Replacing OST caps on Microstar KT6 Delta motherboard

[mariushm]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

I've explained already the above.

p is short for PICOFARADS

475 means 47 x 10 ^5 = 47 x 100000 pF = 4,700,000 pF = 4700 nF = 4.7 uF

It's not necessarily for decoupling, could be for a similar role, to reduce EMI (electromagnetic interference) and make the motherboard radiate less "noise".

You'll never see y5v capacitors on a motherboard, look for X*R.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

Edit (for clarity):
"Seeing 1.26V on the Source of Q28 and 30, rules out the possibility that these [other two mosfets] are shorted between Drain and Sink".

--------------------

There are 2 ceramic caps before the choke, a 1uf and 0.3uf.
then after the choke, are the elecrolytics (the bulk capacitors?).
then, after that, there 2 more ceramic caps, which are 4.7uf (475P)
Would all of these 4 smaller ceramics be decoupling capacitors? (the guy in that decoupling capacitor vid suggested that the first one would be a bulk one, but i don't think so in this instance)


Looked up what the P meant http://www.wescomponents.com/datasheets/capacitor/ and it appears that it is tolerance (+100% -0%)
On this webpage: http://au.rs-online.com/web/c/passiv...ved=4294519046 there isn't any with that tolerance.
Would another one of them do the job?
There is one that is (-20, +80), but that is a Y5V - not an X5R, or X75.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

OK i get it.
A lot clearer now, thanks.
Just 2 little questions about that:
1. "after the input capacitors", the tee off that feeds the controller is not shown on the schematic. Is this the "VCC" terminal 16 on the PWM (HIP6302)?
2. i see that terminals 13 and 8 (Phase 1 & Phase 2) on the driver (HIP6602), go straight to VCORE, what is the function of this path? - It looks like they tweak the thick heavy current that suppies the CPU.

I'm not too keen on desoldering the mosfets if it's not absolutely necessary.
So, i powered up the motherboard (with known good ram and CPU),
then checked voltage at mosfets Q28 and Q30 - the Drain shows 12.2v and the Source shows 1.62V - which looks like the right voltage to power-up the CPU
3. (BTW, what is the function of the other two mosfets (Q27, Q29)? - when they are switched on, they look like they create a path to earth)

Seeing 1.26V on the Source of Q28 and 30, rules out the possibility that these mosfets are shorted between Drain and Sink

So seeing that the voltage was correct at the source of both mosfets (which seems to indicate that the PWM is doing that part of it job correctly),
i then tested both sides of both of the chokes, and got 1.62V there.

4. i also note that there is another path to the VCORE terminal, that comes from the PWM terminal (VSEN) - what does that do - is it a sensor?

Anyway, what would be the next logical step after this testing?
Would it be to remove and test the (some of which are) very slightly bloated 6.3v output caps to make sure that they are not metamorphizing into resistors, or is there a way to go straight to the VCORE terminals (or are they hidden under the CPU housing?) and test voltage there before removing and testing the suspect caps?

Cheers

[mariushm]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

I don't see how it's so hard to understand. We're talking about two different things.

The thick, heavy current flow goes like this :

12v > input inductor > input capacitors > mosfets > output inductor > output capacitors

The mosfets are like on/off switches that have to be turned on or off by something. No matter what state the mosfets are in, you still have 12v after the input capacitors.

So after the input capacitors, a bit of 12v is taken to power up the controller and drivers which then turn on or off the mosfets and monitor the output voltage and adjust how fast the mosfets are turned on or off to maintain the voltage where desired.
The heavy current, the amps on 12v, don't go through these controller+drivers side, for example in theory they could be powered separately from a 9v battery if you'd want to... they just take power to work from the same 12v because it was convenient, already there close to the chips.

So you also have

12v > input inductor > input capacitors > controller (and )
12v > input inductor > input capacitors > driver(s)

and then you have signals like this:

controller ----> drivers ----> mosfets (on/off)

At any time, there's 12v at the input of the mosfets, controller and driver but without signal from the drivers, there's no output voltage on the mosfets, they're switches after all controlled by the drivers.

So I'm talking about how current flows from input to output looking at the controller/driver as something auxiliary, as a sort of "management" circuitry, a "gatekeeper", which just controls the flow.

He's arranging the components in the order of importance, not in the order the components are physically connected on the board. He's basically saying:
First is 12v, inductors and capacitors because nothing works without it.
Next is the controller and the phase drivers because mosfets depend on them to work, without them nothing moves further in the circuit.
Next he puts the mosfets because those let the current flow through them, if they're dead or don't turn on or off, current doesn't flow further.
Next is the output inductor which smooths the current pulses and the output capacitors which further smooth things out.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

That's cool, when/if ever you want
no problem

Regarding the VRM, in terms of the direction of current flow.
The way you describe it in post #17: PSU > 4pin12v > 16v caps > 4 mosfets
The guy in the vid says: PWM > Driver/Doubler/ > 4 mosfets > 2 Phases > inductors > CPU
So you might see where i was getting confused?

i now note that there is two terminals on the scematic called VCC - one at the top left, and another at pin 16 of the PWM.
And also see that the label VCC can refer to one physical terminal with at least these two circuits connected to it.
so i can now see that current flows forward in both of the above descriptions.
is the path that you describe the feedback loop?

Anyway, still have to rewatch those two vids again, and the mofset ones

Powered up the Mobo again, just to see if it would boot, since the melted cap is now removed.
Nope - only CPU fan spins
Both pins show 12v on the multimeter, and the earth pins are good also.
When i pull out the 4pin connector, i can hear the PSU fan speed up,
which, to me, suggests that current is passing through the VRM (when the 4pin connector is plugged in).

off to get some zzzzzzzzzzzzzzzz's now also.

[mariushm]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

HIP6302 :

Microprocessor CORE Voltage Regulator Multi-Phase Buck PWM Controller

The HIP6302 multi-phase PWM control IC together with its companion gate drivers, the HIP6601, HIP6602 or HIP6603 and Intersil MOSFETs provides a precision voltage regulation system for advanced microprocessors.

-
Don't look at the board and think it's layed out exactly as in the schematic. 12v doesn't have to go into the controller, come out and go somewhere else, they're not in "chain" - the controller 6302 and the drivers 6602 can be somewhere further away from the mosfets and capacitors and some tiny traces can send 12v to them to power them up and other tiny traces can go to the mosfets and drivers and so on.

-

The continuity mode on your meter will beep at low resistance (varies from meter to meter, can be anything less than 10 ohm for example).
Just tested on an old board I have around, used continuity mode over one of those capacitors by the cpu, i got 6.28 ohm. You're not supposed to have no resistance and it doesn't tell you the capacitor is shorted, it just means there's something in parallel with the capacitor making the multimeter think there's continuity.


I have to get some sleep now, won't reply for at least 8-10 hours.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

Thanks again,
That's some seriously good info.
i note the two phases on the same mosfet (the two separate drawings on the schematic caught me out for bit)
will be revisiting those first two vids for a recap-itulation.
before i do, just quicky...
1. The "HIP6302 dc-dc converter", that you mentioned, is the PWM?
And the "phase controllers HIP6602" is one component containing two drivers/doublers.
So, looking at the schematic, the current is coming in, getting smoothed by the capacitors, then going into the drivers, and then into the PWM!
But the dude in the video says that current hits the phase controller, and then goes to the drivers/doublers.
2. So is the reason why the current goes to the drivers first - because this is for the feedback loop?

Regarding that meter reading (which indicated continuity), when placed across any of those 12v, or 6.3v, electrolytic caps before the mofsets:
I grabbed another board that was good, and got the same results
So this continuity appears to be normal
Used the digital multimeter ohms beeper and got continuity, when probes were placed across the 6.3v caps, and a brief beep when across the 12v
Which is the same result as for the faulty board

Seems that some current is allowed to flow to earth via those transistors (Q27-Q30), or then through the mofsets, to earth, since the multimeter shows complete continuity or (in the case of the 12v caps) some non-infinite resistance?
Thinking about it now, it seems makes sense - even obvious?

On a separate note,
i'm beginning to wonder if that melted cap, although it looked split in half, was actually a slight short, and therefore not enough current was being supplied to the PWM for the CPU to fire up (should have tested that before cleaning it).
When i tried a good cpu, a week ago, the CPU was cold, but the CPU fan was spinning and it stopped and started once, before i turned off the power.

[mariushm]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

Ok, i can see why you're confused about what i wrote.

When I say dc-dc converter, I generally mean to say the whole circuitry that converts a dc voltage to another dc voltage (so chip, input and output capacitors, inductors, diodes and/or mosfets etc).
Sometimes I refer to a dc-dc converter chip or a dc-dc switching regulator and that means to me a more advanced dc-dc controller, one that often has a built-in mosfet and possibly a diode, parts that are generally required to build a dc-dc converter.
A dc-dc controller means to me just a very basic sort of chip, in the sense that it needs those other parts to build a whole dc-dc controller. In your case, that HIP6302 is just a controller because it relies on those other two tiny driver chips to operate the mosfets, the controller just sends commands to those to increase or decrease the frequency of operation of those drivers.

Now as you figured out, the schematic is for KT6 Delta (codename MS-6590) but there were several revisions so you have to pay attention and follow the traces to see if those part numbers match with what you have on the board.
This being said, that tiny capacitor does look blown up and you did good by desoldering and cleaning that area.

There's only two terminals there, one thick large pad because there's lots of current there and on the other side you may have what looks to be two traces but basically they both go in the same place.

1. What could have caused this?

Why did it blow up? Probably over voltage... ceramic capacitors can short if the voltage goes higher than the capacitors' rating.

Other reasons could be just physical failure in the capacitor due to board bending, from flexing due to heat/cold cycles, from vibrations in the board from cpu cooler etc etc.

Electrolytic capacitors rarely fail shorted or partially shorted, ceramic capacitors and tantalum capacitors usually fall like that.
It's unlikely the electrolytic capacitors are faulty like that, but they could be out of spec.
I would look more towards those mosfets - check for continuity between drain and source, between gate and source etc etc. If the mosfet is blown, usually the source and drain are shorted.
Search on youtube "how to test a mosfet" .. or see these:

https://www.youtube.com/watch?v=gloikp9t2dA
https://www.youtube.com/watch?v=RBJGOOTEwfU
But note this type of testing is kinda problematic when you have those mosfets in circuit, as another mosfet can affect the measurements you make.

2. In such an instance, do those tiny caps explode before the big caps?

Ceramic capacitors are more sensitive to over voltage. Electrolytic capacitors don't pop right away if there's some over voltage for very small periods of time, the electrolyte inside them can start to go bad, the aluminum foil can be damaged, but the capacitor will still work reasonably well.
The electrolytics could be a tiny bit affected inside but they'll still work - they're more resilient.

3. What is the function of these tiny caps that are in parallel with the 16v electrolytic caps, and how necessary are they?

They're there for filtering and for EMI cleaning. Ceramic capacitors are very good at cleaning high frequency noise, electrolytic capacitors handle lower frequencies much better and do the heavy and rough work. Ceramics are for finesse, fine tuning, for decoupling .. see what decoupling means here: https://www.youtube.com/watch?v=mk61DNz27FI

They won't be critical, you can test motherboard without them, but when you manage, you should install them on the motherboard.

4. Does 475P/0805 mean 475 pF, and what does 0805 represent?

0805 is the size , 0.08 inch x 0.05 inch or 2.0mm x 1.2 mm
475p is the size. It may mean 47 and 5 zeroes after it, 4700000pF = 4.7uF. They should be X5R or X75 or C0G/NPO (temperature coefficients)
If it's really 4.7uF, the voltage rating of the capacitors is probably as low as 10v or 16v.
For example, it could be one similar to this: http://uk.farnell.com/kemet/c0805c47...05/dp/2409055?
It could be even rated for 6.3v... in which case that may explain why the big capacitance value - capacitance of ceramic capacitors changes with voltage, a 4.7uF 6.3v rated ceramic capacitors may only be 3-3.5uF at 1.5v.

5. Is it possible to solder a new cap on (with tweezers and an electron microscope)?

It is possible but it may be very hard because one of the traces is a large copper pad which will suck up the heat from your iron tip.
After cleaning up the burned area and the disintegrated cap,

i turned the board over and tested [...]

Not sure how you used the multimeter... but you should know that you really can't get accurate results because all those capacitors are in parallel on the board.
You get the capacitance that's basically the sum of all the capacitors, you get a much lower ESR value that's also affected by those smd capacitors.
Really, you can't test capacitors in-circuit, if they look bad or swollen you would replace those anyway.

If you don't have an ESR meter, you should know that capacitors can go bad and still show good capacitance because multimeters measure the capacitance at low frequencies. The characteristics of a capacitor change when it's working at 100kHz or whatever frequency they're subjected to when motherboard works, that's when the ESR value changes and the circuit can go bad. So if you're not sure about them, just replace them, it's safer.

[Agent24]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

The pad will have burned up along with the capacitor then. If you can see the track or via that it went to, you can solder the good end normally then solder the other end with a thin piece of wire. Tricky but should be doable.

Since you have a few parts in parallel you should be easily able to find a common connection to wire to, nearby.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

ta,
upon further inspection in the daylight, the motherboard now has a pit in it, and the left pad that the cap was soldered to, is no longer there
(have put a better photo on that webpage link)
that vid only applies if there is two pads already there.

i get a continuity between those two metal (they look white on the photo) bits on the left, and the motherboard earth at the metal circles around the board mounting holes

also, i'm now going to remove the 16v and test each of them with the analogue multimeter.

[Agent24]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

C154, that ceramic capacitor could have just shorted by itself and burnt out. Or was stressed due to the faulty electrolytic capacitors not filtering the supply properly and shorted out.

470pF sounds right to me. 0805 indicates the size.

Installing a new one would be easy, following this method: https://www.youtube.com/watch?v=qs7rSM731gk

Suggest you practice on an old board first though. But it's actually quite easy once you get the hang of it. And make sure you have good flux.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

Hi Marishum,
Two things were mentioned in your post: dc-dc controller, and dc-dc converter.
Just to be clear about this,
is the dc-dc converter the one in the PSU?
and the controller, the two mosfets (and the driver) on the motherboard?

i watched those two vids, and will be watching them again
woah - blew my tits off!
but i get the basic idea of what is going on, thanks.

Ok,
i was hunting for position C23 and couldn't find it on the board (maybe it's not there since this is rev 2.0?)
And while i was doing that, i noticed an unusual component, then i realized that it was not an unusual component, but rather a burned-out small capacitor)
1. What could have caused this?
i don't see how a partially-shorted 16v cap could cause this - it's not like that would increase the current from the PSU,
so it seems like this is the result of a suspect PSU, not blowing it's fuse when it should.
2. In such an instance, do those tiny caps explode before the big caps?
3. What is the function of these tiny caps that are in parallel with the 16v electrolytic caps, and how necessary are they?
4. Does 475P/0805 mean 475 pF, and what does 0805 represent?
5. Is it possible to solder a new cap on (with tweezers and an electron microscope)?

After cleaning up the burned area and the disintegrated cap,
i turned the board over and tested (+ve to +ve, -ve to -ve) across all the three 12v caps, and got 2/3 FSD on ohms x1
Then tested across the seven 6.3V caps, and got FSD on all of them
On both tests, the needle did not drop back - indicating current flow through all of them
(Had some 470uf lying around and tested on of them, the needle went up and down quite quickly, so the meter seems good)

Looking at the circuit diagram for the 6.3v caps, if one is shorted, and i measure across another good one, then the multimeter current will be going from one side of the cap, through the motherboard earth, through the faulty cap, along the +ve rail to the other side of the good cap (after the cap has charged).
So, it seems that, in this instance, in-line testing can not identify which cap it is, they will have to be removed, one at a time, until the fault disappears.

Here are the Pics:
http://www.stillflowing.comze.com/mo...melted_cap.htm

[pfrcom]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

No, that's the datecode. The cap was made the 38th week of 2003.

Here's a good replacement for those:
https://www.badcaps.net/store/produc...roducts_id=137

I do ship to Australia.

https://www.badcaps.net/forum/showthread.php?t=13260

[mariushm]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

I also see why the initial pulses from the PSU are 100-120hz. (2 pulses for each AC cycle)

No.

In a classical power supply like you see in audio amplifiers and other devices, you have a large heavy transformer (either toroidal or the classic EI shape one). Input AC voltage (at 80-250v AC , 45-65 Hz) comes in as a sine wave and comes out through the secondary winding at a lower voltage, still at the same frequency.. 45-65 Hz.
Then, a bridge rectifier is used to transform this AC sine wave to a DC wave with twice the number of "lobes" or "peaks" that originally. So now you have a DC output with 90-130 pulses a second. Now you have to use capacitors to smooth out this DC output.

A switching power supply (atx power supply etc) works differently.
The input voltage comes and it's right away rectified using a bridge rectifier to DC voltage and then a capacitor is used to smooth out this DC voltage a bit.
So you had 80-250v AC 45-65 Hz and after rectification you have about 110v - 360v DC with pulses at 90-130Hz.
Optionally, the power supply has an active pfc circuit which boosts this dc voltage up to 420-450v DC. Then, you have a capacitor which only has to make sure there's a minimum voltage all the time, for example let's say 150v DC.
So at any time, the power supply has an input voltage of 150v-450v DC.
Now, there's a controller inside the power supply which sends pulses of energy from this DC input through a high frequency transformer, the controller basically creates a square wave that goes through the transformer and on the secondary winding you have 12v or 5v or other voltages.
By varying the number of pulses and the duration of the pulses, the controller manages to produce at the output of this transformer a relatively stable voltage.
But unlike the classical power supplies, the controller works at high frequency, potentially up to 300-600 kHz, so that's why the transformer inside a computer power supply is so much smaller and lighter, and that's why you need smaller capacitors and so on. You're not working with 60-120 Hz anymore, you're working with thousands of Hz and with square waves, not sine waves.

So the formula that I mentioned above which approximates the capacitance required, it's only valid for the classic power supplies, using big heavy transformers that work with sine waves and low frequencies. It may be "good enough" for up to 400-600 Hz but by no means is it valid for switching power supplies.

So your questions at 1) and 2) makes no sense in the realm of switching power supplies. The ripple of the voltage is dependent on how well the controller reacts at sudden demands of current from components (to change the number and duration of pulses going through the high frequency transformer) and it also varies with the filtering implemented on the output. Lots of power supplies use a classic PI filter ( capacitor - inductor - capacitor) to smooth out the output and make the voltage vary less. Changing the inductor value (and/or its specs) in that filter can make just as much improvement as changing one of those two capacitors in the pi filter, for example.


Now back to your board. Here's a better image I took from google images:



And here's the schematic of your motherboard - note that it may not be the same as your exact board, it may be a different revision, but it's close enough:

http://savedonthe.net/download/561/m...v_100_sch.html

PDF should also be attached below.

If you go on page 39 in the PDF, you'll see the voltage regulator module. You see there voltage coming from VCC (the 4pin 12v connector) and then it goes through an inductor which is the one to the right of the 12pin connector. Then, it goes through several tiny surface mount capacitors and some capacitors that in the schematic show up as 1500uF 16v. The schematic shows 3 but your board could have only 2 x 2200uF or some other values. The capacitors I talk about should be the ones above that inductor, a bit to the right of the speaker jacks and the 4pin 12v connector.
From there, the electricity goes through at least 4 of those black squares with two leads coming out of them (those are mosfets). The schematic shows two phases, each phase having two mosfets.
Each phases' output goes through a choke/inductor (those coils to the right of the mosfets) and then the voltage goes out to capacitors to smooth out the voltage further.

If you check the pdf datasheet, you can see that there's a LOT of capacitors on the output of the regulator - you have there one phase with 2x2200uF+1800uF, one with 3x2200uF and another one with 3x2200uF. The schematic is ideal, it's more than needed, hence why the version of your motherboard only has 3 capacitors installed out of four locations above the processor socket (in my picture) and 2 capacitors to the left of the socket (in my picture) which I believe it's not related to the power going to the cpu, it's more likely to be filtering for the chipset (hence why it's smaller at 1000uF on your board, as chipset doesn't use a lot of power)

And if you're curious, the HIP6302 dc-dc controller and the two phase controllers/drivers HIP6602 (or whatever is used on your version of the motherboard) should be those tiny chips above the capacitors near the connectors, or in top right corner of the motherboard in my picture.

So you have some capacitors rated for 16v at the input, near that 12v 4pin cpu power connector. Those are the bulk storage for the dc-dc converter. dc-dc converter sends pulses of power through those mosfets and you have smaller voltage on those 6.3v rated capacitors after it goes through those chokes/inductors.
The 16v capacitors are needed for moments when the cpu suddenly demands lots of power, for example when it goes from 10w idle to 60w load in a game. The controller has to pull more current all of the sudden, the power supply has to react to this sudden demand and ramps up in a few microseconds or milliseconds, those 16v capacitors are there to provide instant energy until everything recovers or goes up to the new demand.
There's also other technical reasons why those capacitors are needed but the post is long enough already.

If you want to learn more about phases and how dc-dc converters work on motherboards, these videos are VERY informative:

Motherboard VRM Explanation Part1: The VRM and PWM : https://www.youtube.com/watch?v=zDxFbAhu4Bo

Motherboard VRM Explanation Part 2: Digital vs Analog PWM : https://www.youtube.com/watch?v=Viitg4Yoy2Y

[Dan81]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

If they are genuine ones, then it should be. Of course even the good ones don't last forever, so age is a factor.

Sounds like you may have had one that was already recapped? Check the solder joints on the capacitors, do they look any different to other parts on the board?

Strangely,it's not recapped at all,because if it was there would be a small resin mark,but there isn't any.All I know is that board originally came from a T-Systems MT40 machine.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

Very cool, thanks, mariushm!
Thanks for your time and effort.

Have used reference numbers here, 1), 2) etc, to make a reply easier.

Have read and re-read the first section of your post quite a few times now, and a couple of pennies dropped.
You lost me, a bit, when you mentioned "phases (2+1, 4+1, 6+2, 8+2 etc)"
But i sort of get the basic concept of the phasing creating more pumps (less ripple) in-between input cycles.

I also see why the initial pulses from the PSU are 100-120hz. (2 pulses for each AC cycle)

Regarding the formula: ( 2 x AC Frequency x Vripple)
1) How do i find out Vripple? I checked 4 PSU's, and nothing is written on their labels about ripple. Is Vripple the variation in voltage (see my link at the end of this post).

Regarding your comment on low ESR and capacitance.
Let's say the Capacitor was perfect and had no ESR, then the discharge would be completely regulated by the resistance of the circuit.
So lets say, for the existing 2200uf caps, that 2200uf is required because it is above the minimum capacitance required for functional circuit operation, according to the cap's ESR.
A larger cap with the same ESR could be used, but also a smaller one could be used with a lower ESR
Fair enough?
If so,

2) What is the (roughly - speaking) relationship between ESR and required capacitance?
i.e, if it was linear it would be 1/2 ESR = 1/2 capacitance


I note that you said that it was the cap near the 12v 4 pin connector.
3) Is that the one that deals with the output of the PSU and smooths the current to all of the rest of the motherboard?
It is not a "1200-3300uF 16v capacitor", but rather, a 1000uf 6.3v OST capacitor, which is not bulging (i know that that doesn't mean that it is ok)
Did you mean 6.3v, not 16v?
You mentioned a rough minimum requirement being 1000uf, and that's exactly what this cap is.

The caps that i was focusing on were the 6 larger ones, around the CPU.
- a couple of which are not quite flat.
4) What do these capacitors do?
Can i apply the same reasoning for their replacement, as the one at the 12v 4 pin connector.?


Also made a webpage on ripple
5) If someone can check this out to see if it is correct.
http://www.stillflowing.comze.com/capacitor.htm

[Agent24]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

I know this isn't a solution for OP but....
If a board is populated with only Rubycon and Panasonic caps,does that make it good?

I ask this as my ASUS P4P800-VM has only FJ Pannies next to the CPU socket,and the rest is Rubycon ZL,save for a MBZ between DDR slots and IDE ports and 2 small TEAPO caps next to the audio chip in the upper left corner.
However,ANY other version I've seen on the internet has Chemicon KZE instead of the ZL caps and either OST RLX or KZG caps instead of the Panasonics.

If they are genuine ones, then it should be. Of course even the good ones don't last forever, so age is a factor.

Sounds like you may have had one that was already recapped? Check the solder joints on the capacitors, do they look any different to other parts on the board?

[Dan81]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

Those Panasonic FR capacitors have 4000-10000h @ 105c so they'll last for years without any problems on such motherboard.

I know this isn't a solution for OP but....
If a board is populated with only Rubycon and Panasonic caps,does that make it good?

I ask this as my ASUS P4P800-VM has only FJ Pannies next to the CPU socket,and the rest is Rubycon ZL,save for a MBZ between DDR slots and IDE ports and 2 small TEAPO caps next to the audio chip in the upper left corner.
However,ANY other version I've seen on the internet has Chemicon KZE instead of the ZL caps and either OST RLX or KZG caps instead of the Panasonics.

[mariushm]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

yes, capacitance is an important factor, but you have to understand the dc-dc converters work at high frequencies and have multi-phase converters 2+1, 4+1, 6+1 etc so the ripple is very small.

If you work with mains power (110v or 230v and 50-60Hz) you rectify that with a bridge rectifier and you have a dc output with about 100-120 pulses of energy, so you have to smooth that wave out and you need a lot of capacitance depending on how much current you need (approximation for capacitance would be current / ( 2 x AC Frequency x Vripple)

But, on motherboards, you have 12v coming from the power supply which already has about 2200+ uF capacitance to smooth out the ripple. Then, right by the 12v 4 pin connector on the motherboard you have one or several 1200-3300uF 16v capacitors which again act as a bulk storage of energy for the dc-dc converter.
Now you have a dc-dc controller on the motherboard which works with several phases (2+1, 4+1, 6+2, 8+2 etc) ... each phase is formed of a one or several mosfets, an inductor and capacitor.
The controller works at high frequencies 300 kHz, 600kHz, some work even at 1Mhz and basically what they do is turn on mosfets for a brief period of time and then turn the mosfet off.. this can be thousands of times a second or more. The energy that manages to move through the mosfets goes through the inductor and then into the capacitor.
The controller monitors the voltage on the capacitors and adjusts dynamically the turn on and off times of each phase and that's how the voltage remains stable regardless of how much current the processor uses.

When there's several phases, the controller basically (for example) turns on a phase for 2 ms and keeps it for 3ms but 0.5 ms after it turns on the first phase it turns on the second phase and so on. Each phase is slightly delayed from the previous one, to increase the number of pulses of energy that go to the output.
So if a phase turns on 1000 times a second, in a second with 2 phases you'll have 2000 turn on and offs..

So think of whatever is in front of the 6.3v capacitors as a rectified AC voltage that was at thousands of Hz ... instead of just 100-120 big pulses of energy you have thousands smaller ones. So the capacitance needed to smooth those bumps is much smaller. The inductors in front of each capacitor also help a lot - inductors resist current change and for example, if the cpu suddenly pulls a lot of current, some energy from the inductors will go to the cpu until the dc-dc controller increases the turn on time.

-

can you, or someone else, briefly explain how a lower ESR justifies using a significantly lower capacitance replacement. (say, from 2200uf to 1200uf)
i.e. where do you draw the line - why not 600uf?

It doesn't necessarily justifies. The way I understand it (and I may be wrong) .. the esr is .. as the term says .. an internal resistance. You have V = I x R and P = I * I * R .. so if the capacitor has 0.01 ohm esr, and your cpu pulls 1.3v at 10A then you have a potential voltage drop of 0.1v and about 100*0.01 = 1 watt dissipated into the capacitor.
With lower esr you have lower voltage drop which makes it easier to make the output voltage more stable with variations in current, and the capacitors heat less and last longer time.
So paralleling several capacitors reduces this esr, every capacitor heats less, there's more leads and more wider traces to handle those currents etc etc etc

depending on how many phases and how fast the dc-dc controller is on your motherboard, that guides you in choosing the minimum capacitance.. some older intel datasheets/documents were suggesting a minimum of 1000uF capacitance or something like that, but suggested using multiple capacitors for lower esr and longer life (but don't ask me to give you those documents as I don't have them and don't even know what document would that be). Anyway, these were old documents from p4 times, newer processors have different requirements, use less power etc.


--


Also, would it be a better choice to use a polymer as a replacement for these capacitors (reliability, lifespan etc)[/QUOTE]

yes, it would help but for that motherboard it makes no economic sense. They're too expensive. Those Panasonic FR capacitors have 4000-10000h @ 105c so they'll last for years without any problems on such motherboard.

[socketa]

Re: Replacing OST caps on Microstar KT6 Delta motherboard

What designates that capacitor (in the link that you provided) as a Polymer capacitor?
So polymer are a better choice, but cost quite a bit more.
Thankyou very much for the offer, but got enough motherboards etc now, to last a few decades, now that i'm becoming more confident that they can be resuscitated.

Could someone give a reply to post #7?
Is my reasoning along the right lines?