.
Crazy--this is the 2nd time this week I've heard about a vintage PC not booting due to a bad CMOS battery.
Some of the interesting weird behaviors we find in motherboards sometimes can really get our attention and drive us to investigate.
A few minutes after watching this, the new voltage regulators in my Abit AB-SM5 started to work, the Pentium MMX 166 MHz is finally getting hot again ! No POST codes for a while but after I probed the regulators I started to see POST codes and finally a boot screen.. so thanks for the video .. I think it gave me some luck !
It was the auto voltage settings being lost by the dead battery. Thats why it worked after a quick restart, just long enough to remember the settings.
Hey!
That’s the same issue at the Macintosh LC. When the battery is empty you had to turn on and off twice to turn the computer on.
Regards.
when i was in high school back in 2007 i worked as a school based trainee working along side the schools computer tech and the most common thing id see on computers from the socket 7 to socket 370 era is that if the battery was low or dead then the computer would provide almost the exact same issue to what you demonstrated in this video and removing or replacing the battery would resolve the issue every time
I was sure it was the battery as soon as I saw the error on screen. I have seen this before, I have a few motherboards that either refuse to boot at all, or act weirdly at boot if they have a discharged battery.
Like the transistor and Capacitor could be faulty between the CPU and Bridge.... It's a very rare issue that the CMOS battery caused this. Could be corrupting the CMOS chip.
had a packard bell 166 mmx back in the day did weird things like that if the 2032 was almost dead, wouldn't initialize HDD, false start like this one , or it would boot without video, or boot normally if the energy was right, noticed my settings were wonky, not even a low battery warning, changed the backup battery out and problems went away
I have never seen a motherboard that was a that picky about it's coin cell battery. What a strange error/problem indeed. Glad to see you got it back up and running again. Have a good one.
Love those old full towers,we were full of dreams back them. Never seen a fully populated one
I ran into machines that wouldn't boot due to a bad coin cell. If it was borderline, the bios may have been corrupt. Removing it and reinserting it would sometime help. Removing the cell completely would also verify it was the cell
I had a similar problem on Asus P4P800-VM.
I always check the battery first when the computers are not booting because I had this issue before.
I once had an old Shuttle Nforce2 motherboard whose bios battery always drained after a few days when not plugged into the wall. It was a real pain to replace so many batteries and reconfigure the bios. I eventually took it out of the computer case, and noticed there was a sort of sticky substance on the motherboard near the cr2032 battery. I washed the substance off with alcohol and water, and after it dried a few days, it worked perfectly. No more battery drain. There must have been a very minor short to ground from the sticky stuff. Might've been flux or soda.
As soon as you mentioned the battery warning I was constantly yelling at the video, "It's the battery!". Been there done that.
that was a nice simple repair. i have to catch up on these videos i always enjoy them so much
This is a very common phenomenon with various laptops, even much more modern ones. Some appear dead when the CMOS battery is low. Off the top of my head, various Sony models were known for doing that. I've seen it happen with desktop motherboards before, but it is much more uncommon. Some are especially weird and boot fine without a battery installed, but refuse to boot with a low battery like 1V.
Wistron-produced Acers (7551 ,5738,5536, 5740, 7736, 7738, 7740 and there might be a lot more than I know) will definitely NOT boot with no CMOS or low voltage batteries. I've been given a lot of 5738 and a few 5536s, and most of them weren't POST-ing due to that CMOS thingy.
@@NSHG Good to know!
What a beautiful machine!
I have had several motherboards which require a good coin cell before they would power on, i spent hours troubleshooting to determine this!
My DOS machine is a P133 MMX. 384 megs of ram. 2 gb compact flash, dual channel storage. It's so overkill, I love it and getting so much ram into it probably slows it down a little, but it makes me cackle.
I have a really pimped out Mac iici too. I just love retro computers and games.
I love how massive your tower is there! 😍 Thanks for the video!
It's fun to completely pimp out old retro systems. I have a PowerMac G4 400 with 256 MB RAM, running MacOS 9.2. Originally it had 768 MB. I doubt the OS uses more than 8 MB.
Wow I just found a Belgian RUclipsr who makes videos on retro tech as well.
One with 40K subs too!
Awesome
Welkom :)
I've seen similar behaviour on even much more newer systems. Some don't even boot (cold or warm) at all without a full battery.
Could it be that the old coin cell battery is shorted? Maybe the problem was not due to the battery being totally discharged but to it being actually faulty, it's a quite rare occurrance but it could happen that those coin cell batteries develop a short circuit internally.
I had a couple of old XP machines that would post, but not boot into the o/s due to a dead CMOS battery. Apparently Windows didn't like the discrepancy between the original BIOS date and the o/s date. Once I replaced the battery and updated the BIOS time and date, everything was fine. The CPU cooler on that MMX looks so tiny and dainty lol
One possible temporary workaround may have been to boot to another OS via external device (DOS is ideal), change the date, then warm reboot until you had the battery replaced.
Thanks for the filming of the investigation of such a nontrivial culprit. Btw, I'm fine with this kind of non-scripted video.
i'm pretty sure my lga-2011 pc had this issue, it would get stuck in a boot loop, i thought it was a ram issue because i reseated them and it turned on... then i thought it was a psu issue and i replaced the psu... still did it, thought it was caps so i bought a new mobo but then i read a forum post about a guy having the same issue and he replaced the battery and guess what happened? it boots normally... i did the same and guess what happened? it boots normally... now i have an extra computer that i will turn into a NAS or HTPC.
Got a similar thing happening with a am2+ motherboard so even on more modern stuff its an issue, The Asus M3A78.
I have some PCI+ISA 233MHz Pentium and AMD P75 boards that once they warm up they won't re-boot without a full power down. Good find on the battery, I would not have thought it would cause such an issue.
I've seen this happen with moden(-ish) motherboards too.
I would have never expected the battery to be the culprit. I had some issues with a pentium 2 system and it required replacing 3 caps on the motherboard to fix it. It would only boot at the third attempt, the first two times it would beep with a sort of alarm. That is a chaintech board with an Award bios. I don’t mind the off the cuff style, I tend to work without scripts for most of my video’s.
ya seen that on few boards refuses to boot if the batt is dead must be a bios thing
Interesting, I haven't seen a PC before that won't start without the CMOS battery. Although I haven't really had a reason to remove the battery on a "modern" (e.g. post pentium-era) motherboard.
I have seen it on a Mac, my Powerbook 1400c won't boot without the PRAM battery connected. In fact, it won't boot if the PRAM battery is flat. It took me a while to work that out when I first got it home and it wouldn't boot. Powered up, nothing. Tear the thing apart to get to the PRAM battery and it was fully discharged. Wouldn't power on without it either. Put it back together and still won't boot. But I left it plugged in and a while later I'm startled by a loud "BONG" sound, which was the classic Mac boot chime. Turns out you just had to let it charge the battery for a while before it would boot.
Anyway, I don't mind the off the cuff style. Not quite as polished, but I'd say it's definitely still entertaining. About a million times better than I'd do, considering how bad it was when I tried to record a narrated screen capture how-to video for work.
I was thinking that someone put a jumper on the reset button header, and for some reason, when executing the power-off and power-on sequence fast enough, the PC would just start. But that's a good finding. I've seen some other motherboards that do the same. Really annoying
This may sound crazy, but I've seen coin cell batteries get fried after a brown out. Scorch marks and everything, replaced with new. It was a dell optiplex with a core i7 8700. Optiplex is a real workhorse! I've seen many a computer with bios battery messages at boot and/or you have to power on twice...nothing I ever bothered trying to fix beyond trying a battery...in those cases new coin cell didn't fix. It's always old stuff for the most part.
I have the "Baby AT" version of this case. just built a Retro system in it. weird thing is that only the power LED works, likely might have to grab my spare leds and replace them, but still a good solid case, even if a bit yellowed. EDIT: I actually have a variant of this board as well. though my setup runs a k6-2 400.. wow
I have an 98 machine that is doing the exact same thing but it does not give me a low battery warning. That being said, I will change out the battery to see what happens.. Thanks for sharing .
Hot-plugging IDE cables? That is a quick way to cause intermittent errors, chipset or IDE controller errors on many old boards.
Would be nice if you could merge the stereo audio streams. The balance between the left and right channels varies per shot and it can be a little disorienting to the ear.
I agree. You need to merge stereo to mono, otherwise it's unnatural. :)
A nice little feature I learned about Windows 10 (and I'm sure 11 also) a while back was the "Turn mono audio on or off" setting. Go to search and type mono and it will come up. It makes listening to videos with headphones much more tolerable when they have uneven audio or sound on only one channel. Just remember to change it back to stereo afterward.
yea i wonder what was the resistance of the battery the voltage going low than high sounds like an odd short
I'm guessing the IDE cable itself is broken inside and moving it just creates a connection or possibly the ID channel it's plugged into is somehow faulty itself it's only two things I can think of
I would think the BIOS settings got corrupted and when you removed the battery the actually defaulted. I would have tried booting with no battery to make sure.
The keyboard lights in that PC act as a diagnostic code. Check online what the blink pattern means.
many machines, even now require the cmos battery to be working in order to boot.
Would it start with no battery? Not sure how scrambled CMOS would dip the 5V though - I have had one case of scrambled CMOS no boot when I changed a low battery fast enough that the CMOS only partially cleared.
I thought the fitted Intel HSF only applied to the overdrive model
It wouldn’t be caused by scrambled data particularly. Most motherboards will only reply with a pwr_good signal (to the PSU) if every voltage is in fact within margins. With some motherboards, this includes the CMOS battery.
I’ve had the exact same issue with either the:
- Acer (MicroStar) MS-7326, or the;
- MSI 945GCM7-F MS-7507.
It’s a 775 motherboard, but the issue remained identical. Despite the difference in era.
Great vod @Davy!
I'd have thought that a modern CMOS would have a simple checksum to ensure the data wasn't scrambled, and just reset to default values if the checksum didn't match. Maybe I'm assuming too much though.
@@zoomosis data isn’t scrambled in this case. The PWR_GOOD signal isn’t being reported as “good” to the PSU. That’s it. :)
It cannot be stressed enough, weird behavior in POST (in notebooks as well as desktop pcs) is very often associated with a weak or empty cmos battery. Always check for it before taking further measures like disassembling the whole thing.
I've seen that before too.
My Pentium 133 has an issue with detecting a keyboard, whatever I do, it wont detect my keyboard, tried a different one, same issue. I’m kinda at a loss as it worked perfectly before.
I do a simple test that almost work on all motherboards back in the day... When no image i remove the battery and test if no little tickle on my tongue between + and - then i realize that battery is run out, i replace and always work. Is a ultra quick test when you tired to use tester
Perhaps the coincell has shorted internally? Did you measure for that?
I have that exactly same case from 1995. I built 90Mhz pentium inside it back then.
I came across to some Acer laptop few years ago, it came to me for diagnostics because it was only blinking it's lights and nothing else. All it needed was also brand new CMOS battery and it came back to life after that. Seems like some machines need working battery to work properly, some machines do not care at all.
I had the same issue the other day with a Gigabyte motherboard on my main PC, guess those boards are just built that way
Does it start without any CMOS battery? That'd be interesting to test.
That would be a good idea to test that.
I had quite a few mother boards even in the xp or win7 era which did not boot with a dying CR2032, since then I always swap the battery on old motherboards I get.
I had this issue back in the days of P2/3 (I just remember no boot) with some Tomato motherboards (and what made me think of them as low quality...yeah, they were bought because they were cheap). So when I saw the 1st symptom...I was "how's the battery", then when I saw the warning on boot, I skipped the video to find if the battery was indeed the problem. Sorry for the "low watch time", but I would have changed the battery 1st, before removing from the case.
I had a similar fault on a more recent system. Would not boot. Had a power up shut down cycle. Checked the voltage of that battery and it was 3v. A fresh one is 3.3v. System booted when I replaced the battery. I dont remember ever seeing this before In the 20 years I've built/repaired/used computers 🤷🏻♂️
Maybe some quartz generator on the motherboard need full working battery during POST?
Long time ago i experienced similar issue with my Asus laptop : dead battery = dead laptop (only fan noise)
Sorry for my english :/
Great video, nothing wrong with going unscripted. I had a Win98 PC that started doing this. It also turned itself off and eventually wouldn’t turn on at all, even with a new PSU. I should have tried changing the battery before I sold it for spares.
How about they capacitores of they mb?
Easy fix is always good!
The audio......... Maybe those dedicated chest mics...
Yes, it's very strange the mobo needing a new full coin cell.
Do you have ebay or place where you sell your things?
I have a Desktop version with that Case Design. In my PC is a Pentium 166 Mhz with MMX 16MB RAM, 1,6 GB HDD a 2 MB ATI Mach64 and also a Sound Blaster 16
I found that PC on the street and it works with Windows 95 and Office for Windows 95
Just out of interest, does it boot without a battery?
Interesting, ordinary battery and many problems. Without a battery computer boot or not...?
Sorry to say this, but I don't understand why you didn't replace the battery. I've seen so many mainboard reacting weird when the battery is low.
I was thinking the same thing. Until the CMOS battery is replaced, it's bricked. Same thing also happens on old Pokemon cartridges
My guess is, to show the people that are new into retro computing to *always* check the battery first. I would bet the battery was the first thing he tested, then he thought it would make a good educational video that could potentially save a lot of time for some people.
Because it's RUclips and he needs to pad videos out for the views and extra algorithm sightings 🤣
Kidding, I like the diagnosis procedures
EDIT: next thing retrospecter needs to do is go all this video is sponsored by or put a shit ton of ad triggers 🤣🤣🤣🤣
@@adventureoflinkmk2 And what is wrong with that?
@@andreasklindt7144 nothing, I was being sarcastic/smartass
Were Are You Selling that ? for how much im in USA?
how about no battery?
What voltage was the old battery?
Retrospector78 Is the pc for sale online ?
Years ago, a tech at PC Doctors in Huntington WV told me that he made more money from bad batteries than any other hardware problem. I have no reason to doubt him. Over the years, I have had several 486 and Pentium motherboards refuse to post without a battery or with a failing battery, all of them had CPU voltage auto-detection or some other "new tech" that was cutting edge at the time. Enough of them that for a very long time, I refused to buy any motherboard until it was at least revision 2 or higher.
To me, it sounds like the 5v rail might have bad caps. The fact that it takes quite some work to get it to provide stable 5v tells me there is a cap that has trouble charging up, but once it does, it starts working fine.
Had the same motherboard, had to do similar troubleshooting. Took me awhile to realize that the battery is the cause of the problems. It's kind of amusing seeing someone else struggling as I have been.
Very weird... The only thing I can imagine is that it is somehow connected to the auto voltage selection that goes bonkers when the battery is low and might, due to some overvoltage protection, short the 5V (which was used for the VRMs of old computers up to the Athlon and P4) partially to ground.
By the way, the 430TX chipset has a cacheable area limitation: 64MB. Anything larger than this will become awfully slow, when accessing these RAM areas. This is why I chose the 430HX back in these days, despite of this still using the PIIX3 southbridge without UDMA (my HDDs were PIO back in the days anyway, but I was aiming for large RAM), whereas the 430TX already had an UDMA enabled PIIX4.
A word of caution about 430HX motherboards - some cache the lowest 64MB, others cache the entire 512MB, it depends on the size of the tag ram on the motherboard. IIRC many motherboards support either config, just swap tag ram chip and move a jumper. So likely solvable but still a "check this" item if you plan to have more than 64MB on an 430HX board.
IIRC my recollection is that there were some motherboards that used the 430HX system controller (but paired it with the 430TX controller chip (PIIX4/)!, I assume Intel didn't officially support this but it provided the best of both worlds, 512MB cacheable memory and UltraDMA (DMA mode 3). I've found a few forum posts from people who claim they have motherboards like this which suggests my memory might be real.
It would take a pretty large short to drive it down to 4.9 volts. It would think it be enough that the heat would be noticeable if you were looking for it. But the AutoVRM on those boards is absolutly the culprit non-the-less. The ones with the Dallas clock boxes on them don't post when the battery runs down.
As for the 64mb limit. Its not really _that_ bad when you have cas2 ram timings. Its a deficit of about 10-15% performance for things exclusively ran from those blocks of ram. Windows itself will reside in cachable range and most applications will be small enough to be mostly resident there too. On computers with about 128mb of ram the vcache will be taking up the bulk of the upper 64mb anyhow. And this is even less an issue if you are running 96 or 80mb of ram. Now if you pop 256mb ram on them... and then start to do a bit of more intense stuff, then its going to have its effects.
Memory specific benchmarks will show a much larger difference than is actually realized in practice.
I have a Gateway Pentium D system (GX7022E) that will do exactly the same thing if the CMOS battery is low. Over 15 years old and still going strong, but it likes to eat power supplies.
Mmmm yummy power supply mmmm
maby stupid question are you from the netherlands want the windows 98 is in dutch
I have a 486 motherboard that only boots after pressing the reset button on the case. Seems to clear a power good signal or something. But after seeing this, I will double check my CMOS Battery!
This wouldn't be my first encounter with such issues. I had this more than once - a dead CMOS battery should usually be fixed as a first try, since it's a matter of seconds...
BTW, as feedback, I watch YT on a TV with a 5.1 receiver switched to Dolby Pro Logic II, so I get the voice centered. In this, your voice moves and sometimes even comes from the back (which is kinda weird), but is almost never directly from the center. Maybe that's something you want to tackle in your recording setup...
This is a perfectly known thing. For example, it is the same with old laptops. I always change the BIOS battery as the very first thing. :)
For how much do you have it for sale? Thanks.
Try to connect 5v to oscilloskope
Thanks for nice content.
I don't think that is uncommon. I've ran across multiple PCs and laptops that refuse to POST or on on occasion it would fail to even power on-with a dead CMOS battery.
I've had this issue with Socket 478 and 775 motherboards over the years. Failing to detect or boot from IDE drives when the CMOS battery is marginal. You may not get a battery warning from the BIOS but it's still problematic for some reason. Something to do with IDE detection or reading the stored settings from the CMOS to boot correctly. Replace the battery with a new one is standard now with any board I come across that's more than a few years old 😃