Multi Rail Protection Question

[boghubodaghi]

What's all the fuss about multi rail protection systems lately? I don't understand why they are an ultra important feature even in high end PSUs. Sure, yes it is a feature that can save you from certain cases, if you don't know what you are doing. What are the chances that one cable draws more than 60A if you don't do anything weird? I think it's very slim. Moreover, 40A running in one 16 AWG cable is pretty bad already (and multi rail protection kicks in somewhere here). Can anyone change my mind on this? In my opinion, it is just a feature that is nice to have but not necessary. I think less ripple, better voltage regulation, better hold up times (if you get power outtages often), better build quality, should be put over multirail protection anyday.

[Bartholomew]

3 minutes ago, boghubodaghi said:

What are the chances that one cable draws more than 60A if you don't do anything weird? I think it's very slim.

 

Its usually not the user doing something weird, but some component failing, in the psu itself or anything in the other components.

 

Yes, chances are slim. Compared to many other things happening my chances of getting in a serious car accident are slim as well. However I do prefer a car with seatbelts, pretensioners, radar controlled braking and airbags and such.

 

Will the car drive without it? Sure just fine. But in the odd event of catastrophe, the occupants will have a better chance of survival.

 

That said, thats how i view psu protections "in general" in the specific protection you speak of, not sure, perhaps one of the guru's can answer more specifically . So not disagreeing, i just now wonder the same thing as you lol

 

 

[Guest]

^^^

Exactly.  It's not to prevent a user from trying to pull more current down a single cable.  It's to protect the system in the event that a short creates a load that draws too much current.

 

SCP doesn't work if the Output impedance is > 0.1 Ohm.  

[boghubodaghi]

Sorry, I never intended to sound like a smartass (if it feels so). Excuse my language, it's very early in the morning here.

 

But I still wonder. How slim would that be? Non specific car accidents are quite common, so maybe that's a bit irrelevant. But for multirail protection to be useful you have to have a situation that draws between 40A and the OPP or the 12V OCP (whatever that value might be). If it isn't the cable, then what component could potentially be at risk? Isn't 40A on any single motherboard line, or perhaps the MOSFETs on the VRM much worse than on a single 16 AWG cable? Heck, wouldn't it be any different sending 40A to a particular component than let say sending 200A (On 2000W+, single rail protected PSU)? Shouldn't both fry that component no matter what? 

 

To reiterate - I think it's a nice feature to have but the chance of it being useful (over the good old OCP, OPP, OTP, and SCP) is seemingly very low. I still wonder...

[Guest]

Like this:

 

 

 

 

 

 

 

 

 

 

 

[boghubodaghi]

You sure those are caused by OPP/12V OCP not triggering and would be saved by multi rail protection? Are they caused by good PSUs higher than 480W not having multi rail protection?

 

Or since you are the revered Guru Jon maybe you can give us a general idea. How often does this happen on RMx/RM PSUs and not RMi PSUs? 

[Guest]

16 minutes ago, boghubodaghi said:

You sure those are caused by OPP/12V OCP not triggering and would be saved by multi rail protection? Are they caused by good PSUs higher than 480W not having multi rail protection?

Whenever you see a burn on the component side, it's TYPICALLY because something shorted and the PSU kept feeding power to the short until the plastic melted, pins physically made contact and THEN SCP shut down the PSU.  A lower OCP threshold would prevent that from happening.

 

Of course, this can also happen due to a poor connection, but not typically.

[boghubodaghi]

22 minutes ago, jonnyGURU said:

Whenever you see a burn on the component side, it's TYPICALLY because something shorted and the PSU kept feeding power to the short until the plastic melted, pins physically made contact and THEN SCP shut down the PSU.  A lower OCP threshold would prevent that from happening.

How often is this happening long enough to cause catastrophic damage before one of the other protection trips (In case of non-multirail OCP PSUs)? How useless is OPP to not prevent anything other components frying? And if it is possible to save a component just by having a lower OCP triggering point, what would be saved? Isn't something that's shorted broken anyway? 

 

Yes, it is clear that I am somewhat moving the goalpost, but I think it's still relevant to the original question: Why is multirail protection a necessary feature? How often would it be necessary as to put it on top of other parameters that matters all the time? I always wondered. 

 

Also you keep mentioning SCP, but shouldn't OPP trip before SCP trips in the case of "feeding more power to problematic components"? OPP should have a smaller triggering point? 

[Guest]

59 minutes ago, boghubodaghi said:

How often is this happening long enough to cause catastrophic damage before one of the other protection trips (In case of non-multirail OCP PSUs)? How useless is OPP to not prevent anything other components frying? And if it is possible to save a component just by having a lower OCP triggering point, what would be saved? Isn't something that's shorted broken anyway? 

 

Yes, it is clear that I am somewhat moving the goalpost, but I think it's still relevant to the original question: Why is multirail protection a necessary feature? How often would it be necessary as to put it on top of other parameters that matters all the time? I always wondered. 

 

Also you keep mentioning SCP, but shouldn't OPP trip before SCP trips in the case of "feeding more power to problematic components"? OPP should have a smaller triggering point? 

I can't say it's terribly common.  Probably only happens to one out of a few hundred consumers.   But it's still better to have the safety there then not at all.

 

Unfortunately, OPP isn't always a substitution for OCP.  OPP is on the AC side.   It only turns off the PSU if the TOTAL POWER exceeds a certain amount.

[boghubodaghi]

8 hours ago, jonnyGURU said:

I can't say it's terribly common.  Probably only happens to one out of a few hundred consumers.   But it's still better to have the safety there then not at all.

 

Unfortunately, OPP isn't always a substitution for OCP.  OPP is on the AC side.   It only turns off the PSU if the TOTAL POWER exceeds a certain amount.

I think this comes back into my main concern here. Is it often enough to put it above anything else that does not have multirail OCP? It is better to have protection there, but is it enough to be more important than other bonuses that is beneficial all the time?

Consider the AX1000 vs. HX1000. Would the latter be the better option just because it has multirail OCP? Or should I get the former as it performs better, runs quieter/more efficient, etc?

 

If there's short somewhere, shouldn't the PSU provide a lot of power, that should be covered by OPP? I think this should be the case, especially for PSUs lower than 850W.

[seon123]

29 minutes ago, boghubodaghi said:

I think this comes back into my main concern here. Is it often enough to put it above anything else that does not have multirail OCP? It is better to have protection there, but is it enough to be more important than other bonuses that is beneficial all the time?

Consider the AX1000 vs. HX1000. Would the latter be the better option just because it has multirail OCP? Or should I get the former as it performs better, runs quieter/more efficient, etc?

 

If there's short somewhere, shouldn't the PSU provide a lot of power, that should be covered by OPP? I think this should be the case, especially for PSUs lower than 850W.

The HX would be the better option even if it didn't have multi rail OCP, because it has slightly lower ripple on 12V, longer hold up time, lower fan RPM at all loads, a fan that isn't crap, and a lower price. The AX1000 does have better regulation, though.

 

There is the possibility of having some sort of failure that isn't a complete short circuit. The user linked below had a failure in the motherboard's VRM, which did not trigger the PSU's SCP or OPP, but would likely have triggered multi rail OCP.

https://www.overclock.net/forum/31-power-supplies/944707-why-single-rail-not-better-than-multi-rail.html

[Bartholomew]

11 hours ago, boghubodaghi said:

Sorry, I never intended to sound like a smartass (if it feels so). Excuse my language, it's very early in the morning here.

Can only speak for myself but i never took it that way no worries

 

11 hours ago, boghubodaghi said:

 

But I still wonder. How slim would that be? Non specific car accidents are quite common, so maybe that's a bit irrelevant.

Car accidents are irrelevant but the analogy isnt.

 

And i think you underestimate the odds. Pcs are a set of components. What they all share is a psu connection, and most have additional circutry to filter, change or regulate power, consisting of more little components. Lets say that all adds up to a 100 little thingies. Things do break, just read around on the forums here (i had a gpu die on me this week, before that a psu). Considering pretty much each household has 1 or more computers these days I think that on a day to day basis a lot more people have a computer component fail on them than that there are people seeing their airbags deployed

Admitting thats a wild guestimation out of thin air, but dont think thats unrealistic.

 

And not saying you might not have a point, its a additional protection, optional by consumer choice, and if its worth it to you only you (and everyone else) can decide for themselfes.

 

I view things from the "but what if it does happen" perspective. No airbags could result in my face and or ribs smashed by stuff. Lacking psu protections could potentionally result in fire. Odds of either are slim indeed but the impact of when it does happen can be life altering.

 

To be honest though, given the units mentioned above, when spending that kind of money id expect both all possible protections and things like ripple and such to be perfect, nothing less. Especially on high wattage usage, which likely will run multiple high powered components like multiple gpus (since otherwise theres no need for for the high wattage). With a perhaps $10k+ machine like that, and the high currents, id moooost definatly want that protection, no question.

 

 

 

[Juular]

That single-rail master race trend set by Ultra really needs to end. How hard it is to literally just get a supervisor IC that has multiple 12V OCP inputs ? Not to mention that many high-end PSUs already have ICs that support multi-rail OCP but don't use it. Yes, some users would get shutdowns if they use GPUs with improperly designed VRM\power draw profile (looking at you AMD Vega) but this is easily fixed by using two cables on different rails. And on the other end, some users wouldn't get their PCs fried after some unlucky shorts which can't be fixed otherwise. 1kW+ PSUs with single powerful rail marketing is just pure cringe.

[boghubodaghi]

3 hours ago, seon123 said:

The HX would be the better option even if it didn't have multi rail OCP, because it has slightly lower ripple on 12V, longer hold up time, lower fan RPM at all loads, a fan that isn't crap, and a lower price. The AX1000 does have better regulation, though.

Hmm... I guess the HX vs AX example wasn't the exact comparison that I want. I was only going to compare their efficiency/temperature performance but I guess the HX 1000 rather has competitive VR and ripple suppresion. But for the sake of discussion, let's say theoretically the AX has better ripple suppression, has the exact same fan curve, and has the same exact fan (both passes the 18ms+ hold up test, so I doubt an extra ~6ms would help in the case of HX1000). Or let's change the HX1000 to something like let's say the Dark Power Pro 11 1000. How much performance should you forgo for just multirail OCP? Because I would take AX 1000 over the BQ anyday. I hope you know what I meant.

 

The post on overclock.net also said it only very rarely helps, but one thing resonates with the article. The PC in question was broken in the first place and if multirail OCP was there it only seem to stop a bigger problem. With 750W PSUs or less multirail OCP wouldn't make a difference, and only in 1000W this already very rare occurance is a major problem. This is the point that I was trying to explain earlier.

  

1 hour ago, Juular said:

That single-rail master race trend set by Ultra really needs to end. How hard it is to literally just get a supervisor IC that has multiple 12V OCP inputs ? 

I have heard the "ultra setting up the story" (or was it PC P&C?) and no it's not the marketing that bothers me. Now that we have the PSUs to buy in front of us, we can't manually configure non multirail OCP'd PSUs to have them even though it's "easy" as you claimed. The question is a buying decision question. How important is the feature and would you rather have it but have worse performance in anything else (on 850W+ PSUs)? I wonder if it's an overcorrection of that ultra marketing story.

[Juular]

@boghubodaghi So basically what you want to ask is 'at which point the performance difference between two PSUs would be more important than having multi-rail OCP' ?

I'd say, with 1000W+ PSUs as long as it's at least decent, i.e good sub 2% voltage regulation in crossloads, sub 50mV ripple, has all protections (aside of multi-rail OCP) - multi-rail would be more important than having better overall electrical performance. Because slightly worse electrical performance wouldn't hurt your system stability or components longevity but if you'll end up having short, very high OCP tripping point wouldn't save your equipment from frying. With 750-1000W PSUs it's not that definite, i'd probably rather get smth of the level of Corsair RMx\Seasonic Focus vs smth like Corsair CX if it were multi-rail. With even lower, sub 650W wattage multirail OCP becomes pretty much irrelevant.

[boghubodaghi]

15 minutes ago, Juular said:

@boghubodaghiSo basically what you want to ask is 'at which point the performance difference between two PSUs would be more important than having multi-rail OCP' ?

I'd say, with 1000W+ PSUs as long as it's at least decent, i.e good sub 2% voltage regulation in crossloads, sub 50mV ripple, has all protections (aside of multi-rail OCP) - multi-rail would be more important than having better overall electrical performance. Because slightly worse electrical performance wouldn't hurt your system stability or components longevity but if you'll end up having short, very high OCP tripping point wouldn't save your equipment from frying.

I guess that's the question that I was trying to make. Should this imply I should go with the Dark Power Pro 1000 over the AX 1000? If you'd pick the BQ you'll have worse performance in anything else but it's "ok" since it has multirail OCP. Plus, I wonder, why would in the very unlikely case that you get a short an OCP would save your equipment, if the component is broken anyway? What is being saved here (apart from the component not frying)? All these thoughts still makes me adamant that the AX 1000 is the better option. Efficiency, acoustics, VR, ripple supression seems like a better thing to have than multirail OCP.

 

Unless you have the option of something as competitive as the HX 1000, I agree that the HX is a no brainer.

[Juular]

14 minutes ago, boghubodaghi said:

Should this imply I should go with the Dark Power Pro 1000 over the AX 1000?

Assuming Dark Power 10\11 they're pretty good PSUs, definitely yes, OCP would be more important here.

Every protection is a set of 'ifs'. In the case of multi-rail OCP it only prevents any further damage to the PC component from some failed part, making simple repair by replacing that part more possible. And that's only about failed part, maybe it's a short in SATA connector ? Question here isn't why or when multi-rail OCP becomes important but why it's still not required by standard or not implemented as wide in high wattage PSUs by the industry as it should ? Because if we're going to make analogies, why keep a fire extinguisher in your car ? What are the odds of your car catching fire ? Yet you don't want to be helpless if that happens. Sure, a small handheld fire extinguisher wouldn't help in all situations but it's still better than nothing. This is also a matter of price of course, obviously if single-rail 1kW PSU costs like 50$ less than multi-rail one assuming they're both good enough it makes little sense to pay that much.

[boghubodaghi]

24 minutes ago, Juular said:

Assuming Dark Power 10\11 they're pretty good PSUs, definitely yes, OCP would be more important here.

 

In the case of multi-rail OCP it only prevents any further damage to the PC component from some failed part, making simple repair by replacing that part more possible. And that's only about failed part, maybe it's a short in SATA connector ? Question here isn't why or when multi-rail OCP becomes important but why it's still not required by standard or not implemented as wide in high wattage PSUs by the industry as it should ? Because if we're going to make analogies, why keep a fire extinguisher in your car ? What are the odds of your car catching fire ? Yet you don't want to be helpless if that happens. Sure, a small handheld fire extinguisher wouldn't help in all situations but it's still better than nothing.

If prices are equal (I believe their MSRP are quite similar), this implies that multirail OCP is more important than the better performance the AX 1000 has. What about the acoustics, and half less ripple, and so on? Isn't that a good tradeoff considering what the multirail OCP potentially will prevent? It seems to me that you're as helpless with multirail OCP as you are without when you have a broken component already (strictily talking about 1000 W PSUs here).

 

But I think we can agree that it's a more complex issue than just "ohh so you want your components burning" type of discussion. Prices, the actual performance, warranty of PSUs, and so on do matter on the equation. Would it be fine if I disagree with you (or with the opinion of most PSU gurus here?)? I think it's fine.

[Juular]

15 minutes ago, boghubodaghi said:

What about the acoustics

That might be more important for some people but DP11, Corsair AX and Corsair HX have very similiar noise profile.

15 minutes ago, boghubodaghi said:

and half less ripple

When it gets to numbers like 20-30mV vs 10-15mV the difference is quite minor really.

Transient response might be more important and in this regard bq! DP 11 is a bit behind Corsair AX & HX but i'm not sure to which degree it is important, Aris seems to be okay with it's transient response performance. And when it comes to Corsair AX vs HX, the difference here is even smaller.

[boghubodaghi]

59 minutes ago, Juular said:

That might be more important for some people but DP11, Corsair AX and Corsair HX have very similiar noise profile.

When it gets to numbers like 20-30mV vs 10-15mV the difference is quite minor really.

Transient response might be more important and in this regard bq! DP 11 is a bit behind Corsair AX & HX but i'm not sure to which degree it is important, Aris seems to be okay with it's transient response performance. And when it comes to Corsair AX vs HX, the difference here is even smaller.

The AX 1000 should be a lot quieter below 900W, it runs noisy in the 100-110% because the fan is not the best and not configured as best as it can. HX 1000 is about the same. I think the AX 1000 is better in every way than the BQ except for multirail OCP with similar MSRPs. I feel like it makes more sense to go with the AX 1000.

[Juular]

6 minutes ago, boghubodaghi said:

The AX 1000 should be a lot quieter below 900W, it runs noisy in the 100-110% because the fan is not the best and not configured as best as it can. HX 1000 is about the same. I think the AX 1000 is better in every way than the BQ except for multirail OCP with similar MSRPs. I feel like it makes more sense to go with the AX 1000.

https://www.techpowerup.com/review/be-quiet-dark-power-pro-11-1000w/6.html

https://www.tomshardware.com/reviews/corsair-ax1000-80-plus-titanium,5966-2.html

https://www.tomshardware.com/reviews/corsair-hx1000-psu,5214-5.html

 

Technically it is, but anything sub 20dBa you're unlikely to notice over ambient room noise and other PC parts noise output. And bq! DP11 graph isn't very granular as it's quite old review.

[boghubodaghi]

1 hour ago, Juular said:

Technically it is, but anything sub 20dBa you're unlikely to notice over ambient room noise and other PC parts noise output. And bq! DP11 graph isn't very granular as it's quite old review.

The old colours that Aris was using are different. Would you still choose the BQ?

[Juular]

18 minutes ago, boghubodaghi said:

The old colours that Aris was using are different.

No, it's not colors, he just didn't seem to care for sub 25dBa values because that's already pretty quiet but so don't know exact values because of that but i reckon that it should be pretty much on par with both Corsairs there.

19 minutes ago, boghubodaghi said:

Would you still choose the BQ?

If it's cheaper - yes. Otherwise HX > AX > DP11.

[Guest]

8 hours ago, seon123 said:

The HX would be the better option even if it didn't have multi rail OCP, because it has slightly lower ripple on 12V, longer hold up time, lower fan RPM at all loads, a fan that isn't crap, and a lower price. The AX1000 does have better regulation, though.

I agree.  The HX is much better than AX in so many ways. The only thing that's better about the AX is the efficiency (Titanium vs. Platinum).

6 hours ago, Juular said:

That single-rail master race trend set by Ultra really needs to end. 

Umm.... It wasn't Ultra.  It was PC Power & Cooling.

 

When graphics cards started needing the PCIe power connector, PC Power just slapped the connector on one of the existing rails instead of adding another rail.  It's like they never even bothered testing it before shipping it!  Then, after they ended up with a huge number of RMA's, they decided the quick fix was to just pull the OCP off altogether (easier to do than adding a rail) and then put a marketing spin on why they did it.

 

The most insulting marketing for single +12V rail came from XFX.  Man... That was some hot garbage double talk there!  https://www.jonnyguru.com/blog/2012/01/23/xfx-pro-1250w-black-edition-power-supply/

[Falkentyne]

13 minutes ago, jonnyGURU said:

I agree.  The HX is much better than AX in so many ways. The only thing that's better about the AX is the efficiency (Titanium vs. Platinum).

Umm.... It wasn't Ultra.  It was PC Power & Cooling.

 

When graphics cards started needing the PCIe power connector, PC Power just slapped the connector on one of the existing rails instead of adding another rail.  It's like they never even bothered testing it before shipping it!  Then, after they ended up with a huge number of RMA's, they decided the quick fix was to just pull the OCP off altogether (easier to do than adding a rail) and then put a marketing spin on why they did it.

 

The most insulting marketing for single +12V rail came from XFX.  Man... That was some hot garbage double talk there!  https://www.jonnyguru.com/blog/2012/01/23/xfx-pro-1250w-black-edition-power-supply/

I saw the PCP&C guys at the Pomona Computer Show in something around 2005, I think?  Or was it 2004...I don't remember.  They were doing some sort of demo.  I remember their certification thing.  I didn't know the 12v debate or anything about it.  And I couldn't afford their PSU.  I bought some sort of Antec PSU....or was it Enermax?  I forgot. Maybe it was an enermax...and then grabbed a Lian-Li midtower....