Testing VRMs

[momaka]

Re: Testing VRMs

BTW, without the real diagram that matches the board you are working on then you really do not know what point 'A' is really connected to.

It's usually not that complicated for a *desktop* PC motherboard.

Typically the CPU VRM powers the... CPU (surprise, surprise ) and occasionally the NB on a few motherboards. The fact that bianchi77 got a low-resistance after removing the CPU strongly suggests that the NB is powered from the CPU VRM as well. I'm not saying that it can't be shorted ceramic capacitors, but it seems unlikely to me in this case, since the motherboard does turn ON for a few seconds. If any shorted ceramic capacitor was to take that much energy from the CPU VRM for a few seconds, it would have burned for sure.

I didn't see any thick traces going from CPU V_core to NB....
I saw northbridge has its own FET....

Yes, the NB does have its own MOSFET, but your motherboard doesn't have a separate SB, so that means the NB and SB are both in one chip... hence you could have 2 or even 3 different voltage rails going to that chip.

On a typical motherboard with separate NB and SB, the SB usually has two rails: a stand-by rail (usually 3.3V or less) and an active-ON rail that comes ON only when the motherboard is turned ON (and it's usually 1.2-2.5V, depending on motherboard's vintage).

[bianchi77]

Re: Testing VRMs

I got this one :
https://cdn.badcaps-static.com/pdfs/...3a1aafd023.pdf

[bianchi77]

Re: Testing VRMs

BTW, without the real diagram that matches the board you are working on then you really do not know what point 'A' is really connected to.

Where can we find Asus A7S333 motherboard schematic ? In there, we can see where are those capacitors located,
My thinking, they are V_core filter capacitor in each Vcc pins on CPU....

[budm]

Re: Testing VRMs

BTW, without the real diagram that matches the board you are working on then you really do not know what point 'A' is really connected to.

[budm]

Re: Testing VRMs

One info that is still missing is the Output Voltage reading from the power supply. i myself never seen the Vcore pin of the CPU with that <10 Ohms.
Quote:
Originally Posted by bianchi77 View Post
When I tried to measure one of the chip capacitor with my LCR45....I got value >20000uF
That's because of the low resistance of the circuit. It will make your meter think it is filling up a big cap.

That's why it's never reliable to measure anything in circuit, especially if you don't understand the circuit well.

I was going to tell him to get something like 100uF cap in parallel with 100 Ohms resistor and have him test the capacitance to see what reading he will get to proof that reading cap in the circuit is never be reliable, but some people still insist on doing that.

[bianchi77]

Re: Testing VRMs

Any thick traces going from CPU V_core to Northbridge? If yes, that means the NB is powered by CPU V_core as well, and that's what would be giving you the still-fairly-low resistance of 34.6 Ohms.

I didn't see any thick traces going from CPU V_core to NB....
I saw northbridge has its own FET....

so CPU is using V_core itself....

[momaka]

Re: Testing VRMs

I removed CPU resistance to ground = 34.5 ohm...do you mean northbridge sharing the same Vcore ?

Yes. *Possibly*

Is this for the A7S333 motherboard from your other thread? Or a different motherboard?

When I tried to measure one of the chip capacitor with my LCR45....I got value >20000uF

That's because of the low resistance of the circuit. It will make your meter think it is filling up a big cap.

That's why it's never reliable to measure anything in circuit, especially if you don't understand the circuit well.

One test that I can do...is, unsolder all the capacitors about 16 small smd capacitors and test the resistance again....

Yes, do that.
This is by far the best way to isolate a short-circuit: keep removing parts until the short-circuit is gone. Unfortunately, this is also the hardest way. But the reward is the certainty you get.

[momaka]

Re: Testing VRMs

Is it possible that one of the ceramic caps near the CPU is short ? ( on the bottom of the CPU, inside the socket )

Possibly, although I have seen only a few short out with such a high resistance (34.6 Ohms without CPU in socket). Most go short-circuit like a piece of "wire" (i.e. sub-1 Ohm)

without CPU in socket, 34.6 ohm...

Any thick traces going from CPU V_core to Northbridge? If yes, that means the NB is powered by CPU V_core as well, and that's what would be giving you the still-fairly-low resistance of 34.6 Ohms.

I have removed all the output capacitors now....I got only 8 ohms between Vcore and GND...CPU in socket...

If you are trying to find a short-circuit on CPU VRM, *ALWAYS* remove the CPU from the socket. Otherwise the CPU will give you a low resistance reading and you will never know where the short-circuit is coming from.

Something does not make sense here, you had 9 Ohms before the output caps and MOSEFTs are removed, and now you have 7.7 Ohms, and 8 ohms?

Could be his multimeter, if he is using a cheap one. I know my Harbor Freight meters will sometimes show 1 Ohm with shorted leads and other times up to 3 Ohms, depending if I don't play with the rotary dial.

That's why I never take it for granted that anyone here is using perfect test equipment. Always watching out for "weird" and "off" measurements.

[bianchi77]

Re: Testing VRMs

I removed CPU resistance to ground = 34.5 ohm...do you mean northbridge sharing the same Vcore ?

[momaka]

Re: Testing VRMs

It should be very high in the K Ohms range because the devices (Loads) are not active.

In my experience, CPUs and GPUs always measure low resistance, typically under 10 Ohms. So the reading bianchi77 got on the CPU output isn't indicative of anything wrong at all. Now, if he removes the CPU he *may* get a high resistance reading of a few hundred Ohms to a few KOhms, *IF* the Northbridge doesn't use CPU V_core for its power (which is the case only with only *a few* motherboards).

[bianchi77]

Re: Testing VRMs

I've tested just now without NCP5322 and CPU on it....it's running stable not suddenly off...all FETs,Caps and inductors on it....
I'm waiting for my NCP5322 coming now....hopefully it will answer my question

[bianchi77]

Re: Testing VRMs

NCP5322 unsoldered...

[bianchi77]

Re: Testing VRMs

I'm thinking for unsoldering NCP5322 and see what's the response without that chip...
another test...

[bianchi77]

Re: Testing VRMs

I measured from other working motherboard...even lower from Vcore to ground = 2.1 ohm with CPU....

I have a doubt if it's ceramic caps problem....I suspect I have problem with my buck controller....it's generating voltage more than the CPU needs and shut down...

[bianchi77]

Re: Testing VRMs

34.5 ohm photo....without cpu 34.5 ~ 35 ohm....don't worry it's nearly the same...

[bianchi77]

Re: Testing VRMs

From 7 caps I have taken off....value = 5uF and resistance is on level M ohm until OL...
which normal....character of charging of capacitor...I have another 16 to 18 caps to analyze now...hmm...abit fiddly...small on the bottom of the CPU....

[budm]

Re: Testing VRMs

BTW, do you other mother board that you can check the resistance of the Vcore and GND?

[budm]

Re: Testing VRMs

cearamic capacitor never give me a very low resistance: when they fail they will short out or have very low resistance. MLCC cracked easily.

[budm]

Re: Testing VRMs

BTW, you never tell us what dcv you were getting at the output of that power supply.

[bianchi77]

Re: Testing VRMs

yes 35 ohm without CPU.....now 35 ohm only from those small capacitors below the CPU...
do you reckon one or some of them are faulty ? because from my experience...ceramic capacitor never give me a very low resistance....

Have you got one before with SMD MLCC faulty ? overvoltage and overheat from the CPU I reckon ....