Has anyone done a total cap replacement on a board ?

[Vario]

I don't understand how this thread devolved into epeen comparison. Not sure why anyone cares about this? These chips all overclocked differently, there was a lot of variablity. I owned six 939 processors, several venice, san diego, and an opteron. They all clocked differently. The best overclocker had the worst stepping by reputation (opteron CABGE). Bragging about silicon lottery from 12 years ago is a bit much.

How about getting back to the OP @natr0n's recapping? Are you going to recap your board?

As far as recap vs buy another board, all of these boards had defective capacitors. If your board is otherwise good, recap it. The capacitors used on these boards will leak or bulge, it is only a matter of time. If it is a system you care about you should recap it.

https://en.wikipedia.org/wiki/Capacitor_plague

 

[delshay]

Oh really?
How did I do that?

EDIT:
Speaking of benchmarking, it isn't a thread for Prime95 either yet you made reference to that so my post is fair enough, also demonstrates the chip and board are capable of what I could have said but didn't have to, the result says all that needs to be said about it.

All I've done is to show and prove it and that seems to offend you - Don't want to show any proof of your claims on your part?
Get over it.

Yes, cap modding CAN help but it can only make up for so much.

One more thing:
This is getting off topic as you were so quick to point out after the proof I posted was done so..... This is the last reply to this thread from me period since I know you'll try and drag this on but that would make it even further off topic and no need for that.

Everything I am posting is related to this thread, including benchmarks, as it part of a check after most of the motherboard has been changed over to a new type of capacitor.

Somekind of checks must be done, otherwise how do you know what you have done don't have side effects. You are posting something that's not related to capacitor change over.

If you have finished posting in this thread, then good, you may want to take your sidekick with you.

Just for the record, these are part of a small numbers of users that gang-up on me on OCN.

 

[lexluthermiester]

Gonna have to agree with Vario. Enough with the drama. The OP has very good answers, info and perspectives to consider about cap replacement, and even an offer of actual parts for the cost of shipping. Whether or not the board in question is worth the effort is up to no one but the owner of said board. They didn't ask if we thought it was or wasn't worth such an effort, only if it was possible.

 

[natr0n]

The board is an asus a8n-la. This is a prebuilt hp/compaq. I have used a similar chipset in a retail board in the past for a friend which was memorable, so that got me thinking can I fix it.
You know because some of us actually like to fix things to make them functional.That's the idea of this thread. It's not about "I buy/or am given the best motherboards/parts and I look down on you because my time is valuable"

I appreciate all the actual help posted.

 

[Mr.Scott]

Just gonna throw this out there.
It's a 30 piece cap kit for that board. Even at a buck a cap, not including your time, that's 30 bucks.
Those boards go for 25 bucks all day long on ebay.
Just replace the board.

FWIW, I DO recap boards...........if it's worth it.

 

[delshay]

@OP

Because you are doing a number of motherboards, your best friend is going to be a ESR Metre. Part of my changeover, capacitors were remove from dead very expensive motherboards. You can use find capacitors from other dead equipment, just insure their are the same value if you stay with Electrolytic. If you are going polymer you would have found out capacity has changed, like my motherboard has.

I also bought new capacitors from a well known dealer (no fake) & out of a batch of capacitors, a small number of capacitors had very poor reading, but the numbers are very small, this is why I double checked with two ESR Metre. So you can buy bad capacitors even if their are new.

Have fun changing it over, other than buying a more expensive motherboard & messing it up. Practice first on those motherboards you have.
The hardest part is removing the old solder from the holes, but you will soon find smart ways around this.

If you are going polymer, do the VRM first then test from there. Do a number of test including overclocking. If there is a problem sort this out first, don't do changes to other parts of the motherboard without fixing this first, unless you have to change a bad capacitor on other parts of the motherboard.

EDIT: Capacitors removed from dead/old equipment must be tested with a ESR Metre.

Have fun & enjoy.

 

[delshay]

Voltage, farad and polarity. As long as they are matched up, electrolytic's can easily be replaced with solid-state.

UPDATE CORRECTION:

AFAIK the only capacitor that remains the same when changing from Electrolytic to polymer are the 16v 100uf. I have only changed over four of eight (direct swap), as I ran out of this value & a few others, this is why there is a delay in full completion of changeover.

So you were right to some degree.

I have just discovered the thanks icon, I am new to this site, need to find my way around.

 

[delshay]

Motherboard is now 100% Polymer & completed. I will be changing capacitor around chipset again from 4v to 2.5v. The following screenshot was taken before the last 9 capacitor was inserted, It's the motherboard Northbridge IGP overclocked.

 

[delshay]

EDIT: I have found a problem.

Motherboard HDMI output is intermittent. It only applies when you do a reset where it will not display anything on screen. If & when I find what's wrong, I will post location of the fix. GFX card is not effected & always show output.

The good news is image on screen seems clearer, with either IGP or GFX card, both seem to show a sharper image on screen.

 

[jaggerwild]

The board is an asus a8n-la. This is a prebuilt hp/compaq. I have used a similar chipset in a retail board in the past for a friend which was memorable, so that got me thinking can I fix it.
You know because some of us actually like to fix things to make them functional.That's the idea of this thread. It's not about "I buy/or am given the best motherboards/parts and I look down on you because my time is valuable"

I appreciate all the actual help posted.

Ive done a board, can't recall what it was but I got it cheap off Flee Bay. I knew some one would buy it so I bought it recapped it(no CPU to test it with just made sure my caps where properly polarised). Then relisted it on Flee bay for a profit, like used cars(there is an ass for every seat)someone bought it to. The caps where advertised as bad and were not the new style, so it was easy to find n buy new ones.
On a separate topic I saw your micro scope on you tube, must say very impressive!!!

 

[delshay]

After capacitor changes I then found this thread http://www.silentpcreview.com/forums/viewtopic.php?t=41883

FX60 CPU 1.2v set in the BIOS, gain of 0.25mv lower voltage with capacitor changeover.

Below is a screenshot with prime95 running. Note my idle temperature, it has always been this low. This is due to my modified IHS/Heatsink which I already posted info in this thread.

1c improvement over Liquid Metal Compound, but it never needs thermal paste/compound between IHS/Heatsink, it's become obsolete here.

CPU die & IHS is next to be fixed.

Big update on it's way hopefully with a fix for the motherboard HDMI.

 

[delshay]

QUICK UPDATE: HDMI port "is working correctly"

PROBLEM FOUND: The problem was the main memory DIMMs. It's ultra sensitive to DIMM errors when overclocking even thou I have selected the onboard sideband memory & not shared memory. Polymer Capacitors has changed the sensitivity in this area.

If you are doing an all Polymer change-over you may not encounter this problem if you don't have an on-board GFX chip. Your overclocking will be reduced if you are hitting the smallest of memory errors. Before I could overclock very high with a few memory errors & computer will always display an output, but now it can only do this via the 15 pin VGA output, where the HDMI port will flake out.

In short you can still push to the bitter end when overclocking "with memory errors", but you will not get a HDMI output from the motherboard.


As for overclocking the onboard GFX chip, it can set a world record on HWBot, I just did not push-it further, as it did not have it custom heatsink fitted when testing.

New parts arrive early next month. It has been on order from last year December, so their brand new 2018 parts.

OFF TOPIC

Just to make you aware, the CPU (FX-60) also has extra capacitors. The unused pads on the CPU is filled-in with ceramic capacitors. The latest modification to the CPU has just failed. It's supposed to have an extra 54 Tantalum Capacitors (0603). I bought the wrong height (1.00mm). It needs to be below 1.00mm to prevent it from coming into contact with the IHS. So I am now looking at 0.75 to 0.85mm to complete CPU penultimate update, awaiting another experimental final update that also needs to be squeeze under the IHS. ..It's going to be very busy under the IHS, nevertheless it must be done.

 

[delshay]

UPDATE: Capacitors have still not arrived, ordered last year December, so in the mean time here is screenshot of the motherboard with all the capacitor changes. Capacitors have different colours including gold.

OTHER NEW EXPERIMENTAL MODIFICATIONS EXCLUSIVE TO TPU (NEW).

New is VRM cooling. This is the worlds first motherboard to have what I call DFC, "DIRECT FET COOLING" & is exclusive to this 939 motherboard.
It has the highest heat transfer rate of all motherboards made, thanks to it's heatsink being directly soldered to the MOSFET. This does not improve overclocking, but improves stability, all the way to max voltage/CPU speed.

With the heatsink being so small on the MOSFET, it has support on the back for further cooling with it's copper bars which is not utilized yet. There are two options available to send the VRM to an even lower temperature.

WARNING: Do not copy what I have done here, as the heatsinks are "LIVE", this is why MOS-C1-LE are used.

COPPER BARS.

Those with a keen eye will see that this 939 motherboard also has a 8 pin CPU power socket. This also uses copper bars, but its completely hidden out of view, but I can assure you it is there. You can plug in a 4 pin power lead into the left or right socket or use the full 8 pin.

Screenshot is from test bench & not in it's regular case this is why you are see a 20 pin power cord & not a full 24 pin power cord.

 

[lexluthermiester]

UPDATE: Capacitors have still not arrived, ordered last year December

Wow. If it takes longer than 14 days you should always file a claim for refund.

 

[delshay]

UPDATE:

Capacitors have turned up, but I have decided not to fit them to this motherboard, don't want to break it swapping it out again.

New 2.5v capacitors has been fitted to another motherboard, but it's the same type of motherboard as this one, & has been fitted around the northbridge. The motherboard in this thread has 4v capacitors around the northbridge.
What's also changed for this motherboard also is the 3.3v rail. This now has 4v capacitors. This goes against what I said in this thread, but 4v capacitors on the 3.3v rail is fine.

OTHER CHANGES:

Some of CPU capacitors are to be changed, & will move from 2.5v capacitors to straight 2v for both motherboard.

DESOLDERING:

This is not a tip, but my method of getting the solder out of the hole. My soldering iron takes mulit-tips which I can change on-a-fly. I use a sharp pointed tip which goes partly though the hole. This tip is used with a solder mop.

I sometimes desolder with the motherboard in the air with the soldering iron from below. This way the solder is flowing towards the iron.
Be aware you can also do this on the top side of the motherboard which is better, as you will do less damage to the pads (if any) to the pads on the backside of the motherboard. It's important not to damage the pads on the backside of the motherboard, as these are the pads you are going to solder to when new capacitors are replaced.

So in most cases I try to remove what solder I can from the hole from the top side of the motherboard first, then move to the back.
I tend to also coat the solder mop in flux also, as well as the motherboard which sometimes removes the solder from the hole at first attempt.

Another thing to note here, is watch temperature as you solder in the new capacitors. You don't want to overheat the new capacitors.

OTHER MODIFICATION:

Soldered IHS to heatsink has improved. A flaw in the soldering process has been discovered. So this has improved & I have doubled the mounting pressure during soldering process. This has put a Dark Rock heatsink right up there with one of Noctua top-of-the-line heatsink & everything is pointing towards it will outperform it (not confirmed).

 

[delshay]

Just a quick update.

I'v started to change the capacitors on the second motherboard, not much will change when compared to the motherboard in this thread, but I wanted to point out what has changed to the motherboard performance in this thread.

On the first motherboard in this thread with it's heatsink soldered to the MOSFET, @Max voltage with heavy load the motherboard no longer shuts-down. This is new, because it will normally shut down under heavy load due to overheated VRM .

But what has also changed is the CPU fans can now be set at it's minimum fan speed 150RPM. What's different here is @3Ghz with 1.4v, when the CPU goes idle & hits somewhere between 29-31C the CPU fans come to a complete stop for about 5-8 seconds. This is something i'v only seen when I did a experiment with Noctua 24v fans, but never seen the standard Noctua fans do this. The target temperature of CPU is 45c, but with this going on, the VRM is not overheating or shutting down the motherboard. This is impressive, because it's never done this before.

I also forgot to say the heatsink on the MOSFET are always very hot "all the time" with 1.35v+ underload. This is because the maximum heat is always being transferred to the heatsink, which are directly soldered to the MOSFETS.

What I also failed to say in this thread, the soldered "IHS to heatsink" has also changed. The IHS has been lapped in such a way to reduce it's height.
I have no real results on performance yet, but it's looking very impressive. I'm currently working on version two, where the IHS is lapped to more than three quarters, so it's ultra thin. I will have results when I can compare two lapped IHS with different heights.

If I have broken the rules by going a little bit off topic, then MODS can delete just this posting, but not close the thread. It's going to be interesting to know, does the reduce height in IHS improve temperature.

This also opens a question. Those users who lapped their IHS & gained performance, did their gain performance by the IHS being flat, or was is gained by the reduce height of the IHS.

A flaw was also detected during lapping IHS. This flaw can be detected by putting spirit level on a lapped IHS & rotate the IHS but keep the spirt level in the same position. Some of you may see this error, but don't take your computer apart to check it, it only affects me as my IHS are soldered to heatsink.