Which of these 5 PSUs is the best to replace a dead 700W unit?

Toggle sidebar Toggle sidebar
C

curiousff

Prominent
Oct 29, 2017
12
0
510
Hi there, people.

Some days ago I asked about the low voltage my DeepCool Aurora 700W was providing on the +12v rail, got some answers, but sadly the PSU died a couple of days ago before I could start my project. The only thing I did was update my former i3 6100 for an i5 6500 a week ago. I ran Prime95 and other stuff to test the system stability and everything went just fine. I guess my PSU was doomed since day one...

Anyway, back where I live the prices are sky-high and I cannot buy online for several reasons, one of them being the prices, as they are still pretty high because of the taxes and tariffs.

Locally, there are only 5 good options I could find, and they cost about the same ($75 to $85):
EVGA 600W1 (80+ white), EVGA 650 N1 (no certification), Corsair CX 600 (80+ bronze), SeaSonic S12II 620W (80+ bronze), and the Thermaltake Smart 600W (80+ white). I found some info about the Seasonic and the Corsair, but almost nothing about the other three, especially when it comes to the real voltage levels they provide.

Which of those 5 could be considered the best, quality and value wise? I know these are 50W to 100W lower than my previous PSU, but that's fine with me. Any help will be greatly appreciated. Thanks!
 
Sort by date Sort by votes
Seasonic S12II-620, despite it's fully wired design, is best PSU out of the five. Other 4 PSUs won't come close to the good build quality of Seasonic unit.

All my 3 PCs: Skylake, Haswell and AMD are also powered by Seasonic where i have S12II-520 in my AMD build. Full specs with pics in my sig.
 
Upvote 1 Downvote


I read about the S12II being an old design that's still good, but I couldn't find information about the +12v levels under load. I know there is +/-5 tolerance under the ATX specification, but the low +12v levels on my previous PSU and it subsequent death are a concern to me.
 
Upvote 0 Downvote


Thanks for the reply. My objective is to get a "combat proven" PSU that lasts, as warranties and stuff mean squat in my country. It seems Seasonic would be the way to go...
 
Upvote 0 Downvote


Many thanks for that link. I guess I was being too specific on my search for the capacity instead of the model. Those tests and the results are awesome. I assume the 620W model will operate exactly the same as the 520W one.
 
Upvote 0 Downvote


Yep :)
 
Upvote 0 Downvote
Since the chipsets compatible with the i5-6500 will, by default usually, want to use the C6/C7 low power states (All Intel platforms since Haswell) and that group regulated design on the Seasonic S12II platforms do not support it, you will want to make sure that when you first power it up after installing the new hardware you go into the bios and make sure C6 and C7 are disabled in the power options, if they are not, disable them, save settings and restart.

Allowing the C6/C7 states to remain enabled on newer chipsets they want to use it when you have a power supply that doesn't support it can cause problems so it's best to just turn them off and avoid any potential issues.
 
Upvote 1 Downvote
Also, this is the exact same platform as the Seasonic S12II-620w non-modular power supply. Antec high current gamer 620w non-modular, made for Antec by Seasonic. Full review by Aris.

www.techpowerup.com

Antec High Current Gamer 620 W Review

The High Current Gamer series, according to Antec, manages to combine high power and increased efficiency with low pricing. As the naming scheme reveals, HCG PSUs are intended for users that want to feed their precious PC with lots of power, but are on a tight budget.
www.techpowerup.com www.techpowerup.com
 
Last edited:
Upvote 0 Downvote


Wow! Thanks! I had no idea about that. Precious info you just gave me. Just some more questions, please. How can I tell the difference between group regulated/unregulated PSU designs? Which one is better on the long term?

 
Upvote 0 Downvote
Group regulated is a design that used to be used but rarely is anymore except on units using old platforms or low budget models. Now any decent newer model of power supply uses DC to DC regulation.

Basically, if it is group regulated, it cannot support some of the specific low power states required or preferred to be used on newer processing chipset designs, which is why there are some designated as Haswell compliant and other, group regulated models that do not have that designation.

http://www.jonnyguru.com/forums/showthread.php?t=3265


Also:

The issue is specific to certain Haswell-based (Or newer) system configurations.

Haswell introduced new, extremely low-power states called C6 and C7. The processor virtually shuts down in these power states, placing loads as low as 0.05A on the +12V rails. Since it's usually only high-power devices such as the processor and graphics card that draw power from the +12V rail, some low-power desktop builds that rely entirely on integrated graphics pull virtually no power on the +12V rail, while other system components continue to pull significant amounts of power from the +3.3V and +5V rails. This is the cross-loading situation you described in the question.

In group-regulated power supplies, voltage on all rails is regulated based on the total load on each of the rails—the supply compensates for voltage drop on all of the rails as overall load increases (regardless of which rails are under load). Before Haswell, this was never an issue because most older systems placed a nontrival load on all of the major rails. However, in a cross-load situation like the one described above, group regulation can cause the voltage on the +12V rail to be overcompensated to the point where it falls outside of the ±5% voltage tolerance required by the ATX12V standard, exceeding 12.6 volts, while the +5V rail and +3.3V rails may experience excessive voltage drop. The +12V rail may also exhibit excessive noise (ripple) when a group-regulated supply is cross-loaded in this manner.

A properly-designed power supply will detect this overvoltage condition and shut down. However, some very cheap designs may not have this kind of protection and allow the +12V rail to go well out of spec, potentially leading to hardware damage. Even if it didn't go out of spec, a voltage that is consistently way off (not to mention excessive ripple) isn't exactly good for the longevity of the hardware (and remember that office PCs are often idle much of the time, meaning that this cross-loading condition can persist for extended periods of time).

Most modern power supplies use different approaches of generating the +3.3V and +5V rails that eliminate this issue, such as by generating only a +12V rail on the "secondary side" (transformer output) and deriving the +3.3V and +5V rails from the +12V rail through DC-DC conversion, or through independent regulation of each rail. A workaround for this issue on group-regulated supplies is to disable the C6/C7 power states in the system firmware (BIOS or UEFI), but you'll lose the power consumption benefit of these states.

HEXUS has an example where an older group-regulated supply, the be quiet! Pure Power L8 500W, performs poorly under a cross-load test intended to simulate Haswell C6/C7 operation. Notice that the +5V rail nearly falls out of spec at -4.8% while the +12V rail rises to +3.3%:
Click to expand...
 
Upvote 1 Downvote

A very interesting read, thanks. Based on this information, it can be said that it is safer to disable the C6/C7 power states for all group regulated PSUs, regardless of the brand, right?
 
Upvote 0 Downvote
While some Haswell (or newer) systems C6 and C7 sleep states doesn't play well with group-regulated PSU, that doesn't apply to all newer systems.

Since i have Seasonic S12II-520 PSU, i tested it with my Skylake build (i5-6600K) to see if my group-regulated PSU supports my CPU's sleep states or not.
To test it, i hooked S12II-520 to my Skylake build and i booted into BIOS. Then i manually selected CPU's sleep state to be used (available options are: C0, C2, C3, C6, C7, C7s and C8). I started with C6. After then i saved settings and booted into Windows. When in Win, i put my PC to sleep and waited at least 5 mins before i woke up my system again to see if i get any BSOD or instant reboot if the PSU doesn't support that sleep state.
Since i didn't get any issues whatsoever, i went back to BIOS and selected another sleep state until i tested C6, C7, C7s and C8 sleep states. All my tests were successful and didn't face no issues whatsoever.
I concluded that my i5-6600K's C6 and higher sleep states work nicely with my S12II-520, despite the group-regulated design my PSU uses.

It is possible that your i5-6500 and S12II-620 also work together nicely but you need to test it out manually like i did, if you care about using C6 and higher sleep states. If not, set your CPU to use C5 and/or lower sleep state.
 
Upvote 1 Downvote
Thank you Darkbreeze, Aeacus, and Vapour for all your replies and detailed info. I'll go for the Seasonic S12II-620.

I'd like to pick your 3 solutions as the best, as each has its own merits, but sadly, that cannot be done. I appreciate all your help. All the things you wrote were enlight­ening, I definitely want to read more about PSUs. Thanks again.
 
Upvote 0 Downvote


This is wrong. Just because the system doesn't crash, or fail to resume from the low power state, or any other "obvious" problem that is readily apparent, does not mean it should be used. It is still a problem.

This has little to do with the actual problem. The real problem is exactly as outlined in the quote I posted on my earlier post. If you bothered to read it or had researched this previously you would know that while those obvious problems can ALSO occur, the real problem is that even on a good unit group regulation can cause the voltage on the +12V rail to be overcompensated to the point where it falls outside of the ±5% voltage tolerance required by the ATX12V standard, exceeding 12.6 volts, while the +5V rail and +3.3V rails may experience excessive voltage drop.

So even if it doesn't go completely out of spec and cause a shutdown, which a very good group regulated unit SHOULD, but doesn't always, do, the voltage on one or all of the rails can still be consistently and significantly OFF and cause prolonged noise/ripple which is likely to be endured for long periods of time while the power supply is in the low power state, hours, days sometimes, and that is causing cumulative damage to your hardware.

Under normal conditions, a very good group regulated PSU may only slightly shorten the computer's lifetime. It should however be used with caution in DIY computer builds, with good understanding of its limitations. A power malfunction can have extended consequences with such a PSU.
Click to expand...

Also, sample of "one" results should NEVER be considered to be indicative of a successful counter to the results of MANY real world and lab tested systems that have unquestionably shown this to be a problem, regardless that your system didn't explode or slam the door in your face, it does not mean that users should ignore the science and proceed as though they were ignorant of the fact that they are putting their power supply AND their hardware at risk by using them under these circumstances.


SO yes, it does apply to all Intel systems since the release of Haswell. And, the problem is likely to increase due to the likelihood of even further reductions in power on future architectures.
 
Upvote 0 Downvote

I did read your reply thoroughly and i did it twice before i posted mine.

"The issue is specific to certain Haswell-based (Or newer) system configurations."
Certain means some but not all.
Also, reading further from the long quote you posted, one can clearly read that:
"... group regulation can cause the voltage on the +12V rail to be overcompensated to the point where it falls outside of the ±5% voltage tolerance required by the ATX12V standard, ..."

Just because it can cause voltage to go out of spec doesn't instantly mean that it will cause the voltage to go out of spec. This possibility is repeated throughout the quote you posted without any solid conformation that this is so with all CPUs since Haswell family and with all group-regulated PSUs.


I never said that all i5-6600K CPUs are safe with any S12II-520 PSU. I stated that it's safe with my build after i manually tested it. I also said that if OP wants to use C6 and higher sleep states, he needs to test it out by himself. All that is written in black and white if you scroll a bit up to read my reply again.

My issue with this is that something that "can happen" is advertised as "it will happen".
Rather than bluntly say to all who are using group-regulated PSUs: "disable C6/C7 sleep states", one should actually say: "C6/C7 sleep states may not be compatible with group-regulated PSUs and there can be damage to the system in long-term usage".

There are plenty of tech-savvy people out there who are interested to test their systems and to see if it works as it should. Rather than discourage tech-savvy people off from testing their systems (in this case PSUs), i'd rather encourage tech-savvy people to test C6/C7 sleep states with their group-regulated PSUs. Also, it isn't rocket science to take out the multimeter and check the PSU's voltages during PC's sleep to see if the voltage does go out of spec or not.
 
Upvote 0 Downvote
Also note that ALL Intel generations of core processor families support the C6 core and package ACPI state but only 4th gen and newer support the C7 state. I cannot say why Ivy Bridge and older processors that support the C6 state do not have the same issues as 4th gen and newer, except that all the 4th gen and newer CPUs EXCEPT the 5th gen (Broadwell) support C7 as well however it seems Broadwell does not support C7 unless the Intel datasheet is wrong, and the sources I looked at indicated Broadwell has the same issues with group regulated power supplies as the earlier generation Haswell and Haswell refresh families.

I will definitely be looking into this and try to find some answers as this is a relatively important issue that could account for why many systems, past and present, have experiences unusual symptoms that could not be accounted for or explained through any ordinary testing or troubleshooting methodology. I'll do my best to get 100% clarification regarding these. I think perhaps there may also be some indication of earlier Core processors only supporting core C states and not package but this is not clear because I also saw reference to some Gen 2 processors indicating that no core could enter core C6 state without all cores entering core C6 state. This is what I could definitively find though.


8th Gen (S-platform) Intel® Processor Family

• Full support of ACPI C-states as implemented by the following processor C-states:
— C0, C1, C1E, C3, C6, C7, C8
• Enhanced Intel SpeedStep Technology


7th Gen (S-platform) Intel® Processor Family

• Full support of ACPI C-states as implemented by the following processor C-states:
— C0, C1, C1E, C3, C6, C7, C8
• Enhanced Intel SpeedStep Technology



6th Gen (S-platform) Intel® Processor Family

• Full support of ACPI C-states as implemented by the following processor C-states:
— C0, C1, C1E, C3, C6, C7, C8
• Enhanced Intel SpeedStep Technology



5th Gen (H Platform) Intel® Processor Family

• Full support of ACPI C-states as implemented by the following processor C-states:
— C0, C1, C1E, C3, C6
• Enhanced Intel SpeedStep Technology



The Intel Core™ (HEDT) X-Series processor family

• Full support of ACPI C-states as implemented by the following processor C-states:
— C0, C1, C1E, C3, C6, C7, C8
• Enhanced Intel SpeedStep Technology



The 4th Generation Intel Core, Pentium and Celeron processor family

• Full support of ACPI C-states as implemented by the following processor C-states:
— C0, C1, C1E, C3, C6, C7
• Enhanced Intel SpeedStep Technology


 
Upvote 0 Downvote
I wanted to revisit this thread after discussing it in some depth with ko888 and InvalidError, among others.

It turns out that the problem is not EVEN relevant to systems using the C6 low power CPU state, which has been used since the 1st gen Core-i CPUs. The issue regarding the PSU response time is only relevant to systems using the C7 state, and it affects ALL systems, Core-i, Pentium, Celeron, ANYTHING that uses the C7 state MUST be compliant with the new Intel spec for recovery from C7 state. This affects every Intel processor newer than 4th Gen that I've looked at the datasheets for, so basically, all of them.

Intel's Haswell and later processors feature a C7 sleep state that isn't compatible with all power supplies. Haswell and later processor's low power sleep state power draw is substantially lower than that of previous generations, and it can trigger some PSUs' over-voltage or under-voltage protection circuit forcing the need for a PSU reset.

According to Intel's IDF (Intel Developer Forum), Haswell and later processors can enter a sleep state called C7 that can drop the processor VRM's +12V current draw to as low as 0.05 Amp. Even if the sleeping CPU is the only load on the +12V rail, most power supplies can handle a load this low. The potential problem may arise when there is still a substantial load on the power supply's minor rails (i.e. the +3.3V and +5V). If the load on these minor rails are above a certain threshold (which varies by PSU brand and model), the +12V can exceed ATX12V specs (voltages greater than +12.6V) or the voltage on the +5V rail may drop below the 4.75V minimum allowed by the ATX12V specs. If the +12V and/or +5V rail is out of spec when the motherboard comes out of the sleep state, the PSU's OVP (Over-Voltage Protection) and/or UVP (Under-Voltage Protection) circuit may get triggered preventing the PSU from running and will cause the power supply to "latch off". This will require the user to cycle the AC power switch on the back of the PSU to reset the PSU's protection circuit.

This problem may occur with PSUs with a secondary side that uses the group regulated circuit design due to the +12V and +5V rails sharing the same choke. PSUs with their secondary side that utilize independent (indy) regulation or DC-to-DC circuit design don't suffer from this problem.

Prior processor generations to Haswell had a minimum sleep power consumption of 6 Watts. With Haswell the minimum sleep power consumption was lowered to one-tenth of that at 0.6 Watt.

To truly meet Intel's own Haswell PSU testing methodology none of the PSU's output rails should ever go outside of ATX12V specs under any load condition.

Anyone building a Haswell or later system from scratch are better off getting a compatible unit to begin with.

Anyone building a Haswell or later system using their existing group regulated PSU should disable processor C6/C7 power state support in their motherboard's BIOS setup if they encounter the PSU latching off when resuming from sleep state.
Click to expand...
 
Upvote 0 Downvote

Thanks for the update and additional info. :)

From the statement you shared, there's written:
Anyone building a Haswell or later system from scratch are better off getting a compatible unit to begin with.

Anyone building a Haswell or later system using their existing group regulated PSU should disable processor C6/C7 power state support in their motherboard's BIOS setup if they encounter the PSU latching off when resuming from sleep state.
Click to expand...
According to this, there's no requirement to have a compatible PSU for a CPU that has C7 sleep state. While compatible PSU is suggested, using incompatible PSU with C7 sleep state is fine, as long as PSU's OVP or UVP isn't triggered when coming out from C7 sleep state. If PSU's OVP or UVP is triggered, then disabling C7 sleep state prevents it happening again.

I think the high standard of Seasonic PSUs is the reason why my S12II-520 didn't trigger it's OVP or UVP when i tested my i5-6600K's C7, C7s and C8 sleep states. But since not all PSUs aren't created equal, one should test C7 sleep state with group-regulated PSU before enabling C7 sleep state for good.
 
Upvote 0 Downvote
I fully agree with that assessment.

I mainly came back and shared this info because it's INCREDIBLY hard to find any REAL information beyond the obvious systematic belief of C6/C7 = Indy regulated or disable. Information on the specifics were rather hard to come by, even in the datasheets.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom