OC3D Forums

OC3D Forums (https://forum.overclock3d.net/index.php)
-   Cases & Power (https://forum.overclock3d.net/forumdisplay.php?f=26)
-   -   What Advantages Does Great Voltage Regulation Have? (https://forum.overclock3d.net/showthread.php?t=52072)

alpenwasser 27-03-13 08:38 AM

What Advantages Does Great Voltage Regulation Have?
 
Hello everyone,


For the past few years I have spent dozens of hours scouring the web for PSU reviews (mostly just out of curiosity), and while I'm more or less clear (I think) about why you need low ripple and good transient responses, I have not found much useful information on voltage regulation.

To clarify a bit: I am not so much interested in out-of-spec behavior. Naturally, in those cases things go bad rather quickly, from instability, refusal to start to actual component damage.

What I'm interested in much more is the difference in practice between somewhat below average (but still completely in spec) and phenomenal voltage regulation. What advantage(s) does great voltage regulation actually give you as opposed to just something halfway decent?

Since all thorough PSU reviews I've read place a rather high importance on this (meaning great and not just in-spec voltage regulation), I presume it has to be quite important.

NRG! 31-03-13 09:23 PM

When you say voltage regulation, don't you mean ripple? as in: better voltage reg = lower ripple.

Potentially better component lifespan, higher stable overclocks, etc.

OCZ ZX 850w:
my gtx 680 ltg needed +50mV on memory voltage to reach it's max o/c (forgot what it was now)

Corsair AX1200i:
same card doesn't need any memory over voltage to reach max memory o/c

(680 ltg pulls memory power through pci-e 8pins, not pci-e bus)

Rich_ 31-03-13 09:26 PM

Better voltage regulation = less significant 'power spikes' to your components. Theres a select few 'high end' 800+w psu's, i believe Tom did a video on one, and these PSU's will actually kill components in a matter of weeks because of the power spikes.

alpenwasser 01-04-13 09:25 AM

Quote:

Originally Posted by NRG! (Post 629568)
When you say voltage regulation, don't you mean ripple? as in: better voltage reg = lower ripple.

Ah, yes, my apologies, should have clarified better.

I didn't actually mean ripple (I think), what I meant could more accurately be described as "how close are the actual voltages to their nominal values over the power spectrum from idle to full power".

So, for example, when drawing and providing little power (idle), the +12 V rail might supply 12.1 V, and when on full power, the 12 V rail might actually be only 11.9 V.

As an example of what I mean, see the graphs titled "Load Regulation" on this review.

I did indeed watch Tom's vid where the tested PSU blew up, and jonnyguru actually does articles from time to time where they go out and buy a cheap PSU just to see if they can blow it up by testing it to its full specs :lol:

What I conclude from both your answers and my previous knowledge is this:
  • A PSU which provides voltages too high above spec (say, 12.8 V on the 12 V rails :o ) may kill components, or at least shorten their lifespan significantly, same goes for serious ripple and spikes during transient loads (I suspected this already, but I'm writing it here for completeness).
  • A PSU which provides voltages below spec (e.g. 11.8 V on 12 V rails), while not harmful to components, might require more voltage tweaking on your components to reach a certain overclock (for example, the graphics card NRG! mentioned). If so, this is the answer to the question I actually meant to ask (clever me ;) ).

If I've misunderstood anything, I would be grateful for any corrections/clarifications. Thank you both for your help!

Christoi 03-04-13 04:51 PM

From what I've seen, at lot of these tests are to make sure that the power supply doesn't damage the components it's connected to, which can be a problem with low-end units. If the unit is within the ATX spec, it's generally considered to be fine, though naturally lower is better.

By reviews, it's often used to analyse the quality of the output current and compare it with others. Even a lot of low-end units (eg: CX-series, Antec VP-series, CM GX-series) are within safe levels, so most reviews aren't actually testing whether the unit is safe or not. Rather, it's more of a quantitative way of comparing different units.

Do be aware however, that it doesn't really say much about the actual quality and longevity of the unit itself. Units with better regulation tend to have longer lifespans, but it's not a direct trend. A good example of this is the digital power control of the AX-i series, the added circuitry allows for insanely accurate voltage regulation, but that alone does absolutely nothing to the reliability of the unit.

Also, ripple tends to be what can affect stability, as it's about very small deviations of voltage (usually ~0.05v). If this is high enough components like the CPU may suffer given the fluctuating voltage, causing stability issues. Similarly with voltage regulation however, past a point, it really doesn't make much of a difference.

Personally, while I do use this as a point to see how good a unit may be, I really don't see the quality of the output current as much of an issue in a lot of cases. In my opinion, the larger reason to buy a high-end, expensive power supply is the added reliability and longevity, which is achieved by using higher-quality components.

alpenwasser 03-04-13 05:38 PM

Quote:

Originally Posted by Christoi (Post 630372)
From what I've seen, at lot of these tests are to make sure that the power supply doesn't damage the components it's connected to, which can be a problem with low-end units. If the unit is within the ATX spec, it's generally considered to be fine, though naturally lower is better.

By reviews, it's often used to analyse the quality of the output current and compare it with others. Even a lot of low-end units (eg: CX-series, Antec VP-series, CM GX-series) are within safe levels, so most reviews aren't actually testing whether the unit is safe or not. Rather, it's more of a quantitative way of comparing different units.

Yeah, that's generally the impression I got as well.

Quote:

Originally Posted by Christoi (Post 630372)
Do be aware however, that it doesn't really say much about the actual quality and longevity of the unit itself. Units with better regulation tend to have longer lifespans, but it's not a direct trend. A good example of this is the digital power control of the AX-i series, the added circuitry allows for insanely accurate voltage regulation, but that alone does absolutely nothing to the reliability of the unit.

Indeed. I had a professor who was very adamant about keeping systems simple: "Every additional component adds at least two possible points of failure: Itself and any interface/connection to the remaining system. So any advantages it adds must be weighed against that very carefully.", is what he used to say.

Quote:

Originally Posted by Christoi (Post 630372)
Also, ripple tends to be what can affect stability, as it's about very small deviations of voltage (usually ~0.05v). If this is high enough components like the CPU may suffer given the fluctuating voltage, causing stability issues. Similarly with voltage regulation however, past a point, it really doesn't make much of a difference.

Ripple is very important in control engineering, of which I've had two semesters, but naturally we never used it in context with a computer :)

Quote:

Originally Posted by Christoi (Post 630372)
Personally, while I do use this as a point to see how good a unit may be, I really don't see the quality of the output current as much of an issue in a lot of cases. In my opinion, the larger reason to buy a high-end, expensive power supply is the added reliability and longevity, which is achieved by using higher-quality components.

This is why I went for the Enermax Platimax 1200. Voltage control (specifically, voltage drop over the power spectrum) is, while totally within spec, not exactly phenomenal (the remaining performance criteria are very good from the reviews I've read; excellent ripple control, transient response etc.). However, the build quality and component choice are absolutely stunning, and I sort of have a fetish for that :lol:

Thank you for your answer!


All times are GMT. The time now is 09:00 PM.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2020, vBulletin Solutions, Inc.