Was it worth 27 repaired traces? Trash to treasure 486 VLB build. (am5x86 overclocked)
HTML-код
- Опубликовано: 23 фев 2024
- Severely battery damaged Shuttle HOT-419 R1 Motherboard.
Found at Computer Reset repair video here: • 486 VLB board found at...
PCB Prototype the Easy Way. Full feature custom PCB prototype service. www.pcbway.com/
-
Support me on patreon.com/Epictronics
Join me on Twitter: / epictronics1
-
Tools I regularly use
DeoxIT D5 Contact Cleaner
Hanstar 861DW Rework Station
Pro'sKit SS-331 Desoldering Station
UNI-T UT61E Auto Ranging Multimeter
UNI-T UT890D Manual Ranging Multimeter
MESR-100 mk2 ESR meeter
PINECIL Soldering Iron
PinePowerPSU
TS-100 Soldering Iron
AMTECH NC-559-ASM Flux
Kester 951 Flux pen
MaAnt Grinding Pen
Multicore 60/40. 0.38mm and 0.5mm solder
TL866 II Plus Programmer
RIGOL DHO800 70MHz four-channel digital scope
Tektronix 2246A 100 MHz four-channel analog scope
InfiRay P2 Pro Thermal Camera
PCBs from PCBWay.com :)
--------------------------------------------------------------------------------------
patreon.com/Epictronics
Join me on Twitter: / epictronics1
Music by Karl Casey @ White Bat Audio 
Наука
Almost certainly the buck regulator not being able to maintain its output voltage under the full load of the CPU. Notice how the voltage jumped up when the board would hang. I would try a different buck converter. I would have also tested the output voltage with a dummy load while setting it.
I'm gonna try with external power for the next attempt
Also, use that oscilloscope some more - reading the CPU voltage with a DMM only is near as useless, you need temporal information - meaning: does the supply remain stable or does it dip and bounce faster than any DMM is capable of showing. Also, that dummy load would also be of assistance to check those pretty crap DC/DC converters; how much current can it supply before limiting?
That socket blaster from scrap computing is a real nice addition to the retro community
Flashback to me building my VLB computer back in high school. I can remember that whole platform was hell. Finding VLB cards that worked and could be overclocked was a nightmare. VLB video cards were so unstable as soon as you stepped away from stock clocks and VLB controllers were worse with corruptions being a scary reality. First time I played with overclocking a computer. Not too mention I went down the rabbit hole of DX2/66 vs DX/50. Oh those good old days.
The CPU jumper settings for the 5x86 with L1 write back (as it's designed) is the P24D jumper settings. Bus should be set to 33MHz and yes the buck converter isn't helping you at all!! Lol! It requires 3.45 +/- 0.15 to operate correctly. Hope this helps
Thanks! I'll try those settings next. I'm also going to use external stable power in the next attempt.
It's a crime why this channel doesn't have at least 10X number of subscribers. First thing I watch every Saturday evening.
Thanks :)
Indeed!!
I had a 10 5.25" bay full tower back in the 90s.. got it from a friend who's dad worked for the state.. it was STUPID how massive that thing was.. no 3.5" bays, had to use an adapter for the 3.5" obligatory floppy drive.. it had this silly removable plastic rectangle that went around the base to help keep it from falling over it was SO top heavy. What a wonderful time to be alive back then, I've said that before on your videos. I miss those days SO bad, not only for the computers.. it was just such a more innocent time. It was HARD to get products, you had *well I had to go to computer shows, we had one close to me at an old horse racetrack in Cherry Hill, it would be there a couple times a year. I'd go every time and buy stuff.. used, new.. working, not.. ram, cpus.. oh the good days!!
Edit: I will say... like you're dealing with finding manuals for the motherboard jumpers.. I clearly remember back in the good days, you really did have to know what you were doing, you could easily get taken advantage of buying things if you weren't 100% sure about the system you were working on.. you could EASILY mess things up and be quite devastated..
I totally agree, what a fun ride it has been. I'm building a modern PC this week, and I have to say, it's still fun :)
@@Epictronics1 it is still fun, it's just changed a lot! I think plug and play was the major change, honestly. It is exciting, i put an i9 9900k together a few years ago for a friend who was an apple only guy forever but his kids wanted to play games so he finally crumbled. Anyway they love that system, and i put together an AM4 machine 5 years ago and have pretty much upgraded the entire system since, for myself too. I actually think it's great that they've become so much more user friendly AND just about anybody can build one today with a very simple skill set, if any at all!!!
Yeah those computer shows were a lot of fun. I wasn't old enough to buy anything but my dad and brother would take me to them. The only problem was my dad thought he knew more than he did and there was always a few people scamming the crap out of people like my dad. We tried warning him so much but he never listened.
@@adampope5107 hopefully he learned, but yeah it was a pretty rough climate at the shows.. especially if you got there late in the day.. I was born in 1978 so I had to go with my uncle, but I was heavy into PCs back well I started in like 88 building my first system.. I was hardcore back then, still am today though, but wow how times have changed!! Great memories!
I remember reading back in the day how hard the early 486 MBs were to run at 50MHz. This video shows the reality of it. Great to see.... not that I wish fighting a MB on anyone. It explains why later 486 processors never released with 50MHz bus support past the first round of 486s.
Enjoyed this series. I can identify with middle of the road hardware like this with mismatched colors. Definitely what most people's situation was back then. Most people were neither Packard Bell nor IBM, but somewhere in the middle.
Trying to match jumpers without the manual that shipped with the board was always a nightmare. Not counting the built in errors of changes relapsed to production before the manuals would get upgraded. I would shoot for faster RAM chips. I remember when the company I worked for tossed some mainframe memory cards due to production issues and we salvaged 25 ns chips for our home pcs. Not that we could ever overclock out systems to push those chips. Good luck and thanks for another great video.
25ns RAM :o I didn't even know that existed back in the 90s. I may have to solder 72 pin SIMM slots to the board just to give myself a fair chance to find 60 ns modules
Oh man the nostalgia, lol. I had the small tower version of that case.
I’ve got the mini tower (my favorite quintessential 90s AT mini tower!), and the mid tower version. I would love to have the full tower too.
That said, the full tower isn’t very practical. It has a lot of 5-1/4 bays, but no 3-1/2. The mid tower has 3-1/2, so you don’t gain any space if you use smaller drives. And the inside is so big you need really long cables, but it’s not like you can just add more cards on those little boards. ;-)
Still want one though. Haha
If its a 5 pin regulator, the most used are the MICs that are also used on the chicony 486 VLB boards. You only have to check if there is a jumper to set 3.45 / 4volt.
If its a plain simple circuit maybe a fix 3.3 volt one can be used.
If its a 4 pin regulator Sharp is mostly used.
3 pin regulators would have more options.
Thanks!
Forthcoming video on an all-new build? Not the usual stuff, but I'm in!! Maybe precisely because of that: I'd like to watch your take on that kind of thing.
It's video ram speed. Great job! A great save! Be happy at 150 MHz!
Thanks!
I had this motherboard and it cost me a lot of nerves :) I didn't have as much patience as you and sold it :)
There's a very good chance that it's the voltage. While undervolting a CPU can make it run cooler, it may not have the power to run at a higher clock speed.
I'll try with external power in the next attempt
Many VLB video cards don't work with bus speed over 40mhz. Some of the Curris logics 542X and S3 VLB cards tend to play nice with 40 and even 50mhz bus.
Remove the wires at JP64 and solder Jumpers.
The position now is 5V, the other position is 3.3V
Then measure Q2 directly above. There is 5V, GND and a connection to JP64 pin 5/6.
Now solder a 3.3V linear regulator there. Think about it twice as you maybe want a heatsink.
C20/C21 are regular decoupling caps. 6.3V/10µF or similar.
Newer revisions have more space for the heatsink as in this release it's a bit limited!
Thanks!
check the ISA bus clock divider in BIOS. if the mainboard was configured to use 33/66 mhz cpu, then most likely the divider was set to 1/4. try set it to 1/5 if available.
I ran out of time. I'll check those settings in the next attempt. Thanks
Back in the mid 90's, I had a VL-Bus Pentium board. It came with a 60 Mhz first gen Pentium, complete with the math flaw. I had a Tseng labs VL video card with 2MB ram and this super cool Tekram caching VL bus disk controller. I think it had IDE and floppy ports. It had 4 30 pin memory slots, I believe I had 4MB on it. Unfortunately they are all lost to time.
Interesting video. Two days ago I mounted similar system. I used AMD 5x86, FIC 486-VIP-IO2 with PCI and VLB slots, 64 MB RAM and Trident 9440 VLB video card. I could not find jumper settings for CPU but easily found settings for 150 Mhz (3x50). Trident is from my first computer so I like it because of nostalgic reasons and it is my only VLB card which can work with FSB 50Mhz. I don't have voltage regulator for this mobo so I created my own but system was unstable so after long fight I replaced all electrolitic capacitors and few sets of cache but now everything is fine. I must add my old Sound Blaster AWE 32 and CD-ROM also from my first PC and then I will be happy.
Remember the case in our university the server was called IDEAL486! :D
Well, It's ideal if you had the space for it lol
That case is simply awesome. I would like to have a couple of those.
You'd struggle to find room for them lol
Congrats on rescuing this board and managing to even overclock it!
Seems that buck converter can't supply enough current which is why the voltage is dropping, once the CPU crashes the current draw drops and the voltage rises again.
Nice work! I have a 486DX2-66 and needed a heatsink. I found a 3D-printable heatsink holder which took a 40x40x20 heatsink, it sits between the CPU and the socket. It's a very good solution if you can't get a 486 heatsink.
Thanks! Good to know there are 3D-printable solutions out there
Try replacing the pot on the regulator with an appropriate resistor it wont be adjustable but more likely to be stable.
That case is hilariously big. Love it! 👍
A DX2-66 with a descent VESA graphics card ran Doom very well back in the day, so you may wish to try that, an ISA card will kill your fps, at that point in time a card with an S3 chipset even a cheap one were good and much faster than the Cirrus Logic cards, the only issue you may run into with the cheap s3 cards were if you had a com4 - a couple of things to note are I have never seen a VESA card run on a 50mhz bus, the CL card was probably running the ISA side of the card as it started, but as soon as it displays a picture it will bomb out- back in the day if someone wanted a 50mhz machine we always used to steer them towards the DX2's because of problems running the local bus at that speed, so to mitigate it, it meant you had to run your IO and Graphics on a slower ISA bus so what you gained in some ways you lost in others, not only that the ram had to be faster too, I did a lot of experimenting back in the 90's, 40mhz was doable on a lot of boards but it would inevitably lead to hard disk corruption,, a lot of boards with the AMI WinBIOS could be temperamental while in the BIOS and it was common to have experience lockups while in the BIOS, but were stable in normal operations our till had one in for a number of years.
This was a great video. Really enjoyed this one. I hope you manage to hit at least 160mhz. It would be nice to see a 486 run stable at 200 but the only 200mhz 486 I have ever seen was on CPU Galaxy and he was using a peltier cooler with a really loud fan. I don't think anyone would want to run one for a deployed system. Fun for overclocking but not really for regular use.
Thanks!
Perhaps more capacitance in the circuit to stabilize voltage output would help to attain a higher overclock, but that's an awesome improvement!
Thanks. I'll check the ripple and noice with the scope
@@Epictronics1 I'm sure I've seen someone else on RUclips have this exact issue and it was down to a sub-par buck converter and bad ripple, they added caps to smooth it out but, in the end, got a better quality buck converter due to the voltage droop.
Another Tower! Haha I wasn't expecting it so soon. Good work, it looks awesome.
I scored an Aopen full tower the other day from marketplace too, have to work out what to build in it now (perhaps a sleeper with my desktop? It's not the Inwin I'd love for that but beggers can't be choosers haha)
@michaelscarport awesome, I'm still using an Antec P183 as.my daily case, I thinks my tower is an HX45
Thanks!
@@Epictronics1 No worries, I always enjoy your videos.
I just got lucky and the vlb cards I owned worked fine at 50mhz. The video card is a Diamond Stealth 32, which I've never seen another one. The IO card was some generic card, and no idea what motherboard, cache ram or random 8mb 72 pin simm I had. Somehow it just ran and was faster at 150Mhz than 160Mhz in the games I cared about at the time.
Yeah, the thought crossed my mind. It's probably faster at 3x50 than 4x40
it may be your patch Wires
in a clock cycle Wire length Can alter the time it takes to Travel and shape of a bit (Wire Capacitance & ohms Residence can mutate the bits)
Love these videos!
Thanks!
Great stuff absolutely love it!
Thanks!
Nicely done!
Thanks!
Same problem I had with Socket Blasters. The buck modules they're designed for are junk. Going to design my own regulator pcb to use with them.
Do you plan to implement the new regulator on the Socket Blaster or use a separate PCB?
looks old fulltower case some early 90's
That's a big case! I'm not sure I am a fan of that fan controller: the fan will start when the CASE of the PSU is hot? Uhm... Anyways, nice job as usual! This reminds me my recent journey in retro PC with the difference that when you tweak/overclock this is expected I guess! Thanks for the video!
Thanks Tony. Yeah, fan controllers were pretty primitive in 1994. I have a similar solution in my Commodore PC-35 :)
@@Epictronics1 it feels like it's starting the fan when the PSU is already overheating! :)
The takeaway here is; it worked well enough in the past to keep the supply from exploding in it's trousers - so it surely is good enough today.
nice time spending sir
on those video cards that have artifacts try upgrading the video RAM to a faster speed like 60ns or 50ns. Also vlb cards often have some jumpers that you may need to adjust.
I was just thinking the same thought!. I'll check how fast those VRAM chips are.
Thanks! Your use of the osiciliscope helped me fix my own 486 project last night.
Also, what is that mac sitting on top of? I dont recognize that accessorie.
Great! That is the hacked mac from this video : ruclips.net/video/Ug_E9a_HtfQ/видео.htmlsi=mtlkfLGOrdt7xTQj
Artifacts on a modern GPU is usually memory. Also not enough voltage at the memory makes artifacts, ie, we pulldown supply to see which chip causes similar artifacts.
that thing outside power suply is not a sensor if i am correct it is just a thermostat and depending on a type it is either closed or open and nothing in between
check the themperature of it working, should be engraved somewhere on the device
next challenge installing a am5 processor via adaptor
I don’t know about this particular board, but there are often timings in the BIOS that can affect success at 40-50MHz (but I think the cheapo buck converter might be hurting you too. I tried to use the same one in a project this week and failed miserably)
Yeah, the buck converter is at the top of the list to replace for the next attempt. In fact, I'm surprised we got to 150MHz at that voltage :) I ran out of time to test with other timings. We'll try again in a few weeks
As buildzoid would say keep calm and Raize Vcore.
i played with those DC to DC boards before, wasn't impressed, i would try to get something more stable while underload
I'm gonna try with external power for the next attempt
Check your video cards, maybe they have jumpers(something like wait states)
Cirrus was quite a good card for DOS-gaming. Used 5422(ISA) and 5428(VLB) and both were fast especially with some overclocking w/ MCLK. So pure pixelpushing-speed was not a problem. Had to switch to S3's only because GUI-acceleration (drivers 🥵) for Cirrus-cards weren't very good in OS/2 or W95. I'd still choose Cirrus for DOS-era gaming rig, but preferably later 543x-series PCI-one?
from what I have read when was building my 5x86 system, vlb at above 33MHz is a bit of a crapshoot. Many VLB cards wont go above 33MHz bus
(Opti chipset on mine too)
This 5429 seems to be a golden sample :)
Looks like that regulator module just doesn't have enough oomph for the CPU. Maybe try another one. I wonder how awful would that supply voltage look on a scope.
Yeah, I realized too late. I should have checked the ripple and spikes on the scope
50 MHz on the VLB will not work with any GPU. Try some higher V7 Cards e.g. the Mirage or Mercury. If I remember correctly, 50ns RAM was sold for these machines back in the days.
That power supply has a very bad bodge job. That sensor should be on the inside attached to the heatsink, not to the case.
4:00 It's pronounced like foo-you. They're infamous garbage.
Ok, thanks :)
Regarding cooling: what's wrong with regular 486 active coolers that latches directly on CPU? Those should work well, especially when CPU isn't going very hot.
I also think that what is holding you back with CPU speed is either the CPU itself or RAM, or maybe both.
I only have tiny 486 coolers that snap on to the CPU. I'll try to find something larger
VLB is actually a CPU bus, and your CPU is undervolted, so maybe this is the reason most video cards did not work
Yes, very likely the issue. I'll try with external power next
I'm not sure about the 486 but many CPUs are dual-voltage. Even if the CPU core runs at low voltage the bus pins will run at standard voltage.
@@eDoc2020 486 Cpus are not dual-voltage
@@theALFEST I didn't know they were single-voltage only. But since the cards worked at lower clocks with the lower IO voltage I would expect that the same IO voltage is also not an issue at higher speeds. I suspect the boards just have trouble at higher speeds. Increased IO voltage might make the cards run a bit faster but if so you'd definitely be pushing the limits.
@@eDoc2020 VLB is just CPU FSB, so there are some limitations, for example on number of VLB cards depending of bus speed and so on. And it is not just boards have trouble - cpu has trouble too.
Have you checked the temperature of the buck converter?? Underneath the CPU it's pretty snug.
On that note, might want to try a more potent / reliable buck converter. Maybe one that's not mounted on the socket, but externally.
Does the socket thingy come with pins for an external converter??
Yes, I'm going to try with external power for the next attempt
I've been trying to place your accent for almost 2 years now. Are you dutch? :)
On thing, that I certainly don't miss from "ye goode olde tymes" are those horrible cases. Impractical, sharp edges and butt-ugly... and that beige colour... yeah, no.
That said, the period (mid 90ies) saw the dawn of black cases... Dell in 94/95 and IBM in 96.
Even if 96 is a bit late in the 486 game, it's also the beginning of AMD's 5x86. So, in conclusion, you certainly can have a visually more pleasing case from the period (e.g. IBM Aptiva). Or use a spray can.
Voltage could be an issue...
I would guess it's probably the voltage. That's pretty far out of spec.
As your not 100% sure about the jumpers.... Maybe the ISA bus speed isn't right for those graphics cards that don't boot....??!!
No, I'm only sure that they are wrong actually. I'm going to try the settings for P24D next
😂👍@@Epictronics1
Have you checked the 5V line to the socket, also??
No, I never thought to check!
Concerning the 4x Multiplier... does it even work at all, beyond the 33 MHz Bus frequency? Do you get 160 or 200 MHz on the clock pin?
I didn't check unfortunately
You're setting too high a FSB speed, that's why most of your VGA cards don't work or has glitches
I swear i hear a Swedish accent 🤔
You should be using a CRT, the flat screen is stuck at 60 Hz isnt it?
I only use flat screens to spare you guys from the flicker. My camera can only sync to 60 or 80Hz
@@Epictronics1 I am just saying the old CRT is much more tolerant of different frequencies delivered to it from the video card.
@@markae0 Oh, I see. So you think there might have been graphics that the display wasn't able to show? Ok, I'll do some more test with a proper CRT display next
this chinese voltage converter is dangerous crap.
Your cadence makes this unwatchable for me, sorry
I can’t imagine anything would be worth repairing 10’888’869’450’418’352’160’768’000’000 traces…