You only need to ground one pin on the QFP (don't remember which one anymore) and then any CPU will work in that socket.

allright thanks!
so it's this pin according to the ODP datasheet (this pin is N/C on 168pin ODPR 487SX version)

will try that 😀

edit:
Tiido found the 486SX datasheet - it's kinda tricky to get it so I am also linking it here... pin 156 is the #UP pin

https://www.mediafire.com/file/lxn6fznkxon4zo … /486SX.pdf/file

I just finished the upgrade, this trick works and you can install a normal 486 cpu in the 487 socket.
The system went from a 486sx25 to a dx2/50 , sadly even though it uses a VLSI82C480 chipset that TheLastByte supports it cannot be detected and if forced then the system hangs. So no UMB memory for this system 😵

keropi wrote on 2019-08-09, 11:43:

I just finished the upgrade, this trick works and you can install a normal 486 cpu in the 487

Hello, I'm sorry to reanimate an older thread but hoped you could answer a question for me.
I've just got the same PC and am hoping to use the ODP socket to upgrade using a normal DX or DX2 CPU.

Yuu say that the trick worked. Please can I please check exactly what you did? Did you ground the #UP in the ODP socket? If so where did you ground it to? Did you use a wire? Did you end up having to do anything to the onboard 486SX?

If it's not too much trouble, please could I trouble you for a photograph of your mod?

Thanks for sharing the info. you already have.

keropi wrote on 2019-08-09, 11:43:

I just finished the upgrade, this trick works and you can install a normal 486 cpu in the 487 socket.
The system went from a 486sx25 to a dx2/50 , sadly even though it uses a VLSI82C480 chipset that TheLastByte supports it cannot be detected and if forced then the system hangs. So no UMB memory for this system 😵

Try UMB_DRVR I've had better luck using that one with VLSI chipsets than TLB. Have not tried it with a 400 series VLSI though.

micr5l0th wrote on 2020-01-14, 20:26:

Hello, I'm sorry to reanimate an older thread but hoped you could answer a question for me. I've just got the same PC and am hop […]
Show full quote

keropi wrote on 2019-08-09, 11:43:

I just finished the upgrade, this trick works and you can install a normal 486 cpu in the 487

Hello, I'm sorry to reanimate an older thread but hoped you could answer a question for me.
I've just got the same PC and am hoping to use the ODP socket to upgrade using a normal DX or DX2 CPU.

Yuu say that the trick worked. Please can I please check exactly what you did? Did you ground the #UP in the ODP socket? If so where did you ground it to? Did you use a wire? Did you end up having to do anything to the onboard 486SX?

If it's not too much trouble, please could I trouble you for a photograph of your mod?

Thanks for sharing the info. you already have.

yes I have grounded the pin on the backside of the 487 socket... B14 to any ground like A11 (marked Vss on the socket diagram)
do you need a picture of that? I would have to completely disassemble the system to take it - not something I am looking forward too...

you can verify the correct pin with a multimeter, B14 is directly connected to pin156 #UP on the 486sx cpu , once you verify it connect it to whatever ground you want

maxtherabbit wrote on 2020-01-14, 21:53:

keropi wrote on 2019-08-09, 11:43:

I just finished the upgrade, this trick works and you can install a normal 486 cpu in the 487 socket.
The system went from a 486sx25 to a dx2/50 , sadly even though it uses a VLSI82C480 chipset that TheLastByte supports it cannot be detected and if forced then the system hangs. So no UMB memory for this system 😵

Try UMB_DRVR I've had better luck using that one with VLSI chipsets than TLB. Have not tried it with a 400 series VLSI though.

sadly it won't work , already tried it

When you say "no UMB memory", does that include what EMM386.SYS can normally provide?

Anonymous Coward wrote on 2020-01-15, 02:28:

When you say "no UMB memory", does that include what EMM386.SYS can normally provide?

no, that is different kind of UMB memory.
These things like TheLastBYTE, UMBPCI etc turn on the chipset UMB memory mapping. Actually that is the ROM shadowing function, being "abused" for use as UMB.
So this kind of UMB works in Real Mode, whereas EMM386 puts the processor into virtual mode. Which in turn is incompatible with some old DOS extenders. And that in turn makes it desirable to have hardware UMB, not only the speed penalty of about 5% which virtual mode introduces.

keropi wrote on 2020-01-14, 23:25:

yes I have grounded the pin on the backside of the 487 socket... B14 to any ground like A11 (marked Vss on the socket diagram) […]
Show full quote

micr5l0th wrote on 2020-01-14, 20:26:

Hello, I'm sorry to reanimate an older thread but hoped you could answer a question for me. I've just got the same PC and am hop […]
Show full quote

keropi wrote on 2019-08-09, 11:43:

I just finished the upgrade, this trick works and you can install a normal 486 cpu in the 487

Hello, I'm sorry to reanimate an older thread but hoped you could answer a question for me.
I've just got the same PC and am hoping to use the ODP socket to upgrade using a normal DX or DX2 CPU.

Yuu say that the trick worked. Please can I please check exactly what you did? Did you ground the #UP in the ODP socket? If so where did you ground it to? Did you use a wire? Did you end up having to do anything to the onboard 486SX?

If it's not too much trouble, please could I trouble you for a photograph of your mod?

Thanks for sharing the info. you already have.

yes I have grounded the pin on the backside of the 487 socket... B14 to any ground like A11 (marked Vss on the socket diagram)
do you need a picture of that? I would have to completely disassemble the system to take it - not something I am looking forward too...

you can verify the correct pin with a multimeter, B14 is directly connected to pin156 #UP on the 486sx cpu , once you verify it connect it to whatever ground you want

Thanks very much for taking time to reply, that's really useful info. I should be able to have a go now. Perhaps at the weekend I disassemble my machine and try and make the mod on the underside of the motherboard. No need for the photo but thanks again for the helpful advice.

I have a few IBM PS/1’s.
On one of them it had an onboard 486sx-25 ( I think they all come with this as a basic configuration ).
And the motherboard has an over drive socket like yours.
So I added an Intel Overdrive CPU. ( # sz802 ). And it worked fine.
The CPU reads 33 but it actually runs at 66mhz.

On one of the other IBM PS/1’s I was able to use a standard 486dx2-66 CPU but I had to change the jumpers on the motherboard.
You need to change the jumpers for 33mhz and not 25mhz. And you also need to change the jumpers to use a 486dx2 CPU and not so.

You need to find the manual for your model PS/1 as there are about 7 different models.

See my posts for IBM PS/1 hardware upgrades and Manuals.

Link:
IBM PS/1 Multimedia ( max restore ).

@micr5l0th
great, if you need any help just post a pic since you will have the mobo outside the case 😀

@Intel486dx33
yes overdrive cpus work just fine with no mods, what we do here is install regular 486DX2 cpus in the overdrive socket 😉

Thanks again @Keropi for the good, clear advice. Thanks to it I've made some nice progress withmy PS1.

So this weekend I stripped the machine and removed the motherboard. Using the pinouts posted earlier in this thread and a DMM in continuity mode I confirmed the correct pins. I was initially confused because the Overdrive pinout is shown from below, unlike the 486sx.

Initially I tried to solder a short wire from the desired pin to a low source but the only cable I had wasn't solid core and seemed to resist my best attempts to tin it, splaying awkwardly. In fact it initally prevented the board from POSTing at all. I checked my bits bin for something better to no avail and then happened upon a box of Breadboard jumpers, one of which was of more or less prfect length. This soldered easily between the two deired pins. I trimmed both ends with some edge cutters and tried a boot. Success!

For my revision of the board you'll note the positions of jumpers (top left of picture) 23,24 and 25 I'm using. My DX266 is currently functioning well, albeit as a DX2/50. I can't find any decscription of those jumpers beyond Page 236 of the linked mobo manual. I'll try and tinker and se if some combination can achieve a 33 insted of 25Mhz speed.

The DX2 runs quite hot so I added a passive heatsink with some thermal adhesive tape. I've upgraded the 4Mb RAM to 16MB (2 x 8MB 72 pin parity) from ebay. For storage I struggled a bit. I initally tried to use one of my IDE XT cards but it seems imossible to actually disable the onboard IDE, any connected device is always autodetected at subsequent boots. I worked through several DOMS, IDE Flash modules etc until I settled upon an old spinning Quantum Fireball - 15Gb and as the secondary device an IDE --> CF adapter with an 8GB CF Card.

As is my habit for many of my systems I first used the EZ-Drive DDO software, used it to boot a dos floppy image on my GOTEK clone and then used the excellent XFDISK software to create 2 x Primary and one Extended (with 2 drives in) partitions. EZ-Drive and XFDISK seem to play well together, I get upto 4x2GB partitions per drive and a boot menu that allows me to multi boot.

Currently the machine multiboots between Dos 6.22 (WFWG3.11) and Dos 7 with two 2GB data drives shared.

I finished the install with a couple of ISA cards, a 3com etherlink 3 for network access and an Avance Logic ALS 100 + sound card.

I like the front case flap and note that as well as the 3.5" and 5.25" drive bays there's another cut away on the left that seems just to be designed to save metal that should allow me to front mount the IDE / CF card with a small cut to the front plastic.

The onboard VGA is a Cirrus Logic model that I think connects via VLB. Sadly it only has 512k , although it looks like there might be pads for a second video memory chip? The only other upgrade I'd like to make is some external cache chips. As time allows I'll hunt throughmy spares and see what I can find.

All in all I'm quite leased with the PS1 and was thinking about tinkering with OS/2 on it as it's an OS I've almost no experience of. Once again, big thanks to @Keropi for taking time to help me out.

excellent! nicely done!
my PS1 also has a 25mhz fsb so DX2/50 is the best it can do... still it's a great and cheap speed boost
enjoy! I must return to the system at some point, atm it's packed away in storage...

Just a quick note here, a 487 has one signal in a different place than a 486DX CPU - FERR. So this method might work or not depending on the BIOS code, and then some programs that make use of the FPU and do ecounter numeric exceptions might crash or hang the system. If it does boot though and FPU is unsused (or does not ecounter exceptions) it should be stable.
Also, in the same way it might be possible to use an SX2 chip - this depends on how the mobo handles NMI (on typical PCs it's only used for reporting critical errors like memory parity mismatch). I haven't tried SX or SX2 in a 487 socket but it does boot on my mobo in a standard 486 socket configured for DX CPU.

Hi. I have IBM PS/1 type 2168 model 452. The mobo installed is soldered SX25 / two 72 pin SIMM / 1+4 L2 cache sockets. It looks like this owned by @micr5l0th.
Previously I've installed regular DX2-66 (running at 2*25 mHz) with pin B14 at CPU pulled to ground (pin A11 at CPU) and it succesfully does boot into Win3.1.
Now I've tried overdive DX4ODP75 (P23T) but got "129 Cache Error".

Spoiler

Don't ask me why but it made me test to which points at mobo the j23, j24, j25 pins are connected.
All socket pins numbered as for CPU (not as labeled at mobo):

j23 pins:
1 - connected to j24.3
2 - B15 at socket: MNI for almost all 486 chips, except SX. For SX chips - NC.
3 - NC
j24 pins:
1 - pins 77 and 78 at soldered SX. For soldered intel 486sx25 is NC. It is wiered. Maybe it for other manufacturers' QFP chips.
2 - A15 at socket: IGGNE# for almost all 486 chips, except SX. SX chip have NMI at this pin.
3 - connected to j23.1
+ pin 100 at soldered SX - it is NMI.
+ pin 80 of VL82C480-FC1 (chipset).[/m]
So j23, j24 pins 1-2 should be short for almost all CPUs installed. And 2-3 short for SX (not for overdrive chips marked as SX). But it looks strange to put the SX processor if it is already soldered in the motherboard..

j25 pins:
1 - C14 at socket: INC or NC for SX, 487SX and ODP (P23T). FERR# for DX, DX2, DX4 and ODPR.
2 - pin 81 of VL82C480-FC1 (chipset)
+ pin 81 at soldered SX. For soldered intel is NC. It is wiered. Maybe it for other manufacturers' QFP chips.
3 - A13 at socket: FERR# for SX, 487SX and ODP (P23T). For other intel's chips - NC (for not intel's some other siglas may be here).

Deunan wrote on 2020-01-22, 10:26:

Just a quick note here, a 487 has one signal in a different place than a 486DX CPU - FERR. So this method might work or not depending on the BIOS code, and then some programs that make use of the FPU and do ecounter numeric exceptions might crash or hang the system. If it does boot though and FPU is unsused (or does not ecounter exceptions) it should be stable.
Also, in the same way it might be possible to use an SX2 chip - this depends on how the mobo handles NMI (on typical PCs it's only used for reporting critical errors like memory parity mismatch). I haven't tried SX or SX2 in a 487 socket but it does boot on my mobo in a standard 486 socket configured for DX CPU.

So for FERR#:
- leave j25 at 1-2 if you installed regular DX or DX2 (with B14 pulled to ground)
- or put j25 to 2-3 if you installed overdrive marked as ODP or 487 chip.

All pinouts for 486 found here.

Lets back to my overdive DX4ODP75. It has 16k internal L1 cache and released in October, 1994 (all DX4 intel chips was released start from middle of 1994).
CMOS setup on my mobo reported BIOS dated 01/22/93:

So I think it can't regognize correctly 16k L1 cache and disables it following the error 129.
I found dump "P/N 52G2974, dated 26/04/93, for model 2133-451 (EMEA)" here.
My BIOS ic labeled as "52G2974 LE[C]IBM 08/09/93" but as i wrote the date reported in CMOS setup is 01/22/93. So i suppose it is the same BIOS version. I will try it later to be sure for 100%.

Anyone have ROM BIOS dump dated later than January, 1993 for mobo which I have (IBM PS/1 type 2168 model 452)?

keropi wrote on 2019-08-06, 18:51:

allright thanks! so it's this pin according to the ODP datasheet (this pin is N/C on 168pin ODPR 487SX version) […]
Show full quote

allright thanks!
so it's this pin according to the ODP datasheet (this pin is N/C on 168pin ODPR 487SX version)

will try that 😀

edit:
Tiido found the 486SX datasheet - it's kinda tricky to get it so I am also linking it here... pin 156 is the #UP pin

https://www.mediafire.com/file/lxn6fznkxon4zo … /486SX.pdf/file

Thanks, that datasheet was missing from my library

In Case this is still relevant, I have the same issue on a 2133-451 and just used this tool with a 486obd75

Cache Enable/Disable Utility for DOS
http://vogonsdrivers.com/getfile.php?fileid=74

Socket 1/2/3 Voltage Interposer Tweaker (Alpha)

If you guys have any suggestions for anything more to add to this project let me know.

I’ve included instructions for the b14 mod found here on it.