J1=SHORT
J3=SHORT 1-2
J4=SHORT 2-3

JV4=2-3
JV3=2-3
JV2=NC
JV1=NC
JV0=2-3

Any other combo might cause you problems.

What's the model of VRM that cooked?

TOBOR wrote:

J1=SHORT J3=SHORT 1-2 J4=SHORT 2-3 […]
Show full quote

J1=SHORT
J3=SHORT 1-2
J4=SHORT 2-3

JV4=2-3
JV3=2-3
JV2=NC
JV1=NC
JV0=2-3

Any other combo might cause you problems.

So you think the auto jumper settings may be selecting the wrong voltage? I was wondering if it might be that.

gdjacobs wrote:

What's the model of VRM that cooked?

On this board it is really hard to read. I will try to make it out tomorrow. I now have two of these boards and they look like they have a few duplicate mosfets so I was thinking about moving some over and retesting with the slotket hard jumpered to the exact voltage of the CPU which is 1.75. If it is that it surprises me the auto voltage setting was such a catastrophic failure.

gdjacobs wrote:

What's the model of VRM that cooked?

Ok. I got the best photos I could for you. I am not an electronics expert by any stretch of the imagination, but I can pump out a decent job with a solder station. If I use incorrect terminology or call something that it isn't please correct me. Learning as I go.

Photo 1 shows the first board that blew and it looks like to me the best candidate for repair as the traces look in tact (blue circle). The board looks to use the same mosfets around the CPU slot which I circled in red. Hard to make out for me, but maybe you can make sense of it.

Photo 2 shows the second board that blew and the locations of what appears to be the duplicate mosfets. Blue circle is the damage and this blow looks like it had collateral damage.

Photo 3 shows a clearer image of the mosfets being used around the CPU slot on the second board.

Photo 4 is yet another zoom in on the other mosfet on second board.

Photo 5 is just to show off the aftermath from the eruption on board 2 before I cleaned it up a bit..

TOBOR wrote:

J1=SHORT J3=SHORT 1-2 J4=SHORT 2-3 […]
Show full quote

J1=SHORT
J3=SHORT 1-2
J4=SHORT 2-3

JV4=2-3
JV3=2-3
JV2=NC
JV1=NC
JV0=2-3

Any other combo might cause you problems.

Both times I had my jumpers set to the following so it would auto detect the CPU:

J3=SHORT 1-2
J1=SHORT
J4=SHORT 1-2

I set J1 & J4 to this as I thought "CPU define" would be an auto detect setting (as the box shows). but maybe that is a problem?

JV4=1-2
JV3=1-2
JV2=1-2
JV1=1-2
JV0=1-2

Set JV0-JV4 to this for auto detect voltage.

Either auto detect sucks on this slotket and doesn't work or I am misunderstanding what "CPU define" means on J1 & J4. Hopefully that clarifies.

Do these mainboards officialy support Slotted Coppermine-core Processors? Coppermine core normally requires voltages of 1.65 to 1.75V. But early Slot-1/Socket 370 motherboards precede this requirement.

Do these mainboards UNofficialy support Slotted Coppermine-core Processors? Which usually means: Does the actual VRM (I do not mean the blown component, but the 'ICS' branded chip) support voltages below 1.8Volt?

gerwin wrote:

Do these mainboards officialy support Slotted Coppermine-core Processors? Coppermine core normally requires voltages of 1.65 to 1.75V. But early Slot-1/Socket 370 motherboards precede this requirement.

Do these mainboards UNofficialy support Slotted Coppermine-core Processors? Which usually means: Does the actual VRM (I do not mean the blown component, but the 'ICS' branded chip) support voltages below 1.8Volt?

Good point and I don't know. The P-III 450 that was in there was 2v and I assumed in my ignorance if it could handle a 2v CPU it could handle a lower voltage 1.75v Coppermine. Would a lower voltage CPU cause this sort of issue?

I never dared to try that. My guess is: on a poorly prepared motherboard one of the Voltage request signals is unsupported and gets ignored, so what you may get instead is a totally unexpected core voltage. On a properly prepared mainboard it would detect 'voltage request is unsupported' and will refuse to boot.

It is about the 'VID' (voltage Selection) pins on the processor. These are connected (usually directly) to the VRM Chip. Attached is VRM specification 8.4 which was made to support coppermine core. One thing to do is to look up the datasheet of your particular VRM chip. If voltage bottoms out at 1.8V a common trick is to select 1.8V with the slotket, which is pretty close to 1.75V.

When replacing the blown transistor (SOT-23, I think), you have to match up the part codes. A20, for instance, is a JFET and would have to be replaced by the same channel type JFET.

gerwin wrote:

I never dared to try that. My guess is: on a poorly prepared motherboard one of the Voltage request signals is unsupported and gets ignored, so what you may get instead is a totally unexpected core voltage. On a properly prepared mainboard it would detect 'voltage request is unsupported' and will refuse to boot.

It is about the 'VID' (voltage Selection) pins on the processor. These are connected (usually directly) to the VRM Chip. Attached is VRM specification 8.4 which was made to support coppermine core. One thing to do is to look up the datasheet of your particular VRM chip. If voltage bottoms out at 1.8V a common trick is to select 1.8V with the slotket, which is pretty close to 1.75V.

So let me ask you this. If the CPU voltage request is possibly being ignored and I set the slotket jumpers from auto to hardwired 1.75v will that possibly do the trick? I have a newer Socket 370 D815EEA Intel board, but to my surprise it performs worse than the older Intel 440BX board. I don't know if the newer 815E chipset is at fault or because the older board is auto detecting the CPU and it may be unintentionally overclocking it and that's why I get better performance hence it burning up after awhile? Trying to decode the puzzle. I thought I was pretty knowledgeable at hardware after decades of doing this, but during the time Slot 1 and later was coming out I was strictly using AMD CPUs back in the day. I feel like I am coming into this completely new again. It feels refreshing and lame at the same time, but at least I am learning something.

No problem, but I think my previous post answered your voltage question already? All this is speculation, which may or may not apply to your motherboard, you have to first find the motherboard documentation and VRM datasheet.

As for the performance, I don't know. The i440BX was told to be a fast chipset, though not that it should be really noticable compared to the i815.

gerwin wrote:

No problem, but I think my previous post answered your voltage question already? All this is speculation, which may or may not apply to your motherboard, you have to first find the motherboard documentation and VRM datasheet.

As for the performance, I don't know. The i440BX was told to be a fast chipset, though not that it should be really noticable compared to the i815.

I just went through my emails and I remember now that luckybob had warned me I might have to overvolt the Coppermine to 1.8v and that should not be a problem. From reading parts of the following doc I guess I should have applied 1.8v to the slotket. What I don't understand is why 1.75v, being a lower voltage, would fry that component, but I'm guessing a chain of events happen and it is not as simplistic as I am thinking. Page 5 sort of indicates this if I am understanding it right. I believe this Intel board adheres to the chart on page 29 of the PDF link attached below.

"Another side effect of lowering voltages of some components is the existence of
multiple voltages within the system. On a basic Slot 1 processor-based system board there
will be 1.5V for GTL+ termination, 1.8V to 2.8V for the processor, 2.5V for CMOS non-GTL
signals, 3.3V for the chipset and the L2 cache, and 5V for other components. The possibility that
any of these voltages may come up before another must be taken into account. This is discussed in
Section 3.3."

http://db.zmitac.aei.polsl.pl/Electronics_Fir … ts/24333202.pdf

I looked it up for you, the VRM does not mention Pentium III/VRM-8.4 compliance, but can supply 1.75 Volt nonetheless.
"SC1185CSW" -> http://www.semtech.com/images/datasheet/sc1185.pdf -> "1.3V to 3.5V Pentium ll microprocessor supply"
So the Vcore problem should not apply to your motherboard. (To be 100% sure you have to measure Vcore with a multimeter while running the system.)

gerwin wrote:

I looked it up for you, the VRM does not mention Pentium III/VRM-8.4 compliance, but can supply 1.75 Volt nonetheless.
"SC1185CSW" -> http://www.semtech.com/images/datasheet/sc1185.pdf -> "1.3V to 3.5V Pentium ll microprocessor supply"
So the Vcore problem should not apply to your motherboard. (To be 100% sure you have to measure Vcore with a multimeter while running the system.)

Interesting you started a thread years ago that was very similar.

(im)Possibilities wit Slotket CPU adapters?

The post from Tetrium in that thread helps the most. Looks like Intel boards do not play well with slotkets and I have also read the voltage regulators are stressed more with slotkets and some mobos don't last long before they blow. All starting to sound familiar. I am going to stick with a Socket 370 at this point and skip Slot 1 all together.

(im)Possibilities wit Slotket CPU adapters?

It could just be a matter of Vds-on * Icore being too much for those little transistors with a higher TDP Coppermine chip.

After talking with a few people who PM'd me I guess they also have 440BX boards that play fine with certain Slotkets so I am at loss what the issue is. Could be an Intel branded mobo thing. I will clarify this is the MSI revision 2 Slotket that is supposed to be much more compatible with more motherboards. I do appreciate everyone's input, but at this point the D815EEA will have to do for my Windows 98SE build. I really wanted the single ISA slot on that 440BX board for driving a couple of floppy drives (mobo BIOS only supports one floppy drive annoyingly), but I'll have to pony up some cash and build a proper period correct DOS machine down the road. I have an almost factory mint ITT Xtra XP machine I am testing that I want to exchange data with easily so having a proxy machine with proper floppy drives will make that job easier (may setup a serial connection instead). Makes me regret selling my pristine full tower DX4-100 decades ago. Oh well, as they say, hindsight is 20/20.

I had experience with a number of these socket-to-slot converters at the time. One thing I remember well is that quality varies greatly. As they weren't officially endorsed by Intel, care is required. We had so many failures with the cheaper units, that we started refusing to sell anything but the more expensive units from the likes of Asus or Abit. Those were usually pretty good, and produced stable systems; even when the others appeared to work initially, the systems would almost always fail to pass pre-shipment testing. In the cases where I was ordered to release systems without testing, they always came back. Save yourself the trouble, and use only the best adapters, or don't bother with them at all.

Use the jumpers to set what you can, without depending on auto-detection. If they can still be had, find an adapter with its own voltage regulation and quality capacitors on-board. I don't remember the part numbers, but they used to be available.

I chose the MSI MS-6905 Revision 2 board because I had read it was a high quality conversion board. A member from this forum sent me this attachment with Intel's recommendations of tested SSAs. Found it interesting. If I do decide to give it a go again I will not be using an Intel board and I agree with you 640K!enough, I will not be relying on any auto jumper settings anymore just to be safe.