VOGONS


First post, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie

WARNING: long post ahead! 😁

I've had an ASUS VL/I-486SV2G REV. 1.8 motherboard for a while, and after reading a couple of threads here I realized it was possible to convert it to the X4 version that supports lower voltage CPUs. Also, I noticed that most of the official and unofficial information about jumper settings refer to revisions 2.x of the board, which have significant differences to 1.x (in fact, I don't know if user manuals for revisions 1.x exist).

So, I figured it would be useful to have a comprehensive guide on how to do the mod for revision 1.8 as well as a detailed description of what each jumper does. My main sources of information were these two great threads:

Conversion guide for rev. 2.0, which includes many useful details, especially regarding voltage detection
Conversion guide for rev. 1.7, very similar to my 1.8 but does not do voltage detection and I noticed several differences in the jumpers

And also Jan Steunebrink's 5x86 upgrade page, which was extremely helpful when trying to determine what each jumper does with regards to L1 WB cache, VOLDET and clock multiplier.

So let's get started!

Components required

Here's an image of the board before the mod, with the locations where the new components will go, highlighted in color:

The attachment before_small.png is no longer available

List of components:

  • LT1085CT or equivalent adjustable voltage regulator [PURPLE]
  • N-channel power MOSFET [BLUE], I used a STP16NF06L as suggested by user d0pefish in his 2.0 conversion
  • 2 x 10uf 16V and 2 x 1uf 16V tantalum capacitors [RED]
  • 330uf 16V electrolytic capacitor [YELLOW]
  • 7407 hex buffer [GREEN]
  • Two jumper wires to remove [LIGHT BLUE] and one to add [ORANGE]

This is how it looks after adding the components (don't forget to remove the sticker to reveal the full name! 😎 ) :

The attachment after_small.png is no longer available

Voltage detection
An output of the 7407 is connected to the MOSFET gate, while the drain is connected to +5V and the source is connected to the regulator output.

To enable CPU voltage detection, place a jumper cap over pins 4-5 of JP20. This will connect the VOLDET pin of the CPU to the corresponding input of the 7407, so a lower voltage CPU such as DX4 or 5x86 which ground VOLDET internally will use the 3.4V regulator output, while 5V CPUs will cause the regulator output to be shorted to +5V.

Leaving this jumper cap off will provide a fixed +5V to the CPU.

Write-back L1 cache
For some reason ASUS only provided L1 WB settings for 2.x boards, but placing jumper caps over the following pins will enable it on 1.8 (make sure to have the latest BIOS!):

  1. JP16 pins 1-2 (this connects the WB/WT# pin of the CPU to a pull-up resistor)
  2. JP17 pins 2-3 (this connects the W/R# output to the INV input of the CPU)
  3. JP18 pins 2-3 (this connects the HITM# input of the chipset to the HITM# output of the CPU)
  4. Finally, set both JP5 and JP6 to 1-2 for the proper L1 WB trap setting of the chipset

For WT L1, just leave the JP16, 17 and 18 jumpers off and set JP5 to 1-2 and JP6 to 2-3.

Clock multiplier
For CPUs that support the CLKMUL signal (such as DX4 and 5x86), the multiplier is controlled by pins 4, 5 & 6 of JP17:

  • Leaving these 3 pins unconnected will cause CLKMUL to float, which sets the default 3x multiplier
  • Placing a cap over pins 5-6 drives CLKMUL low, this sets the multiplier to 2x (4x for the AMD 5x86)
  • Placing a cap over pins 4-5 connects CLKMUL to the BREQ pin, this should set 2.5x if the CPU supports it (untested)

Other jumper settings

  • To enable power management on CPUs that support it, set JP16 3-4 (SRESET) and 5-6 (SMI ACT), as well as JP18 4-5 (SMI)
  • For CPUs without NPU (SX) set JP19 to 2-3, the default is 1-2 for DX type
  • Be sure to set JP20 to 2-3, setting it to 1-2 apparently connects the CACHE# output of the CPU to the corresponding chipset input for L1 WB, but doing so causes the system to hang during POST, so it looks like the CACHE# line is not used by the chipset on this board revision (also, it's not mandatory for L1 WB implementation according to Steunebrink)

Well, I hope this is a good source of info for owners of the 1.8 revision of this great board, as it was frustrating to me that pretty much all documentation I could find was on 2.x!

Reply 1 of 9, by Iano

User metadata
Rank Newbie
Rank
Newbie

I am working on a conversion of a 486SV2 to GX4 REV 1.7 and I can't get anything but 5v (4.8-4.9v actually) output from the regulator. I was hoping someone with some electronics knowledge can help me troubleshoot this. I wonder if the LT1085CT that I bought are not working right or if the board is not providing the right conditions for 3.x volts. Can someone tell me how these regulators work? I know there is Vin and Vout and Adjust but I'm not sure how to test that it is doing what it's supposed to. I bought the parts through Aliexpress so who knows what I really got.

This is what I know so far.
First I placed all the parts on the board (the 10uf, 1uf and 330uf caps, 7407, 16NF06 and the LT1085) and powered up without CPU and tested Vout (the middle pin of the regulator as well as Vcc at the socket) measured 5v volts. JP20 4-5 is jumpered to engage voltage regulator via VOLDET. Got 5v. I shorted VOLDET to GND at the socket, and I get 4.8v. I placed two different 3.3v CPU in the socket (AMD DX266 and AM5x86-133) which should signal 3.3/3.45 v via VOLDET. Still get 4.8v. I the switched out the 7404 and the regulator. still same results.

When I removed just the regulator (LT1085) and place a 5v CPU (Intel DX266) I get 5v and system boots up because the 16NF06 is bypassing the regulator. When I place a 3.3v CPU in the socket (above AMD CPUs) I get no power up, Measuring 1.xx volts at Vcc (measured at Vcc where the middle regulator pin would be). If I remove JP20 4-5 the system boots up and Vcc is again at 5v (with the 3.3v CPU installed). This tells me that the board is correctly switching between 5V and 3.3v regulator depending on the CPU installed. So the 7407 and 16NF06 are working. When I put the regulator back in I get only 4.8v from the regulator output again.

So how do I test this? How do I know the board is sending the right signal to the 1085? and how do I know the regulator is doing what it should? Can someone tell me how to test the regulator independent of the board?

Looking at the data sheet for this 1085 the Vin is on one side, the adjust is on the other side, the Vout is the middle (Vcc). If Vin and Vout are connected to the board (and 5v is supplied) but adjust is left disconnected... what should the output be? (I still get about 5v). What should I be measuring at the adjust pin?

With the regulator taken off and 3volt CPU installed. I measure 5v at Vin and I measure 1.2v at Vcc and 0.7v at adjust. Obviously no power to the CPU without the regulator engaged. With regulator back on the board I measure 4.8v at Vout and 3.0v at adjust. When I take JP32 off I get 4.7v at adjust but Vout still does not change (stays fixed at almost 5v). What makes the regulator produce 3.3V?

I just don't know enough to identify the problem here.

(EDIT: Initially I reported 5v from the regulator. I want to clarify (without rounding) that I am getting about 4.8v from the regulator... and 5v when bypassing the regulator. I edited my numbers reported above)

Last edited by Iano on 2021-04-03, 02:05. Edited 1 time in total.

Reply 2 of 9, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie

If you have a 5V supply, you can test the regulator separately by following this diagram (look up "voltage divider" if you want more details on how this works):

The attachment LT1085CT.png is no longer available

According to the datasheet, the VREF voltage between the output and adjust terminals should be 1.25V and the IADJ current is negligible, which simplifies the VOUT formula to:

VOUT = 1.25 * (1 + R2/R1)

In the case of the ASUS board, R1 would be R35 (120 ohm) and R2 is R34 (210 ohm) if I'm not mistaken, which produces a 3.4V output (there are other resistors for multiple adjustment options via jumpers on later versions of the board).

When testing the regulator on its own, remember that you need input and output capacitors for stability. Values are specified in the datasheet.

Reply 3 of 9, by Iano

User metadata
Rank Newbie
Rank
Newbie

Thanks for the quick reply. I must have the wrong component for the LT1085CT. I measured resistance on the board between Vout and Adjust at 120 ohm, and between Adjust and Ground at 210 ohm. Vin is 5v. I don't think there is a problem there.

I tested the regulator on a simple homemade circuit following that schematic. Vin was 4.9v. I had 120 ohm and 220 ohm resistors for R1 and R2 which should produce about 3.5 volts from the regulator Vout. I measured 4.72v. I switched R2 for a 150 ohm resistor, which by that formula should produce about 2.8v. Still 4.72v (which is about the same voltage I measured when it was installed on the board, just below 5v Vin). Vref was measuring 2.12v, not 1.25v. (I hope I am testing this correctly). I even tried a 12 Vin 12.38v in and got 12.1v out regardless of what I did with the resistors (Vref was about 5v on that try). I bought 5 of these and they all behave the same way. It's unlikely I got 5 bad components but I have to assume either they are not what I think they are or I'm doing something wrong (probably). Either way I ordered some different 1085CTs from an ebay vendor. This should hopefully answer my question.

Here is a picture of my circuit. If you have any other suggestions of how I can verify this component is the correct part please let me know. I have a hard time believing this is just a bad part (mislabeled maybe). I'm almost sure this is user error but I can't think of how to determine what is wrong. If I can't simulate correct function of the regulator off the board then I've got to assume that's the problem.

Reply 4 of 9, by Iano

User metadata
Rank Newbie
Rank
Newbie

Just to follow up... New 1085 regulators work as expected. I went through two orders that did not work correctly. Lesson learned, get this part from a reputable vendor.

Reply 5 of 9, by arncht

User metadata
Rank Oldbie
Rank
Oldbie
TheMobRules wrote on 2019-01-26, 06:20:
WARNING: long post ahead! :-D […]
Show full quote

WARNING: long post ahead! 😁

I've had an ASUS VL/I-486SV2G REV. 1.8 motherboard for a while, and after reading a couple of threads here I realized it was possible to convert it to the X4 version that supports lower voltage CPUs. Also, I noticed that most of the official and unofficial information about jumper settings refer to revisions 2.x of the board, which have significant differences to 1.x (in fact, I don't know if user manuals for revisions 1.x exist).

So, I figured it would be useful to have a comprehensive guide on how to do the mod for revision 1.8 as well as a detailed description of what each jumper does. My main sources of information were these two great threads:

Conversion guide for rev. 2.0, which includes many useful details, especially regarding voltage detection
Conversion guide for rev. 1.7, very similar to my 1.8 but does not do voltage detection and I noticed several differences in the jumpers

And also Jan Steunebrink's 5x86 upgrade page, which was extremely helpful when trying to determine what each jumper does with regards to L1 WB cache, VOLDET and clock multiplier.

So let's get started!

Components required

Here's an image of the board before the mod, with the locations where the new components will go, highlighted in color:

before_small.png
List of components:

  • LT1085CT or equivalent adjustable voltage regulator [PURPLE]
  • N-channel power MOSFET [BLUE], I used a STP16NF06L as suggested by user d0pefish in his 2.0 conversion
  • 2 x 10uf 16V and 2 x 1uf 16V tantalum capacitors [RED]
  • 330uf 16V electrolytic capacitor [YELLOW]
  • 7407 hex buffer [GREEN]
  • Two jumper wires to remove [LIGHT BLUE] and one to add [ORANGE]

This is how it looks after adding the components (don't forget to remove the sticker to reveal the full name! 😎 ) :

after_small.png
Voltage detection
An output of the 7407 is connected to the MOSFET gate, while the drain is connected to +5V and the source is connected to the regulator output.

To enable CPU voltage detection, place a jumper cap over pins 4-5 of JP20. This will connect the VOLDET pin of the CPU to the corresponding input of the 7407, so a lower voltage CPU such as DX4 or 5x86 which ground VOLDET internally will use the 3.4V regulator output, while 5V CPUs will cause the regulator output to be shorted to +5V.

Leaving this jumper cap off will provide a fixed +5V to the CPU.

Write-back L1 cache
For some reason ASUS only provided L1 WB settings for 2.x boards, but placing jumper caps over the following pins will enable it on 1.8 (make sure to have the latest BIOS!):

  1. JP16 pins 1-2 (this connects the WB/WT# pin of the CPU to a pull-up resistor)
  2. JP17 pins 2-3 (this connects the W/R# output to the INV input of the CPU)
  3. JP18 pins 2-3 (this connects the HITM# input of the chipset to the HITM# output of the CPU)
  4. Finally, set both JP5 and JP6 to 1-2 for the proper L1 WB trap setting of the chipset

For WT L1, just leave the JP16, 17 and 18 jumpers off and set JP5 to 1-2 and JP6 to 2-3.

Clock multiplier
For CPUs that support the CLKMUL signal (such as DX4 and 5x86), the multiplier is controlled by pins 4, 5 & 6 of JP17:

  • Leaving these 3 pins unconnected will cause CLKMUL to float, which sets the default 3x multiplier
  • Placing a cap over pins 5-6 drives CLKMUL low, this sets the multiplier to 2x (4x for the AMD 5x86)
  • Placing a cap over pins 4-5 connects CLKMUL to the BREQ pin, this should set 2.5x if the CPU supports it (untested)

Other jumper settings

  • To enable power management on CPUs that support it, set JP16 3-4 (SRESET) and 5-6 (SMI ACT), as well as JP18 4-5 (SMI)
  • For CPUs without NPU (SX) set JP19 to 2-3, the default is 1-2 for DX type
  • Be sure to set JP20 to 2-3, setting it to 1-2 apparently connects the CACHE# output of the CPU to the corresponding chipset input for L1 WB, but doing so causes the system to hang during POST, so it looks like the CACHE# line is not used by the chipset on this board revision (also, it's not mandatory for L1 WB implementation according to Steunebrink)

Well, I hope this is a good source of info for owners of the 1.8 revision of this great board, as it was frustrating to me that pretty much all documentation I could find was on 2.x!

The wb cpus came out just in the late 95, you need:
* mid 95 bios
* proper jumpering

The earlier bioses also supporting the “wb” option, but practically the cpu works in wt mode. Worth to check the result in dos.

But i have question… i would like to build an authentic dx4 rig from 1994. Previously i had an early svg0gx4, the bios does not support the lba and the dx4 marked as “p24t” on the bios screen, which is correct, but not nice 😀

I have more sv2gx4, but with 2.x. These boards sold (and mostly manifactured) in 1995.

So… the original bios on these 1.x boards support lba and the dx4 properly?

My little retro computer world
Overdoze of the demoscene

Reply 6 of 9, by TheMobRules

User metadata
Rank Oldbie
Rank
Oldbie
arncht wrote on 2024-02-27, 06:04:

But i have question… i would like to build an authentic dx4 rig from 1994. Previously i had an early svg0gx4, the bios does not support the lba and the dx4 marked as “p24t” on the bios screen, which is correct, but not nice 😀

I have more sv2gx4, but with 2.x. These boards sold (and mostly manifactured) in 1995.

So… the original bios on these 1.x boards support lba and the dx4 properly?

My board is from early '94 according to the datecodes on the components (around April-May), the BIOS that came with it (ver. 0204) works just like your SVGOGX4 in the sense that there's no LBA support and the DX4 is supported but detected as P24C. Also it only supports 1 IDE channel instead of 2 like later BIOSes.

There are at least a couple other newer BIOSes from 1994, ver. 0303 and 0304 which detect the DX4 properly and I think they have LBA support. However those two seem to have introduced a bug that prevents stuff like MIDI cards from using IRQ2/9 and that was finally fixed in the first BIOS from 1995 (ver. 0305) that was likely released alongside the 2.0 boards. This is mentioned in the notes for that BIOS:

1. Turn on IRQ9 interrupt mask bit for some bus mouses,sound cards, and IBM 3278/79 emulation cards.

Reply 7 of 9, by CoffeeOne

User metadata
Rank Oldbie
Rank
Oldbie
arncht wrote on 2024-02-27, 06:04:
The wb cpus came out just in the late 95, you need: * mid 95 bios * proper jumpering […]
Show full quote
TheMobRules wrote on 2019-01-26, 06:20:
WARNING: long post ahead! :-D […]
Show full quote

WARNING: long post ahead! 😁

I've had an ASUS VL/I-486SV2G REV. 1.8 motherboard for a while, and after reading a couple of threads here I realized it was possible to convert it to the X4 version that supports lower voltage CPUs. Also, I noticed that most of the official and unofficial information about jumper settings refer to revisions 2.x of the board, which have significant differences to 1.x (in fact, I don't know if user manuals for revisions 1.x exist).

So, I figured it would be useful to have a comprehensive guide on how to do the mod for revision 1.8 as well as a detailed description of what each jumper does. My main sources of information were these two great threads:

Conversion guide for rev. 2.0, which includes many useful details, especially regarding voltage detection
Conversion guide for rev. 1.7, very similar to my 1.8 but does not do voltage detection and I noticed several differences in the jumpers

And also Jan Steunebrink's 5x86 upgrade page, which was extremely helpful when trying to determine what each jumper does with regards to L1 WB cache, VOLDET and clock multiplier.

So let's get started!

Components required

Here's an image of the board before the mod, with the locations where the new components will go, highlighted in color:

before_small.png
List of components:

  • LT1085CT or equivalent adjustable voltage regulator [PURPLE]
  • N-channel power MOSFET [BLUE], I used a STP16NF06L as suggested by user d0pefish in his 2.0 conversion
  • 2 x 10uf 16V and 2 x 1uf 16V tantalum capacitors [RED]
  • 330uf 16V electrolytic capacitor [YELLOW]
  • 7407 hex buffer [GREEN]
  • Two jumper wires to remove [LIGHT BLUE] and one to add [ORANGE]

This is how it looks after adding the components (don't forget to remove the sticker to reveal the full name! 😎 ) :

after_small.png
Voltage detection
An output of the 7407 is connected to the MOSFET gate, while the drain is connected to +5V and the source is connected to the regulator output.

To enable CPU voltage detection, place a jumper cap over pins 4-5 of JP20. This will connect the VOLDET pin of the CPU to the corresponding input of the 7407, so a lower voltage CPU such as DX4 or 5x86 which ground VOLDET internally will use the 3.4V regulator output, while 5V CPUs will cause the regulator output to be shorted to +5V.

Leaving this jumper cap off will provide a fixed +5V to the CPU.

Write-back L1 cache
For some reason ASUS only provided L1 WB settings for 2.x boards, but placing jumper caps over the following pins will enable it on 1.8 (make sure to have the latest BIOS!):

  1. JP16 pins 1-2 (this connects the WB/WT# pin of the CPU to a pull-up resistor)
  2. JP17 pins 2-3 (this connects the W/R# output to the INV input of the CPU)
  3. JP18 pins 2-3 (this connects the HITM# input of the chipset to the HITM# output of the CPU)
  4. Finally, set both JP5 and JP6 to 1-2 for the proper L1 WB trap setting of the chipset

For WT L1, just leave the JP16, 17 and 18 jumpers off and set JP5 to 1-2 and JP6 to 2-3.

Clock multiplier
For CPUs that support the CLKMUL signal (such as DX4 and 5x86), the multiplier is controlled by pins 4, 5 & 6 of JP17:

  • Leaving these 3 pins unconnected will cause CLKMUL to float, which sets the default 3x multiplier
  • Placing a cap over pins 5-6 drives CLKMUL low, this sets the multiplier to 2x (4x for the AMD 5x86)
  • Placing a cap over pins 4-5 connects CLKMUL to the BREQ pin, this should set 2.5x if the CPU supports it (untested)

Other jumper settings

  • To enable power management on CPUs that support it, set JP16 3-4 (SRESET) and 5-6 (SMI ACT), as well as JP18 4-5 (SMI)
  • For CPUs without NPU (SX) set JP19 to 2-3, the default is 1-2 for DX type
  • Be sure to set JP20 to 2-3, setting it to 1-2 apparently connects the CACHE# output of the CPU to the corresponding chipset input for L1 WB, but doing so causes the system to hang during POST, so it looks like the CACHE# line is not used by the chipset on this board revision (also, it's not mandatory for L1 WB implementation according to Steunebrink)

Well, I hope this is a good source of info for owners of the 1.8 revision of this great board, as it was frustrating to me that pretty much all documentation I could find was on 2.x!

The wb cpus came out just in the late 95, you need:
* mid 95 bios
* proper jumpering

The earlier bioses also supporting the “wb” option, but practically the cpu works in wt mode. Worth to check the result in dos.

But i have question… i would like to build an authentic dx4 rig from 1994. Previously i had an early svg0gx4, the bios does not support the lba and the dx4 marked as “p24t” on the bios screen, which is correct, but not nice 😀

I have more sv2gx4, but with 2.x. These boards sold (and mostly manifactured) in 1995.

So… the original bios on these 1.x boards support lba and the dx4 properly?

I recommend to always to latest patched bios: SV2G4fix.bin, it is derived from 4.02.1 (patched by Feipoa as far as I know). I can confirm it works with a SV2GX4 Rev. 2.0, but also with SVGOX4 Rev. 1.1.
It supports all CPUs, and switches the L2 to 7+1 mode, when Write Back is configured.

EDIT: Do you think updating the Bios removes authencity? OK, then it is not an option.

Reply 8 of 9, by arncht

User metadata
Rank Oldbie
Rank
Oldbie
TheMobRules wrote on 2024-02-27, 06:28:
My board is from early '94 according to the datecodes on the components (around April-May), the BIOS that came with it (ver. 020 […]
Show full quote
arncht wrote on 2024-02-27, 06:04:

But i have question… i would like to build an authentic dx4 rig from 1994. Previously i had an early svg0gx4, the bios does not support the lba and the dx4 marked as “p24t” on the bios screen, which is correct, but not nice 😀

I have more sv2gx4, but with 2.x. These boards sold (and mostly manifactured) in 1995.

So… the original bios on these 1.x boards support lba and the dx4 properly?

My board is from early '94 according to the datecodes on the components (around April-May), the BIOS that came with it (ver. 0204) works just like your SVGOGX4 in the sense that there's no LBA support and the DX4 is supported but detected as P24C. Also it only supports 1 IDE channel instead of 2 like later BIOSes.

There are at least a couple other newer BIOSes from 1994, ver. 0303 and 0304 which detect the DX4 properly and I think they have LBA support. However those two seem to have introduced a bug that prevents stuff like MIDI cards from using IRQ2/9 and that was finally fixed in the first BIOS from 1995 (ver. 0305) that was likely released alongside the 2.0 boards. This is mentioned in the notes for that BIOS:

1. Turn on IRQ9 interrupt mask bit for some bus mouses,sound cards, and IBM 3278/79 emulation cards.

thx, it was useful 😀 i also use a midi card in the dx4, so i stuck with the 95 boards.

My little retro computer world
Overdoze of the demoscene

Reply 9 of 9, by arncht

User metadata
Rank Oldbie
Rank
Oldbie
CoffeeOne wrote on 2024-02-27, 17:45:
arncht wrote on 2024-02-27, 06:04:
The wb cpus came out just in the late 95, you need: * mid 95 bios * proper jumpering […]
Show full quote
TheMobRules wrote on 2019-01-26, 06:20:
WARNING: long post ahead! :-D […]
Show full quote

WARNING: long post ahead! 😁

I've had an ASUS VL/I-486SV2G REV. 1.8 motherboard for a while, and after reading a couple of threads here I realized it was possible to convert it to the X4 version that supports lower voltage CPUs. Also, I noticed that most of the official and unofficial information about jumper settings refer to revisions 2.x of the board, which have significant differences to 1.x (in fact, I don't know if user manuals for revisions 1.x exist).

So, I figured it would be useful to have a comprehensive guide on how to do the mod for revision 1.8 as well as a detailed description of what each jumper does. My main sources of information were these two great threads:

Conversion guide for rev. 2.0, which includes many useful details, especially regarding voltage detection
Conversion guide for rev. 1.7, very similar to my 1.8 but does not do voltage detection and I noticed several differences in the jumpers

And also Jan Steunebrink's 5x86 upgrade page, which was extremely helpful when trying to determine what each jumper does with regards to L1 WB cache, VOLDET and clock multiplier.

So let's get started!

Components required

Here's an image of the board before the mod, with the locations where the new components will go, highlighted in color:

before_small.png
List of components:

  • LT1085CT or equivalent adjustable voltage regulator [PURPLE]
  • N-channel power MOSFET [BLUE], I used a STP16NF06L as suggested by user d0pefish in his 2.0 conversion
  • 2 x 10uf 16V and 2 x 1uf 16V tantalum capacitors [RED]
  • 330uf 16V electrolytic capacitor [YELLOW]
  • 7407 hex buffer [GREEN]
  • Two jumper wires to remove [LIGHT BLUE] and one to add [ORANGE]

This is how it looks after adding the components (don't forget to remove the sticker to reveal the full name! 😎 ) :

after_small.png
Voltage detection
An output of the 7407 is connected to the MOSFET gate, while the drain is connected to +5V and the source is connected to the regulator output.

To enable CPU voltage detection, place a jumper cap over pins 4-5 of JP20. This will connect the VOLDET pin of the CPU to the corresponding input of the 7407, so a lower voltage CPU such as DX4 or 5x86 which ground VOLDET internally will use the 3.4V regulator output, while 5V CPUs will cause the regulator output to be shorted to +5V.

Leaving this jumper cap off will provide a fixed +5V to the CPU.

Write-back L1 cache
For some reason ASUS only provided L1 WB settings for 2.x boards, but placing jumper caps over the following pins will enable it on 1.8 (make sure to have the latest BIOS!):

  1. JP16 pins 1-2 (this connects the WB/WT# pin of the CPU to a pull-up resistor)
  2. JP17 pins 2-3 (this connects the W/R# output to the INV input of the CPU)
  3. JP18 pins 2-3 (this connects the HITM# input of the chipset to the HITM# output of the CPU)
  4. Finally, set both JP5 and JP6 to 1-2 for the proper L1 WB trap setting of the chipset

For WT L1, just leave the JP16, 17 and 18 jumpers off and set JP5 to 1-2 and JP6 to 2-3.

Clock multiplier
For CPUs that support the CLKMUL signal (such as DX4 and 5x86), the multiplier is controlled by pins 4, 5 & 6 of JP17:

  • Leaving these 3 pins unconnected will cause CLKMUL to float, which sets the default 3x multiplier
  • Placing a cap over pins 5-6 drives CLKMUL low, this sets the multiplier to 2x (4x for the AMD 5x86)
  • Placing a cap over pins 4-5 connects CLKMUL to the BREQ pin, this should set 2.5x if the CPU supports it (untested)

Other jumper settings

  • To enable power management on CPUs that support it, set JP16 3-4 (SRESET) and 5-6 (SMI ACT), as well as JP18 4-5 (SMI)
  • For CPUs without NPU (SX) set JP19 to 2-3, the default is 1-2 for DX type
  • Be sure to set JP20 to 2-3, setting it to 1-2 apparently connects the CACHE# output of the CPU to the corresponding chipset input for L1 WB, but doing so causes the system to hang during POST, so it looks like the CACHE# line is not used by the chipset on this board revision (also, it's not mandatory for L1 WB implementation according to Steunebrink)

Well, I hope this is a good source of info for owners of the 1.8 revision of this great board, as it was frustrating to me that pretty much all documentation I could find was on 2.x!

The wb cpus came out just in the late 95, you need:
* mid 95 bios
* proper jumpering

The earlier bioses also supporting the “wb” option, but practically the cpu works in wt mode. Worth to check the result in dos.

But i have question… i would like to build an authentic dx4 rig from 1994. Previously i had an early svg0gx4, the bios does not support the lba and the dx4 marked as “p24t” on the bios screen, which is correct, but not nice 😀

I have more sv2gx4, but with 2.x. These boards sold (and mostly manifactured) in 1995.

So… the original bios on these 1.x boards support lba and the dx4 properly?

EDIT: Do you think updating the Bios removes authencity? OK, then it is not an option.

for me yes 😀 i like to do time capsules from every perspective, tools, drv, etc.

My little retro computer world
Overdoze of the demoscene