Post a pic of the CPU, the SECC carts came in two versions and the early PIII CPUs used both, SECC1 was annoying to disassemble requiring a torx screw driver to undo the 4 studs, SECC2 was far easier to disassemble.

If you need to replace the fan it might be cheaper to buy a new cooler for it, the twin fan versions are pretty nice and dont cost much.

As I have the habbit of changing Slot1 CPUs often, I mounted the fan on the board. I used a bigger fan to get away with less noise. The CPUs only have the cooler without fan. That way, I can switch them fast while keeping near silent cooling.

Trashbytes wrote on 2024-08-12, 11:26:

Post a pic of the CPU, the SECC carts came in two versions and the early PIII CPUs used both, SECC1 was annoying to disassemble requiring a torx screw driver to undo the 4 studs, SECC2 was far easier to disassemble.

If you need to replace the fan it might be cheaper to buy a new cooler for it, the twin fan versions are pretty nice and dont cost much.

It's a SL37D exactly like this one:

The fan side unclips easy, it has push tabs on the ends that you squeeze to make it release.

It's the back side I have trouble with. There are these 4 little pins in there that I can't really see well. I tried jamming an ink rod in there in hopes the tube would squeeze any barbs, but no luck. They're definitely not screws! They're plastic pins of some kind! There are no screws in sight anywhere.

r00tb33r wrote on 2024-08-12, 11:35:

It's a SL37D exactly like this one: https://pcrebuilding.altervista.org/images/resized_my_pcrebuilding/58eb68a887164_p1200190.jp […]
Show full quote

Trashbytes wrote on 2024-08-12, 11:26:

Post a pic of the CPU, the SECC carts came in two versions and the early PIII CPUs used both, SECC1 was annoying to disassemble requiring a torx screw driver to undo the 4 studs, SECC2 was far easier to disassemble.

If you need to replace the fan it might be cheaper to buy a new cooler for it, the twin fan versions are pretty nice and dont cost much.

It's a SL37D exactly like this one:

The fan side unclips easy, it has push tabs on the ends that you squeeze to make it release.

It's the back side I have trouble with. There are these 4 little pins in there that I can't really see well. I tried jamming an ink rod in there in hopes the tube would squeeze any barbs, but no luck. They're definitely not screws! They're plastic pins of some kind! There are no screws in sight anywhere.

https://www.reddit.com/r/retrobattlestations/ … t_sink_removal/

Hope you have a torx screw driver, since you need to push the pins into the cart. I guess any flat headed item may work as long as it fits the small holes.

Here is another guide

http://www.fat-rat.com/fluff/Heatsink_info/

Once you have it off you may want to seriously consider an after market cooler instead of putting the infernal thing back together.

Trashbytes wrote on 2024-08-12, 11:39:

https://www.reddit.com/r/retrobattlestations/ … t_sink_removal/

Hope you have a torx screw driver, since you need to push the pins into the cart. I guess any flat headed item may work as long as it fits the small holes.

Oh, pfft!

Hexagon screwdriver, it has a wide flat head that we can push with

It's not used as a screwdriver, just something to push with, and yes, I have plenty.

When you said torx screwdriver to me that meant screws. No screws here 🤣.

I guess we push...?

r00tb33r wrote on 2024-08-12, 11:42:

Oh, pfft! […]
Show full quote

Trashbytes wrote on 2024-08-12, 11:39:

https://www.reddit.com/r/retrobattlestations/ … t_sink_removal/

Hope you have a torx screw driver, since you need to push the pins into the cart. I guess any flat headed item may work as long as it fits the small holes.

Oh, pfft!

Hexagon screwdriver, it has a wide flat head that we can push with

It's not used as a screwdriver, just something to push with, and yes, I have plenty.

When you said torx screwdriver to me that meant screws. No screws here 🤣.

I guess we push...?

There are models that have torx screws in them, they are attached to metal posts which got replaced by the nasty plastic pins of the model you have. .. I have a number of them here .. horrible nasty things to disassemble as the torx screws tend to corrode.

Trashbytes wrote on 2024-08-12, 11:43:

r00tb33r wrote on 2024-08-12, 11:42:

Oh, pfft! […]
Show full quote

Trashbytes wrote on 2024-08-12, 11:39:

https://www.reddit.com/r/retrobattlestations/ … t_sink_removal/

Hope you have a torx screw driver, since you need to push the pins into the cart. I guess any flat headed item may work as long as it fits the small holes.

Oh, pfft!

Hexagon screwdriver, it has a wide flat head that we can push with

It's not used as a screwdriver, just something to push with, and yes, I have plenty.

When you said torx screwdriver to me that meant screws. No screws here 🤣.

I guess we push...?

There are models that have torx screws in them, they are attached to metal posts which got replaced by the nasty plastic pins of the model you have. .. I have a number of them here .. horrible nasty things to disassemble as the torx screws tend to corrode.

Got it apart, thanks!

Wow, that required more force than I was comfortable with...

It shall receive fresh thermal compound and I already ordered a Noctua NF-A4x20 FLX for which I will print a shroud.

ux-3 wrote on 2024-08-12, 11:32:

As I have the habbit of changing Slot1 CPUs often

I can't really think of any reason to desire something else. I feel like it would be a diminishing return, as any game that needs a faster processor will run fine on modern systems (with patches or tweaks).

This machine is twin duty, scratches my DOS itch, and runs Glide games with the Voodoo 3.

I run this one at 600MHz.

Recently someone developed a HALT pin pulsing circuit that allows throttling the processor all the way down to like 8088 speeds and everything in between... I haven't seen it for Slot 1 yet, but since it was shown on Socket 370 I should be able to figure out the pin. Meaning I wouldn't really need a slower Slot 1 processor for anything either.

r00tb33r wrote on 2024-08-12, 11:51:

It shall receive fresh thermal compound and I already ordered a Noctua NF-A4x20 FLX for which I will print a shroud.

Have you got a link to the shroud?

zuldan wrote on 2024-08-12, 12:11:

r00tb33r wrote on 2024-08-12, 11:51:

It shall receive fresh thermal compound and I already ordered a Noctua NF-A4x20 FLX for which I will print a shroud.

Have you got a link to the shroud?

It's in my brain at the moment 🤣, will be drawn up in SolidWorks later today.

So my printer is a little on the old side of things, or maybe I'm just not great at tuning the slicer, but it has dimensional swell, generally 0.4-0.5mm on all boundaries horizontally. Accuracy is fine vertically. Meaning I have to account in my models for the dimensional swell of my print. Meaning it's hard for me to share my models easily for anybody, but if there is interest I can produce a version with exact dimensions for anyone that has a printer with good dimensional accuracy.

(It probably needs extrusion adjusted, but also old CURA was not exactly a high-end slicer to begin with, and when I played around with adjustments back in the day less extrusion resulted in weaker prints, but yeah, I should dedicate some time to it at some point.)

Do you just want to see a picture of the model or do you want the file?

r00tb33r wrote on 2024-08-12, 12:07:

Recently someone developed a HALT pin pulsing circuit that allows throttling the processor all the way down to like 8088 speeds and everything in between... I haven't seen it for Slot 1 yet, but since it was shown on Socket 370 I should be able to figure out the pin. Meaning I wouldn't really need a slower Slot 1 processor for anything either.

Yeah it's Kermit the Frog. And his invention should work on slot 1 as well. I got the parts and I'm planning to build a throttle blaster for a slot-1 Pentium2 machine.

wbahnassi wrote on 2024-08-12, 17:28:

r00tb33r wrote on 2024-08-12, 12:07:

Recently someone developed a HALT pin pulsing circuit that allows throttling the processor all the way down to like 8088 speeds and everything in between... I haven't seen it for Slot 1 yet, but since it was shown on Socket 370 I should be able to figure out the pin. Meaning I wouldn't really need a slower Slot 1 processor for anything either.

Yeah it's Kermit the Frog. And his invention should work on slot 1 as well. I got the parts and I'm planning to build a throttle blaster for a slot-1 Pentium2 machine.

I have a Pentium II or a Pentium II Celeron up in the attic. Any reason I'd want to use that? I never really thought about it much.

r00tb33r wrote on 2024-08-12, 12:19:

zuldan wrote on 2024-08-12, 12:11:

r00tb33r wrote on 2024-08-12, 11:51:

It shall receive fresh thermal compound and I already ordered a Noctua NF-A4x20 FLX for which I will print a shroud.

Have you got a link to the shroud?

It's in my brain at the moment 🤣, will be drawn up in SolidWorks later today.

The face side looks boring because it is designed to be printed with minimum support and no raft, so it lays flat on the print bed. (I generally design to use no support at all.) The only support needed here is the fan cavity.

Among other things this also results in a pretty reasonable print time.

Whoa dang, it actually fits right off the printer.

The fan arrives later today. Looks like I'm all good to finish.

The 20mm fan sticks out 7mm more than Intel original, 10mm fans would be 3mm slimmer than Intel's. Ether way there's plenty of clearance.

I am somewhat disappointed with the noise level of the Noctua fan. The noise level it produces at the very least inline with other fans I've had in this size from other reputable brands like Papst and Sunon, though IMO the Papst fans I have actually have less whirring for more airflow. I'm kind of torn there, as I was planning to use the 5V version on my Voodoo cards, and now I wonder if I should check out the 10mm version or look for another brand. In terms of airflow this 40x20 from Noctua is "just enough" for this application, but I have yet to do actual testing. I have Fluke IR thermometer I can take measurements with.

So overall my print performed better than expected while the Noctua product performed worse than expected. So on average...?

zuldan wrote on 2024-08-12, 12:11:

r00tb33r wrote on 2024-08-12, 11:51:

It shall receive fresh thermal compound and I already ordered a Noctua NF-A4x20 FLX for which I will print a shroud.

Have you got a link to the shroud?

Here's your link:

🤣 so VOGONS had me ZIP up the STL file because apparently somehow that's better??? /rant

Test results:
NFS HS replay loop: 42°C hottest point
Prime95 stress test: 40°C hottest point

IMO acceptable numbers. Exactly how the game is hotter than Prime95 I have no idea, but that's what it is.

I'll be making fan mounts for my Voodoo 3 cards next. By the way, if anyone has a community project that needs printable parts designed I'm up for it. I enjoy this more than playing games anyway.

wbahnassi wrote on 2024-08-12, 17:28:

r00tb33r wrote on 2024-08-12, 12:07:

Recently someone developed a HALT pin pulsing circuit that allows throttling the processor all the way down to like 8088 speeds and everything in between... I haven't seen it for Slot 1 yet, but since it was shown on Socket 370 I should be able to figure out the pin. Meaning I wouldn't really need a slower Slot 1 processor for anything either.

Yeah it's Kermit the Frog. And his invention should work on slot 1 as well. I got the parts and I'm planning to build a throttle blaster for a slot-1 Pentium2 machine.

r00tb33r wrote on 2024-08-13, 05:43:

I have a Pentium II or a Pentium II Celeron up in the attic. Any reason I'd want to use that? I never really thought about it much.

I retrieved it from the attic, it's a 333MHz Pentium II, SL2S5, with what looks to be an aftermarket cooler.

So what's the big whoop, besides being able to select the multiplier?
Any reason I'd want this over a P3?

r00tb33r wrote on 2024-08-14, 04:38:

I am somewhat disappointed with the noise level of the Noctua fan.

That is part of the reason why I am using a larger fan at lower RPM. It is less audible.

r00tb33r wrote on 2024-08-18, 09:14:

I retrieved it from the attic, it's a 333MHz Pentium II, SL2S5, with what looks to be an aftermarket cooler.
So what's the big whoop, besides being able to select the multiplier?
Any reason I'd want this over a P3?

Have you checked if the multiplier is unlocked? From SL2S5, I would suspect it isn't unlocked. Try or check the date code. Perhaps there is no option to slow down this way.
There are not many reasons you would want this. You can be slow enough for the turbo pascal bug, but there are too many speed issues that you won't be able to defeat with either CPU.
Some games will play OK if you run on 133 or 166 MHz. This you can do with the P2. But many more need even less speed and will cause trouble with either CPU.

At the moment, a P3 or P2 isn't ideal as an aio DOS CPU, but with new slow techs, it might be great.

ux-3 wrote on 2024-08-18, 14:26:

Have you checked if the multiplier is unlocked? From SL2S5, I would suspect it isn't unlocked. Try or check the date code.

I couldn't figure out an obvious date except the copyright prints which are '94 and '96, which I presume are not useful.

Now that you mentioned it I did plug it into my ASUS P3B-F set to jumperless mode, and the multiplier appeared to be locked at 5.0x.

I thought I read somewhere that Intel locked down the multiplier at like 450MHz models, but that must be incorrect.

Is there any way to unlock it on the processor PCB, like by modifying resistors or something?

Either way I just don't really see a point of pairing the P2 with the Voodoo3 as opposed to a P3.

With a ThrottleBlaster I don't think the lower multipliers are needed anyway. I didn't know about the "Turbo Pascal bug", I figured it's some sort of timing problem, and appears there are binary patches that take care of the problem anyway.

I haven't looked further into ThrottleBlaster beyond the initial announcement some months ago, I don't know if there are any caveats with that, like if it causes any problems with other hardware while the processor is halted. I suppose it would be possible to crash if interrupts aren't serviced?