Half-Saint wrote on 2021-07-26, 10:12:

Nice work on this computer! Really nice work!

Regarding the BIOS, you should be able to flash it via software just fine. Simply make a bootable floppy with a BIOS flashing utility and the ROM that you want to install.

What kind of a 3D printer have you got? I've been thinking about buying one for ages.

I don't think that bios can just be flashed. I think I need a programmer and a fresh chip (even if I can overwrite my current one, I wouldn't, just in case)

I bought an ender 3 v2 from creality's website, the honey plugin got me a good discount on the price.
It has been heavily modded since then, new PSU fan, new mainboard fan (bit of a required mod because by default it's tied to the part cooling fan) dual 5015 on a custom hotend mount, bi-metal heat-break from slice engineering, bltouch, triangle lab extruder, ender 3 screen to replace the silly one it comes with and Klipper firmware.

I don't really know what to recommend TBH, If you're someone who likes tinkering maybe the ender 3 pro or the v2 would be a good starting point for a printer. The stock ender 3 v2 has poor cooling for the mainboard and PSU, so keep that in mind. If you got the cash to splash, maybe a Prusa.
Unless you got a really good deal on an ender 3, I don't think it's worth it today, noisy steppers, poor PSU and the pro + v2 around the same price point.

I love mine, I love making things with it and I love modding it but a cheaper one is not something for everyone. It has tested my patience more than any inkjet I've had before.
-Lime

Tiido wrote on 2021-07-26, 09:18:

That BIOS chip is possibly an actual mask ROM but it would be 27C512 compatible.

Very nice work on this computer ~

Thanks mate.

Your pfp looks really familiar, I can't quite place it. I wonder if we have crossed paths before somewhere.

CalamityLime wrote on 2021-07-26, 01:22:

Out of curiosity; I yanked out the bios chip from the board to try to identify it but I have no idea what it is. […]
Show full quote

Out of curiosity; I yanked out the bios chip from the board to try to identify it but I have no idea what it is.

The top of it says: mb512sg2
The bottom of it says: um5197 9509s1 2c008

xgpro doesn't have it in it's database. I'm guessing it's a 27C512 compatible?

Yes, this BIOS chip is probably a 27C512 compatible PROM or EPROM. If it has a window on top, like on the picture below, it is an EPROM which can be erased by UV-light. A PROM is essentially the same chip, but without the window and can therefore be programmed only once and never erased.
Both these chips need to be programmed in an EPROM programmer.

The adapted REV. J.1 BIOS has the same 64KB size as the original REV .G2. BIOS so it will still fit in a 27C512 EPROM.

Later 486 en Pentium boards use a flashchip to store the BIOS. This is much easier because flashchips can be electrically erased, so a BIOS update can be done via software and the chip doesn't need to be removed for reprogramming.
Most 486 boards however still use EPROMs...

Chkcpu wrote on 2021-07-26, 20:28:

Yes, this BIOS chip is probably a 27C512 compatible PROM or EPROM. If it has a window on top, like on the picture below, it is a […]
Show full quote

CalamityLime wrote on 2021-07-26, 01:22:

Out of curiosity; I yanked out the bios chip from the board to try to identify it but I have no idea what it is. […]
Show full quote

Out of curiosity; I yanked out the bios chip from the board to try to identify it but I have no idea what it is.

The top of it says: mb512sg2
The bottom of it says: um5197 9509s1 2c008

xgpro doesn't have it in it's database. I'm guessing it's a 27C512 compatible?

Yes, this BIOS chip is probably a 27C512 compatible PROM or EPROM. If it has a window on top, like on the picture below, it is an EPROM which can be erased by UV-light. A PROM is essentially the same chip, but without the window and can therefore be programmed only once and never erased.
Both these chips need to be programmed in an EPROM programmer.

The adapted REV. J.1 BIOS has the same 64KB size as the original REV .G2. BIOS so it will still fit in a 27C512 EPROM.

Later 486 en Pentium boards use a flashchip to store the BIOS. This is much easier because flashchips can be electrically erased, so a BIOS update can be done via software and the chip doesn't need to be removed for reprogramming.
Most 486 boards however still use EPROMs...

Hey, thanks for the clarification.

Thankfully it looks like the 27C512 eprom chips are cheap enough and I have two strong UV lights that should do the job for blanking an eprom. One with a UV bulb and one I made with leds. I'll ask around, I have a short list of names of people who might have a programmer handy.
All else fails, I'll order a programmer when I have the spare cash. I'll probably have to order tbh.
I'll probably be a cheapskate and get the cheapest 866 I can find.

-Lime

Well if it looks like a long boring rainy weekend, you can print out the code and program a ROM manually on a breadboard 🤣

Are you using a OTI video card ? that's a turtle.... put something faster, much faster i your system.

There's no law stating that usage of old OTI or Trident video cards is verboten. I say, use whatever makes ya happy (or is affordable).
ISA VGA cards haven't been getting any cheaper lately, especially none of the "fast" ones.

BitWrangler wrote on 2021-07-27, 02:51:

Well if it looks like a long boring rainy weekend, you can print out the code and program a ROM manually on a breadboard 🤣

Eep386 wrote on 2021-07-27, 22:32:

There's no law stating that usage of old OTI or Trident video cards is verboten. I say, use whatever makes ya happy (or is affordable).
ISA VGA cards haven't been getting any cheaper lately, especially none of the "fast" ones.

It's just the card that was inside of the machine when I got it. TBH I find the old isa vga cards and vlb vga cards very confusing. I am keeping an eye open for vlb cirrus logic whatever on ebay, to my limited understanding is that it'll be faster than the OAK one I got currently. It would be nice if the machine could play doom decently well, just for a bit of fun.
The jump from the dx2 to the dx4 made a difference in doom.

So a sort of update, someone local is selling a Willem Eprom Programmer (specifically its a PCB5.0E-2018) and I do have an older windows 7 pc with a parallel port, so I'm going to try to nab that for 20 quid then source some chips from somewhere.
That should do the trick. I read that the power circuitry is hit or miss but look, it'll work for one and that's all I need.

Wish me Luck

Update:
Nabbed it, hopefully it'll be here before the weekend

Another small update.

I nabbed a 10 new M27C512-10F1 on ebay, here is the link for another 20 quid
It's coming from Spain so it'll take at least a week to get here. I found one Irish seller but they seem to be away or something.

Hopefully I'll be able to nab some card without a bios on discount or something. It would be nice.
Welp, this should be fun.

This is a nice build.

I would personally run a fan on a DX4-100 if it were me though.

Well those shortarse ones aren't much good on anything without a fan, double the height sinks should be acceptable on a 3.3V DX4-100 without a fan, fastening methods are difficult though on 486... Just coming up with something in my head that might work with cocktail-sticks/toothpicks and fishing line on those that have lost their slide in clips.... will have to prototype that.

Great job - props!

Part 17: Willem Programmer

Yeah, an entire part just for a programmer!

So, as stated in a previous update, I ordered a Willem Programmer for 20€ from a local seller. It arrived with USB and parallel cable. For these cheaper ones the actual programming is done over parallel with the USB for power. While you can buy a PCI-e to parallel card, I'd recommend grabbing an older Windows XP machine with on-board parallel. I really don't think you'll get a better deal from a PCI-e card and Willem programmer VS a USB TL866, but it's up to you.
I went the the Willem Programmer because I already have a machine with a parallel port and I don't mind setting it up for a quick flash.

It took a bit of effort to get it working, more than I thought would be required to be honest. So in order of issues, here is a bit of a log.

• Issue 1: Bare Naked board.
  The Willem programmer doesn't come with a case, so that was the first thing to fix. Considering that there are jumpers and ports all over the top of the board, I decided to go with an iPhone box style of case. The top lid just lefts off and out of the way, with the bottom just to protect the thing from short circuits. I'll wrap it with masking tape for storage. I'm happy with how it came out and that it used the last of two rolls of filament that I've been looking to finish!
  STL's to print your own case are attached.

• Issue 2: Software.
  I did not get software with this programmer so I had to resort to the internet for help. My god, the top results give you so much non-sense. I took some googling and getting some dubious links to find what I was looking for.
  For the latest version go here to ezoflash look for 0.98d12c3.
  Since the latest version doesn't support all revisions of the board, here is Sivavas website from the wayback machine which has a larger list of software versions.

• Issue 3: Windows XP
  The software seems to be happiest with Windows XP, I recommend going to Windows XP instead of Windows 7 for reasons below but even Windows XP seems a little iffy out of the box.
  Sivava recommends installing the older 0.97ja version of the program first before using the newer version and turning on DisableWarmPool under the parport section in the registry. It doesn't seem to appear so you'll need to use a reg file.
  There is a video by Sivava explaining this here!
  I might be worth enabling legacy plug and play under, Port settings of the LPT in device manager.
  Regfile contents

  Copy code to clipboard

  1REGEDIT4
  2
  3[HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Services\Parport\Parameters]
  4"DisableWarmPoll"=dword:00000001
  

• Issue 4: Windows 7
  So Windows 7, I set up the 64bit variant which always makes old software and hardware a pain to work with.
  First road block will be the following error message: "Unable to start the IO.SYS service. Error code = 1275", followed by a "privileged instruction" error which is not a lack of admin permissions but a result of io.sys not working.
  Thankfully, someone has already written a blog about this issue and how they fixed the problem. Check his blog!
  Essentially you're replacing the stock io.dll with a wrapper. The result does work but seems to be a bit slower compared to Windows XP, I mean unless you're in an absolute hurry, I don't think that'll matter.
  
  After installing the software that the blog mentions, Windows 7 began boot looping with the programmer plugged in on start up. I checked the DisableWarmPoll variable and it was already set to 1. Turns out the fix is to either plug in after the pc is turned on, which may or may not work very well OR Stop Windows from trying to initialize the parallel port.
  
  I might still be worth enabling legacy plug and play under, Port settings of the LPT in device manager.
  Stop Windows initializing the Parallel Port

  1: Open Regedit
  2: Go to HKEY_LOCAL_MACHINE\System\CurrentControlSet\Services\Parport
  3: Double-click on the REG_DWORD “Start”
  4: In the pop-up window replace 3 with 4 and click OK.
  5: Reboot

I didn't really plan on posting an update just for the programmer but given the issues I had an how long it took me to resolve them. I thought it might help someone else to get their Willem Programmer working on Windows 7 64bit and Windows XP machines. It seems to work okay, I'd still recommend using Windows XP with this if you can. The fix for Windows 7 64bit might translate to Windows 10, I have not tested this.

Thank you for reading, your comments and your interest.
-Lime

PS: For those commenting about CPU fan; I do have a fan blowing air across the CPU heat sink which seems to be doing a good job. I am keeping an eye on it though.
PPS: I messed up the part cooling settings so the lettering on the case looks a bit... bad. That's my fault though.

Part 18: BIOS

WOOOO the 27c512 chips arrived today in nice little plastic tubes. Seller included a free pen!

So BIOSs I have at hand; the rev G2, the rev J1 (thanks to Chkcpu for that one) and the rev G3 for the soyo-25p2. The rev G2 is from my own bios chip that I dumped. The G3 variant was a shot in the dark recommended by someone on this site.
All the BIOSs mentioned are included in the attached ZIP file.

Tools

• EPROM flasher, see previous update.
• Bent flat head screw driver. An idea from CPU Galaxy (I think) I just bent a cheap one that came in a kit with a set of pliers. Works a treat.
• UV light, didn't actually need it since the EPROMs were brand new and blank.

Since I hadn't given the Willem programmer a good test, I flashed an EPROM before dumping my own bios chip.

• Rev G3
  I decided to start with the rev G3 option since I expected that one to work. Given that the p and n variant of the boards are basically the same but the p variant includes the 3.3volt transistor which I modded into my board.
  The PC gave me a happy post beep and the bios screen reported that it was using the rev G3 bios and that the CPU is a DX4 instead of a DX2 like the old BIOS said.
  I had a guess, the datasheet I found for the bios said that EDO ram was supported but the EDO ram I had didn't work originally. I gave the EDO ram a try and no dice with that.
  Hoping into the BIOS settings the first thing I noticed was that it supported 2 IDE channels and still reported my 4GB CF card as a 2GB card. Other then that, it was basically the same. Nothing too exciting.

• Rev J1
  Moving on to try the J1 revision from Chkcpu.
  One happy beep later and I was greeted with the BIOS boot up screen saying that it was using the REV J1 bios and that the CPU was a DX4.
  A few BIOS features options were missing, back to 1 IDE channel but it did report my 4GB cf card as a 4GB card. At first I was a little unsure what to think but this is where things get unexpected.

Weirdness
So someone has been private messaging me about a Doom benchmark so I had Doom set up and ready for benchmarking. It gave me the idea to run the benchmark on each BIOS to see if there was any difference for some reason. I set the bios settings to default for each bios, hoped into dos and ran the Doom benchmark. For the Rev G3 it preformed about the same as it did in the Rev G2, a little better of a result but nothing really noticeable however for the Rev J1, there was a massive improvement in performance.

Here are a few results from benchmarks

I'm guessing that this is all a result of timing settings and maybe it's increasing the clock on the ISA bus a bit. Especially considering that the CPU alone scores the same in HWINFO. I'm still very new to this so it's just a guess. Still, it does work, it seems to be happy working like this so I think I'll leave in the J1. It would be nice to have support for 2 IDE channels but that's not the end of the world.
Since I did say that there are some options missing from the BIOS Features Setup screens, here is a list. I don't think any of them really matter but correct me if I'm wrong.

So this update has been a massive success and it has been a lot of fun getting this far in tinkering with the sad and lonely 486 left to rot under some stairs. If anyone has any ideas for further things to with this old beast, let me know. I believe that I have proven that I will actually do them.
A massive thank you goes out to Chkcpu for their efforts in making a custom BIOS image for these Soyo boards. This update wouldn't have happened without you.

Thank you for reading, your comments and your interest.
-Lime

Chkcpu wrote on 2021-07-25, 18:01:

Hi CalamityLime, […]
Show full quote

Hi CalamityLime,

In addition to my previous reply, I have some further BIOS info.

If you are looking for an updated Soyo 25K2/25N2 BIOS to fix the bugs I mentioned, I have one attached below.
It is a 11/28/95-SIS-85C471B/E/G-2C4I9S21-00 / REV. J.1 BIOS for the Soyo 25JKL series and I'm confident it will work on your 25N2 as well.

This 11/28/95 BIOS is milenium compliant, supports HDD's up to 8GB, and supports all enhanced Intel/AMD/Cyrix 486/5x86 CPU's including L1 cache WB mode.
Note that this BIOS only shows the “L1 cache: WB/WT” option in the BIOS Setup for the P24T and for Cyrix CPUs.
For the P24D (486DX2WB), Am486DX4WB and Am5x86, this 12/28/95 BIOS can detect if these CPUs are in WB mode and programs the chipset registers accordingly. It then hides the “L1 cache: WB/WT” option, because user interaction for this automatic function is not required.

I took this up-to-date SiS471G BIOS from a Gemlight board and carefully adapted it for these Soyo boards, including the default chipset register programming from the original REV .G2. BIOS.

Cheers.

I thought I'd give you a reply message if you're interested in this update.

Hi CalamityLime,

Thanks for sharing your experiences with this 486 board in detail. I read the parts about the Willem Programmer and the BIOS with interest.

Great to hear that the REV. J.1 BIOS works so well. 😀
Yes, a lot of the 1994/1995 Award BIOSes supported only one (primary) IDE channel, so only 2 drives (master and slave) could be attached. But a few of these BIOSes did also support a secondary IDE channel and if the board has 2 IDE connectors, an extra 2 drives can be hooked up.
Strangely enough, the 09/26/94 REV. G.2 and 11/28/95 REV J.1 BIOSes only support a primary master and primary slave drive, but the 03/08/95 REV. G3 BIOS supports the secondary master and secondary slave as well.

I will look into this G3 BIOS for bugs and proper Am5x86 support, and report back on this.
Maybe this BIOS is a better candidate to patch for your 25N2 board, in case you need 4 drive support.

About the list of missing options in the BIOS FEATURES SETUP menu, they are actually all there in the REV. J.1 BIOS!
But I see them missing (hidden) in the REV. G3 BIOS... 😉

Later, Jan.

Chkcpu wrote on 2021-08-09, 14:25:

Hi CalamityLime, […]
Show full quote

Hi CalamityLime,

Thanks for sharing your experiences with this 486 board in detail. I read the parts about the Willem Programmer and the BIOS with interest.

Great to hear that the REV. J.1 BIOS works so well. 😀
Yes, a lot of the 1994/1995 Award BIOSes supported only one (primary) IDE channel, so only 2 drives (master and slave) could be attached. But a few of these BIOSes did also support a secondary IDE channel and if the board has 2 IDE connectors, an extra 2 drives can be hooked up.
Strangely enough, the 09/26/94 REV. G.2 and 11/28/95 REV J.1 BIOSes only support a primary master and primary slave drive, but the 03/08/95 REV. G3 BIOS supports the secondary master and secondary slave as well.

I will look into this G3 BIOS for bugs and proper Am5x86 support, and report back on this.
Maybe this BIOS is a better candidate to patch for your 25N2 board, in case you need 4 drive support.

About the list of missing options in the BIOS FEATURES SETUP menu, they are actually all there in the REV. J.1 BIOS!
But I see them missing (hidden) in the REV. G3 BIOS... 😉

Later, Jan.

Hey There!

Great to hear back from you.
I had found the G3 bios about 2 or 3 days before the eproms arrived. I thought it was interesting that it had 2 IDE channel support.

Since the BIOS is for the P variant of the board and the P variant can do 3.3v, I'm guessing the compatible CPU list for the G3 would all be listed in the manual here. That is just a guess though, the manual also says it supports EDO ram which doesn't seem to work in the system.

Do let me know how you get on with the G3 bios.
4 drive support could be handy but isn't too important for me, I don't think it is anyway. If there happens to be a simple toggle or hidden switch in the j1 to enable 4 drive support, please let me know.

Thanks Again
-Lime

Hey there,

Long time no see!
Despite not posting much I have been tinkering with things that I do mean to talk about when I get the chance to.
One big project was a GBSControl that kinda takes the old VGA signal from the 486. There are some sync issues I want to try to resolve but I got side tracked with something a bit more important.

A mouse.
I had a look around the forums for various solutions. Matze's, Lameguy64's and Necroware's PS/2 to serial boards popped up as did that one solution that used a nullmodem cable + a pi to go from USB to Serial.
The null modem and pi would have been great if I had a null modem laying around but I don't, so a bulky solution like that just isn't worth it to me. So I decided to go with the PS/2 to Serial route.
I took a bit of time to compare the different solutions and decided to go with Matze's offering, for the reason that I felt that modifying and making the board would be the most fun out of all of them.

I downloaded Matze's through hole variant of his board, modified it a bit in KiCad, gave it a nicer silkscreen, moved a few traces around, modified some footprints. All that fun stuff done in some attempt to learn the basics of KiCad.
I ordered 10 of those boards and the parts to make 5 of them. I plan to keep one and try to sell off the other 4 with a 3d printed case. Why not, after all it didn't cost too much more than the parts for 1.

A USB mouse?
So while I was waiting for the boards to arrive and parts to arrive, I was looking at the circuit and the code.
A thought was forming, "why can't I do this with a USB mouse?". I have 3 Pi Picos laying around and I have begun digging into any code examples of the pi pico acting as a USB host. Turns out that there's very slim pickings for any examples code and I'm pretty much suck with the C SDK since Adafruit doesn't have TinyUSB hosting ported over to Arduino or CircuitPython, MicroPython doesn't seem to have it either.

I haven't used C since I was first learning to code, there seems to be a breaking change with TinyUSB so the old USBHost examples don't work any more, the new examples are all Midi rubbish. After some headbanging and lots of confusion I got a cheap USB mouse to talk to the pi pico. I attached a GIF of me clicking a button on the mouse and a red LED turns on and off. This was using the first bit of "this could actually work" code. Since then it got a bit better and I'm basing the actual "to serial" part off of Necrowares code since that was written for Arduino which is much closer to C then the assembly of Matze's code.
I have gotten it to the point where I am ready to hook it up to the PC for testing but I'm still waiting on the parts to arrive from LCSC. I ordered the boards and parts on the same day from LCSC and JLC respectively, the PCB's have already arrived and LCSC has yet to ship my bits.
Considering that I've been wanting to order some more solder, I placed an order with Farnell today with the max232 and a cheap serial connector that will let me go further with this project when they arrive. Hopefully next week I'll be able to continue with this project.

I thought about making a new thread for this topic but since I have yet to have a working prototype, I decided to post here for the moment to see if there is any interest in this project at all. Personally I'm a lot more confident that I can actually do this now than I was at the beginning.

If anyone is interested in the code, I'll be happy to share it on GitHub or something when it either works or I admit defeat.
If the interest is there, I might make a PCB to go with it and maybe sell a few ready made ones to anyone who wants one. No idea on the time frame for that kind of thing but it could happen.

Thank you for reading, your comments and your interest.
-Lime

hmmm