Anyone ever have to replace capacitors?

Message boards : Number crunching : Anyone ever have to replace capacitors?
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5

AuthorMessage
Profile skildude
Avatar

Send message
Joined: 4 Oct 00
Posts: 9541
Credit: 50,759,529
RAC: 60
Yemen
Message 1181377 - Posted: 29 Dec 2011, 16:41:43 UTC - in response to Message 1181374.  
Last modified: 29 Dec 2011, 16:43:59 UTC

L3 cache really doesn't help all that much for crunching though.

I thought it was interesting that a couple Mobo manufacturers were claiming their AM3 boards could handle the AM3+ chips when the AM3+ chips have a different number of pins in a different arrangement from the AM3


In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope
ID: 1181377 · Report as offensive
tbret
Volunteer tester
Avatar

Send message
Joined: 28 May 99
Posts: 3380
Credit: 296,162,071
RAC: 40
United States
Message 1181381 - Posted: 29 Dec 2011, 16:53:45 UTC - in response to Message 1178777.  

I should have supersized the caps in the speaker crossover networks........



Nah. Those are probably electrolytics. What you should do is next time use some poly caps, preferably 250v+, but be aware that the speakers will probably get brighter. The cause of this is still under fierce debate, but the consensus, and one I don't concense-to, is that the difference is ESR, poly caps having almost none and electrolytics having up-to a significant fraction of an ohm.

The reason I think that's a false cause and effect is that you can't change this "brightness" with an L-pad adding back resistance; the octave over octave "voice" is changed permanently, but then it changes when you use non-identical electrolytics, too, and you can change it again by adding a small "by-pass capacitor" on the tweeters. (Can eliminate "grain" in the top-end.)

I participated in a year plus long study of these things and we did find an effect I'm not free to disclose using both homemade bridges and HP lab equipment (with EE grad students involved at a university you would recognize). Because we could not connect the dots between measurements and subjective observation, and because we lacked a rather specialized lab to quantify the observed results, the PhD in question would not publish what we did find.

Ultimately the "team" concluded that the results would mean nothing to manufacturers and that most speaker listeners would never know the difference since the room has more bearing on the ultimate sound than does the cap and almost nobody would ever have the chance to A/B electrolyics vs poly.

In fact, we found different production runs of the same equipment had as much ESR variation (design changes using available parts on hand; paralleling caps for larger values, as one for instance among many) as echanging a NP electrolytic for something else. It was observed that dumping ESR by using many parallel caps did NOT produce the same result as using a poly capacitor.

We only observed this difference when the capacitor was part of a filter that carried audio at a significant voltage. Line level in-circuit designs (like a preamp) weren't tested for the same thing because the differences, if any, were not generally observed.

In fact, we found several things that should have been identical, really weren't. But, again, the conclusion was that nobody would care because changes were too subtle to detect except under very tight A/B comparisons. The effects which were easily swamped by room acoustics or slight variations in manufacturing / materials were not observable in an either / or comparison given 10 minutes to "forget" the comparitive differences, but were once again easily detected if a new A/B was performed.

Conclusion: you listen with your brain, not your ears and your brain forgets tiny nuances (or compensates).

I believe the observable effect has to do with the lobing of in and out of phase areas of sounds in the crossover region which would be affected by the "time to charge" different frequencies which changes not only with the materials used but the physical size of a capacitor. I cannot prove that, but it is about all that was left to blame. (Phase has been repeatedly shown not to matter; i'm talking about areas of destructive interference in the crossover region - this much more obserable in a three way or four way system than a two way and not observable at all using a single driver as reference. I am NOT referring to timealignment.)-
ID: 1181381 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1181547 - Posted: 30 Dec 2011, 5:17:26 UTC - in response to Message 1181374.  

::random angry shouting:: Newegg deactivated the 1100T in favor of FX-6100. I was so close to having the money to order the new parts, and they're killing off Thuban. Guess it's not really all that terrible. Same clock speed, more L2 and L3, lower wattage.. oh, and lower price.

After doing some quick research, it looks like they really are just about identical in performance, too. Some tests Thuban has a slight lead, others Zambezi has a slight lead. Guess I'm going with the future on this one.


Of course you have to change your mainboard as well.

How so? The one I picked is an AM3+ board, and has all of the new FX-xxxx CPUs in the support list. It is bulldozer-ready since the first (and only as of present) BIOS build for it.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1181547 · Report as offensive
Profile arkayn
Volunteer tester
Avatar

Send message
Joined: 14 May 99
Posts: 4438
Credit: 55,006,323
RAC: 0
United States
Message 1181548 - Posted: 30 Dec 2011, 5:36:03 UTC - in response to Message 1181547.  

::random angry shouting:: Newegg deactivated the 1100T in favor of FX-6100. I was so close to having the money to order the new parts, and they're killing off Thuban. Guess it's not really all that terrible. Same clock speed, more L2 and L3, lower wattage.. oh, and lower price.

After doing some quick research, it looks like they really are just about identical in performance, too. Some tests Thuban has a slight lead, others Zambezi has a slight lead. Guess I'm going with the future on this one.


Of course you have to change your mainboard as well.

How so? The one I picked is an AM3+ board, and has all of the new FX-xxxx CPUs in the support list. It is bulldozer-ready since the first (and only as of present) BIOS build for it.


Forgot about what board you ordered.

ID: 1181548 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1188713 - Posted: 27 Jan 2012, 1:30:55 UTC

So I have another update about this situation. I got my new parts last week and the install went fine and everything but then the sound started acting up, so I RMA'ed the board and am waiting for the replacement to show up. In the meantime, I put the old rig back together and am using it in the meantime. I did notice when I was taking the old rig apart that there was one more cap that I hadn't seen before that was bulged. It was right next to the 8-pin CPU power connector. Of course the giant 3300µF cap next to the second RAM bank was yet again bulged and even oozing this time.

I decided before putting the old rig back together to go ahead and replace those two caps. Fortunately, I still had an 830µF cap left over from the first batch of them I did on this board, but since I started taking old mobos apart and taking all of the useful stuff off of them (caps, fan headers, rear panel connectors, attempted an SD-RAM slot..didn't work, ATX connectors, battery sockets, etc), I had a dozen or more 1500µF caps, which is what the big one next to the RAM was originally. Granted, it was off of a 10-year-old board, but it wasn't bulged, and it wasn't oozing, so chances are it is still reasonably reliable.

Replaced those two caps, put it all back together and gave a file transfer to the array a try. 4GB copied just fine. Tried an 8GB transfer and it finished successfully, but when I went to select those three files and delete them, the system didn't respond to that action. Ended up hitting the reset button.

So I figured the caps weren't the problem and the board is just getting old and tired. Understandable. Well on the new board, I noticed the voltages were really low. 12v was reading 11.4, 5v was reading 4.1-3, 3.3v was reading 2.9. I figured maybe it just isn't reporting properly. Well after being back on the old setup for three days, I decided to pull up Everest and see what it reports for the voltages. Same thing. Okay, so PSU is just getting weak. Swapped it out with the one that was in the single-core machine and 12v now reads 11.98, 5v reads 5.03, and 3.3v reads 3.37.

Booted back up, copied 4GB to the array. Was fine. Copied 12GB, was fine. Deleted those four files and copied them again. Fine. Did it one more time just to make sure. Fine. So I guess with a load, you get voltage drop, and since the voltage was already low, that makes the amperage go up, which will make voltage drop even more under a load. Oh, and the high-pitch capacitor squeal/whine that was present.. that's gone now. So that old 600-watt PSU is about to go. It does have over-current protection though, so when it goes, it shouldn't take anything else out with it, but that's also why it's in the single-core machine now. That machine isn't really that important to me, so if the whole thing melts down, oh well.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1188713 · Report as offensive
.clair.

Send message
Joined: 4 Nov 04
Posts: 1300
Credit: 55,390,408
RAC: 69
United Kingdom
Message 1188976 - Posted: 27 Jan 2012, 22:30:05 UTC

When salvaging components from boards that i will bin afterwards i firmly mount the board so i can heat it from below with a butane torch,
Just use wood screws or nails through the normal mounting holes into some 3x2 inch timber, and then work fast,
i dont care if i burn the board as it is going in the bin anyway.
heat the whole area around the parts i want off and just pick the parts off the top with tweezer`s or pliers.
with this method it is possible to remove PCI slots, north bridge`s, cpu socket`s the lot,
Give it a go on a junk board first and let us know how you get on :¬)
Expect smoke and flame from underneath,
good ventilation is required,
don`t set fire to the house,
In fact, dont do it in the house :¬))
ID: 1188976 · Report as offensive
Profile Paul D Harris
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 1122
Credit: 33,600,005
RAC: 0
United States
Message 1189054 - Posted: 28 Jan 2012, 0:52:51 UTC - in response to Message 1188976.  

Won't that over heat or burn the salvaged parts?
ID: 1189054 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13398
Credit: 208,696,464
RAC: 304
Australia
Message 1189087 - Posted: 28 Jan 2012, 2:45:35 UTC - in response to Message 1189054.  
Last modified: 28 Jan 2012, 2:45:57 UTC

Depends on how you do it- a couple of compents at a time won't be a problem.
Cook the whole board in one go will kill many.
A good way to remove sockets etc. Not so good for semi conductors.

Soldering is about providing enough heat to melt the solder & have it attatch the the component & board correctly. For that the board & component also have to be at the right temperature. Do things too quickly, the component & the board won't be hot enough. Take too long & things cook.
Grant
Darwin NT
ID: 1189087 · Report as offensive
.clair.

Send message
Joined: 4 Nov 04
Posts: 1300
Credit: 55,390,408
RAC: 69
United Kingdom
Message 1189110 - Posted: 28 Jan 2012, 4:01:41 UTC

It is a matter of timeing and experience to get the amount of heat correct,
which is why i say to practice on a junk board first.
and what i forgot to say is dont turn the flame up anywhere near max,
Probably only about 10% of what it could do and only the tip of the flame just touch`s the board to give it a hot flash,
and keep hold of the component while heating so to get it off as soon as possible,
I took the PCI slot`s off just to see if it was possible,
and expect to melt the plastic on two out of three and so render them useless,
But it is possible, not that i have any use for them,

If i am correct when boards are built, a sticky solder paste is first put onto the copper pads where all the components are to be fitted and then the components are stuck to the paste by machine/robot then the whole board is heated from below to melt the paste/glue, my idea is to reverse the process,
It is far from perfect but if you have ever tried to get something like, for example, a PCI slot off a board with `normal` methods like a solder sucker and braid you will likely know it is almost impossible,
As far as some of the larger chips are concerned (northbridge etc) that have no visible connection to the board because all of the connections are underneath the chip, there is no way that conventional methods can work.
ID: 1189110 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1189288 - Posted: 28 Jan 2012, 6:22:25 UTC

Yeah, when I was removing stuff from the old boards, I was using a propane torch and it was turned down just about to off. Needed heat but not a full-on inferno. Worked pretty well. Fan headers came off pretty easily, as well as things like USB headers (the 2x5-pin ones) and so on. I got a RAM slot about halfway off, but half of the pins were staying put and just being pulled out of the plastic. Got a couple of SATA headers though.

I was mostly after the ATX power connectors. the first one I pulled, I took all the pins out of it except for the two that you use for hot-wiring a PSU, and soldered a jumper lead between the two. Now I can just plug my hot-wire connector into any PSU and it turns on. No more shoving jumper wires into the harness.

It's a maybe stupid idea.. but I want to try to see if I can build my own HyperDrive5. I have a feeling a DIY version can be significantly cheaper.. it's just a matter of figuring out what all is needed and then getting the components. Of course since I have tons of it and it's basically worthless these days, I'm going to start with PC-100 32 or 64mb modules. I've got like 100 of them, so if I fry a few, oh well.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1189288 · Report as offensive
Profile Paul D Harris
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 1122
Credit: 33,600,005
RAC: 0
United States
Message 1189382 - Posted: 28 Jan 2012, 11:39:35 UTC - in response to Message 1189110.  
Last modified: 28 Jan 2012, 11:40:31 UTC

It is a matter of timeing and experience to get the amount of heat correct,
which is why i say to practice on a junk board first.
and what i forgot to say is dont turn the flame up anywhere near max,
Probably only about 10% of what it could do and only the tip of the flame just touch`s the board to give it a hot flash,
and keep hold of the component while heating so to get it off as soon as possible,
I took the PCI slot`s off just to see if it was possible,
and expect to melt the plastic on two out of three and so render them useless,
But it is possible, not that i have any use for them,

If i am correct when boards are built, a sticky solder paste is first put onto the copper pads where all the components are to be fitted and then the components are stuck to the paste by machine/robot then the whole board is heated from below to melt the paste/glue, my idea is to reverse the process,
It is far from perfect but if you have ever tried to get something like, for example, a PCI slot off a board with `normal` methods like a solder sucker and braid you will likely know it is almost impossible,
As far as some of the larger chips are concerned (northbridge etc) that have no visible connection to the board because all of the connections are underneath the chip, there is no way that conventional methods can work.


It seems reasonable for large through the hole components but what about all the surface mounted components since about 99% of boards are surface mount parts.
ID: 1189382 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1189385 - Posted: 28 Jan 2012, 11:57:02 UTC

99% of SMDs are useless and can be bought in quantities of hundreds for just a few dollars. It's not worth it at all to try to remove those from boards. Likewise, most of the larger through-hole components can also be acquired for really cheap, especially in bulk.

Removing them from a board serves a few purposes: the entertaining nature of general destruction, the use of fire, the challenge of seeing how much you can remove and how little of it is destroyed in the process, and lastly.. general nerd/geek activity.

There really isn't any reason at all to remove anything from any board.. we do it because it's fun. At least I do anyway.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1189385 · Report as offensive
Profile Paul D Harris
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 1122
Credit: 33,600,005
RAC: 0
United States
Message 1189389 - Posted: 28 Jan 2012, 12:14:50 UTC - in response to Message 1189385.  

I have salvaged parts for many reasons like hard to find and not near parts store and emergency repair and some times I am broke especially now since I am retired and money is not easily acquired now than before when I was working and money was not a big deal. I could salvage surface mount parts if I had too since I had built with them before and they are easier than the through the hole parts they are just so tiny.
ID: 1189389 · Report as offensive
.clair.

Send message
Joined: 4 Nov 04
Posts: 1300
Credit: 55,390,408
RAC: 69
United Kingdom
Message 1189399 - Posted: 28 Jan 2012, 13:27:16 UTC

Mostly i do this kind of `nutcase` activity if i have several of the same board or components all with different faults and on a cure or kill basis where i have nothing to loose,
just start swapping things about till it works,

I do not like being part of the `throw away society` and however pointless it may be,
it gives me a good feeling to repair something that the manufacturers welded shut and landfill was its only fate.
And sometimes end up with a pice of kit that i could not afford just for the sake of swapping a component that costs less than a peanut :¬)
ID: 1189399 · Report as offensive
Profile Paul D Harris
Volunteer tester

Send message
Joined: 1 Dec 99
Posts: 1122
Credit: 33,600,005
RAC: 0
United States
Message 1189413 - Posted: 28 Jan 2012, 14:28:29 UTC - in response to Message 1189399.  

I also salvage whatever parts I can and put them in my parts bin before I trash them.
ID: 1189413 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1190055 - Posted: 30 Jan 2012, 12:55:19 UTC

So.. interesting story. After my last post about low voltages being the cause of probably all of my problems, I got a PM from somebody telling me they also had low voltages and was wondering why they didn't have any errors like I did. I explained that it can be random, and if the voltages get too low, you'll either start getting freezes/hangs, or the system will just simply power off. 10 minutes later.. their system powered off. At least they had a PSU upgrade on stand-by.

So I swapped my failing PSU into the single-core machine and the voltages read correctly, but there was the unmistakable squeal/whine from one of the caps inside. Was only a matter of time. Well the single-core machine has been fine for three days with it, and I just tried remote-desktoping to it.. nothing. I figured it was off. Nope, it was still on and running, but frozen. Dug around in my pile of PSUs and grabbed another one from the carcass pile and that's back up and running now.

I was looking through the back of the old PSU, and yep.. there it is: a large cap that is bulged. And... just checked.. yep, only a 1-year warranty and I bought it a little over 5 years ago. Time to open it up and see if it's one that I can replace.. or just take most of the pieces out. I really like the harness on it (Rosewill RD600-2DB-SL).

So far, the 550 that I'm using now is doing fine. No squealing, and the voltages are spot on. I've been playing games and making my array do some hard work the past two days and it's all rock-solid, so I know I'll have a few months before needing to get a new PSU.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1190055 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1190820 - Posted: 2 Feb 2012, 0:43:22 UTC

So far, so good. My first AP went pretty quick, but it turns out it had 30 repetitive pulses, so its duration was shortened. It looks like the normal time for APs on this setup is going to be somewhere in the mid 15-hour range.

I did have an issue with that whole "no heartbeat from core client for 30 seconds..exiting" issue, but I figured out it was because of vista/7's automatic daily defrag. Disabled that, but I keep getting them.

System logs show 'amdsata' is issuing a reset to \Device\RaidPort0 (the OS drive) every couple of hours, and has to issue the reset three times, which takes 60 seconds each time. Not sure what that's all about. I don't have RAID enabled. The onboard controller is set to AHCI single-disk mode.

It is similar to an issue I had with the previous setup. I couldn't have my DVD burner plugged into the onboard SATA controller because the system would lock up due to a supposed "failed parity check" for a DVD burner. Plugged it into a Siicon Image 3132 controller and all was well. I don't think this is the same situation though.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1190820 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5

Message boards : Number crunching : Anyone ever have to replace capacitors?


 
©2022 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.