Looks like a promising build! Since you are using a Tualatin with a 440BX, you will be overclocking the AGP bus so be careful there - you may find some cards are more sensitive at the higher bus speed and I don't believe these 440BX chips have a divider. Some people have reported no issues getting video cards to work, but I had to try several before I found one stable enough to get through some torture tests. Yeah, I hear ya - gotta watch those Abit boards too - they tended to cheap out on their caps. Both of my Abit SA6R (i815e) boards won't power up right now and also need recapping.

You can see my thread here: Artex's Build of the Week: Altered Beast

Artex wrote:

Looks like a promising build! Since you are using a Tualatin with a 440BX, you will be overclocking the AGP bus so be careful there - you may find some cards are more sensitive at the higher bus speed and I don't believe these 440BX chips have a divider. Some people have reported no issues getting video cards to work, but I had to try several before I found one stable enough to get through some torture tests.

I found that these Compaq 3500 cards are quite tolerant of 133MHz bus speeds. The 2000 and 3000 Voodoo3s are less cooperative, in my experience, but then this might vary between individual cards, even. I guess with nVidia there are even better options but I really like the Voodoo's 2D performance in DOS so I'd prefer to stick with it.

Artex wrote:

Yeah, I hear ya - gotta watch those Abit boards too - they tended to cheap out on their caps. Both of my Abit SA6R (i815e) boards won't power up right now and also need recapping.

I'm doing some tests on these busted caps right now. Stay tuned... 😀

Okay, time for some measurements: let's find out what the effect of bad caps is on the CPU voltage. Right at the CPU socket is probably the best place to measure this:

At 66MHz the board starts up and seems to work okay. Noise level is around 350 mVpp. Note that I've already replaced the burst caps on the 5V input circuit with new Panasonics here.

At 100MHz, the board POSTs most times but get stuck during boot about 8/10 times. Noise has increased to about 400 mVpp.

At 133MHz, all we get is a black screen. Not surprising, the noise level is nearly 550mVpp! To put that in perspective, the CPU core voltage is constantly jumping up and down between about 1.45V to 2.0V here. Looks like this with the scope set to DC: http://i.imgur.com/VqWNJBP.jpg

Let's compare this to an Asus P3B-F, with stock Rubycon and Sanyo capacitors, at 100MHz. Noise level is approx. 250mVpp:

Now here's an Asus P2B with brand new Panasonic electrolytics. About 50mVpp. Honestly, now that I'm writing this, this seems so low I'm starting to doubt if I really did set the scope to 50mV/div.

Next up, I have this ESR meter, and a transistor/component tester, both from China (ebay). No idea how accurate these are. Let's pull some caps from the board and see what they say.

Three popped 1500µF caps from the 5V input circuit: 4.03, 4.22 and 4.48 ohms. YES, FOUR OHMS! (component tester: http://i.imgur.com/dRyfEan.jpg)
One green 1500µF cap from the CPU voltage regulator: 0.06 ohm (component tester: http://i.imgur.com/J2Fun21.jpg)
One green 1000µF cap from the 3.3V regulator: 0.05 ohm (component tester: http://i.imgur.com/3WgFAhf.jpg)
One black 1500µF cap found near the DIMM slots: 0.06 ohm (component tester: http://i.imgur.com/rkfGgbb.jpg)
One black 1500µF cap which I presume is from either the AGP or Vtt voltage regulator: 0.06 ohm (component tester: http://i.imgur.com/A1A3sxj.jpg)
One brand new Panasonic FR, 6.3V 1500µF: 0.03 ohm (component tester: http://i.imgur.com/zrYP2iq.jpg)

Uh, yeah. Not sure if I can draw any conclusions from this. ESR on all tested caps is at least twice as high as the new cap (or more, according to the transistor tester), but not quite as high as I expected with the massive CPU voltage ripple going on. I'm also not sure what Vloss on the transistor tester means, or if this is any indication of failure. Is that something like a leakage current (a resistor in parallel)?
Oh well, at least both testers confimed: the popped caps are really popped. 😀

I've ordered new caps and intend to repeat this test once I receive them.

Here's one of the heatsinks I got:

This is a Thermalright HR-05-SLI. It's actually marketed as chipset heatsink, however its size seems appropriate for a Pentium 3 too.

Will have to find some way to mount this properly. It's much thicker than a normal S370 heatsink so the standard clips won't fit. Oh well. Let's quickly hack something together for now, just to see how it would fit.

Not too bad, right? Almost looks like it's made for it. Nicely fills up all that empty space and still leaves enough clearance to access the DIMMs:

Now on a Voodoo3 it's much easier to mount. The included clips fit rather well:

But I wouldn't call it very practical, though...

🤣. That thing is a monster.

All I can say is, Wow! 😮

And what kind of chipset is that heatsink meant to cool?

You tell me! 🤣 Even the latest chipsets don't dissipate much more than 5-15W, with the most power-hungry barely reaching 30. I'm sure this heatsink is absolute overkill.

I have another heatsink here that's even more impressive... or laughable, your choice 😀
But I'll save that for another time, got new caps today for the mainboard so that's main priority for now.

Pay close attention people. THIS is how you prepare an old system if you want to use it seriously. Old capacitors are a big problem, especially off-brand ones.

By the way, I had an Abit BF6. Was the same thing as a BE6-II, but without the on-board RAID controller. Great board, noticeably faster than my BX6 2.0. Unfortunately, I was using an inadequate PSU at the time (2004), and the motherboard had bulged caps on the VRM, which I ignored. One day, it just didn't boot up, and that was the end of that.

mockingbird wrote:

Pay close attention people. THIS is how you prepare an old system if you want to use it seriously. Old capacitors are a big problem, especially off-brand ones.

By the way, I had an Abit BF6. Was the same thing as a BE6-II, but without the on-board RAID controller. Great board, noticeably faster than my BX6 2.0. Unfortunately, I was using an inadequate PSU at the time (2004), and the motherboard had bulged caps on the VRM, which I ignored. One day, it just didn't boot up, and that was the end of that.

I hear ya - Abit really had problems with bad caps it seems. I need to find the time to get my two boards recapped. I don't have the patience, or the skillz to do a good job.

Finished replacing the capacitors, and gave it a quick test run (and then I ran adtrack2.exe and kinda lost track of time...)

So here's the result:

These are all 6.3V, 1000µF Nichicon FPCAP ("Functional" Polymer, whatever that means). Looks funny, doesn't it? 😀
Yep, that's less than the original 1500µF, but I'm hoping the massive reduction in ESR will make up for the loss in capacitance. These supposedly have an ESR of only 0.007 Ohm! 😳

Haven't done any extensive tests yet, but it seems to boot fine now at both 100 and 133 MHz. I even tried 140 briefly and it seemed to work too. Quite an improvement over 66, I'd say.

Artex wrote:

mockingbird wrote:

Pay close attention people. THIS is how you prepare an old system if you want to use it seriously. Old capacitors are a big problem, especially off-brand ones.

By the way, I had an Abit BF6. Was the same thing as a BE6-II, but without the on-board RAID controller. Great board, noticeably faster than my BX6 2.0. Unfortunately, I was using an inadequate PSU at the time (2004), and the motherboard had bulged caps on the VRM, which I ignored. One day, it just didn't boot up, and that was the end of that.

I hear ya - Abit really had problems with bad caps it seems. I need to find the time to get my two boards recapped. I don't have the patience, or the skillz to do a good job.

Have to agree as well, I lost my Abit KT7A for the same reason (the board went up in smoke, literally 😵 ). Finally found someone local to recap my Soyo P4 board; got it back in its original box, ready to go and be repaired 😀

jwt27 wrote:

Finished replacing the capacitors, and gave it a quick test run (and then I ran adtrack2.exe and kinda lost track of time...) […]
Show full quote

Finished replacing the capacitors, and gave it a quick test run (and then I ran adtrack2.exe and kinda lost track of time...)

So here's the result:

These are all 6.3V, 1000µF Nichicon FPCAP ("Functional Polymer", whatever that means). Looks funny, doesn't it? 😀
Yep, that's less than the original 1500µF, but I'm hoping the massive reduction in ESR will make up for the loss in capacitance. These supposedly have an ESR of only 0.007 Ohm! 😳

Haven't done any extensive tests yet, but it seems to boot fine now at both 100 and 133 MHz. I even tried 140 briefly and it seemed to work too. Quite an improvement over 66, I'd say.

Ooohhh nice, solid caps! They actually look right at home to me ^^

Beautiful recap job jwt27. Still, you're right; more testing is needed to make sure the BE6-II is stable and works as expected.

I bet those North Bridge heatsinks were for those crazy 775 Nvidia chipsets that got super hot. Like the 680i

JayCeeBee64 wrote:

Artex wrote:

I hear ya - Abit really had problems with bad caps it seems.

Have to agree as well, I lost my Abit KT7A for the same reason (the board went up in smoke, literally 😵 ).

Abit boards are kinda notorious for this.
However, my old rev. 1.0 Abit KT7A-RAID doesn't have any capacitor trouble for some reason, unlike any other KT133A board I've ever seen.

@jwt27: Nice recapping job, looks awesome! Where did you buy the caps? I have a hard time finding any polymers that aren't SMD 😕

jwt27 wrote:

Finished replacing the capacitors, and gave it a quick test run (and then I ran adtrack2.exe and kinda lost track of time...) […]
Show full quote

Finished replacing the capacitors, and gave it a quick test run (and then I ran adtrack2.exe and kinda lost track of time...)

So here's the result:

These are all 6.3V, 1000µF Nichicon FPCAP ("Functional" Polymer, whatever that means). Looks funny, doesn't it? 😀
Yep, that's less than the original 1500µF, but I'm hoping the massive reduction in ESR will make up for the loss in capacitance. These supposedly have an ESR of only 0.007 Ohm! 😳

Haven't done any extensive tests yet, but it seems to boot fine now at both 100 and 133 MHz. I even tried 140 briefly and it seemed to work too. Quite an improvement over 66, I'd say.

That is some superb work there! Looks absolutely beautiful. I look forward to hearing how stable it is and how it performs with those solid caps. This meme pretty much sums up my reaction when I saw it:

mockingbird wrote:

Pay close attention people. THIS is how you prepare an old system if you want to use it seriously. Old capacitors are a big problem, especially off-brand ones.

By the way, I had an Abit BF6. Was the same thing as a BE6-II, but without the on-board RAID controller. Great board, noticeably faster than my BX6 2.0. Unfortunately, I was using an inadequate PSU at the time (2004), and the motherboard had bulged caps on the VRM, which I ignored. One day, it just didn't boot up, and that was the end of that.

It's not too late to fix that, if you still have the board. I don't believe popped caps can do any damage to the rest of the board. If you replace them I'm sure it'll work as new again.
Besides, have you ever tried the 150+ FSB speeds this board has to offer? What's your experience with that?

Artex wrote:

I hear ya - Abit really had problems with bad caps it seems. I need to find the time to get my two boards recapped. I don't have the patience, or the skillz to do a good job.

Best way to learn IMO, is to practice: Get an old scrap board, pull all the caps out, and put them back in. Repeat until you feel comfortable fixing a valuable board 😀

That said though, I did kinda botch this at one point: the solder from the temporary caps I installed previously was really hard to remove from the holes, for some reason. I ended up damaging the copper plating a bit while trying to poke a hollow needle through them. In the end I didn't really trust the solder joints on the new caps so I bridged them with a leftover cap lead like this: http://i.imgur.com/rZIecOm.jpg
Clearing solder from the holes manually is by far the most tedious and annoying part when replacing caps, not to mention easy to screw up. Maybe investing in an automatic desoldering iron wouldn't be a bad idea after all...

ODwilly wrote:

I bet those North Bridge heatsinks were for those crazy 775 Nvidia chipsets that got super hot. Like the 680i

Could be, I've never seen those in action before, and I can't find any TDP specs on them. Can hardly believe these took much more than 20-30W though...

5u3 wrote:

Nice recapping job, looks awesome! Where did you buy the caps? I have a hard time finding any polymers that aren't SMD 😕

Got them from Mouser, they have a large selection of both electrolytic and polymer caps (and basically anything electronics-related 😉)
One annoying thing though is that for some reason you can't filter by lead spacing in the polymer category on their website, which is rather important when looking for replacements. Also make sure to double-check the datasheet before ordering, I found several caps listed with 3.5mm lead spacing which were really 5mm according to their datasheet.

jwt27 wrote:

5u3 wrote:

Nice recapping job, looks awesome! Where did you buy the caps? I have a hard time finding any polymers that aren't SMD 😕

Got them from Mouser, they have a large selection of both electrolytic and polymer caps (and basically anything electronics-related 😉)
One annoying thing though is that for some reason you can't filter by lead spacing in the polymer category on their website, which is rather important when looking for replacements. Also make sure to double-check the datasheet before ordering, I found several caps listed with 3.5mm lead spacing which were really 5mm according to their datasheet.

Thanks! I've searched at some german distributor's websites before, they have a much narrower selection of parts. And their website search engines are absolutely horrible! I didn't know Mouser also operates in Europe.

Can I say that that recapping job is sexy as hell! It's like a classic car being restored. Anyone one here know anyone state-wide that could recap a motherboard like this?

jwt27 wrote:

It's not too late to fix that, if you still have the board. I don't believe popped caps can do any damage to the rest of the board. If you replace them I'm sure it'll work as new again.
Besides, have you ever tried the 150+ FSB speeds this board has to offer? What's your experience with that?

That board has long since been thrown out... Unfortunately...

Is there a VRM high on those Slot 1 boards, or does the VRM work differently from modern boards?

Beautiful job BTW. I think they also have 1500uF in 10mm diameter, but you're right, 1000uF is more than adequate with polymers.

ODwilly wrote:

I bet those North Bridge heatsinks were for those crazy 775 Nvidia chipsets that got super hot. Like the 680i

jwt27 wrote:

Could be, I've never seen those in action before, and I can't find any TDP specs on them. Can hardly believe these took much more than 20-30W though...

The Asus board that I had with the 680i chipset had a massive copper heatsink w/ a 60mm cooling fan attached and STILL got hot. From what I understand reading up online the 670i was a big disappointment performance wise so Nvidia's solution was to add another chip along the Northbridge to do something or another with the end result being two hot running NB chips in one! 😊 this was awhile ago so some of the details might be blurred mind you.