VOGONS


486 VLB UMC-Chipset, what is it?

Topic actions

Reply 60 of 75, by bnbjaj

User metadata
Rank Newbie
Rank
Newbie

Gerwin, you are a life saver! Last week I scored an 486 system on Ebay. The motherboard simply stated MB456V1.1 which gave no google results. Then I searched the UMC chip set and hit the jackpot with this post. Your package for TK8498F motherboard and VGA+Multi-I/O gave me all the info I was looking for. It seems this UCM motherboard design is available with a variety of different markings on the board. As already mentioned in this post, my 486 UCM rig also includes the UM85C418 VGA/Multi-IO VLB board. Thanks a million for your detailed posts, it has become my 486-bible.

Reply 61 of 75, by gerwin

User metadata
Rank l33t
Rank
l33t

Good to hear that this topic was helpful for you. I suppose quite a few typical 486 problems and solutions were encountered during the tweaking of this 486. Makes me wonder, which problem did I miss out on. Well... Fake cache for example 😉.

If you have time, please share some more details and/or photos of the system you obtained. I wonder what the CPU it came with, and wheter or not the VRM is present. Also, are you planning to use it as it is, or do you want to upgrade parts of it?

Yesterday I slightly updated the package on vogonsdrivers: Corrected typos in the CPU jumper table, Put the right jumper setting for 20MHz FSB in the layout and removed the duplicate VGA manual.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 62 of 75, by bnbjaj

User metadata
Rank Newbie
Rank
Newbie

The 486 was extremely dusty. I did not take any pictures when it came in. I bought it because I needed the VLB video card, controller card and audio for another board. After I found this post, I played with the MB for a while, trying out some combinations of RAM, CPU and cards. In the end I put it aside due to its multiplier limitation. It came originally with an Intel 486DX2-66 S (heatsink glued to the CPU) but in the end I used an AMD 80486DX2-80. If found one lying around but, it too, had the heatsink glued. Since AMD does not mark the bottom for these CPU's I pried loose the heatsink but it took some of the top surface with it. With the help of cpu-collection.de I figured it was an AMD A80486DX2-80NV8T.

Here is the motherboard with the original 486DX-2 66-S. I had already removed the 2x8Mb EDO and replaced it with double sided 2x32Mb EDO. Cache shows all banks 32Kx8 20Ns for a total of 256Kb.

BAE73321-36C1-481C-B338-C6E89146A633.JPG
Filename
BAE73321-36C1-481C-B338-C6E89146A633.JPG
File size
2.77 MiB
Views
2830 views
File license
Fair use/fair dealing exception

Backside of the CPU

A714F102-1AF1-4B1A-889E-B771FED2F112.JPG
Filename
A714F102-1AF1-4B1A-889E-B771FED2F112.JPG
File size
2.6 MiB
Views
2830 views
File license
Fair use/fair dealing exception

Board marking

8990706E-D99B-4854-8CDC-81D8B866E862.JPG
Filename
8990706E-D99B-4854-8CDC-81D8B866E862.JPG
File size
1.36 MiB
Views
2830 views
File license
Fair use/fair dealing exception

TK85C418GVIO-4J-D08 for which I found the jumper settings in your documentation! I disabled the HDD/FDC as it was not stable and used a seperate VLB I/O card.
This card ended up on the other 486 board (ZIDA 4DVS, AMD486DX4-100)

4B04502E-6FFC-4794-99FE-D0542D8B35B1.JPG
Filename
4B04502E-6FFC-4794-99FE-D0542D8B35B1.JPG
File size
2.89 MiB
Views
2830 views
File license
Fair use/fair dealing exception

The Cirrus Logic GD5422 I used for the benchmarking

5134C443-A7A4-407A-88D5-39DC65195087.JPG
Filename
5134C443-A7A4-407A-88D5-39DC65195087.JPG
File size
2.73 MiB
Views
2830 views
File license
Fair use/fair dealing exception

Reply 63 of 75, by bnbjaj

User metadata
Rank Newbie
Rank
Newbie

Continued:
The I/O board used for the benchmarks. It's an UMC card with UM82C863F/UM82C865F and branded MIO-VL P103.

1026BA9A-4E98-4178-A596-0AB1CFEB500C.JPG
Filename
1026BA9A-4E98-4178-A596-0AB1CFEB500C.JPG
File size
2.84 MiB
Views
2829 views
File license
Fair use/fair dealing exception

The Crystal CX4235-XQ3 that also ended up on the new board. I was unable to get it working in pure DOS. No issues with WIN95.

D09A8F29-54D5-4A94-A693-C58856254FDA.JPG
Filename
D09A8F29-54D5-4A94-A693-C58856254FDA.JPG
File size
2.79 MiB
Views
2829 views
File license
Fair use/fair dealing exception

The AM486DX2-80 after I removed the heatsink. It's the AMD A80486DX2-80NV8T as it has the 'N' is barely visible under X. In addition it shows the 'B' under A80, which is part of the B 9520 APD T marking.

F7FDD747-B516-4CB4-98E4-CD1B38974DAF.JPG
Filename
F7FDD747-B516-4CB4-98E4-CD1B38974DAF.JPG
File size
2.01 MiB
Views
2829 views
File license
Fair use/fair dealing exception

AWARD BIOS

AE910ADF-3090-4F30-9145-65D63249DB2B.JPG
Filename
AE910ADF-3090-4F30-9145-65D63249DB2B.JPG
File size
1.94 MiB
Views
2829 views
File license
Fair use/fair dealing exception

Excellent jumper markings on the board

37F8EF66-7F53-4B70-977C-2B7E64583461.JPG
Filename
37F8EF66-7F53-4B70-977C-2B7E64583461.JPG
File size
2.69 MiB
Views
2829 views
File license
Fair use/fair dealing exception

Reply 64 of 75, by bnbjaj

User metadata
Rank Newbie
Rank
Newbie

Continued:

Test setup: I used an ATX power supply with the ATX-to-AT connector, a CF card for DOS 6.22 and benchmark programs, the AT-to-PS2 Keyboard adapter and a passive heatsink on the AMD486DX2-80.

31C5B7EC-941B-4338-B083-3BBD3B17A023.JPG
Filename
31C5B7EC-941B-4338-B083-3BBD3B17A023.JPG
File size
2.54 MiB
Views
2824 views
File license
Fair use/fair dealing exception

VOGONS has an awesome collection of benchmark programs. A big Thank You to Phil and others for initiating and uploading these. As I am rather lazy than tired, I simply use Phil's DOSBENCH which I downloaded from his site.

Results:

Results.JPG
Filename
Results.JPG
File size
55.44 KiB
Views
2824 views
File license
Fair use/fair dealing exception
95B93694-D216-4034-A377-CC7CCFE4DBA8.JPG
Filename
95B93694-D216-4034-A377-CC7CCFE4DBA8.JPG
File size
1.47 MiB
Views
2824 views
File license
Fair use/fair dealing exception

Although the battery clearly needs replacement, it doesn't leak and after a couple of hours of operation seems to gather enough energy to retain changes in the BIOS for a couple of days. So, if I use the board regulary, there is no need to setup the BIOS each time I start the machine. It's a fun board and I am happy to have it in my collection.

Reply 65 of 75, by bnbjaj

User metadata
Rank Newbie
Rank
Newbie

Forgot: as can be seen above, I was playing with Write Back and Write Through. Speedsys 4.78 above shows the WT results. With WB the outcome for Speedsys is slightly higher. In the Excel overview different outcomes show and not all results are higher for WB.

Speedsys for WB

IMG_7822.JPG
Filename
IMG_7822.JPG
File size
1.65 MiB
Views
2822 views
File license
Fair use/fair dealing exception

After tweaking the BIOS a bit more, all WB totals got even better.

IMG_7823.JPG
Filename
IMG_7823.JPG
File size
2.1 MiB
Views
2822 views
File license
Fair use/fair dealing exception

Speedsys for WB after BIOS tweak.

IMG_7824.JPG
Filename
IMG_7824.JPG
File size
1.51 MiB
Views
2822 views
File license
Fair use/fair dealing exception

Reply 66 of 75, by gerwin

User metadata
Rank l33t
Rank
l33t

Thanks for sharing these good photos and the background story!
Funny that it came with EDO SIMMS instead of FPM.
The Crystal CX4235-XQ3 card should work fine in DOS. Like with this set of Drivers. But this particular CS chipset has malfunctioning FM music, or so I was told.
I actually gave up on Write-Back cache. It was unreliable and also 'Anonymous coward' mentioned that WB does not work properly with VLB in general.

Two questions:
-What do you mean with "its multiplier limitation"?
-Did you use the original BIOS?, would it be possible to make a BIN file of the BIOS?

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 68 of 75, by gerwin

User metadata
Rank l33t
Rank
l33t

I used the settings from Total Hardware '99 for the identical board F4DXL-UC4. All is in the package I uploaded to vogonsdrivers before.

CPU Type     JP11         JP12         JP13     JP17     JP18     JP19     JP20
AM486DX4 2-3 2-3 1-2,3-4 1-2,3-4 1-2 Open Open

Note that the above does not match the silkscreened settings for P24C, which is the DX4 codename.

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 69 of 75, by feipoa

User metadata
Rank l33t++
Rank
l33t++
gerwin wrote:

I actually gave up on Write-Back cache. It was unreliable and also 'Anonymous coward' mentioned that WB does not work properly with VLB in general.

WB seems to work with the Asus VL/I-486SV2GX4 and the Am5x86-133 when the BIOS is updated.

I also noted that the mod with 3V CR2032 doesn't always work well with time counting when connected to the EXT BATT jumper header. This is because this jumper header usually goes through two series diodes (~1.5 V drop) before reaching the RTC when compared to the rechargeable solder point. Even if you disconnect the positive "charging" lead to the rechargeable battery and connect the CR2032 holder, it doesn't always keep good time after many months. So what I have started doing is using a double-stacked CR2032 battery holder and connect this to the EXT BATT header. Many EXT BATT headers request 6 V as noted in the manuals. I've had better luck with the double-stacked CR2032 holders. I recently ordered more from digikey.

Plan your life wisely, you'll be dead before you know it.

Reply 70 of 75, by marmotta

User metadata
Rank Newbie
Rank
Newbie

I have the same board (without vrm), for the battery you can remove the R3 or CR4 (5v volts charging line) and you can use the lithium battery on pcb 😉

(I have yet to try)

Update: tested, work! 😀

Reply 71 of 75, by Jed118

User metadata
Rank Oldbie
Rank
Oldbie

I just got one of these boards - NOS it seems, no ISA slot scarring, no missing tin from around the mount screws, hell, it even smells unused.

BUT - it has the Dallas RTC - fortunately, the part I have to drill into is facing outwards at the very edge of the board, so no need to mess about with unsoldering it. It does POST but won't let me past the CMOS battery error.

What kind of board is it? (*edit NVM Jed118 usedRead, it was very effective.

Noyce, this board can take 3.45 v chips (mine has a different regulator but it's there) - I think I'm gonna throw my ITS 486ST 100 MHz or my 80 MHz DX2 from TI.

Last edited by Jed118 on 2020-11-03, 21:43. Edited 1 time in total.

Youtube channel- The Kombinator
What's for sale? my eBay!

Reply 72 of 75, by matze79

User metadata
Rank l33t
Rank
l33t

Wow that’s very slow for a 80Mhz 486

https://www.retrokits.de - blog, retro projects, hdd clicker, diy soundcards etc
https://www.retroianer.de - german retro computer board

Reply 73 of 75, by gerwin

User metadata
Rank l33t
Rank
l33t

I can finally post an update about this 486 VLB system. It was in limbo for a long time. But I recently decided to just finish things in a quick way. Especially the interfacing at the front of the case, so I can put the cover on the case. I cheated using quick tie-rib fixtures.
What changed since 2016: Different memory modules. Different VLB Video card. Different SD Card. Fixed the BIOS regarding >2GB HDD-size display (attached).

486 VLB - Current Config
- Motherboard: Original TK8498F VLB Motherboard, with UMC 498F Chipset + 256kB Cache. Added heat-sink to VRM. Replaced battery.
- PSU: Original one, Replaced Fan and connected it to just 5V.
- Case: Original one.
- BIOS: Award 1995 2.21r BIOS with >2GB HDD-size display fix. Stored in new 64kB EEPROM, with 5V jumper mod.
- CPU: Cyrix 5x86-100MHz (adjustable to 33MHz and 66MHz as well, with mechanical switch and/or SetMul)
- Memory: 2x16MB FPM 60ns.
- Multi-I/O: Boca Research IDEVL2M with Atronics 2015PL chipset. A VLB Multi-I/O without separate BIOS.
- Video card: Diamond Stealth 64 DRAM VLB. S3 Vision864 chipset. 2MB.
- Monitor: TFT DELL E151FP (2002), D-Sub input only. 1024x768 native, auto-upscale.
- Sound: Acer Magic S23, CS4232 + OPL3. Using IRQ 7. Combined with a Korg 4MB Midi daughter board.
- Storage: 8GB SD Card PNY (C10/U1) on KTC FC1306T adapter, on 90-degree angle bracket. With SD extension cable.
- Mouse: PS/2 optical mouse from Logitech, Type M-SBF96. With Matze79's PS/2 to serial mouse adapter. Using COM 1.
- Keyboard: Samsung SEM-K 20S with DIN connector.
- OS: Windows 95 OSR2 with MS-DOS 7.
(Currently no floppy disk or CD drive. Disabled the COM 2 port and the parallel port)

I have a Goldstar Prime 2C ISA Multi-I/O card that, according two to benchmarks, performs a bit better in DOS: 1566 kB/s compared to a depressing 1323 kB/s for the above VLB card. On the other hand, the VLB card does much better in Windows 95 with its driver.
In practice with the VLB card in DOS; loading up Doom 2 to menu screen takes 15 seconds. With smartdrv 14 seconds. I suppose the SD card's access time is helping out here.
The VLB Multi-I/O card FAQ says, that one is not supposed to use its DOS sys driver when LBA scheme is enabled. LBA is essential for interfacing the 8GB Drive.
Another thought is that the AMI BIOS "IDE 32-Bit Mode" toggle is surely not available, nor working with the Award BIOS that I am using. And I just found that exact concern mirrored in this recent topic pages 6 and 7.

EDIT: First I found that the BIOS-setting "Local Ready Delay Setting" should be put on "No Delay". Basically all BIOS delays and wait-states are now at the absolute minimum. This improves disk access a little. Next I messed around with the DOS VLB IDE driver: This card also has an updated DOS driver which is told to work with LBA enabled, contrary to the older one. But whatever the settings it makes disk access slower! At best it resulted in around 1000 kB/s. What I found though, is that the driver does enable something good on the IDE card, which remains active even after a warm reboot. So now I made this SPEED.bat file that loads the 2015PL.sys device driver using CTLOAD.com, then warm-reboots using REBOOT.com. This way I get 1725 kB/s IDE speed, without any driver loaded. It does take a warm reboot to get there (A warm reboot being like Ctrl-Alt-Del, but not like the case Reset button). That 1725 is a big improvement over 1323.

Last modification, before finally closing the case-cover: Lacking case airflow, the heat of the CPU and VRM was just being trapped. I don't like these old parts heating up, so I added a small 12V fan, on an improvised bracket. Also added small Gelid-branded fan-controller. 7V is seemingly sufficient to reliably start the fan, whilst decreasing the rpm to make it silent.

Some benchmarks comparable with the table on page 2 of this topic, now with S3 Vision864 graphics:
3D Bench = 83,3
3D Bench 1.0c = 81,1
PCP Bench VGA Mode = 17,5
PCP Bench 640x400, using UniVBE 5.0 = 7,5
Doom v1.9 Demo3 + SFX = 3863 / 3041 * 35 = 44,46 frames/sec
SpeedSys v4.78 = 52,23
Duke Nukem 3D, 320x200, SFX+Midi, Full HUD, initial view E1M1 = 29 frames/sec

Just noticed that this video card has tantalum capacitors. Not sure I trust that for the coming years. It seems to be standard issue for such a card.

Attachments

--> ISA Soundcard Overview // Doom MBF 2.04 // SetMul

Reply 74 of 75, by douglar

User metadata
Rank Oldbie
Rank
Oldbie

Can any of you 85C418 owners share your card BIOS?

My board has an eprom with these two option roms in it:

**** UMC SuperVGA UM85C4X8  70Hz - 1991, 1992 ****
BIOS VERSION 1.05, BOARD VERSION VL-1AV 11/22/93
(C) COPYRIGHT AWARD SOFTWARE, INC. 1991
(C) COPYRIGHT UNITED MICRO. CORP. 1989
GCW757VL VESA local bus IDE BIOS 
TEST ROM FOR UM82C418
Copyright 1992 UMC CO.,LTD
Date:6/8/1993

º VESA SUPER I/O Card Installed º
º ROM BIOS FOR UM82C418 º
º VERSION 1.02 º

The VGA BIOS doesn't seem to report the amount of installed vram correctly and the IDE bios prevents overlays and other option roms working with drives > 512MB.

Looking for something newer if it exists.

I know, it's strange that the IDE ROM says "GCW757VL" which is a different card, and also says "TEST ROM". The newest dos driver says VERSION 1.07, so maybe there is a newer ROM out there.

Reply 75 of 75, by douglar

User metadata
Rank Oldbie
Rank
Oldbie

@Gerwin, Thanks for the updated BIOS.

I put it here: https://www.vogonsdrivers.com/getfile.php?fileid=2143

**** UMC SuperVGA UM85C4X8 70Hz VESA V1.2 BIOS - 1994 ****  
BIOS VERSION 1.22, BOARD VERSION VL-1AV Date: 03/18/94