Jump to content
Corsair Community

AX1200i OCP - does it protect against a short?


BorisTheSpider

Recommended Posts

I have asked this elsewhere, but I wanted to make a separate thread for it so it didn't get mixed-up with other issues or cause confusion.

 

I asked earlier, in http://forum.corsair.com/v3/showthread.php?p=606307 whether or not it was necessary for the Corsair Link software to be running for multi-rail mode on the AX1200i to be active. RAM GUY confirmed that the software does have to be running.

 

He also said:

 

If you start the system with another O.S. other than Window's the PSU will default to its profile as defined by the factory defaults.

 

Now I'm concerned about something I've been seeing. I have been having numerous false trips of OCP because of erroneous current readings in the software.

 

When OCP has been tripping in error for me, rather than powering off, the PSU has been resetting the PC, so current continues to flow - it just stops momentarily, and the PC resets as though the reset button on the case had been pressed.

 

I want to clarify if I'm getting something wrong here, because my understanding at present of what would happen if a real overcurrent happened (a short for example) is this:

 

1. Overcurrent happens (something shorts out)

2. OCP triggers

3. PC resets

4. OCP is now not active (corsair software not running)

5. 100+ amps goes through the short

 

Have I got something wrong here?

 

Is the AX1200i supposed to shut off when OCP trips, or is it supposed to reset like I'm seeing?

Link to comment
Share on other sites

OK, I have something to add that may be relevant to why this is happening.

 

I shut down to triple-check that I had the "Off after AC power loss" setting switched correctly in the BIOS, so that power would stay off. It was set correctly as I had recalled.

 

Then I pulled the plug for a few seconds, reinserted and the system stayed off.

 

Then I powered on, pulled the plug quickly, and reinserted it before the power had drained from the motherboard/psu (the few seconds when the LEDs etc. stay on after wall power has been switched on).

 

When the plug is reinserted before power has fully drained from the system, it just powers back on.

Link to comment
Share on other sites

This is not a double post.

 

This thread is about an entirely separate issue.

 

You are referring to http://forum.corsair.com/v3/showthread.php?t=111660 where you said you are looking into reports of the current readings being incorrect.

 

This thread is about the fact that multi-rail OCP does not appear to do anything useful. If it trips, it resets the PC, and hence disables multi-rail OCP.

 

Is that, or is it not the case?

Link to comment
Share on other sites

  • Corsair Employee
The ATX design provides for Short detection but only when in soft off and before the system is powered on. If something in the circuit shorts after the power in initiated the full power of the respective Rail will be available. So yes there is short protection but it is comprised of the MB and PSU working together.
Link to comment
Share on other sites

So are you saying that if a short did occur, that the OCP would trip, that would reset the system, and that there would then be short circuit protection in place?

 

I'm still confused as to why the OCP tripping is resetting the system rather than turning it off - surely if multi-rail OCP is to have any benefit at all it has to turn the system off, not reboot it????

Link to comment
Share on other sites

  • Corsair Employee
So are you saying that if a short did occur, that the OCP would trip

Yes and it should throttle down the rail that is drawing too much power, and send the shut down signal the MB.

But what you have described sounds like there is or may be some other issue with your system. Can you test another PSU on your system?

Link to comment
Share on other sites

Yes and it should throttle down the rail that is drawing too much power, and send the shut down signal the MB.

But what you have described sounds like there is or may be some other issue with your system. Can you test another PSU on your system?

 

Well I don't know what I could test - the only reason OCP has tripped for me is because of the faulty/spurious current readings that I've been getting (along with the other users that have reported the same faulty readings). If I tried another PSU, it's overcurrent protection wouldn't trip (because there is no real overcurrent happening).

 

The only option I can think of that would help would be to put my PSU in a different system, then if that exhibited the faulty current readings, make it trip the OCP and see if it reset, or shut off. Unfortunately, I don't have access to a system I could do that on.

 

I also get the impression that the multi-rail type of OCP in the AX1200i isn't happening in the PSU itself (as the overall 100 amp OCP does), but rather is done through the link software. The reason I get that impression is because the multi-rail OCP doesn't work if the link software isn't running, so I am guessing that what happens is that if the link software reads a current higher than what is set as the OCP trip point for the rail, it does something in software (presumably via the USB link dongle) to tell the PSU to shut down - is this correct?

 

If we could determine exactly what is supposed to happen when the multi-rail OCP is tripped by the link software (eg. what is it supposed to do exactly in order to shut the PSU down, does it tell the PSU to break the pin-16 PS_ON circuit?) that would help, I think, to track down whether this could be my system, or whether I perhaps have a faulty PSU.

 

I've asked in another thread if a user experiencing faulty readings could try setting an OCP trip point to force OCP to kick in and see if the system rebooted (like I am seeing) or shut down (like I expect it should), but I didn't get a reply. The thread is this one: http://forum.corsair.com/v3/showthread.php?t=111660 where jan54 referred to getting odd power readings, but I don't know if he's actually seeing the same as me in terms of the PCIe rails over-reading, because his screenshots don't show that.

 

Have you been able to reproduce the stuck (overreading) pcie rails problem at Corsair? If so, perhaps you could set an OCP trip point and see if the test system reboots vs. shuts off?

Link to comment
Share on other sites

  • Corsair Employee

We have not seen this issue in our testing so no we have not duplicated the issue. I am waiting for more feed back from Engineering at this point.

And so far you are the only user reporting this issue so beyond that I would suggest using a different PSU to test the system.

Link to comment
Share on other sites

We have not seen this issue in our testing so no we have not duplicated the issue. I am waiting for more feed back from Engineering at this point.

And so far you are the only user reporting this issue so beyond that I would suggest using a different PSU to test the system.

 

Well that's odd, because I found this video of a Corsair employee, George Makris, promoting the AX1200i to a reviewer, where the issue is happening.

 

 

Here's a screenshot from the video, where you can see over 56 amps reported on PCIe1 (over 670 Watts) while the total system power draw is reported at around 150 Watts.

 

http://i38.photobucket.com/albums/e119/bit_monkey/Exclusivehands-onwithCorsairsAX1200idigitallypowercontrolledPSU-YouTu_2012-11-07_23-36-02_zps24754b55.jpg

 

http://i38.photobucket.com/albums/e119/bit_monkey/Exclusivehands-onwithCorsairsAX1200idigitallypowercontrolledPSU-YouTu_2012-11-07_23-36-02_zps24754b55.jpg

 

If you watch the video, you can see that rail jumps around a lot, but is also stuck a lot at around 56 amps, and is also stuck a lot at 21 amps (still more than the total system power draw reported in the video). Here's a link straight to one point where there's a particularly clear close-up in the video:

#t=3m12s

 

Here's another video, of it doing it in front of another Corsair employee, PR Director Rick Allen

 

 

In fact, at 2:37 he is pointing to these faulty readings while discussing how you can see the PCIe current delivery:

 

#t=2m37s

 

http://i38.photobucket.com/albums/e119/bit_monkey/pointing_zps2626a78a.jpg

 

Did no-one notice that a single PCIe cable was supposedly carrying 670 Watts?

 

As you can see in this screenshot at 2:28, it's doing it on PCIe1, PCIe2 and PCIe3 with 55.8 Amps reported on each (a total of over 2 kilowatts) while total power output reads 174 Watts.

 

http://i38.photobucket.com/albums/e119/bit_monkey/Computex2012CorsairzeigtAX1200i-YouTube-MozillaFirefox_2012-11-08_00-21-23_zps1a870e39.jpg

 

Surely someone ought to be fetching a fire extinguisher?

 

 

These videos were posted in June, and I find it difficult to believe that your testing of a new product has been so limited that you hadn't noticed a glaring issue that was right in front of you during the filming of a promotional video 5+ months ago, nor can you seem to reproduce it now...

Link to comment
Share on other sites

  • Corsair Employee

Ok.. everyone stop....

 

OCP does NOT reset the computer!

 

If OCP trips, the PSU SHUTS DOWN!!!!!!!

 

The shut down is actually within the PSU. The only way to reset an OCP shut down in the PSU is to flip the switch on the back of the PSU off and then back on again. So even if your motherboard is set to "power on after power failure" set in the BIOS, it would not power back on or reset, unless you cycle the switch on the back of the power supply first.

 

I agree with RAMGUY: "...what you have described sounds like there is or may be some other issue with your system." RAM would be my first guess, but the best thing to do is make sure Windows is not set to "Automatically Restart" during a BSOD, which is something you would set in "Startup and Recovery" under "System Properties".

 

I repeat: a restart is NOT indicative of OCP, OPP, a short, etc. A complete shut down would be and nothing more.

Link to comment
Share on other sites

Jonny,

 

Just to clarify, let me make this absolutely clear - the system does not reboot any other time. I have had no reboots, crashes or any other problems with the system in the last couple of weeks since I turned off the OCP tickboxes in the Corsair Link software. None.

 

If I were to go into Corsair Link now, where I have a PCIe rail stuck reading 56.1 amps, and tick the box to apply 40 amp OCP on that rail, the system would instantly reboot, just like the reset button on the case had been pressed. Not shut down, not power off. Reboot.

 

This won't happen any other time, it _only_ happens when OCP is set in Corsair Link, and when a reading exceeds the set OCP trip point.

 

If the OCP should trip _inside_ the PSU, and shut it down hard so you have to cycle wall-power to it before it will come back on, how can my RAM, a windows setting or any other setting or hardware over-ride that? If the PSU did what you say it _should_ do, and cut the power completely, it would not be possible for it to reboot.

 

 

OCP does NOT reset the computer!

 

If OCP trips, the PSU SHUTS DOWN!!!!!!!

 

The shut down is actually within the PSU. The only way to reset an OCP shut down in the PSU is to flip the switch on the back of the PSU off and then back on again. So even if your motherboard is set to "power on after power failure" set in the BIOS, it would not power back on or reset, unless you cycle the switch on the back of the power supply first.

 

Exactly what I would expect. As you said, OCP should happen _inside_ the PSU. However, RAM GUY told me in http://forum.corsair.com/v3/showthread.php?t=111112 that Corsair Link has to be running for the multi-rail OCP trip points to be active, and said that in Linux, at the BIOS or any other time that Corsair Link isn't running, the PSU reverts to what he called "the default profile" which I took to mean it becomes 100 amps single-rail.

 

If the software has to be running for multi-rail OCP to work, that suggests that the multi-rail OCP does not happen entirely inside the PSU, but rather has something to do with the software.

 

Perhaps you are able to clarify exactly how the multi-rail mode on the i-series works, ie. what exactly "trips" the OCP - if I set a 30 amp OCP trip point in the software, and that rail detects 35 amps of current, what exactly does the detecting, and how exactly does the protection kick in - does the software have anything to do with tripping that "OCP", and if it doesn't, why must it be running as RAM GUY told me it must?

 

My guess (and it may be wrong) is that the link software detects the current exceeding the set trip point, and signals the PSU via the link dongle to activate OCP. Just a guess, but it's the best I can figure out based on what I've been told.

 

I agree with RAMGUY: "...what you have described sounds like there is or may be some other issue with your system." RAM would be my first guess, but the best thing to do is make sure Windows is not set to "Automatically Restart" during a BSOD, which is something you would set in "Startup and Recovery" under "System Properties".

 

Again, if OCP should happen _inside_ the PSU, and should hard power-off the system such that you can't restart it without cycling power to the PSU itself, then it's impossible for this to be related to a windows setting, or to my RAM isn't it?

Link to comment
Share on other sites

  • Corsair Employee

I didn't start at Corsair until after the AX1200i.

 

But I do know the AX1200i and the OCP works the same way as any other OCP. The only difference is that it's a DSP microprocessor versus a supervisor IC.

 

You set OCP in the link software, and this is set in the DSP, but if this OCP is tripped... like I said, and just like a PSU with a supervisor IC instead of an OCP.... THE PSU WILL HARD TRIP. OCP on ANY PSU doesn't work ANY OTHER WAY. The DSP shuts down the PSU. Not just the mobo via soft off.

 

The difference between a supervisor IC and DSP when it comes to settings for OCP, OPP, etc. is that the settings to the DSP can be dynamic. The settings for a supervisor IC are done at the factory. This does NOT, however, change how they react.

 

If the computer is doing a soft off or reset, the problem is something other than the PSU.

Link to comment
Share on other sites

Thanks for the reply.

 

It's a great help to have someone technical in terms of PSU design get involved. Hopefully you can help me get to the bottom of this, and I do appreciate you getting involved.

 

I am not meaning to be difficult here, just to clarify, but I do have to ask, especially since you said that you weren't involved with the AX1200i design - are you absolutely certain that the AX1200i, when the selectable OCP trips, turns off via the DSP in the "hard-off" manner you've described. I mean, we both agree that is what _should_ happen, but can you be absolutely certain about the implementation in the AX1200i?

 

Just to elaborate on the symptoms I am seeing, if I was to turn on the OCP tickbox on my over-reading PCIe rail, the PC would reset. It would reboot, and as soon as it got to Windows (either at the logon screen, or right after I entered a password, I can't exactly remember which) it would reboot again - right when the link software should load.

 

If I were to stop the boot (say by entering the BIOS) it would run fine for an indefinite period of time, until I let the boot continue, then as soon as it got to the point where the corsair software would load, it would reset again.

 

If I disconnected the link dongle from the PSU, it would stop resetting and I could then get into windows, untick the OCP tick-box, and reconnect the cable between the link dongle and PSU.

 

I've seen that exact, boot-loop behaviour several times when the PCIe rails have started misreading.

 

I mean, if that's not the PSU - what the hell is it?

 

Why would it stop when I disconnected the cable from the link dongle to the PSU if it wasn't the PSU!

 

Bear in mind especially, that it would stop if I disconnected the link dongle from the PSU. I wouldn't have to disconnect the link dongle's USB cable to the motherboard, merely disconnect the PSU from the link dongle.

 

I mean, not that I want to seem difficult, but that behaviour cannot be ascribed to anything other than the PSU surely.

 

If the computer is doing a soft off or reset, the problem is something other than the PSU.

 

Yeah, I'd like to think so, but as I said it _only_ happens when the selectable OCP trips. Always and reliably it will happen if I turn on OCP when one of the PCIe rails is over-reading, any other time it never happens.

 

And I do mean never. I haven't seen this behaviour once, other than when OCP has been selected on in corsair link with a rail over-reading such that OCP should trip.

 

- If multi-rail OCP is switched off, it won't do it.

- If multi-rail OCP is switched on, but there isn't a bogus reading that exceeds the selected trip point, it won't do it.

- If multi rail OCP is switched on, and it's been repeatedly rebooting (because of a bogus current reading), and I disconnect the cable from the link dongle to the PSU, it will stop rebooting.

 

Really, given those symptoms, it's tough to see how it can be anything other than the PSU.

Link to comment
Share on other sites

  • Corsair Employee

Ah! I didn't know about it coincide with the software running and/or the dongle being plugged in!

 

To answer your first question: You want OCP to shut down the PC instead of restart the PC in the case that the OCP is tripping do to a short. See, there really isn't such a thing as "SCP" by itself. SCP relies on the other protections, OCP included, to be in place. If there's a short, current rushes somewhere, overloads the OCP and the PSU shuts down to protect the computer.

 

If the PC is shutting down only when it hits Windows and Link2 has OCP set, then it's definitely an issue with the PSU, the dongle or the Link2 software. But you're not coming anywhere close to actually tripping the OCP set on the DSP within the PSU. I have an AX760i and have the OCP turned on on everything and have all the numbers turned down to 20A and I never trip, even with an OC'd i5, SLI and a third card for Physx. Furthermore, on my load tester if I have the Link2 software running on a computer and then intentionally overload the rail with my load tester, the PSU shuts down and all of the voltage readings on my PSU go to zero.

 

The first thing I would do is uninstall Link and make sure you have the newest version installed. If it does the same thing after that, maybe there's something wrong with the dongle, but I would just return the whole PSU because you really don't know if it's the dongle or the DSP reporting to the software via the dongle.

Link to comment
Share on other sites

Yes, it's already been replaced. Previous RMA was 1281311

 

That RMA was because of the faulty power readings on the PCIe rails, the same as is happening in the videos I linked to a few posts back.

 

Whether the first RMA'd unit exhibited the reboot (rather than shut-down) behaviour when the readings exceeded the OCP trip point set, I cannot recall - if it did, I don't think I noticed at the time.

Link to comment
Share on other sites

you're not coming anywhere close to actually tripping the OCP set on the DSP within the PSU.

 

So you think, that even though it is rebooting when the set OCP trip-point is exceeded (albeit by a bogus reading rather than a real overcurrent), that the reboot is not being caused by the OCP within the DSP, but rather by something else, perhaps in the software?

 

on my load tester if I have the Link2 software running on a computer and then intentionally overload the rail with my load tester, the PSU shuts down and all of the voltage readings on my PSU go to zero.

 

That is reassuring to know. At least it seems then that the behaviour with a real overcurrent is as it should be.

 

The first thing I would do is uninstall Link and make sure you have the newest version installed. If it does the same thing after that, maybe there's something wrong with the dongle, but I would just return the whole PSU because you really don't know if it's the dongle or the DSP reporting to the software via the dongle.

 

Yeah, I've reinstalled (including clean reinstall, removing the dongle driver and deleting the remnants of the Corsair directory in program files) numerous times.

 

PSU (and link dongle) has also been replaced in the previous RMA as detailed above. In any case, the videos linked to (the ones with George Makris and Rick Allen demo'ing the AX1200i at Computex) point to a general problem with the AX1200i (or software) rather than a particular faulty unit.

Link to comment
Share on other sites

  • 3 months later...
  • Corsair Employee

Sorry. I didn't realize you were waiting for a response.

 

So I did some more research. If there is an actual short, the PSU would shut down and stay shut down.

 

We were trying to figure out how long and what gauge wire would prevent OCP from working, so I actually tested this by putting a wire between +12V and ground and then powering up the PSU. In a split second, the PSU shut off and stayed off.

 

Now, if I put that short on a momentary switch and the motherboard is set to "turn on after power failure", the PC simply recycles and powers back up.

 

While the OCP for the individual +12V rails is programmed in Link and may not be enabled by default, the PSU has an overall OCP on the +12V rail that should shut the PSU down in the event of a short. That said, if the short is momentary and the motherboard is set to restart after power failure, the PC will boot back up again.

Link to comment
Share on other sites

Sorry. I didn't realize you were waiting for a response.

 

So I did some more research. If there is an actual short, the PSU would shut down and stay shut down.

 

We were trying to figure out how long and what gauge wire would prevent OCP from working, so I actually tested this by putting a wire between +12V and ground and then powering up the PSU. In a split second, the PSU shut off and stayed off.

 

Now, if I put that short on a momentary switch and the motherboard is set to "turn on after power failure", the PC simply recycles and powers back up.

 

While the OCP for the individual +12V rails is programmed in Link and may not be enabled by default, the PSU has an overall OCP on the +12V rail that should shut the PSU down in the event of a short. That said, if the short is momentary and the motherboard is set to restart after power failure, the PC will boot back up again.

 

Originally Posted by jonnyguruThe only way to reset an OCP shut down in the PSU is to flip the switch on the back of the PSU off and then back on again. So even if your motherboard is set to "power on after power failure" set in the BIOS, it would not power back on or reset, unless you cycle the switch on the back of the power supply first.

 

So since these two posts of yours contradict themselves, can we clarify - if the motherboard were set to restart after power failure, and a permanent short (your piece of wire) were put between two pins, will the PSU keep cycling on and off repeatedly, putting current through the short?

 

Also, my prior post was not necessarily aimed at you in particular, but more at the fact that absolutely nothing has been done about my problem (the misreading PCIe rails making multi-rail OCP useless) since this thread started, nor since I was contacted on OCN asking for details of my system so the problem could be troubleshooted.

 

As I posted on OCN, I was told on here that you (Corsair) had never seen the misreading PCIe rails problem before, then when I posted videos of it happening to Corsair employees (Rick Allen and George Makris) at product demonstrations, I was simply fobbed off with excuses to the effect that the problem has been fixed - clearly it hasn't since it is still happening to me on a second RMA'd PSU, and it seems no-one at Corsair knows how to fix it.

 

Has anything actually been done to find out why it's happening? It's obvious that it would be possible to replicate it, since it has been admitted that it happened on your own systems.

Link to comment
Share on other sites

  • Corsair Employee

Yes, they do contradict and I'm sorry for that. I actually did some testing before my last post to confirm some things, so please consider my latest post the one that is most factual. ;)

 

If there was an actual short on the PSU, like the wire, the PSU should stay off regardless of how the motherboard is set (turn on after power failure, etc.)

 

The OCP worked on my AX1200i and it's working now on my AX860i. But I haven't tested it for shorts. These are in my personal build and will only shut down the PC in the event of an overload.

 

I can suggest that you make sure you have the newest version of Link. It's on v2 now:

 

http://www.corsair.com/us/media/cms/drivers/CorsairLinkInstall_2.2.0.exe

Link to comment
Share on other sites

  • 1 month later...
  • Corsair Employee

Boris,

 

I've recently heard of an issue with Link and AI Suite running on the same PC. If you're running AI Suite, do you mind uninstalling Link and AI Suite and then reinstalling Link to see if the problem corrects itself.

 

I've done a few blog posts for Corsair using my personal system (all Gigabyte, not Asus so I can't really reproduce your problem), but if you look at my posts, you'll see that I'm not seeing the same problem you're seeing. (I have no idea what mobo George was using in that video, but if it's an Asus using AI Suite.. hey! Why not?)

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...