An expert among retro mainboards (Part 3): adding 3V CPU support
Вставка
- Опубліковано 4 жов 2024
- Some time ago I got one very exciting mainboard DataExpert EXP4349. In this last part I'm going to add a voltage regulator to support 3V 486 CPUs and update my results #isadoom25fps challenge.
The Retro Web entry:
theretroweb.co...
486SocketBlaster Project by @scrap_computing
• 486SocketBlaster: A DI...
github.com/scr...
Patreon:
/ necroware
Interested to see more of this board? Absolutely! I look forward to part 4.
Absolutely yes. 👍
i concur
I'm waiting part 4.
part 4 will be good
Those variable resistors in that regulators are very sensitive as you already told, and also it could change its value without notice, you would better replace it for the apropiate smd resistor to get the 3.3v and be on the safe side that it will not change resistance by its own... Great video as always! And make part 4 of course!
I was thinking the same thing. Either using a smaller variable resistor to limit the maximum voltage spread with fixed resistors to limit the min and max range (instead of 0v to 5v have it variable only from 3v to 4v) or replacing the variable resistor with a multi turn (probably the easiest solution). I would also add extra filter capacitors with low ESR on the regulator output and possibly directly across the CPU power pins (he did mention adding filtering).
Of course... We need part 4. I'd love to see if you can make it run at 40Mhz FSB. Really enjoy watching your videos!
I recommand! ❤😅
LOL :)
Would love to see scope of ripple at 120mhz. I am guessing that is causing it not to post. Great content walking through the process :)
Let’s see an overclocked AM5x86!
God imagine if all boards were as great as this one back then. Seriously impressive at the capabilities and features this thing has, as if it were intentionally designed to be a test bench.
Yeah, this board is really exceptional and I'd love to see further testing and tweaks.
You were probably getting too much ripple current when trying to push the 486DX4. Try it again after adding some capacitor to the input & output of the regulator.
This motherboard is quickly becoming one of those Subaru WRX cars people keep on tinkering with, adding all sorts of components that regular people can only dream of.
We need a mad-lad from the community to make universal motherboard that accepts all CPUs for the past 4 decades. 🤣 Just imagine all the jumpers!
A separate jumper daughterboard lol
Please make a part 4! Very interesting!
Dude... you know we're interested! Please, by all means, continue.
Old tech holds a special place in my soul. Love your content my man.
Yes, yes, yes - I want more about this board
For sure i need to see part4 next week
Yes, more please!
Maybe the real parameters of those chinese regulators are lower than declared ones. I would try it with linear regulator first, because output is much cleaner without ripple. Power dissipation is not so high for 486. (5.0V - 3.3V) * 2A = 3.4W max.
Thanks for keeping the best geek UA-cam content!
Just imagine how it will be awesome to watch your videos in 4K 60fps HDR ;)
To be really geeky, it should've been VGA ;)
This board is awesome... Part 4 plz
I would love to see a continuation on this board.
Seriously you have to ask. Yes. And again I would love to see the 486 DLC. With VLB
I wonder if this board could be configured to run the original 386 from 1985 at its correct 12 MHz 🤔
That board is a BEAST!!! 10 YEARS!!!! We're lucky if we get 3 these days!!!
I wouldLOVE to see you come back to it ini the near future, I really wanna know why its coming up at 75Mhz in the POST.... but 100 in DOS.... very strange, but YOU WILL FIGURE IT OUT!!! :D
Most likely some processor identification microcode that the BIOS is lacking.
It'd be awesome if they made a motherboard nowadays that could last that long
What a terminator of a board. Awesome series!
I want more testing on this board!
Great video, as always! And thanks for the mention!
I was not aware that older 486s had a different multiplier pin, thanks for sharing this information! This makes adding a Write-Back cache jumper to the 486SocketBlaster even more useful, as it will also help control the multiplier in these older CPUs.
This really reminds me Fagear's marathon on getting ASUS P2B-F board to run everything from the slowest Pentium II to the fastest Pentium III, but such hardcore mods like dedicated VRMs weren't done. :D
I think it be funny for April fool's replace the wolf sound in intro with chicken or random noise
Never thought I would enjoy a series about 386 and 486 stuff, but this got me hooked. If a motherboard needed soldering I would probably just give up. If a power rail had ripple I would never find out. You solve the problems and go on to overclock it. I've always thought that overclocking the PCI bus (or ISA bus in this case) is foolish and impossible, but you even got that to work. I don't mean to make you blush or anything, but this is just so fun to see. Thanks for making these videos :)
Thank you very much :)
Looking forwards to part 4 ;)
It's nice to see when your work is going to right direction. Also today I receive my package with parts to my 386/486/VLB MoBo on USA SC-9204-A chipset and now, it's working. Now I'm owner of two, working 386/486/VLB MoBos :). It left for me to make some tests and compare them to your results. At the end I'm glad to see you're back to project VRM for socket 7 MBs. I'll keep my eye on it. ;)
I love this video series.
I remember many years ago having a mother board that was permanently set to a CPU voltage of 5v. It got its power directly from the AT power supply. I cut the CPU power lines and added diodes in series until I got a voltage drop that was close enough to what I needed.
I'd like to see another video about this board. I'm interested in seeing how far you can go with fixing and upgrading this board and how high you can go with processors.
I also vote for part four of this video.
What an incredible journey following all 3 videos so far! This board has had so many problems I almost wondered if it was some kind of prototype. Keep up the great work, these videos are fun to watch!
To do all this and still have this great music going on. Indeed, this is magic. These UA-cam uploads are like short sci-fi movies. Thank you, Sir!
Can't wait for the next part on this
This is the best series of a board ever seen, please continue to explore this board!
Great stuff as always. Nice to watch someone who has the knowledge to take those retro boards to the next level while I can only manage to use them stock
I was missing your videos there for a while, this board is an awesome comeback! Would love to see more
Awesome work.
12:33 Lol, "Please, don't question him, he is an expert! It says DataExpert here."
Excellent work Necroware, I really like your teaching style, I think you would be a very good computer and/or electronics teacher. Greetings from Chile.
I actually worked in university and in my career I was trainer in software engineering ;) So, may be this is where I have this habit from.
@@necro_ware Great job my friend!!!
This is a great series. Love the music too, specially when he jacked up the VRM to 4v
Excellent video series. Love it.
By the way, the P24C is a DX4 with write-through enabled
But of course, continue experimenting with this board, because it shows your thought process and for us enthusiasts it teaches us. So by all means, continue experimenting!
Amazing. I am honestly hocked to this series. Please do a part 4.
Very interesting. Nice video.
More about the board!
Great video as always!
this board is cool
You are amazing. Respect for your knowledge and passion.
Great video, ive been binge watching your videos since finding and subscribing to your channel recently. We had a dx4 100 as our first family pc in the mid 90s.
the silver pads is where the 80MHz xtal is supposed to mount
That KTS KDN 0147C IC is the PLL clock generator. The control pins are driven by the chipset instead of setting them manually with jumpers.
Since it generates correctly other frequencies, I suspect a bad solder joint or a cracked trace.
Since I did not found any datasheet for it (only hints that kdn 0147c is part of a series of PLL clock generators), I would set every possible frequency and check the pins to try and make a truth table. So can (maybe) figure the setting for 40MHz and find the fault.
You've inspired me to try and get my old 486 board going it looks very similar to this one even
Love the content! your video's are always a way to make my day much better!
Great stuff
This has been a really great series! Thank you!
Motherboard: it's been 30 years, please let me die in peace. Necroware: no.
It has been about 25 years since I was over clocking dx4 cpus, but if I remember I had similar issues. I recall changing the voltage regulator, as it needed up to 5 amps. I also remember needing exceptionally clean power for it to remain stable at and above 120mhz.
I have a couple of 486 off the dump recently with some ripped off or non functioning 3v regulators......only since your videos i actually thought about upgrading some of the others that just left the 3v area of the board unused.
Cant wait for part four! been loving these videos
The worst thing about these videos is learning that we could have done so much more with our old computers back in the day; things that we would have never dared to do like this literal hardware hacking.
Please do more videos on this it's so interesting!!!
From back in the day when motherboards like that were the bleeding edge, I often dreamed about hacking one like this so I could upgrade accordingly. Then the slot 1 P2 came out and blew my mind. I remember the P3 chip when it broke 1Ghz for the first time. We've come so far over the last few decades. I'd actually like to see if it's possible to hack a slot 1 daughterboard to socket a different processor. Like put a socket on it and see what's possible.
Mmmm, I had this board with all VLB slots used and Yamaha wavetable sound card. Nice times!
I remember the madness combination with jumpers XD that was terrible but satisfy.
Love your work. Keep it up!
As always a great video! Those hybrid boards are always something special! BTW: Received a Shuttle Hot-317H today and just reviewed your repair video but the damage on my board is more excessive. The traces are almost gone and one tantalum cap is ripped apart. Hope that i can fix it.
That is a great board. Thanks for sharing all your work and adjustments. cheers.
Amazing work!
Great video once more. I watch your content regularly and find it extremely inspiring. Your videos motivated me to start learning to use tools like Multimeters, Soldering Irons and Hot Air Stations. Also very educational and relaxing to watch you explain stuff and work on parts. Glad you have the spare time to make videos regularly again :-)
If you upload videos, by my opinion, your channel will be reached most of the views these days I guaranteed about it😊
I used to have a 486 DX4 @ 133 MHz overclocked to 166 MHz. But it was a more modern board with PCI.
I wouldn't expect a linier reg to get that hot going from only 5V to 3.3V, but then again it depends on currant draw.
Loving this
Awesome board
You were SO close the 40mhz multiplier, i was thinking that perhaps a .1v drop or raise might have made it post fully without crashing, but you must have thought the same thing since you mentioned being curious about ripple coming from the VRM
More!!!
Great video, would love a part 4. Can that vlb card take advantage of more memory?
8:07 Trim pots can settle after they are adjusted. I'd check in a couple of days to see if the voltage is still 3.3v
Stupid algorithm didn't show these videos to me and just noticed them. Hope to see a part 4 with both AMD and cyrix 5x86 offerings to see how they perform and if this system can reach 150 and 160mhz speeds.
a triple overclock from 33mhz to 100mhz . astonishing , it did get a little choppy trying to bit bang the data it looked like. This board with your modification is ridiculous , 5 plus different types of cpus so far as well as adding the fpu.
There are at least 2 other jumpers which the board just straps with a wire. I bet that one of those is important for unlocking things that don't currently work as expected
Greate job!
Love it! Can you try out the Cyrix/TI 386 CPU that has built in cache? I bet you can put up a decent benchmark with Doom.
It looks like a choosen voltage regulator is not powerful enough to handle overclocking. As an option ripple may become an issue at higher loads. It wasn't great from the beginning.
did you make a part 4? i love this board!
I noticed the new music and there's no mention of the usual bandcamp, I'm curious
Also I've never owned a retro board so I'm not sure how the chips behave but I got kinda scared when you did 4v instead of trying to tackle the ripple.
With a compatibility range like that, it seems very similar to the AM2 series of sockets.
I got a bag full of old tantalums, so the offer still stands...
;)
What's the advantage of tantalums vs electrolytic?
@@darthtripedacus1 They're a lot more volatile...
@@darthtripedacus1 In the last video, I suggested a reverse recap, putting in capacitors that are ready to fail, to punish the motherboard for being such a clusterfuck, including the egregious ripple voltage that made it so unstable.
nice video, bur time for anither board. this one needs some rest.. ahahahah
Is it possible to replace the current clock generator with a more modern and capable programmable clock generator?
Would be an awesome hack for overclocking experiments, allowing you to set the front side bus clock speed in smaller (say 1mhz) increments.
Bonus points if you can find a clock generator that allows (cleanly) adjusting the clock speed while the CPU is running.
Maybe replacing the variable resistor for a fixed value that delivers the 3.3V required could improve stability
May I ask where you acquire some of your spare parts are they from donor boards.
As I have a socket 7 MB with some damage memory clips.
Those regulators are horrible with wild voltage swings depending on load so it wasn't too surprising to see the old 486 go berserk at higher frequencies. You can't do the same you did with the socket 7 board? :)
I'm going to guess the board didn't like the 120 MHz fsb because the cache couldn't keep up. I could be wrong but it's a valid guess
It is a valid guess, indeed, however I upgraded the cache in the last part to 15ns and it should be fine. I think, it's either voltage ripple, or there is something wrong with the clock generator, since it also made problems to deliver 40/80MHz for the 386. May be it is related, since at 120MHz we again have 40MHz FSB. Against this theory is that it ran fine with DX2-80 at 40MHz FSB. Well, I'll have to investigate it in another part ;)
@@necro_ware That's true. I forgot you upgraded the cache. Also thank you for the smooth correction. I got the bus speed and the clock speed mixed up. I can't wait for part 4