ITTSB.EU Blog Forum
ITTSB Blog Specific => ITTSB - Small projects by Kiriakos & User reviews => Topic started by: Kiriakos GR on July 06, 2020, 09:54:30 PM
-
New gift in my hands ... an damaged CORSAIR CX750 CP9020015 :)
This seems as Quality Monster ;)
Inner capacitors are Panasonic at the important parts of the circuit and also Capxon at lesser of importance circuits.
The exception in this PSU this is the addition of DC to DC converter which control voltage stability.
Such circuit will not aloud voltage sag under load by the offering of instant output correction.
These are new technological additions of 2012, but these smart additions they also contribute in to a greater complexity and degree of difficulty, when this thing it must be repaired.
Within one hour of diagnosis I did found the problem and also the reason which caused it.
Here are some pictures from it for now.
-
Gentlemen I found the smoking gun 8)
A faulty CPU connector over a motherboard, this caused sparks and arching, this is an occasion very RARE in which the power supply electronic protection circuit cannot handle or resolve.
This electrical incident forced several APEC 9990GH N-CH 60V 100A( its one ) MOSFET's them to collapse.
These are mounted over a PCB this called as SYNC-Card.
Molex the manufacturer of such plugs, informs us that its female pin this can handle 9A Max.
I did order the necessary spare parts, now I am assuming after a month of waiting (Speed-Pack CN) that I will see this little monster having a second life :)
In the past 48 hours I did inspect by my LCR meter all the capacitors, all are at 100% of health and specifications.
Additionally for confirmation purposes I did visit the website of Capxon, the capacitors series in use this has 5000 Hours specification.
The replacement series of the 2200uF 16V Capxon this can be the Panasonic KZE.
One small problem if these capacitors require replacements, this is the diameter of 10mm, all newer revisions they are now manufactured at 12.5 mm.
The installation of three 12.5 mm electrolytic capacitor ( one next to another) this is possible but not an easy job.
Either way I am now convinced that the specific PSU, this has not operate more than a few months.
-
Few words about this PSU electronics design history, education combined with entertainment. :)
CWT manufacturer this build several PSU for other brands.
The specific PCB and circuit named as CWT PUQ-G this is printed at the PCB of LEPA G850-MAS PSU , this is identical to CORSAIR CX750 - CP9020015
https://www.techpowerup.com/review/lepa-g850-mas/4.html
Even so, the CORSAIR CX750 - CP9020015 this has an additional heatsink over the hart of the system that is two powerful Mosfet over the lengthier heatsink on this PCB.
Now it is obvious that CORSAIR - CP9020015 this is a down-rated 850W PSU with optimal cooling than the overrated LEPA at 850W with lesser heatsink surface.
Therefore the CORSAIR - CP9020015 this is a better build in the details, but it is rated at just 750W, in order this to provide an extended life cycle, because any wise person he will not stress it more than that.
Now while LEPA G850-MAS this is identical PCB , they did use Panasonic capacitors at 12V Rails, they did that so to improve mostly the impression created at the eyes of product reviewers. ;D
Panasonic KZE this is also a 5000 hours capacitor, as is the Capxon, but the Panasonic has Lower ESR, which this might help just a bit at very high Loads at 850W.
Now the LEPA G850-MAS seems to were stressed up to 1150W until circuit protection engages.
Here is another tip, CORSAIR did not aloud at CWT them to stamp this PCB as CWT PUQ-G because CORSAIR wanted this detail to stay hidden.
Today it looks as funny, but even the newer Corsair Gaming Series GS800 this is based at the identical PCB too. :)
It is obvious that a few first Revisions of CX750 were not that perfect, until the CWT PUQ-G get in to make the job, but the fame of this model (CX750) were injured due the product reviews.
And now Corsair returned back by the good PCB by using a new naming as Corsair Gaming Series GS800 among with an DC fan with LED lights.
https://hexus.net/tech/reviews/psu/42425-corsair-gaming-series-gs800/
In both reviews the Identical PSU design shown worthy and a great performer!
The new name of GS800 .. July 2012 .. this now restored the true named capacity of the circuit it self at 800 Watt continues when the Max headroom stops at 1150W.
Most Gamers PC, this will never use over 350W continues, but this is another bed time story. :)
-
All right !! Today I did put back together all the parts and wires which I did remove under the electronics fault diagnosis stage.
Back at desoldering stage one electrolytic 470uF 16V this came out damaged, also at the specific PCB it was planed to be three of them, so now I though as good idea to add and the third.
But now I have to handle an small new issue, the ones that I have in stock they are a little bit taller, now I am planning a small modification which will solve this minor issue too. :)
-
This is a reference relative to DC Fan speed control
The DC fan this is 140 x 140mm.
Over the Sync Card at 12V output there is a Thermistor TTC103 ( this has operating temperature minus 30C and up to 115 Celsius )
According to Junction Temperature characteristics of the AP9990GH MOSFET it temperature it should retained at 50C up to 100C for long lasting operation.
AP9990GH MOSFET this can handle Max temperature of 180 Celsius
And therefore the Thermistor TTC103 this will set the DC Fan at Full speed when the Corsair Sync card PCB this is at 115 Celsius ;)
In other words ... the DC Fan speed control this is triggered according the 12V Rail load.
Now in the scenario in which the DC Fan cannot cool-down the Sync PCB at lower than 115C I am assuming that over temperature protection circuit will kick-in and the PSU will be forced to shutdown.
-
Another recorded adventure of Kiriakos :)
Therefore the incoming gift, this has reputable background as electronic design.
As soon this comes back to life, what are your plans about it ?
-
Another recorded adventure of Kiriakos :)
Therefore the incoming gift, this has reputable background as electronic design.
As soon this comes back to life, what are your plans about it ?
Hi mate!
Well as soon I will verify this PSU good functionality by my electronic DC load https://www.ittsb.eu/forum/index.php?topic=1316.0 among with my oscilloscope https://www.ittsb.eu/GW-Instek-GDS-2102A-MSO-Review.html then this will become a keeper PSU as spare for my main PC system.
Since beginning of 2009 I am using a HiPER HPU-4M780-PE 780W this is a CWT Taiwan design, imported - distributed by a brand name Hiper-Group in UK.
Hiper brand were active and even honor any warranty plans for a few years, but they get out of the market by their own decision.
HPU-4M780-PE stats after eleven years in use.
After 10 Years + five months, I had to replace the 12V Rail electrolytic capacitor.
The YATE LOON DC Fan ( D14SH-12 ) this is running problem free all these years.
CORSAIR CX750-CP9020015 this is a more modern CWT design in comparison to HiPER 4M780 but equally powerful in Watts.
Actually both of these PSU, they were priced above the 120 Euro mark, and as such their quality standards are high among with their reliability and longevity.
-
Few pictures from the CP9020015 DC fan this is also one YATE LOON D14SH-12 ;)
What now empowers my curiosity this is the secondary code
At the year 2008~2009 this were L-SO1 and at 2016 this become L-SSS
Either way the CORSAIR CX750 Repair project this is now On-Hold until spare parts arrival. ;)
-
I wish to thanks AKASA in Taiwan and David Tai for the provided information of what is the AWG of AKASA AK-CB8-8-EXT ATX 12V CPU extension cable !
Originally I though that in order of me to replace the damaged connector (ATX 12V CPU) both 4Pin plugs, I would have to follow the expensive route of buying plastic housings, metal pins, special crimping tool.
But now there is a much better alternative !! 8)
AKASA AK-CB8-8EXT extension cable this is at reasonable pricing at below 5 euro and it has 18 AWG wires that is exactly of what this Corsair 750W PSU also using.
By getting one of them I can do a simpler wire to wire connection and use the new perfectly crimped connectors as 100% reliable spare-part. ;D 8)
( I thought also taking the old 12V ATX CPU connector from a dead - old - low in wattage PSU, but this is WRONG and Stupid plan to follow when you are repairing a high performance PSU ) .
-
Shopping from local market has the benefit of speed regarding delivery.
AK-CB8-8EXT arrival.
18 AWG this also confirmed.
Speaking of Molex plugs these are not from Molex brand, the product seems assembled in China and at the 8 Pin plug there is ECI as brand logo.
I did a quick search about ECI ( brand or factory in China ) no results shown.
By having at hand my NEW dual ATX plugs replacement, I am now free to focus at the next step of this repair process. :)
I am now waiting for Mosfet's arrival ...
-
The last piece of this puzzle arrived today. :)
I feel that I have good chances for success, because I own Hot-Air rework and even a soldering station, both be at the above of average quality.
-
;D 8) ;D 8) ;D 8) ;)
Now I am working on a few FRESH ideas about special electrical test and measurement harness for PSU voltage measurements.
And more tests under load will follow. 8)
-
I can now announce this repair as truly finished and tested. :)
1) I did install the male plug AK-CB8-8EXT at the PSU
2) I came with the idea to use the female plug AK-CB8-8EXT as DC load harness, because of that smart idea ... I tested and the PSU among with the quality of my own wire to wire connection.
3) The test run for approximately two hours, but the presented measurements they are for the last 1H 20 Minutes.
The 12V rail this is capable for 62A according the specifications, my DC load this is capable of 110Watt and the max stress was at 9.5A , this is 16% of load.
The amazing part this is the Min-Max voltage deviation ( with load - No Load ).
Average value 12,188V
Max value 12,191
Min Value 12,183
This is just 8 mV DC drop, that is totally insignificant.
The theoretical variation at 12V rail this can be + / - 2% = 12,240 OR 11,995 V
I did also test the +5V VSB Stand-By rail (PSU sleep mode) , from the named 3A the 5V stayed at the specification of 4,90V (-2%) up to 2.5A of load !! ;)
5,00V at 1A and 5,054V at 200mA, No load 5,070V.
In all those two hours the PSU and the cable harness were frozen, among the wire to wire connection which I did with lots of attention to detail.
The final conclusion this is that the specific CORSAIR CX750 - CP9020015 - this is now ready to serve it purpose with max reliability as this is restored up to factory quality standards 8)
By this new experience I am also now aware of max load that its cable harness this can handle.
For example when I did stress with 9.5A two pairs of cables the voltage drop this were a bit higher.
The PCI-E connector this using three pairs and this helps for transfer of up to of 10A with a minimum voltage drop under load.
One demanding hard-drive will use 750mA and in RAID mode this will be double as 1.5A per single pair of wires with four Molex or SATA connectors.
The new conclusion coming from that math, this is that single pair this is calculated to tolerate 3A ~ 4A with minimum voltage drop.
CPU harness this will aloud 12A with minimum voltage drop.
Motherboard 12V (24 pin) rail this is two pairs = 6A
The final estimate of wires sizing and expected nominal load in a chart, this is:
1 x PCI-E = 10A
2 x PCI-E = 20A
CPU 12V Power = 12A
Motherboard 12V = 6A
HDD single wires pair with four plugs SATA = 3A ~ 4A
CDRW or DVD recorder this is 2A its one (12V) for Molex four pin (At DVD to DVD recording mode the sum this will become 3A)
Three DC fans = 1.5A
-
UPDATE (Year 2022): I am now a believer, even the quality monsters they have a date that they must get retired.
I did use the CX750 with my Intel P45 motherboard along Q9560 3GHz quad core, with no problems.
This month I did switch to Intel Z87x motherboard along i7 4770 (4C/8T) and at heavy graphics load the CX750 this simply shutdown
I did some basic power consumption measurements, max load on average 300 watts, the GPU this is GTX1060 6GB, I am now assuming that this produces DC current spikes at 30% higher consumption of what I can measure with an economy (slow) power meter.
My verdict is that my CX750 of 2013, this is unable today(2022) to handle DC current spikes because it electronic protection circuitry, this is today in huge error, according to original factory calibration.
This is a problem which slowly develops at passive components, and especially at the side of resistors.
PSU current shunt this gets out of specification, because possibly the solder in use at the PCB this gets old, and it develops a higher resistance.
Or other SMD resistors at the control circuit, they get out of specification, and the protection control circuit, this receives wrong feedback, and activates prematurely.
We may be able to repair larger power electronics (replacing Mosfet and DC bridges and capacitors), but to restore the functionality of the electrical protection circuitry to factory standards, this is micro surgery that personally I will never get involved.
So I am now forced to declare CX750 as out of duty, when this is to serve i7 Haswell along an GTX1060.