Can't find a datasheet for the SC1152CSW, but I'm assuming it's like the SC1153CSW except for the Power Good pin being inverted. Wouldn't it just be easier to mod the slocket with an adjustable linear regulator, though? I mean, current draw is tiny.

The C3 can be a bit of a bugger as far as motherboard compatibility is concerned. See this thread :via c3 nehemiah motherboard compatability
Out of 4 socket 370 motherboards only a SiS 630 chipset based one would support the nehemiah properly for me.
I built a machine around this board and CPU last year out of surplus parts, it is a good little machine, thread here:My new VIA C3 "Nehemiah" build, the weird and the wonderful!
I found performance of a C3 nehemiah at 1200MHz to be about equivalent to a Celeron coppermine at 600MHz. The above machine now has a PIII copermine at 733MHz installed and is far snappier for everything. Of course the C3 has the advantage of being able to down clock & lower the multiplier with setmul.
I found the older C3's much easier to get working in the average motherboard, be warned though that the original Samuel without L2 cache is VERY,VERY slow.
Best,
Chris

gdjacobs wrote:

Can't find a datasheet for the SC1152CSW, but I'm assuming it's like the SC1153CSW except for the Power Good pin being inverted. Wouldn't it just be easier to mod the slocket with an adjustable linear regulator, though? I mean, current draw is tiny.

Yeah, if I'd have to do more than simply replacing the voltage controller, modding the slocket would definitely be the better option. Leaving the mobo unmodded is a good thing if possible, and it would also give me more future options. I mean, a slocket is only a slocket, it either works or not, and if it does, there's no point in replacing it later. A mobo is a completely different thing, there might be plenty of other reasons to replace it in the future.

Couldn't find that datasheet either.

BSA Starfire wrote:

The C3 can be a bit of a bugger as far as motherboard compatibility is concerned. See this thread :via c3 nehemiah motherboard c […]
Show full quote

The C3 can be a bit of a bugger as far as motherboard compatibility is concerned. See this thread :via c3 nehemiah motherboard compatability
Out of 4 socket 370 motherboards only a SiS 630 chipset based one would support the nehemiah properly for me.
I built a machine around this board and CPU last year out of surplus parts, it is a good little machine, thread here:My new VIA C3 "Nehemiah" build, the weird and the wonderful!
I found performance of a C3 nehemiah at 1200MHz to be about equivalent to a Celeron coppermine at 600MHz. The above machine now has a PIII copermine at 733MHz installed and is far snappier for everything. Of course the C3 has the advantage of being able to down clock & lower the multiplier with setmul.
I found the older C3's much easier to get working in the average motherboard, be warned though that the original Samuel without L2 cache is VERY,VERY slow.
Best,
Chris

Ah, that's some good info, will read up more on it once I'm back home. Strange that Via chipset based mobos causes trouble with C3:s by the way 😜

Yeah, that's what I want, the possibility to control the CPU from software. Setmul + a boot menu makes for a very versatile machine. As I mentioned, I've already got two performance oriented rigs, if I get this project up and running, it'll ultimately replace my Pentium 200 rig for those speed sensitive games 😀

There are a surplus amount of cheap slockets/ slotkets with on-board VRMs on eBay right now:
http://www.ebay.co.uk/itm/MSI-MS-6905-MS6905- … kIAAOSwpkFY43VT

That's just one listing with a 100 MSI Master MS6905 adapters, and there are other listings; might be easier to pick one up.

The Ezra-T VIA C3 has the following speed profile range:
download/file.php?id=25324&mode=view

The Nehemiah can't match this flexibility, especially in the Speedsys range from around 15-100 (486 to Pentium 100). This is because the Nehemiah behaves too similarly to a Coppermine Pentium III. It does have a lower bottom and higher top end though.

If you are replacing a Pentium 200, then ensure you have tested it with Setmul and Test Registers because they work well, too.

Pretty much every 440BX board with the latest BIOS should work with VIA C3s.

kaputnik wrote:

gdjacobs wrote:

Can't find a datasheet for the SC1152CSW

Couldn't find that datasheet either.

*yawn* http://datasheet.datasheetarchive.com/origina … /DSA0078277.pdf 😁

j^aws wrote:

The Nehemiah can't match this flexibility, especially in the Speedsys range from around 15-100 (486 to Pentium 100). This is because the Nehemiah behaves too similarly to a Coppermine Pentium III. It does have a lower bottom and higher top end though.

If you are replacing a Pentium 200, then ensure you have tested it with Setmul and Test Registers because they work well, too.

Pretty much every 440BX board with the latest BIOS should work with VIA C3s.

My understanding is that the Nehemiah can do the equivalent of ~486DX-40 or so before the next speed step in the Pentium range. Is this accurate?

Stiletto wrote:

*yawn* http://datasheet.datasheetarchive.com/origina … /DSA0078277.pdf 😁

Right you are, although it seems that chip is just a form factor change on the 1152CS. Still 1.8V minimum Vcore.

j^aws wrote:

There are a surplus amount of cheap slockets/ slotkets with on-board VRMs on eBay right now ...That's just one listing with a 100 MSI Master MS6905 adapters, and there are other listings; might be easier to pick one up.

That is not a VRM there, it is a voltage clamp. All slotkets except the Powerleap PL-iP3 are dependant on the motherboard VRM to supply the proper core voltage.
That voltage clamp is used for I/O voltage conversion. Slotkets without voltage clamp are actually out of spec, and some motherboards may not work with them.

For what it is worth; In the past I replaced the VRM on two motherboards: Asus P2B rev 1.04 (HIP?->HIP6019BCB) and a Shuttle HOT-661 v1.1 (HIP6004ACB->HIP6004BCB). It works as intended.

kaputnik wrote:

The mobo's max multiplier is 8. Planning to run FSB at 100 MHz to avoid OC'ing the AGP bus, so max clock will land at 800MHz.

Don't worry about it. That maximum does not mean anything. VIA C3 will totally ignore the Intel Multiplier Request signal. (For a Pentium II/III that limit does not mean anything either. )

gdjacobs wrote:

Stiletto wrote:

*yawn* http://datasheet.datasheetarchive.com/origina … /DSA0078277.pdf 😁

Right you are, although it seems that chip is just a form factor change on the 1152CS. Still 1.8V minimum Vcore.

Strange. Read up some on different revisions of the MS-6119, and it was explicitly said that the versions with the SC1152CSW controller could supply voltages down to 1.3 volts. Would make it much harder to get hold of the right chip if there are different versions with the same model number.

j^aws wrote:

There are a surplus amount of cheap slockets/ slotkets with on-board VRMs on eBay right now: http://www.ebay.co.uk/itm/MSI-MS-69 […]
Show full quote

There are a surplus amount of cheap slockets/ slotkets with on-board VRMs on eBay right now:
http://www.ebay.co.uk/itm/MSI-MS-6905-MS6905- … kIAAOSwpkFY43VT

That's just one listing with a 100 MSI Master MS6905 adapters, and there are other listings; might be easier to pick one up.

The Ezra-T VIA C3 has the following speed profile range:
download/file.php?id=25324&mode=view

The Nehemiah can't match this flexibility, especially in the Speedsys range from around 15-100 (486 to Pentium 100). This is because the Nehemiah behaves too similarly to a Coppermine Pentium III. It does have a lower bottom and higher top end though.

If you are replacing a Pentium 200, then ensure you have tested it with Setmul and Test Registers because they work well, too.

Pretty much every 440BX board with the latest BIOS should work with VIA C3s.

No matter if I'll still have to replace the voltage controller on the mobo, I should pick up a few of those MS-6905:s while they're abundant and cheap, always a good thing to have options. Thanks for the heads up 😀

Oh, that table looks nice, virtually no "holes". Seems like the Ezra-T is the better option then.

I'm not replacing the P200 as in getting rid of it. Only got space for two retro rigs on my desk though, so I aim for having one performance and one compatibility/flexibility oriented rig wired up and ready to use at a time. If the P200 turns out to still be the better option for flexibility in the end. I'll just keep wired up, be happy I had a fun time building the C3 rig, and keep the latter for future use.

Good to hear. Then it might just be a question of getting the right supply voltage.

gerwin wrote:

That is not a VRM there, it is a voltage clamp. All slotkets except the Powerleap PL-iP3 are dependant on the motherboard VRM to […]
Show full quote

j^aws wrote:

There are a surplus amount of cheap slockets/ slotkets with on-board VRMs on eBay right now ...That's just one listing with a 100 MSI Master MS6905 adapters, and there are other listings; might be easier to pick one up.

That is not a VRM there, it is a voltage clamp. All slotkets except the Powerleap PL-iP3 are dependant on the motherboard VRM to supply the proper core voltage.
That voltage clamp is used for I/O voltage conversion. Slotkets without voltage clamp are actually out of spec, and some motherboards may not work with them.

For what it is worth; In the past I replaced the VRM on two motherboards: Asus P2B rev 1.04 (HIP?->HIP6019BCB) and a Shuttle HOT-661 v1.1 (HIP6004ACB->HIP6004BCB). It works as intended.

kaputnik wrote:

The mobo's max multiplier is 8. Planning to run FSB at 100 MHz to avoid OC'ing the AGP bus, so max clock will land at 800MHz.

Don't worry about it. That maximum does not mean anything. VIA C3 will totally ignore the Intel Multiplier Request signal. (For a Pentium II/III that limit does not mean anything either. )

Yeah, being that there are different sub-revisions of the MS-6119 v1.1 board with different voltage controllers, it's at least not impossible that it might work to just replace it, as long as there are no electrical differences between the revisions.

Ah, great. Not that I'd mind being capped at 800 MHz, but if the CPU can run at full speed, even better 😀

I had a look. Pinout, electrical characteristics, filter specs, and output voltages are all identical.
http://pdf1.alldatasheet.com/datasheet-pdf/vi … +/datasheet.pdf
http://datasheet.datasheetarchive.com/origina … /DSA0078277.pdf

Having a look at the datasheet of the SC1153CS:
http://pdf.datasheetcatalog.com/datasheet/Sem … orp/mXrzzss.pdf

I think this is what you want. 1.3V minimum Vout with otherwise identical recommended configuration for passives, and it seems the inverted output they show for PWRGOOD is a documentation change, not a difference in usage. The logic behavior they describe in the pin table is the same. However, please someone double check me before any board modding. Semtech doesn't show a timing diagram to make this explicit.

@gerwin: Thanks for that clarification.

gdjacobs wrote:

My understanding is that the Nehemiah can do the equivalent of ~486DX-40 or so before the next speed step in the Pentium range. Is this accurate?

Using the Speedsys scale, the Nehemiah drops verys sharply using ICD (Instruction Cache Disable), whilst the drop for Ezra/ Ezra-T is not as significant. Which means the Nehemiah misses a range from around a slow 486 to a fast Pentium 100.

The drop falls to around Speedsys 15, so whatever the equivalent 486 speed is at that range. Because of this sharp drop, the Nememiah isn't as 'tweakable' around 386/486 speeds compared to the Ezra-T, which means you can't fine tune speed sensitive games such as Test Drive 3 and Wing Commander.

You also can't tune games falling into the Speedsys range of 15-100, such was Ultima 7 and Magic Carpet (you can increase resolution, but you lose the low-res option).

When it come to 'dial-a-speed', the Ezra-T is a better choice than the Nehemiah.

Okay, that's what I thought. You essentially lose 486DX/2, 486DX/4, and P75 equivalent performance ranges, presumably even with multiplier manipulation taken into account.

^^ Yes, even with multiplier manipulation taken into account. Nehemiahs Speedsys profile is similar to an unlocked Pentium III, except it sharply drops to around Speedsys 15, rather than around 8. Which makes it more usable in the 486 range.

gdjacobs wrote:

I had a look. Pinout, electrical characteristics, filter specs, and output voltages are all identical. http://pdf1.alldatasheet. […]
Show full quote

I had a look. Pinout, electrical characteristics, filter specs, and output voltages are all identical.
http://pdf1.alldatasheet.com/datasheet-pdf/vi … +/datasheet.pdf
http://datasheet.datasheetarchive.com/origina … /DSA0078277.pdf

Having a look at the datasheet of the SC1153CS:
http://pdf.datasheetcatalog.com/datasheet/Sem … orp/mXrzzss.pdf

I think this is what you want. 1.3V minimum Vout with otherwise identical recommended configuration for passives, and it seems the inverted output they show for PWRGOOD is a documentation change, not a difference in usage. The logic behavior they describe in the pin table is the same. However, please someone double check me before any board modding. Semtech doesn't show a timing diagram to make this explicit.

Yeah, seems like that would work. If I can't get it confirmed, I believe I could just do a replacement and see, worst case scenario should be that it'll just work exactly like the SC1152CS, and not give me the voltages below 1.8V.

If your board uses DIP switches or jumpers, you can manually use VID combinations in Table 1 of the PDF. Firmware controlled voltages would be another story.

My primary concern is with that one control pin. If it's active low and the board is designed for an active high POWER GOOD signal, it will have problems.

gdjacobs wrote:

If your board uses DIP switches or jumpers, you can manually use VID combinations in Table 1 of the PDF. Firmware controlled voltages would be another story.

My primary concern is with that one control pin. If it's active low and the board is designed for an active high POWER GOOD signal, it will have problems.

Sadly, it's firmware controlled, but as mentioned, there are newer boards of the same revision that can supply lower voltages, and they're using the same bios. As long as there are no electrical differences between the sub-revisions, it might just work.

Well, if it's even possible to order one of those SC1153CS chips, it's easy enough to try. If I have to invert that control pin signal to get it working, I'll just put the original controller back, call it a day, and resume the project once I get another slot 1 board 😀

This might be interesting for other owners of the MSI MS-6119.

Ordered a couple of those MS-6905 slotkets with voltage clamp jumpers j^aws suggested back when I started this thread. Tonight I rigged up my MS-6119 board for another purpose, wanted some pre y2k HW benchmarks. Put a Coppermine 1000EB in the one of those MS-6905 and clocked it down to 840MHz@112MHz FSB, which would be very close to the fastest pre y2k PIII. After doing the benchmarks, I started playing around with the voltage jumpers, and it seems that the SC1152CS VRM can output lower voltages than 1.8V after all. Booted the underclocked Coppermine at these voltages:

1.7V
1.6V
1.5V
1.4V
1.3V

Skipped the 1.x5V settings, but I'm sure those works too.

A very neat feature of the MS-6119 is that it reports rail aswell as core voltages after POST. CPU-Z reports the same core voltage.

Now to find out how to measure vcore with a multimeter (anyone happen to know a good test point btw?), to double check and make sure it's for real, and not just display.

Seems like there's hope and that I won't have to mod the board after all 😀