Announcement

**stj** · 02-14-2019, 04:23 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

chemicon psg series maybe.
or kemet?
https://uk.rs-online.com/web/p/alumi...itors/1739437/

**ChaosLegionnaire** · 02-14-2019, 06:32 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

careful! this board may not take kindly to polymodding due to the coils in the cpu vrm. the vrm coils look like they have 6 or more turns which means the vrm uses a low switching frequency and this sort of vrm design prefers high capacitance over low esr.

i recommend u recap using 10k hour electrolytic caps but if u insist on polymodding, u assume all responsibility and bear the cost of paying for the costly poly caps that unfortunately failed to work...

if polymodding:

OST RLS 3300uf 10v 10x25mm -> use chemicon npcap psg series (15k-20k hour cap series) 2200uF 16v 10x20mm like stj suggested. the nichicon rr5 series u linked is only a 2000 hour cap series.

if that is not available, u can try panasonic oscon sepc 2700uF 2.5v 10x13mm; nichicon fpcap nu series 2700uF 2.5v 10x12.5mm; chemicon psc series 2700uF 2.5v 10x11.5mm. u have to keep the capacitance as close to the originals as possible and do not downrate due to the low switching frequency of the vrm.

its okay to go lower on the voltage of the caps because this is for the cpu vrm output. the cpu core voltage for athlon xp cpus is typically 1.75v at most so 2.5v caps will have no issues there even if u overclock and overvolt the cpu to 2v or more.

OST RLX 1500uf 16v 10x20mm -> this is for the cpu vrm input which takes 12v. use chemicon npcap psg series 1500uF 16v 10x16mm like stj suggested.

if that is not available, u can try panasonic oscon sepf 1000uF 16v 10x13mm or nichicon fpcap ne series 1000uF 16v 10x12.5mm. u have to keep the capacitance as close to the originals as possible and do not downrate due to the low switching frequency of the vrm.

OST RLX 1500uf 10v 10x16mm -> these are for general filtering of the minor rails and wont see more than 5v. use chemicon npcap psg series 1500uF 16v 10x16mm like stj suggested.

if that is not available, u can try panasonic oscon sepc 1500uF 6.3v 10x13mm; nichicon fpcap nu series 1500uF 6.3v 10x12.5mm; chemicon psc series 1500uF 6.3v 10x11.5mm.

its okay to go lower on the voltage as these caps wont see more than 5v. the stock caps are overrated to 10v because its a trick that manufacturers use to get junk caps to last longer i.e. higher voltage junk caps tend to last longer than lower voltage rated junk caps.

OST RLP 1000uf 10v 8x13mm -> these are for general filtering of the minor rails again and wont see more than 5v. this one is tricky, due to height restrictions, u have to downrate and use chemicon npcap psg series 560uF 16v 8x11.5mm. any caps taller than 13mm may not fit and may cause an obstruction when u try to insert an agp or pci card. so for the four caps that are around the agp slot, use the above ones i mentioned. for the other 6 caps around the dimm slots, u can use chemicon npcap psg series 1000uF 16v 8x20mm.

however, there is a cap that is next to the southbridge heatsink almost in line with the agp slot. if an agp card is inserted, that cap may collide with the video card. so kindly check the exact position of that cap if it may collide with a long agp card. if so, u have to use the shorter 560uF 16v 8x11.5mm cap there.

if that is not available, u can try panasonic oscon sepc 1000uF 6.3v 8x7mm or nichicon fpcap nu series 1000uF 6.3v 8x11.5mm.

if sticking to electrolytic caps:

OST RLS 3300uf 10v 10x25mm -> use panasonic fr, rubycon zlh or chemicon kzm (these are 6k to 10k hour electrolytic cap series) 2700/3300uF 6.3v 10x20/25mm caps.

OST RLX 1500uf 16v 10x20mm -> use panasonic fr, rubycon zlh or chemicon kzm 1500uF 16v 10x20mm caps.

OST RLX 1500uf 10v 10x16mm -> use panasonic fr, rubycon zlh or chemicon kzm 1500uF 10v 10x16mm caps.

OST RLP 1000uf 10v 8x13mm -> use panasonic fr, rubycon zlh or chemicon kzm 820/1000uF 6.3v 8x11.5mm caps.

**Rawrl** · 02-14-2019, 03:10 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

Thanks for the replies.
I'm not married to the idea of poly modding it, I just figured they'd last longer. This is an Athlon XP, Nvidia chipset, and high-end video card (probably gonna go with a Radeon 9800 or X800) all crammed into a tiny box and cooled by a single 80mm fan. Polies are supposedly more heat tolerant, and I certainly don't want to have to do the same work twice.

From what I was reading, 8-10 turns was the indicator for a low frequency VRM. If 6 is a better rule of thumb, I'll keep that in mind for the future.

For the other caps on the board, I'm aware of the voltage substitutions some manufacturers do and why, I was more concerned with the impact of drastic changes in ESR. I know that failed (and thus presumably severely mismatched) caps in the minor VRMs can kill the northbridge and RAM on these, which was mainly why I hesitated. Similarly, most of the RLPs look like filters, as you say, but there are a couple in close proximity to regulators near the AGP slot (1.5v supply?), so the same concerns apply.

It certainly would be easier to just throw a bag of Panny FRs at it...

**ChaosLegionnaire** · 02-15-2019, 03:38 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

yes i understand your plight. it really is a crapshoot if it should be polymodded or not. but from the pics u posted, nearly all of the caps are in potential hotspots around the board. u have caps near the nb and sb heatsink, caps near the cpu socket and caps around the agp slot since u intend to use a hot high-end agp card in there. the radeon 9800 series of cards has a tdp of around 50w, almost as hot as a cpu and the x800 around 50-65w depending if its the pro or xt. there have been cases on here of bloated caps around the agp slot due to usage of a hot high-end agp card.

there are only 4 caps on the left side of the picture around the dimm slots that arent considered in hotspots around the board. if that was the case, prolly just better to polymod the whole damn thing if most of the caps are in hotspots!

Originally posted by Rawrl View Post

From what I was reading, 8-10 turns was the indicator for a low frequency VRM. If 6 is a better rule of thumb, I'll keep that in mind for the future.

well lately it has become a concern because of this thread where someone polymodded an Abit NF7-S 2.0 which is also an athlon xp board. the cpu vrm has around 5-6 turns on this board similar to yours and someone polymodded it and reported a loss of overclocking, about 100 mhz loss of stable overclocking. so it seems the decrease in capacitance or maybe the heavy decrease in esr contributed to some stability issues in the cpu vrm albeit it is a very tiny minor issue: 100 mhz loss of overclocking. however, no major problems like the board catastrophically failing, explosions, burning or "popping" were observed which is what u were worried about.

so if u arent overclocking, just go for the polymod. personally, trying to overclock such a hot running system already would be suicidal, hardware wise...

**stj** · 02-15-2019, 09:07 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

keep the capacitance the same.
the halving of it was an experiment on apple shit.

**momaka** · 02-28-2019, 01:12 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

Originally posted by Rawrl View Post

Also, on a related note, I'm going to order a Shuttle SilentX PC40 250w power supply to replace the weedy 200-watter the system came with, and I'll need to recap it (with electrolytics, naturally!). Has anyone done one of these, and do you still have your notes? It would be nice to order everything for the board and the PSU at the same time, not to mention not having to chase down replacements for whatever junk caps are in it. (Heck, if someone has a 300w PC50 lying around, I'd gladly take it off your hands!)

I have a PC40N250EV out of a Shuttle SB83G (other part numbers on PSU are 9TPWR2503V and 88-220-250170). Has been many years since I connected power to it or tested it, so probably not a good idea to sell it to anyone. But I can get those values for you.

Here are the caps found inside:

2x OST RLS, 16V, 2200 uF, 10x25 mm (both for the 12V rail)
4x OST RLS, 10V, 2200 uF, 10x20 mm (2x for 5V rail, 2x for 3.3V rail)
2x OST RLS, 6.3V, 1000 uF, 8x13 mm (both for 5VSB)
1x OST RLS, 6.3V, 1000 uF, 8x13 mm (both for 5VSB)
2x Ltec ___, 25V, 10 uF, 4x8 mm
3x Ltec ___, 25V, 100 uF, 6.3x11 mm (1x is the PSU "start-up" cap; 2x are for -12V rail output, one of which can accommodate 8 mm cap)

Notes:
1) For the 16V, 2200 uF OST RLS, the max height restriction is about 32 mm.
2) For three of the 10V, 2200 uF OST RLS, the max height restriction is about 32 mm. Only one of them near the secondary heatsink has a max height restriction of 25 mm.
3) For the 6.3V, 1000 uF OST RLS, the max height restriction is about 20-25 mm if they are slightly bent away from control daughterboard.
4) The 2x Ltec 25V, 10 uF caps are likely general filters for the daughterboard, so ESR is not critical. Can easily be replaced with standard 5x11 cap and there is no height restriction for these caps.
5) For the Ltec 25V, 100 uF caps on the 12V rail, height restriction is again about 32 mm. One of these caps can also have 8 mm diameter (PCB has silkscreen for it.) As for the "start-up" cap - it's under a heatsink, so max height is probably limited to 20 mm max. Shouldn't be an issue, though to find a small 25V 100 uF small cap.
6) Main input cap on primary side on my PSU is a Hitachi HP3. Not sure about the voltage and capacitance spec, as it is wrapped in tape and hard to read without taking out the cap. But this cap shouldn't need replacement anyways, unless yours is a CapXon or some other TW/CN crap. In any case, size for this cap is 25x30 mm in case anyone needs to replace one in their PSU.

**momaka** · 02-28-2019, 01:42 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

Originally posted by Rawrl View Post

From what I was reading, 8-10 turns was the indicator for a low frequency VRM. If 6 is a better rule of thumb, I'll keep that in mind for the future.

There isn't really any rule of thumb.

Generally, the only low-frequency VRM switching motherboards to be aware of are those with TL494/DBL494/KA7500 chips driving the VRM - some of which existed up to the late s478/s462 era (and those were mainly "lower-tier" mobos like ECS and PCChips). But just about everything else wouldn't really care too much about a slight decrease in capacitance (up to 30% loss probably won't make a difference at all.)

Originally posted by Rawrl View Post

Similarly, most of the RLPs look like filters, as you say, but there are a couple in close proximity to regulators near the AGP slot (1.5v supply?), so the same concerns apply.

Yes, that would be the AGP power supply rail. That rail is almost always linear type, so it's a good idea to keep the capacitance closer to the original. Too low of an ESR generally shouldn't be an issue for it either... but in a marginal design, it could be (though, I've personally never seen any linear regs on mobos oscillate due to using polymers.)

Originally posted by Rawrl View Post

It certainly would be easier to just throw a bag of Panny FRs at it...

That should be fine, as Panasonic FR are quite tough. Will likely easily last at least 10 years in there without issue, if not 15. So I wouldn't sweat it about doing a complete poly recap. If heat really is an issue inside this case, it would probably be a better idea to invest time into adding more ventilation/holes in the case rather than messing with polymers. After all, those hot temperatures will probably impact the lifespan of the chipset too. Speaking of which, consider adding a small fan on the NB heatsink.

Originally posted by ChaosLegionnaire View Post

well lately it has become a concern because of this thread where someone polymodded an Abit NF7-S 2.0 which is also an athlon xp board. the cpu vrm has around 5-6 turns on this board similar to yours and someone polymodded it and reported a loss of overclocking, about 100 mhz loss of stable overclocking. so it seems the decrease in capacitance or maybe the heavy decrease in esr contributed to some stability issues in the cpu vrm albeit it is a very tiny minor issue: 100 mhz loss of overclocking. however, no major problems like the board catastrophically failing, explosions, burning or "popping" were observed which is what u were worried about.

It's likely the decrease in capacitance that caused it, because the lower the VRM switching frequency is, the bigger effect capacitance will have on ripple and noise suppression. This is why in linear circuits (like those simple oldschool iron-core line transformer "wall-warts" / adapters) you always need fairly big caps even for very small currents.

Think of this as battery for your cell phone: you can either have a small battery (smaller capacity) power your cell phone or a large one (large capacity). The difference will be that with the small battery, you will have to charge it more frequently if you don't want your phone to lose power. With the smaller battery, the time it takes to charge the battery will also likely be shorter. Whereas with the bigger batt. you can charge it less frequently, but when you do, it will take longer too (given the same charger anyways). Now, as to where ESR comes into play... that's not something that can be illustrated easily in the above particular example.

Anyways.
A 100 MHz decrease in overclocking should NOT be a concern at all.

**stj** · 03-01-2019, 05:27 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

the only thing you should "overclock" in a shuttle is the fan!!

**Rawrl** · 03-04-2019, 08:37 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

Thanks for the additional information, momaka.

I've been meaning to update this thread with my results. I finished the recap at the beginning of last week, but still wanted to finalize the build, and then I spent the weekend stress-testing it.

For the motherboard, I wanted to go full poly on it, but luckily common sense reasserted itself when I started pricing everything out, and I wound up doing polies on the VRM and FRs everywhere else.

Code:

OST RLS 3300uf 10v 10x25mm -> Chemicon PSG 2200uf 16v
OST RLX 1500uf 16v 10x20mm -> Chemicon PSG 1500uf 16v
OST RLX 1500uf 10v 10x16mm -> Panasonic FR 1500uf 10v
OST RLP 1000uf 10v 8x13mm -> Panasonic FR 1000uf 6.3v*

For the PSU, I mostly went with Panasonics, except for the 25v 100uf caps, where I used Nichicon UHE. I ordered replacements for all three 100ufs, but the startup cap is trapped under a heatsink and boxed in by the daughterboard, so it's staying put unless it becomes a problem. Likewise I skipped the 10ufs, my understanding being that caps that small are seldom an issue.

Code:

OST RLS 2200uf 16v -> Panasonic FS 2200uf 16v
OST RLS 2200uf 10v -> Panasonic FR 2200uf 10v
OST RLS 1000uf 6.3v -> Panasonic FR 1000uf 6.3v*
LTEC TK 100uf 25v  -> Nichicon UHE 100uf 25v

Everything went back together over a couple nights, and the system seems to be working fine. Final specs are:

Code:

[b]AMD Athlon XP-M 2500+[/b] //2.2GHz @ 1.55v - Kinda bad for the stepping (IQYHA), I was hoping for lower vCore. Might see what it'll do at 1.5v.
[b]2GB DDR400[/b] //stock voltage/JEDEC timings, I'm aware of the vDimm issue on these boards.
[b]WD 320GB HDD[/b] //single-platter, for reduced heat and power
[b]ATI Radeon X800 Pro[/b] //I was considering some sort of Geforce 6800 as well, but R420 has better power figures, and with the 250w PSU...
[b]Sony DRU-820a[/b]
[b]System fan replaced with a Panaflo H1BX.[/b] //If this turns out not to be enough, I have a 4800rpm/72CFM Thermaltake Smart Case Fan II I can throw in there.

Sure, it's technically overclocked, but I'm getting 3200+ speeds at a lower TDP than the 2600+ that was originally in it. I'm not looking to break any records with this, just get as much performance for as low voltage as I can.
At idle after boot, it hovers around 41C. Load temps stabilize around 51C in Prime95 with the BIOS SmartFan massaged such that it stays on the second lowest speed (~1500rpm) at that temperature. Quitting Prime95, things eventually bottom out at 44C. As for the GPU, it stays around 38-39C idle with the system cover on. The "system" temp (I'm assuming it's the NB, but it might just be the hardware monitor chip or something) never gets above 44C. Not bad overall for a tiny, cramped shoebox.

I ran it for 3 hours on Prime95 in-place large FFTs, then again for 12 hours the next day, which it pulled off without a hitch. I still need to check how it handles an extended 3DMark load, as well as hook up a Kill-a-Watt to get a vague idea of what I'm doing to the poor PSU with all of this, but so far I'm pretty happy with how it turned out. This is my first major recap, not counting a S370 board I did years ago, and I've gained a lot of confidence to do some of the other stuff I've been sitting on to "someday, eventually" recap.

Thank you all so much for your help with this.

*As an aside, can I mention how much BS it is that Digikey doesn't carry 1000uf 6.3v Panny FRs? I wanted to do the entire order there, especially since they're cheaper, but noooooo. I have to make a whole separate order with Mouser and eat another 8 dollar shipping charge 'cause they don't feel like stocking an extremely common value in an extremely useful capacitor line. Jerks.

**ChaosLegionnaire** · 03-05-2019, 03:24 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

well not sure which version of the xp-m 2500+ u have as there are three different versions. there is a low voltage version with 35w tdp, a mainstream version with 45-53w tdp and a desktop replacement version with 72w tdp. but i guesstimate the tdp of the overclocked 2500+ to be anywhere around 70-80w assuming the worst case scenario.

the x800 pro has a tdp of 50w but if overclocked it can hit around 55-60w tdp. and if we add another 60w tdp for the mobo, hard drive and the other peripherals, the total max load system power draw should be around 200w. its pretty close to the 250w power limit for the psu but hoping that shuttle doesnt overrate their psus, it should be fine even though ati recommends a 300-350w psu for the x800 pro. the x800 pro has the same tdp of the 9800 pro which i am very familar with the power requirements for.

**Rawrl** · 03-05-2019, 03:34 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

It's the 45W version with 1.45 vCore, AXMH2500FQQ4C. I'm running stock speeds on the X800.

Originally posted by ChaosLegionnaire View Post

i guesstimate the tdp of the overclocked 2500+ to be anywhere around 70-80w assuming the worst case scenario.

What? Nahhh, no way it's pulling that much. The desktop 3200+ is rated at 76.8w TDP at 1.65v vCore[1]. I'm running a tenth of a volt less than that at the same speed.

I hooked up the Kill-a-Watt last night to get some ballpark power draw figures. After a brief heart attack moment where it kept crashing and restarting in the middle of 3DMark (which turned out to be the video driver[2]), and a much longer time finding a working driver [3], I finally got some numbers.

At idle sitting at the desktop, it pulls just over 100W. Prime95 in-place large FFTs raised that to 124W. In 3DMark03, both at default settings and with all of the goodies turned up, I observed a maximum power draw of 163W at the wall. Weighing the inaccuracy of a consumer power meter against the mediocre efficiency of the PSU[4] (72% under full load), I feel pretty good about where I'm at power consumption-wise. Obviously they're synthetic benchmarks, and I should probably throw Doom 3 or Far Cry at it to get real-world numbers, but I can't see those being too far off of what I have here.

1. Keep in mind, AMD's TDP specs from that era denote the theoretical maximum power draw, as opposed to Intel's vague approximations.
2. Good old ATI Quality Software™
3. Did you know, ATI had an entire string of drivers in the late 2000s where AGP cards just didn't work in Direct3D? F***ed up but true!
4. Meaning the actual draw of the components is even less, of course.

**ChaosLegionnaire** · 03-06-2019, 03:49 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

well thats quite a way below the 200w estimation so its good. i think the issue was busting the 250w limit of the psu which prolly wont happen. 163w is barely 65% of the power limit of the psu so its fine with what we have here.

some psus will overheat or have their lifespan shortened if u run it too close to its limit like about 80-90% of the limit. not sure if that shuttle psu u have is affected but since it wont see more than 70% of its power limit or so that wont be a problem. i guess that pc will last several years or more of use without problem which is good.

**momaka** · 03-06-2019, 09:56 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

Awesome work! Great to hear it's up and running again.

Originally posted by Rawrl View Post

For the motherboard, I wanted to go full poly on it, but luckily common sense reasserted itself when I started pricing everything out, and I wound up doing polies on the VRM and FRs everywhere else.

Should be more than fine.
After all, if the OST RLX caps lasted there for over 5 years, chances are the FRs will make it at least 10-12 years without any degradation, if not more. So full poly recap isn't really needed unless you just have the money and don't care.

Originally posted by Rawrl View Post

I ordered replacements for all three 100ufs, but the startup cap is trapped under a heatsink and boxed in by the daughterboard, so it's staying put unless it becomes a problem.

Yes, it can be a little hard to get to that one. With curved tweezers and some creative techniques, though, I've removed and installed caps from much tighter spaces than that.
But like you said, it shouldn't be a problem that you left it out. If the PSU ever acts "dead" (no 5VSB, no power), then you know it's time to fix that cap. That said, perhaps take the bag with the 100 uF cap and tape it to the side of the PSU with a note inside. That way, you will have the part right then and there instead of wondering where you might have placed it or what could be wrong with the PSU.

Originally posted by Rawrl View Post

Likewise I skipped the 10ufs, my understanding being that caps that small are seldom an issue.

Indeed.
While small caps like that do fail quite often (especially from Ltec), the circuits they are used in aren't usually very sensitive. Of course if those 10 uF caps were used on something important, like the startup cap or BJT bias/drive in H-bridge PSUs, you could be in for some trouble.
In your case here, though, I think those 10 uF caps are just general filters for the ICs on the daughterboard, so not that important really.

Originally posted by Rawrl View Post

This is my first major recap, not counting a S370 board I did years ago, and I've gained a lot of confidence to do some of the other stuff I've been sitting on to "someday, eventually" recap.

Thank you all so much for your help with this.

Yup, once you get started, it's much easier to continue.

You're welcome!

**ChaosLegionnaire** · 03-07-2019, 03:45 AM

Re: Poly modding an old Shuttle, question about non-VRM caps.

Originally posted by momaka View Post

perhaps take the bag with the 100 uF cap and tape it to the side of the PSU with a note inside. That way, you will have the part right then and there instead of wondering where you might have placed it or what could be wrong with the PSU.

lol thats like a psu spare part time capsule idea. i think that idea's excellent lol!

**momaka** · 03-07-2019, 06:07 PM

Re: Poly modding an old Shuttle, question about non-VRM caps.

^ It was actually quite a common practice on some old appliances. For example, last year I decided to "deep clean" our gas stove/oven and removed a panel on the bottom to clean under it. Then I saw this metal tab with four nozzle-like things and a label hanging by them. Checked the label and it said these were indeed gas nozzles - just spare ones in case the original ones wear out / clog.... which is funny, because this oven/stove is about 25 years old and has never been serviced or needed servicing. I suppose if they were to clog today, I can change them and expect another 25 years of life out of it?

I guess they don't make them like that anymore.

Announcement

Poly modding an old Shuttle, question about non-VRM caps.

Poly modding an old Shuttle, question about non-VRM caps.

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment