Sign in with
Sign up | Sign in
Your question

Corsair 620HX and other PSUs - 8800 G92 Power Issues?

Last response: in Components
Share
February 1, 2008 7:39:57 PM

So I got my new build together this week, and everything went great on the first boot. But after installing the Nvidia driver CD for my EVGA 8800GTS 512 (G92) card, on the reboot I got a Nvidia Sentinel message that the card was receiving insufficient power and that the settings would be scaled down to prevent damage to the card. It had me scratching my head since the Corsair 620HX PSU I installed is more than enough power, and I ran the PCI-E power cable directly out of the PSU to the card.

So I started doing some Google searches and found this is FAR from an isolated problem. Here are some reference threads:

http://www.asktheramguy.com/v3/showthread.php?t=66051

http://www.jonnyguru.com/forums/showthread.php?t=3584

http://www.hardwarecanucks.com/forum/power-supplies/412...

http://www.hardforum.com/showthread.php?p=1031921280

http://www.yougamers.com/forum/showthread.php?t=73754

Notable posts:

An email from Antec customer support:
Quote:
We originally noticed this problem with the 8800GTX video cards. Nvidia tested our power supply with the cards to certify them for SLI testing and they passed with flying colors. We sent out a fix when we saw there was a problem, the fix worked and now we are noticing that the newly released 512mb GTS and GT cards are not working with this new revision.

We currently do not have a working version of the TPQ-850, and TPQ-1000 that will work with the Nvidia 8800GT and GTS 512mb video cards. Our engineers are working on a fix for this issue and it is a high priority for us.

When we get the replacements that work we will swap the units out for customers experiencing the incompatibility with the 8800GT and 8800GTS 512MB video cards.

It now seems that this problem is not only with Antec. Other manufacturers are having problems with the G92 card, and their power supplies. Nvidia is now helping Antec, and these other companies to find a solution to the problem.

The power supply will work with any other video card besides the G92 based 8800 series cards.

We apologize for the inconvenience this may have caused.

Antec Customer Support


From a Corsair Rep:
Quote:
We are working with nVidia on this and it's tough to track down because it doesn't follow any logical progression. We can't say "all people with Gigabyte motherboards" or "all people with EVGA graphics cards" or "all people with Corsair PSUs" have this issue, in fact it doesn't seem to be tied to any one component.

The only thing that's clear is that it's some mix of the motherboard, video card, and power supply somehow. I say this because some people RMA the PSU and get a new one and it fixes it, other people RMA the video card and it fixes it, and yet other people change out their motherboard and it fixes it.

The only thing that's clear is that it is ONLY a known issue with G92-based cards so far, but it's not the G92 architecture, because it works fine like 99% of the time.

My theory is that it's a component variance issue, and probably on the voltage monitoring hardware on the GPU itself. The G80 cards have no problems at all, but the G92s do, so I'm trying to work with nVidia to determine what the changes in voltage/power monitoring hardware in the two architectures were and how that could affect us. We've only been able to duplicate it in the lab with a couple of bad video cards that reproduce the error on ANY PSU (enermax, antec, corsair, fsp, whatever) so it can't be just that.

The 7900 series had this issue a lot, too, with some PSU designs, but the 7900 series had a reputation for random failures on some online forums, so it's hard to gauge what's rumor and what's fact.


From Ram Guy (Corsair Product Guru)
Quote:
So far all the cards that have been posted as having this problem are using the G92 core (8800GT 512MB and 8800GTS 512MB), which consume less power than the older 8800 (GTX, Ultra, GTS 320/640) cards which use the G80 core. We are currently working with Nvidia to find out what might be causing these issues so that we can resolve them.


From a Silverstone Rep:
Quote:
Manufacturer Response:Thank you very much for purchasing a SilverStone and I apologize for the inconvenience you have experienced. The new 8800GT and 8800GTS 512MB cards have unique power draw timings that are different from the older cards such as 8800Ultra/GTX/GTS 320MB/640MB. Usually this problem can be solved immediately by connecting the PCI-E cable to another one of four different PCI-E connector slots on the ST1000. If doing this still does not work, we have a new version of ST1000 with adjusted timing that will be available from the week of February 4 for free exchange.



Apparently, the problem boils down this this ...

Quote:
"Okay a little more info... apparently it is a timing issue, when your computer first turns on the PSU sends power down each line and then backs off waiting for the computer to start the boot process. With the G92 cards they are expecting to see this at a different time than when they are actually getting the signal and it causes them to think they are underpowered so they clock themselves back"


Soooooo .... has anyone else been dealing with this or heard about it? I was surprised not to have heard about this when I was researching the components for the new system I built. You'd think such a big compatibility issue would be more "out there" and in your face.
February 1, 2008 7:55:38 PM

Double post
February 1, 2008 11:09:27 PM

I have the one down from you, the HX520. And it powers mine fine. System is:

Intel C2D e6750
MSI P35 Neo2-FR
2GB Geil Black Dragon 6400
Point of View 8800GT 512mb
2X250GB SATA Drives
1X500GB SATA Drive
Related resources
Can't find your answer ? Ask !
February 3, 2008 8:27:45 AM

The hx520 is also affected by this problem.It strikes randomly on many psu's which is making it very difficult to nail the problem down.I have this same problem with my hiper type r 580w.Ive had this problem for 4 weeks and have tryed everything i could to resolve it.I would like to buy a new psu but will be waiting a little bit more to see if a fix comes.I think you should give corsair a call since they are aware of this and they seem to me to be more than helpfull.I think this is the most important thing when you have these problems as when i called hiper they just told me to buy a new psu which definetly wont be theirs.
February 4, 2008 5:19:49 PM

Here's another setup which doesn't work:

Asus P5KC + Inno3d 8800 GT 512Mb + Antec TPQ 850

Looking forward to the fix!
February 4, 2008 6:05:44 PM

Jaywill said:
So I got my new build together this week...EVGA 8800GTS 512 (G92) card, on the reboot I got a Nvidia Sentinel message that the card was receiving insufficient power and...the Corsair 620HX PSU I installed is more than enough power



While your PSU may have enough WATTS to power your rig, the EVGA 8800GTS 512 requires at least 26amps on the 12v rail. Per the EVGA website...





Requirements

[img said:
http://www.evga.com/images/common/blt_subnav.gif
Minimum of a 400 Watt power supply.
(Minimum recommended power supply with +12 Volt current rating of 26 Amp Amps.)
Minimum 450 Watt for SLI mode system.
(Minimum recommended power supply with +12 Volt current rating of 30 Amp Amps.)

An available 6 pin PCI-E power connector (hard drive power dongle to PCI-E 6 pin adapter included with card)]


Requirements


Minimum of a 400 Watt power supply.
(Minimum recommended power supply with +12 Volt current rating of 26 Amp Amps.)
Minimum 450 Watt for SLI mode system.
(Minimum recommended power supply with +12 Volt current rating of 30 Amp Amps.)

An available 6 pin PCI-E power connector (hard drive power dongle to PCI-E 6 pin adapter included with card)
[/img]

According to Corsair, the 620HX only has 18amps available on the 12v rail.





CORSAIR CMPSU-620HX - +3.3@24A,+5V@30A,+12V1@[color=#d4002a said:
18A,+12V2@18A,+12V3@18A,- 12V@0.8A,+5VSB@3A
]


CORSAIR CMPSU-620HX - +3.3@24A,+5V@30A,+12V1@18A,+12V2@18A,+12V3@18A,- 12V@0.8A,+5VSB@3A


Sorry Charlie, your PSU is inadequate. That's a tough lesson to learn but it helps to research these things before making the purchase.
[/color]
February 4, 2008 6:16:23 PM

I believe video card recommands the power supply requirement is for total current on 12 volt rail, not the single rail. I found out the G92 card draw less power than 80w( it is abot 7A on 12 volt rail).
February 4, 2008 6:31:26 PM

chunkymonster said:
While your PSU may have enough WATTS to power your rig, the EVGA 8800GTS 512 requires at least 26amps on the 12v rail. Per the EVGA website...

According to Corsair, the 620HX only has 18amps available on the 12v rail.

Sorry Charlie, your PSU is inadequate. That's a tough lesson to learn but it helps to research these things before making the purchase.

This just goes to show you that someone can have well over 1000 posts and still not know WTF he is talking about :sol: 
February 4, 2008 6:33:29 PM

Quote:
Usually this problem can be solved immediately by connecting the PCI-E cable to another one of four different PCI-E connector slots on the ST1000.

This is why I pay the extra money for a well regulated single rail PSU and only build with PC Power & Cooling.
Even the $109 USD 610 puts out a -constant rated- 49 Amps on the 12V line with it all being usable unlike multi rail systems.
February 4, 2008 6:34:18 PM

Quote:
Usually this problem can be solved immediately by connecting the PCI-E cable to another one of four different PCI-E connector slots on the ST1000.

This is why I pay the extra money for a well regulated single rail PSU and only build with PC Power & Cooling.
Even the $109 USD 610 puts out a -constant rated- 49 Amps on the 12V line with it all being usable unlike multi rail systems.
a b ) Power supply
February 4, 2008 7:29:06 PM

I have this issue when my PC comes out of sleep mode. For now i kind of ignor it and continue with what ever i want to do with my pc.

During normal power-up i dont encounter this problem or message from my video card:

Video card : BFG 8800GT OC 512
CPU: Q6600:
MB: Asus Maximus Formula
Memory: 4 GBYTE
PS: ANTEC QUATRO 850: I know i have sufficient PS for this card.
OS: Vista Ultimate 64

Any recommendation?


February 4, 2008 8:26:14 PM

chunkymonster said:
While your PSU may have enough WATTS to power your rig, the EVGA 8800GTS 512 requires at least 26amps on the 12v rail. Per the EVGA website...

According to Corsair, the 620HX only has 18amps available on the 12v rail.

Sorry Charlie, your PSU is inadequate. That's a tough lesson to learn but it helps to research these things before making the purchase.


FYI - the 3 rails are not fully independent, the maximum output rating is ~50 Amps at 12v (620 W). 18.5 amps/rail is the ATX PSU standard to prevent wire overheating.

homerdog said:
This just goes to show you that someone can have well over 1000 posts and still not know WTF he is talking about :sol: 


No sh*t :sarcastic: 
February 4, 2008 8:42:32 PM



[b said:
badgtx1969 wrote :

FYI - the 3 rails are not fully independent, the maximum output rating is ~50 Amps at 12v (620 W). 18.5 amps/rail is the ATX PSU standard to prevent wire overheating. ]

badgtx1969 wrote :

FYI - the 3 rails are not fully independent, the maximum output rating is ~50 Amps at 12v (620 W). 18.5 amps/rail is the ATX PSU standard to prevent wire overheating.
[/b]


Uhh, NO! Wrong answer!



These are the so-called "multiple power supply rails". They are not
fully independent; they are all connected to a single high-current 12V
source inside the power supply, [b said:
but have separate current limit
circuitry. The current limit groups are documented so the user can
avoid placing too many high-current loads in the same group.
]


These are the so-called "multiple power supply rails". They are not
fully independent; they are all connected to a single high-current 12V
source inside the power supply, but have separate current limit
circuitry. The current limit groups are documented so the user can
avoid placing too many high-current loads in the same group.

[/b]
Seperate current limit circuitry. Hmmm, I wonder what that means?!




This just goes to show you that someone can have well over 1000 posts and still not know WTF he is talking about
said:



This just goes to show you that someone can have well over 1000 posts and still not know WTF he is talking about

Assume I have a 200amp 120/240v 1phase service running from the street to my house. By your logic, a single circuit running thru my house can pull all 200amps from the electrical panel. But I'm pretty EFFING sure that the 15AMP CIRCUIT BREAKER will trip and/or fail by "breaking" the circuit when that circuit gets over 15amps of load.

Whatever...stoopid newbs...FAIL!

To the OP, be careful with what some folks here tell you. Nothing replaces taking the few extra minutes to research the topic. Google and Wikipedia are quick and easy ways to get 80% of the answers you need.
February 4, 2008 11:51:41 PM

chunkymonster said:
To the OP, be careful with what some folks here tell you.

You got that right ;) 

To the OP: Single 12V rail PSUs are preferable, but a well made multi-rail unit such as the Corsair HX620 is perfectly fine. In fact, the HX620 is more than fine; it is one of the highest quality PSUs on the market.
February 5, 2008 1:07:46 AM

homerdog said:
You got that right ;) 

To the OP: Single 12V rail PSUs are preferable, but a well made multi-rail unit such as the Corsair HX620 is perfectly fine. In fact, the HX620 is more than fine; it is one of the highest quality PSUs on the market.

Other than the obvious manufacturing problems stated in op, sure.
February 5, 2008 2:44:39 AM

joseph85 said:
Other than the obvious manufacturing problems stated in op, sure.

You might want to reread the original post.
February 5, 2008 11:44:28 AM

if this kind problem only happen with corsiar 620 and ANTEC QUATRO 850, I believe it is design problem, since both psu are made by seasonic.
when OP find out the problem, the video card is on 2D mode. so the current it draw is way less than it running on 3d mode. so "the psu is inadequate" does not mmake any sense. The video card must has a sensor to reading the voltage from the psu, if the senor does not read correct data on time, it will give the error.
I just read the whole OP post. it is clear a timing problem.
February 5, 2008 12:13:39 PM

chunkymonster said:
Uhh, NO! Wrong answer!

Seperate current limit circuitry. Hmmm, I wonder what that means?!


Assume I have a 200amp 120/240v 1phase service running from the street to my house. By your logic, a single circuit running thru my house can pull all 200amps from the electrical panel. But I'm pretty EFFING sure that the 15AMP CIRCUIT BREAKER will trip and/or fail by "breaking" the circuit when that circuit gets over 15amps of load.

Whatever...stoopid newbs...FAIL!

To the OP, be careful with what some folks here tell you. Nothing replaces taking the few extra minutes to research the topic. Google and Wikipedia are quick and easy ways to get 80% of the answers you need.


I have the one down the Corsair 520 HX and the exact same Graphics card, with a E6750, GA-DS3L 2gb Crucial Ram etc etc.. and have no problems what so ever. I highly doubt that the PSU is at fault..UNLESS it's faulty. My machine runs flawlessly and has a nice big overclock on it. I would say follow the info you've been given and either RMA the PSU or Card and see if it resolves the problem.
February 5, 2008 12:24:11 PM

ZOldDude said:
Quote:
Usually this problem can be solved immediately by connecting the PCI-E cable to another one of four different PCI-E connector slots on the ST1000.

This is why I pay the extra money for a well regulated single rail PSU and only build with PC Power & Cooling.
Even the $109 USD 610 puts out a -constant rated- 49 Amps on the 12V line with it all being usable unlike multi rail systems.


You do realize that the PCP&C 610w is virtually Identical to the Corsair 620w.
They are both produced by Seasonic on the same platform as Seasonics own PSUs.
February 5, 2008 12:29:41 PM

homerdog said:
You got that right ;) 

To the OP: Single 12V rail PSUs are preferable, but a well made multi-rail unit such as the Corsair HX620 is perfectly fine. In fact, the HX620 is more than fine; it is one of the highest quality PSUs on the market.


Actually, the Corsair really is a single rail.
The box reads "3" rails, but in fact it's a single rail.

JonnyGuru has disected, analyzed, explained, and even confirmed with the manufacturer.
The Seasonic, Corsair(HX), and PCP&C in the 500-600 watt range are all built on the same platform and are just fine.
February 5, 2008 12:45:26 PM

zenmaster said:
Actually, the Corsair really is a single rail.
The box reads "3" rails, but in fact it's a single rail.

JonnyGuru has disected, analyzed, explained, and even confirmed with the manufacturer.
The Seasonic, Corsair(HX), and PCP&C in the 500-600 watt range are all built on the same platform and are just fine.


I was wondering about that. Nice to know. Pretty happy with my 620HX.
February 5, 2008 1:52:31 PM

Some hx620's work fine with the g92 cards but some dont.Only the g92 cards are affected by this problem.Corsair is currently working on a fix with nvidia so im sure its not just a problem of ampage.
February 5, 2008 2:54:07 PM

zenmaster said:
Actually, the Corsair really is a single rail.
The box reads "3" rails, but in fact it's a single rail.

JonnyGuru has disected, analyzed, explained, and even confirmed with the manufacturer.
The Seasonic, Corsair(HX), and PCP&C in the 500-600 watt range are all built on the same platform and are just fine.

That's good to know. Seems Corsair has been up to a bit of "creative labeling" to stay within Intel's recommended 240VA per rail limit :D 
February 7, 2008 8:02:59 AM

homerdog said:
That's good to know. Seems Corsair has been up to a bit of "creative labeling" to stay within Intel's recommended 240VA per rail limit :D 
Corsair isn't the only offender, it seems like the mislabeled single rail is the norm. I just looked at the JonnyGURU Antec True Power Trio 650W(Trio is for three rail) a little while ago and it is a single rail, as are the Earthwatts. Which is actually preferable IMO, but I wish they would label them correctly. All Corsair HX and VX series are single rail. The TX750 is also a single rail, but it is labeled correctly.

Jaywill's problem was well stated by him in the OP. It isn't a total current problem, it's that the card isn't waiting long enough for the current to be applied fully and thinks it's underpowered, as near as I can tell. If it was me I would take it back and get something else. It looks like Nvidia screwed the pooch this time.
February 7, 2008 8:52:34 AM

No, I said it was a single rail, which is actually a good thing. It's just that they mislabeled it, calling it a triple rail. Why don't you click on the link in my post above and read the review. That should put your mind at ease.
February 7, 2008 11:42:27 AM

Zorg said:
Corsair isn't the only offender, it seems like the mislabeled single rail is the norm. I just looked at the JonnyGURU Antec True Power Trio 650W(Trio is for three rail) a little while ago and it is a single rail, as are the Earthwatts. Which is actually preferable IMO, but I wish they would label them correctly. All Corsair HX and VX series are single rail. The TX750 is also a single rail, but it is labeled correctly.

I was aware that most PSUs that claim to be multi rail are actually single rail with OCP on a per rail basis, but I'm surprised to see that in some cases there isn't even per-rail OCP! Maybe the marketing departments hadn't gotten wind of the shift away from multi-rail when these models were released.
February 7, 2008 5:25:17 PM

ah i see, ok thanks
February 7, 2008 11:11:30 PM

homerdog said:
I was aware that most PSUs that claim to be multi rail are actually single rail with OCP on a per rail basis, but I'm surprised to see that in some cases there isn't even per-rail OCP! Maybe the marketing departments hadn't gotten wind of the shift away from multi-rail when these models were released.
Right it used to be that 99% were really single rail with OCP (Over Current Protection) making multiple pseudo rails. Now it's not some cases of no OCP per "rail", but many/most cases. The marketing department knows better and should be slapped hard.

The reason that they are doing this is because the new VGAs are such power hogs and they want to be able to power them without shutting down due to the OCP on an overloaded "rail". So it makes perfect sense. I bought my Silencer610 because it was a single rail and is not modular.

Also, AFAIK a dual rail PSU, to be ATX compliant, needs one rail to be dedicated to the ATX12V connector, so this "locks" up unused current that would otherwise be usable.

It's safe to say that ATX12V compliance is out the window.
February 7, 2008 11:26:20 PM

Quote:
The official ATX 2.0 standard suggests it for safety reasons, but some manufacturers DO NOT HAVE IT. Nor is it a requirement (with properly gaged wire).
I agree, but they are obligated to the consumer, and probably by law, to put an accurate specification sticker on the outside of the case.
a b ) Power supply
February 9, 2008 4:24:49 PM

chunkymonster said:
While your PSU may have enough WATTS to power your rig, the EVGA 8800GTS 512 requires at least 26amps on the 12v rail. Per the EVGA website...




According to Corsair, the 620HX only has 18amps available on the 12v rail.




Sorry Charlie, your PSU is inadequate. That's a tough lesson to learn but it helps to research these things before making the purchase.

Inadequate in that it can run a 939 Dual core rig, a pair of 8800GT Alpha Dogs in SLI along with five HD's, DVD/RW and five case fans (all at 100% due to no controller) quite comfortably without even getting warm. Sorry chunkymonster, but your quote really does sum it up. :pfff: 
!