Badcaps.net Forum
Go Back   Badcaps Forums > Troubleshooting Hardware & Devices and Electronics Theory > Troubleshooting Power Supplies and Power Supply Design
Register FAQ Calendar Search Today's Posts Mark Forums Read

 
Thread Tools Display Modes
Old 02-21-2014, 04:03 AM   #1
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Antec Earthwatts EA-430D's OSTs killed my Conroe?

I ran my heavily featured Foxconn P9657AB-8EKRS2H for several years with a Cedar Mill SK9KE Pentium® 4 651 2M Cache, 3.40 GHz, 800 MHz FSB that I upgraded 19 months ago with an eBay SL9ZF Conroe Core™2 Duo E6700 4M Cache, 2.66 GHz, 1066 MHz FSB. I don't remember what PSU I started out with, but when it was young I bought my first ever ATX Antecs, EA-380 and EA-430D models. Later I tried what I expected to be an EA-500 that turned out to be EA-500D. After several swaps with Newegg it gave me my money back, and I left the EA-430D in the P9657AB.

One AM last month at about 04:00 I was awakened to the beeps of 7 UPSs. After determining that the P9657AB's UPS battery wouldn't likely survive the outage, I shut it down. Since then, until today, I've never been able to get it to run more than 10 minutes or so before freezing, except if all it was doing was running memtest86+, which it would do for at least 6 hours without errors no matter what RAM I tried.

Soon after realizing something had gone wrong, what seemed to be on account of the forced shutdown, I opened up the EA-430D to find two slightly swollen OSTs, 3300uf/16v & 1000uf/16v. I replaced them with what I had on hand, Nichicon UVZ1C332MHD and United Chemi-Con ELXZ160ELL102MJ20S. No improvement. I tried a couple of other power supplies, including an EVGA 100-W1-0430-KR for $35 with $10 rebate available that I had received only about a week earlier, but still no improvement.

I also tried via BIOS raising various voltages without apparent effect. As I've had several disruptive experiences with LGA775 cooler attachments, during cap replacement I purposely left the Freezer Pro7 cooler installed. Only when done soldering while in the process of cleaning residual flux, the white component of two of the plastic cooler retention feet completely failed. I was able to replace them by conscription from an OEM Intel cooler. The failure made me wonder if either or the pair had partially failed much sooner, leaving the contact reasonable on the other side, allowing overheat on the failed side, with whatever mechanism senses and warns of CPU overheat being fooled into satisfactory status, and overheat being the actual culprit for the lockups.

I eventually gave up doing anything more without some Googling first. About the P9657AB I found nothing except problems with POSTing, nothing about locking up. About Earthwatts I didn't even think to look. Eventually I did some more reading here and found the big red warning about KZG. Since the P9657AB had four as shown in that post's photo, I replaced them all with Nichicon, 3 UHM1C152MPD purchased 2008/07, 1 UHV1C152MPD just received. No improvement.

Today I finally got around to trying the P9657AB's original Cedar Mill. It ran fine for over 7 hours, 4 hours of memtest, over an hour of just running top, later running zypper dup of openSUSE 13.2 to give it a bit of an I/O workout for half an hour or so. After rebooting the new kernel it ran top for over an hour before I shut it down.

As the Cedar Mill and the Conroe use different FSB, 800 vs. 1066, I proceeded to swap a Conroe from another system into the P9657AB and try the apparently damaged Conroe in that other system to confirm it's the Conroe that's failed and not the P9657AB. First try was another 965 system (Dell GX745), with another Conroe, a 20% slower E6400. Both systems behaved nicely for over an hour with the swapped CPUs. I then swapped them back. The Dell @2.13GHz is behaving normally. The P9657AB @2.67GHz again locked up, this time after 14 minutes of running top, plus whatever makes up the sleeping 134 tasks idling openSUSE 13.2. I have another P965 system with another E6700 I will try swapping with after getting some sleep.

I can't make out what most remaining caps are on the EA-430D without removing them. One for sure is a 220uf/16v OST. Another of yet another spec I can't read without removing it may also be OST, but none exhibit any apparent leaks or swelling. Should I replace all but the giant anyway before returning it to duty?

Any comments about the UHM1C152MPD mixed with UHV1C152MPD replacing the KZGs? The P9657AB has about 20 OK looking 1000uf/6.3v green Lelons. The only large electrolytics the board has besides those I replaced are a pair of Rubycon MCZ 3300uf/6.3v with no evidence of leaking or swelling. Any chance they might no longer be up to task with the faster of the two Conroes after having been subjected to the bad OSTs in the EA-430D?

Other comments about any of this situation?
Attached Images
File Type: jpg p9657ab-009.jpg (187.3 KB, 122 views)
File Type: jpg p9657ab-001.jpg (32.1 KB, 80 views)
File Type: jpg p9657ab-002.jpg (38.8 KB, 77 views)
mrmazda is offline   Reply With Quote
Old 03-03-2014, 10:56 PM   #2
momaka
master hoarder
 
momaka's Avatar
 
Join Date: May 2008
City & State: VA (NoVA)
My Country: U.S.A.
Line Voltage: 120 VAC, 60 Hz
I'm a: Hobbyist Tech
Posts: 10,860
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Most likely the CPU failed because of the overheating. While those OST caps were probably on the 12V rail (which is used for the CPU VRM to power the CPU), the CPU itself is usually well-protected by the VRM. So it's unlikely that those OST caps damaged it.

As for the caps choices - Nichicon VZ is not suitable for either PSU or motherboard repair. Nichicon HM and HN, on the other hand, are well-suited for motherboard repair. Not so much for power supplies, though. For power supplies, that Chemicon LXZ you used is a very fine choice. Panasonic FC, Chemicon KY, and Nichicon HE are also the preferred choices.
momaka is offline   Reply With Quote
Old 03-04-2014, 12:50 AM   #3
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Thanks momaka, all good to know.

It turns out my 2nd paragraph should have been extended, or a following paragraph added. On first attempt to boot after the shutdown, the BIOS reported out of the ordinary messages which at this point in time I cannot remember. The gist was that what was reported made me think the 2032 battery was too weak. I changed the battery, fixed the date and time, then thought consciously no more about it due to the frustration that seemed to be a hardware failure.

After more reflection I remembered that when the P9657AB was young it had had RAM trouble, in addition to Earthwatts trouble. I had originally bought CL4 Crucial Ballistix Tracer for it. Due to the trouble I was having, Crucial exchanged them twice. When the third set proved to be no better, I again contacted Crucial, which at my urging exchanged the Ballistix for a lower voltage CL6 RAM model that proved compatible. After that, all seemed rock solid for years, until January's shutdown.

While in BIOS setup in January before starting to suspect hardware trouble, I hadn't paid much attention to its overclocking section, labeled "FOX Intelligent Stepping". I left it on "Auto", same as I thought it had been from the start. After weeks of suspecting but not isolating hardware issues I took a closer look at this BIOS section and noticed that RAM voltage shown following the "Auto" section head was not set on +0.000v". Whether at that time it was showing +0.200v or +0.600v I can't recall, but the short story is that as of Friday before last it has been showing +0.600v, and the board, RAM and CPU as they were before the shutdown have since been solid performers again. Apparently on my first passes through the BIOS in January, which included multiple instances of load optimal defaults and load failsafe defaults, and in first ensuing weeks, I must have been forcing through manual setting RAM voltage somewhere in the +0.000v to +0.200v range. This was obviously enough for the RAM to survive multiple Memtest passes with no errors, but apparently not for a normally running system.

My luck with cap replacement has been considerably less than stellar. Since spending time in the FAQ here I realize I've been lucky I haven't had less success, using 25W and 35W irons and a pin-vised wire drill that may have been crucifying vias, among other no-nos.

Now that all seems good, do I dare replace that one VZ for a HM or HN, or should I leave it as is until such time as any symptoms of malfunction develop?
mrmazda is offline   Reply With Quote
Old 03-04-2014, 03:43 AM   #4
c_hegge
Badcaps Veteran
 
c_hegge's Avatar
 
Join Date: Sep 2009
City & State: North Coast, NSW
My Country: Australia
Line Voltage: 240V 50Hz
I'm a: Professional Tech
Posts: 5,051
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

You sure that's a 430D Power supply? Those are made by Delta, who don't use OST. It's the original Seasonic built EA-430 (the non-D variant) that uses them.
__________________
I love putting bad caps and flat batteries in fire and watching them explode!!

No wonder it doesn't work! You installed the jumper wires backwards

Main PC: Core i7 3770K 3.5GHz, Gigabyte GA-Z77M-D3H-MVP, 8GB Kingston HyperX DDR3 1600, 240GB Intel 335 Series SSD, 750GB WD HDD, Sony Optiarc DVD RW, Palit nVidia GTX660 Ti, CoolerMaster N200 Case, Delta DPS-600MB 600W PSU, Hauppauge TV Tuner, Windows 7 Home Premium

Office PC: HP ProLiant ML150 G3, 2x Xeon E5335 2GHz, 4GB DDR2 RAM, 120GB Intel 530 SSD, 2x 250GB HDD, 2x 450GB 15K SAS HDD in RAID 1, 1x 2TB HDD, nVidia 8400GS, Delta DPS-650BB 650W PSU, Windows 7 Pro
c_hegge is offline   Reply With Quote
Old 03-04-2014, 05:46 AM   #5
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Bitten by fallible aging memory again. I thought I had only ever bought 2 EA-380, 1 EA-430D, and 1 EA-500D, exchanging the 500 twice before getting a refund. It turns out I bought 2 430, the second a D, and the OSTs came from the older non-D. The inside of the D looks good. Delta did a good job of orienting and bunching various size and color caps to hide their manufacturers. One other than the really big one out in the open is a Nichicon. I guess I probably need to open up the 2 380s for inspection.
mrmazda is offline   Reply With Quote
Old 03-08-2015, 06:43 PM   #6
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Events transpiring since last activity here have made me rethink the whole history. Since last, it ran stably with one of my two E6700 Conroes for about 10 months. After 6 months I bought a new CORSAIR CX series CX430 430W power supply to replace the recapped EA-430. The first clue to trouble came 4 months later, while I was upgrading the RAID from 320G HDs to 1TB HDs. The first OS installation, openSUSE 13.1, went without a hitch. I added two more, openSUSE 13.2 and Tumbleweed. That produced so much trouble that I filed an openSUSE bug.

A couple of weeks later I decided I was tired of having only 4G of RAM, so ordered 6 new 16-chip 2G PC2-6400 sticks. Trying any 4 of them produced lockups in 13.1 similar to those that were the reason for my bug report. Putting the original RAM back in failed to restore stability to 13.1.

As the motherboard is for this desktop/LAN server/web server, I swapped in a motherboard from one of my test/backup systems. That made the bug go away in all three OS, so I resolved the bug invalid.

The whole process made me think back to when the board was new, and I had Crucial replace the original (2.2V Ballistix Tracer 4-4-4-12) RAM 3 times before having it replaced with a different RAM model (1.8V CT25664AA800.I16F 6-6-6-18). I now think the board has been on the border of defective since it was new, or quickly became so due to power supply deficiency.

Its BIOS has various overclocking features, among which include:
Vcore voltage
RAM voltage
MCH voltage
CPU termination voltage
Speedstep enable/disable
TM enable/disable
C1E auto/disable

I remember long ago (since new IIRC) being lucky to simply get it to POST unless I raised the RAM voltage, so ever since I've kept it raised. I don't remember any more how high I ever tried it, but +0.2V comes to mind as where I've usually kept it.

Thursday, early in the day, I blasted it with compressed air (new pic is before cleaning), flushed it with 91% isopropyl alcohol, blowed it dry, and warmed it ~+30°F for about 3 hours. During the wait I found the 4 1500uF 16V UC KZGs originally installed in EC15, EC16, EC17 & EC18 (before I had an ESR tester), and measured them. uF was in the 1520-1545 range. ESR on one was 0.01, the other three, 0.00. In the process I found 4 used Sanyo WG 1500uF 16V and checked them. They measured uF from 1899 up to 2088, and ESR 0.00. I found no more HM like I installed in place of the KZGs, but my remaining Ruby MCZs, purchased here, like the one I put in EC18, measure 1470-1488 & 0.01-0.02.

In the evening I hooked it up on the bench, and have been running it that way ever since. The differences between installed configuration and bench condition, besides no case controlling airflow, are: 1-no case fans, 2-no connected multi-USB device reader, 3-no OM device connected, 4-3 connected HDs instead of 2 (but #3 external powered, connected to onboard eSATA port), 5-non-RAID boot HD (with far more installed operating systems to boot from), and 6-different PS, a long ago recapped Rosewill AP-450X "400W maximum power", which in its output table lists 370.5W max, 30A on +3.3V, 40A on +5V, 18A on +12V, 1A on -5V, 1A on -12V, 2.5A on +5Vsb.

For the first 20-30 bench hours, among which included about 6-7 hours of Memtest86+ 4.2, I fiddled with various BIOS settings, and rebooted a lot. Eventually I decided it seemed that Speedstep, something under ultimate control via voltage manipulation, was the root problem. I ran Youtubes on Firefox that way, with the only other deviation in overclock settings from defaults being RAM +.05v, about 14 hours straight, then it locked up. Kernel Watchdog reported CPU0 was hard locked, same as I had seen with the lockups reported in the bug.

Only at this point did I consider the possibility that the PS might be a problem, so I found and connected the recapped EW-430 that I had no reason to replace when I did, beyond making sure the new Corsair worked before the free return period expired. Since then, I've run openSUSE 11.4, 12.1, 12.2, 12.3, 13.1, 13.2 and Tumbleweed, Mageia 4 and Cauldron, and Fedora 21, 22 and Rawhide, roughly 15 different kernels. Several I exercised by doing zypper/urpmi/yum updates/upgrades. I kept it I/O heavy using rsync to/from same disk a couple or maybe three hours. 13.1 and 13.3 I ran much longer, either Youtube in Firefox (about 14 hours in 13.1), or Smplayer playing DVB streams saved to the eSATA HD (13.3 most of time since). Last lockup was on the other power supply, now approaching 32 hours ago.

Does anyone here care to comment generally, or suggest some action to take that might be useful? I'm thinking more PS might be a way to ensure better stability, but maybe not before trying something else. I'm wondering based on overall history whether maybe 1500uF/16v in EC15-18 could have been short-sighted originally, whether maybe substituting those higher uF Sanyos, or some 1800uF/16V HMs (with their nominally higher ripple) I have could provide extra stability. Is there significant likelihood the prior use with a marginal PS weakened those 3300uF 6.3V MCZs in EC19 & EC20? I have both Panasonic FM and Nich HM I could try there, if someone were to suggest there might be any point trying. Then there are those available overclocking options. Should I try adding voltage somewhere? If so, where, and how much?
Attached Images
File Type: jpg p9657ab-015-1500.jpg (185.2 KB, 51 views)

Last edited by mrmazda; 03-08-2015 at 06:44 PM..
mrmazda is offline   Reply With Quote
Old 03-08-2015, 09:44 PM   #7
Wester547
-
 
Join Date: Nov 2011
City & State: CA.
My Country: USA.
Line Voltage: 120-125VAC 60Hz.
I'm a: Knowledge Seeker
Posts: 1,265
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

It sounds like those Sanyo WGs are probably going leaky. They are outside the 20% tolerance (on the upper side) and the fact that ESR reads 0.00 probably means that they're going leaky (partially shorted). I wouldn't trust those WGs though the WG series itself has a very good track record minus the tendency to seldom fail. I don't know the history of those used caps so I couldn't say why they failed. MCZ or HM should be fine for the VRM input. Some of those KZGs might be starting to go leaky too (0.00 ESR reading) but at least one of them is still in good shape (though I could be wrong about that as those values may be displaying a reading beyond the limitations of the meter, IE between 0.00 and 0.01). On the other hand, it might just have to do with the length of the leads. The longer the leads, the higher the ESR is. Also, 1500uF (or 6,000uF in total with four of them) should be fine for the VRM input. Switching (or buck regulators, in this case) regulators like those in the CPU VRM don't need capacitors with bulk capacitance in very high amounts, they need capacitors with low ESR and good ripple ratings. That's how you see some modern boards get away with only 270uF - 470uF polymers in the VRM input. But more capacitance is better for transients and voltage regulation, though higher capacitance does mean that the capacitors will take longer to charge and discharge.

Last edited by Wester547; 03-08-2015 at 09:48 PM..
Wester547 is offline   Reply With Quote
Old 03-09-2015, 06:48 PM   #8
momaka
master hoarder
 
momaka's Avatar
 
Join Date: May 2008
City & State: VA (NoVA)
My Country: U.S.A.
Line Voltage: 120 VAC, 60 Hz
I'm a: Hobbyist Tech
Posts: 10,860
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

If there is anything that might be underspecced on that motherboard, it's those green Lelon RGA caps. They are not even entry-level low ESR! If you got bad power going to the Northbridge or Southbridge, that can cause all sorts of issues as well.
momaka is offline   Reply With Quote
Old 03-10-2015, 04:37 PM   #9
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Wester547, those .00 readings I wrote are what the meter displays on brand new low ESR caps, so presumably with .00 displayed the caps are somewhere between .000 and .005.

Momaka, until this AM there were 22 8x11.5 RGAs, 3 470uf 6.3V in the USB/SATA port area, and 19 1000uF 6.3V scattered elsewhere. As I had enough 1000/6.3 HM to replace all 22, I started removing all, and only in the process of measuring each on removal did I realize they weren't all the same. The removed RGA 1000/6.3 averaged 1099.8uF and .081ESR. I initially put back 2 470s I had removed after measuring them at 511uF .12 ESR and 497uF .14 ESR and finding I had no matches in stock. The closest matches I had to the 3 470s were 8x11.5 FR 470uF 16V and 8x11.5 HN 680uF 10V, so it was only after replacing the 19 1000/6.3, then running memtest for 5 passes in 6 hours without changing any BIOS CPU or OC settings and having Stresslinux 0.7.106 lockup first try about half way through init, then on second try, after bumping RAM +.2V, post-login initializing sl-wizard.sh, that I put in 3 FRs. Checking those 3 RGA 470uF 6.3Vs produced uF 477-491 & ESR .18-.22 readings.

With next Stresslinux boot, I got all the way through sl-wizard.sh the first try. I then started through the usage examples on http://www.linux.com/learn/tutorials...-your-hardware , checked core temps on tty12, and it locked up on the 2nd test: 'stress --cpu 2000 --timeout 30s --verbose'. As I had gotten though all these tests yesterday with Speedstep disabled, without apparent failures, I went back to BIOS and turned off Speedstep again. On reboot BIOS Health Status reported VCC: 3.26V, Vcore: 1.35V, VCC1.8V: 1.96V, VCC1.25: 1.27V, +5V 4.79V, +12V 12.05V, +3.3V 3.18V, VSB 3.24V, battery 3.13V.

I went in and out of POST and BIOS several times next trying different OC voltages trying to get VCC up, but none of Vcore Over Voltage, MCH Voltage or CPU Termination Voltage had any effect, and besides RAM, there are no other voltages adjustments I can find. I moved RAM back to +.1V, then booted back to Stresstest. This time 'stress --cpu 2000 --timeout 30s --verbose' completed normally, sending both CPU cores from 49C idling to 57C. All subsequent sample commands completed successfully, so I repeated 3 times in short succession 'stress --cpu 2000 --timeout 30s --verbose', all successfully, sending current CPU coretemps to 63C and peaks to 78C.

I'm now wondering if:
1-Given the POST failure history of this model in web forums dating back to when these motherboards were new if improvement of a design shortcoming mightn't be had by filling that empty TC1 with a cap matching those polys in TC2-TC6, allowing Speedstep to be enabled without lockups resulting?
2-What's going on here is actually a Linux kernel regression WRT Speedstep, nothing principally to do with hardware; maybe a BIOS bug not hit using a single core CPU or the slower Conroes available when these Foxconns were current products?
3-Lasting damage to the MCZs in EC19 & EC20 occurred while a marginal PS was in use? The only thing I have on hand that I could substitute is FR, but their ripple is only 76% of what those MCZs spec at.
4-Some combination of 1, 2 &/or 3 apply?
mrmazda is offline   Reply With Quote
Old 03-10-2015, 05:28 PM   #10
RJARRRPCGP
Badcaps Veteran
 
Join Date: Jul 2004
City & State: North Springfield, Vermont
My Country: USA
Line Voltage: 118-127V 59-63.5 Hz-> actualizo: pérdido de voltaje
I'm a: Knowledge Seeker
Posts: 6,083
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Quote:
Originally Posted by mrmazda View Post
Wester547, those .00 readings I wrote are what the meter displays on brand new low ESR caps, so presumably with .00 displayed the caps are somewhere between .000 and .005.

Momaka, until this AM there were 22 8x11.5 RGAs, 3 470uf 6.3V in the USB/SATA port area, and 19 1000uF 6.3V scattered elsewhere. As I had enough 1000/6.3 HM to replace all 22, I started removing all, and only in the process of measuring each on removal did I realize they weren't all the same. The removed RGA 1000/6.3 averaged 1099.8uF and .081ESR. I initially put back 2 470s I had removed after measuring them at 511uF .12 ESR and 497uF .14 ESR and finding I had no matches in stock. The closest matches I had to the 3 470s were 8x11.5 FR 470uF 16V and 8x11.5 HN 680uF 10V, so it was only after replacing the 19 1000/6.3, then running memtest for 5 passes in 6 hours without changing any BIOS CPU or OC settings and having Stresslinux 0.7.106 lockup first try about half way through init, then on second try, after bumping RAM +.2V, post-login initializing sl-wizard.sh, that I put in 3 FRs. Checking those 3 RGA 470uF 6.3Vs produced uF 477-491 & ESR .18-.22 readings.

With next Stresslinux boot, I got all the way through sl-wizard.sh the first try. I then started through the usage examples on http://www.linux.com/learn/tutorials...-your-hardware , checked core temps on tty12, and it locked up on the 2nd test: 'stress --cpu 2000 --timeout 30s --verbose'. As I had gotten though all these tests yesterday with Speedstep disabled, without apparent failures, I went back to BIOS and turned off Speedstep again. On reboot BIOS Health Status reported VCC: 3.26V, Vcore: 1.35V, VCC1.8V: 1.96V, VCC1.25: 1.27V, +5V 4.79V, +12V 12.05V, +3.3V 3.18V, VSB 3.24V, battery 3.13V.

I went in and out of POST and BIOS several times next trying different OC voltages trying to get VCC up, but none of Vcore Over Voltage, MCH Voltage or CPU Termination Voltage had any effect, and besides RAM, there are no other voltages adjustments I can find. I moved RAM back to +.1V, then booted back to Stresstest. This time 'stress --cpu 2000 --timeout 30s --verbose' completed normally, sending both CPU cores from 49C idling to 57C. All subsequent sample commands completed successfully, so I repeated 3 times in short succession 'stress --cpu 2000 --timeout 30s --verbose', all successfully, sending current CPU coretemps to 63C and peaks to 78C.

I'm now wondering if:
1-Given the POST failure history of this model in web forums dating back to when these motherboards were new if improvement of a design shortcoming mightn't be had by filling that empty TC1 with a cap matching those polys in TC2-TC6, allowing Speedstep to be enabled without lockups resulting?
2-What's going on here is actually a Linux kernel regression WRT Speedstep, nothing principally to do with hardware; maybe a BIOS bug not hit using a single core CPU or the slower Conroes available when these Foxconns were current products?
3-Lasting damage to the MCZs in EC19 & EC20 occurred while a marginal PS was in use? The only thing I have on hand that I could substitute is FR, but their ripple is only 76% of what those MCZs spec at.
4-Some combination of 1, 2 &/or 3 apply?
78 C is unusually high for core temps, even with Linpack. Especially a dual core.
Check the push pins.
__________________
ASRock B550 PG Velocita

Ryzen 5 "Vermeer" 5600X

32 GB G.Skill Trident Z Neo F4-3600C18D-32GTZN

Sapphire Nitro+ Radeon RX 6750 XT

eVGA Supernova G3 750W

Western Digital Black SN850 1TB NVMe SSD




"¡Me encanta "Me Encanta o Enlistarlo con Hilary Farr"!" -mí mismo

"There's nothing more unattractive than a chick smoking a cigarette" -Topcat

"Today's lesson in pissivity comes in the form of a ziplock baggie full of GPU extension brackets & hardware that for the last ~3 years have been on my bench, always in my way, getting moved around constantly....and yesterday I found myself in need of them....and the bastards are now nowhere to be found! Motherfracker!!" -Topcat

"did I see a chair fly? I think I did! Time for popcorn!" -ratdude747
RJARRRPCGP is offline   Reply With Quote
Old 03-10-2015, 05:56 PM   #11
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

Quote:
Originally Posted by RJARRRPCGP View Post
78 C is unusually high for core temps, even with Linpack. Especially a dual core.
Check the push pins.
No push pins to check. For bench testing I use a low tech 100% aluminum Maasco cooler with a back bracket that connects to cooler using screws. When in case I use a radically more efficient Freezer Pro 7, which is much easier to bump into and dislocate if used for bench testing.

78C was the peak that must have been reached with those sequential cpu 2000 tests that failed with Speedstep enabled. It was down 15C between the announcement of end of tests and when I get switched to tty12 to observe the temp readings.
mrmazda is offline   Reply With Quote
Old 04-17-2015, 12:44 PM   #12
eccerr0r
Solder Sloth
 
eccerr0r's Avatar
 
Join Date: Nov 2012
City & State: CO
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 7,211
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

That Foxconn G9657MA board sounds familiar... I think I have one of them, except I had a Conroe E6700 in there from the get go, probably on the order of a decade ago. I've been running that machine almost 24/7 since when I got it, and the caps in the PSU (Leadman custom low profile micro atx) blew up once. I replaced those and got the PSU to work again, but had noticed the caps on the motherboard had also started to leak...

So far I haven't done anything to the leaking caps on the motherboard, but the machine is still running... wonder when it'll be time to swap those caps out...

And coretemp in Linux reports the cores at 60C right now... under no load...
eccerr0r is offline   Reply With Quote
Old 09-19-2015, 10:36 PM   #13
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

In June this P9657AB-8EKRS2H ceased to post. Yesterday, upon encountering some 1800uF 16V MCZ Rubies, I gave its behavioral history more thought, and used 3 of them to replace the 3 1500uF 16V HMs that previously replaced the OEM KZGs in EC15, EC16 & EC17, leaving the 1500uF 16V Ruby that replaced the HV that replaced the other KZG in EC18. That made it POST. I reset the BIOS, which doesn't any more seem to have any way to disable Speedstep while the 2.67GHz E6700 is in use. After BIOS reset it ran MEMTEST86+ 4.20 flawlessly with the 4 2GB generic RAM sticks for about 3 hours before I turned it off for a break to sleep. Since, it has been running 6 different Linux installations, without apparent flaw, for about half the waking hours since. CPU temp in BIOS PC Health Status is 54C. Voltages running one rotating rust are: VCC 3.3=3.28; Vcore=1.35: VCC 1.8V=1.79; VCC 1.25=1.27; +5V=4.83; +12V=12.05; +3.3=3.20; VSB=3.28. PS is re-capped non-D EA-430.

Writing this I got 'stress --cpu 2000 --timeout 30s --verbose' to run to success. Then I ran stress --cpu 2000 --timeout 20s and stress --cpu 2000 --timeout 90s to success. 'burnP6' sends core0 to 80C and core1 to 85C, while "high" seems to be stuck on 78C. Reboot into BIOS shows CPU back at 54C. If this ultimately goes back into my server, it'll be getting fresh cooler contact surface cleaning, fresh thermal paste, and more effective Freezer 7 Pro cooler in place of the noisy OEM Intel cooler on it now.

Looks like maybe the ripple requirement from EC15-EC18 may have been borderline. Anyone else think so?
mrmazda is offline   Reply With Quote
Old 09-20-2015, 03:35 AM   #14
Per Hansson
Super Moderator
 
Per Hansson's Avatar
 
Join Date: Jul 2005
City & State: ----
My Country: Sweden
Line Voltage: 230v 50Hz
I'm a: Knowledge Seeker
Posts: 5,232
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

As my crystal ball seems to not be working could you tell us where those caps are, CPU VRM input?
Maybe you have some pictures to help?
__________________
"The one who says it cannot be done should never interrupt the one who is doing it."
Per Hansson is offline   Reply With Quote
Old 09-20-2015, 04:00 AM   #15
mrmazda
MrMazda
 
Join Date: Sep 2013
City & State: Nature Coast, FL
My Country: USA
Line Voltage: 120VAC 60Hz
I'm a: Hobbyist Tech
Posts: 100
Default Re: Antec Earthwatts EA-430D's OSTs killed my Conroe?

First pic in post #1 shows EC15 HM at right, EC16 HM to its left, EC17 HM to its left and EC18 MCZ to its left, all marked with red nail polish. EC19 is MCZ 3300uF 6.3V north of EC15. EC20 is MCZ 3300uF 6.3V north of EC17 & EC18.
mrmazda is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump



Badcaps.net Technical Forums © 2003 - 2022
Powered by vBulletin ®
Copyright ©2000 - 2022, Jelsoft Enterprises Ltd.
All times are GMT -6. The time now is 02:55 AM.
Did you find this forum helpful?