The Glade 4.0

Both Silverstones offered by CyberPower lack one of the most important features possible: their own power disconnect switches.

_________________

Why would you need a switch immediately adjacent to a removable power plug?

_________________
In time, this too shall pass.

Capacitors.

_________________

I don't know why it would make any difference, but I suppose that's just one more reason I should have majored in electrical engineering instead of switching to psychology after a semester.

Anyhow, I did notice that was a cricticism, but the reviewers did not seem to think it was a major issue. I chalk it up to "you can't have everything."

(I probably would have gone with Khross's SeaSonic recommendation had it been available, but it wasn't, and I just couldn't find a website that would give me the same value for the entire machine as Cyberpower. )

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

The capacitors in your power supply will hold a charge for about 3-5 minutes after you pull the plug. A power switch breaks the circuit between the plug and anything else on a good supply, so you don't zap yourself removing the cord or risk a discharge that spikes your motherboard under 0 load. (At least, that's how it was explained to me).

_________________

Ahh, ok. Yeah, I did know capacitors did that but then, I've never heard of this actually happening so I'm not too worried about it.

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

With respect to capacitor aging...

Bear in mind the source of the information. I would not trust Thermaltake's (or almost any other vendor's) PSU calculator to give you honest, realistic answers. I've got good money that says they're cooking the numbers to push you into buying an oversized, more expensive PSU. To be blunt, the PSU market is and always has been full of snakes. The advent of the Internet and its myriad review sites has made it more difficult for them to outright lie to your face about their PSU wattage ratings. But they're more than happy to invent new bullshit to convince you that you need a 5000W Megabeast(TM) PSU 9000+ Gold Edition.

Don't get me wrong, capacitor aging is a real thing, but the effect is:
1) logarithmic and
2) generally way overstated if you buy a quality PSU

Caveat emptor.

_________________
Sail forth! steer for the deep waters only!
Reckless, O soul, exploring, I with thee, and thou with me;
For we are bound where mariner has not yet dared to go,
And we will risk the ship, ourselves and all.

I suspected that was some of what was going on, since they want to "recommend" their power supplies to you based on the results. However, I did find that it gave me results that seemed to fall more or less in line with what the reviews were calling for.

It didn't appear from just eyeballing it that the capacitor aging was logarithmic, and that was the part that was most suspicious to me. I eventually went with the 1000 watt because I figured that hell, it's got a warranty anyhow, and if I have to replace the PSU in 3 or 5 years, no big deal.

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

A quick and dirty search came up with this link: http://engknowledge.com/power_supply_design.aspx

In a nutshell, here's what's going on. The top image is what comes out of your wall. The green line in the middle is 0. We run that through a series of diodes to make all of the voltages positive, causing the current to continually flow in the same direction. That's the second image.

Next, we put a capacitor across the load. Once the capacitor gets charged up, it runs the device (in this case, your computer) whenever the voltage from the wall is below a certain level. That's the purple line on the third image. Finally we put on voltage regulators, which basically protect the device from the spikes and dips that come along the line provided by your imperfect electric company.



If you'll look at the purple line again, you'll see it's wobbling up and down between the peak voltage, and somewhere in the curve a little bit below that. Whenever the purple line is sloping down, that's when the capacitor is running your computer. When the purple line is following the yellow power signal up, the capacitor is charging while the electrical outlet is powering your device. When you pull the plug or turn your machine off, there is still electricity stored in that capacitor. When it falls below a certain level, the capacitor can't supply enough voltage to run your machine, but it can still discharge and damage components elsewhere in the network.

The chance for that to happen is pretty small, but Khross thinks it's high enough that he wants a switch to cut off his computer from that capacitor. Considering that such switches are cheap, and computers generally cost over a thousand dollars, Khross is perfectly reasonable to ask for one.

Now, onto capacitor aging:

http://www.johansondielectrics.com/technical-notes/general/ceramic-capacitor-aging-made-simple.html

I wouldn't particularly worry about this for a power supply. If you take a look back up at the graphs I posted, the capacitor is determining how far down that purple line dips before it hits the yellow line. That's the ripple voltage Khross referred to on the first page. We can't actually make a perfectly flat line. The graph you're looking at is highly exaggerated, and it's meant to provide a concept of what's going on. Ripple voltage is tiny.

As your capacitor ages and the dielectric degrades (not all capacitors experience this), the capacitance value goes down. As that happens, that purple line gets steeper. It doesn't make it as far across before it hits the yellow line and starts charging back up. In other words, the voltage ripple gets bigger.

Let's assume that losing 1% capacitance increases your ripple voltage by 1% (this is incorrect, but I don't have a better figure to pull out of my ***). Suppose your power supply loses 10% capacitance due to capacitor aging. The ripple voltage increases from 0.000010V to 0.000011V. Let's say you're running a iPhone that probably takes a 5V power supply (based on running off a USB port). That means you're old, shitty power supply is providing power between 4.99989V and 5.000011V.

In other words, capacitor aging doesn't totally destroy your power supply and the machine you run off of it. Furthermore, it's the dielectric that's degrading, and not the entire capacitor. There is a minimum value that your capacitor can not fall below.

_________________
Buckle your pants or they might fall down.

Thanks Corolinth, that was a good read.

_________________
"Dress cops up as soldiers, give them military equipment, train them in military tactics, tell them they’re fighting a ‘war,’ and the consequences are predictable." —Radley Balko

I feel smarter.

Very informative Coro, thanks.

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

My flux capacitor provides 1.21 gigawatts of electricity.

Now I you could only get it up to 88 miles an hour...

_________________
In time, this too shall pass.

re: Capacitor Aging

I tested this out in the lab last night. Cutting the size of the capacitor in half doubles your voltage ripple. In hindsight, I suppose I could have just done the math, but the equation for voltage ripple is based upon approximations that are only valid under certain conditions. Anyhow, Khross's power supply would go from 5µV to 10µV if the capacitor "aged" to half capacitance. Put into perspective, on the 3.3V terminal (where ripple voltage matters the most), Khross will have gone from a 3.299995-3.300005V range to a 3.299990-3.300010V range. That really doesn't look all that catastrophic. It gets better.

According to the Johanson Dielectric website I linked, it takes a Class IV capacitor 114 years for that to happen! A Class II capacitor takes much, much longer. Between one year and eleven years after purchasing the device, your capacitor is aging between 65% and 58% capacitance going by the capacitors with the fastest rate of decay. That's putting you between 1.54x and 1.70x the ripple voltage specified by the manufacturer over the lifespan of your computer, assuming you keep the computer for eleven years (which I find unlikely). Other stuff in the power supply is going to fail long before capacitor aging ever becomes an issue. It's also possible you could have Class II capacitors in the power supply, at which point the eleven year time span takes you out to 1.12x the ripple voltage specified by the manufacturer.

_________________
Buckle your pants or they might fall down.

Heheh, now I want to know what class of capacitors I have in my PSU.

_________________
"Dress cops up as soldiers, give them military equipment, train them in military tactics, tell them they’re fighting a ‘war,’ and the consequences are predictable." —Radley Balko

That, I don't know. The only thing I know about actually making a capacitor is from my electromagnetics course. It depends on how big they are. According to that link I posted, the Class IV can be made to have a higher capacitance than the Class II. However, since they decay faster, you probably want to use a Class II if you can make one big enough. I also don't know what the actual values are that are being used without taking one apart and measuring them.

By "big" I'm talking about the capacitance value, not the physical size.

Now, if I had to make an educated guess, I'd say both. There are other capacitors in your power supply than just the smoothing capacitor. The smoothing capacitor is probably the biggest one in the case, and the only one that would actually need to be Class IV. A Class II might even suffice for the smoothing capacitor, but I suspect power supplies in televisions and computers were the impetus for inventing the Class IV.

_________________
Buckle your pants or they might fall down.

I don't know what class they are, but there did seem to be a lot of bragging about how they were 'Japanese' capacitors rather than 'Taiwanese' and that's supposed to be better (Racism, Iknorite?)

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

Japan manufactures a lot of electronics. I would imagine that the distinction between being made in Japan vs. Taiwan refers to the reliability and tolerance rather than the class.

Give you an example - I have an old video card that went bad because all of the capacitors burst. The entire line of video cards got a bad batch that turned out to be defective. That would be less likely to happen with Japanese units.

_________________
Buckle your pants or they might fall down.

The implication did seem to be that the Japanese ones were more reliable. I also noticed that all the Japanese manufacterers were companie I'd heard of, like Matsushita (sp?) while the Taiwanese ones I hadn't.

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

Well... "none of the above". PC Power supply units don't use ceramic capacitors -- they use electrolytic aluminum oxide capacitors. That actually changes the game considerably. The real issue isn't so much capacitor aging as it is capacitor life expectancy. In the case of elcaps, this is highly dependent on operating temperature (amongst other things).

Here is a good article on the specific nature of of electrolytic capacitor (elcap) aging:
http://www.emersonnetworkpower-partner.com/ArticleDocuments/Capacitors%20Age%20and%20Capacitors%20Have%20an%20end%20of%20Life.pdf.aspx

The bottom line is this: for about every 10C increment below the maximum ambient temp rating of the capacitor, its expected lifetime roughly doubles. The max ambient temp for caps used in PSUs is usually either 85C or 105C. With a cheap PSU, a cramped case, and mediocre cooling solutions, you can easily spend a lot of time baking your PSU at ~65C. If you run the math on this assuming caps with a 5k hour max condition life expectancy, you're looking at PSU burn-out in a little over 2 years.

Take-away lesson:

• Invest in a decently sized case and good airflow solutions. Consider the fact that for most cases, the PSU intake fan is directly above the CPU exhaust. The difference between an intake temp of 45C and 25C is a 4x increase in your PSU's life expectancy (ok, not quite really. But it's still very significant).
• Invest in a good PSU that uses better caps. All other things being equal, 105C max temp caps are going to live 4x as long as 85C, and a 10k hour @ 105C cap is going to last longer than a 5k hour @ 105C cap.
• Invest in a good PSU with good air flow & cooling. Even just looking at mid-range or better PSUs, the intake-to-exhaust temp gradient can range from +6C to +20C. Which is again, potentially a 2-4x increase in life expectancy.

Or to boil down to an even simpler everything-you-really-need-to-know-about-cases-and-PSUs:

Cases: On a budget? Antec. Not on a budget? Lian Li.
PSU: Buy a Seasonic.

_________________
Sail forth! steer for the deep waters only!
Reckless, O soul, exploring, I with thee, and thou with me;
For we are bound where mariner has not yet dared to go,
And we will risk the ship, ourselves and all.

There's a difference?

_________________
"Hysterical children shrieking about right-wing anything need to go sit in the corner and be quiet while the adults are talking."

In the sense that the life expectancy is not purely a function of time, yes. To me, "aging" implies that it's purely a function of the age of the capacitor.

Oh, and with respect to Taiwan vs. Japan...

The Great Capacitor Plague of the early 2000s has a lot to do with that. Faulty electrolytic formulas were the culprit in that whole fiasco. Accusations have long been floating around that a number of Taiwanese capacitor manufacturers were using incomplete formulas stolen from the Japanese, the missing component being an anti-corrosion agent. Whether this is true or not...who knows.

The capacitor plague has pretty well run its course, but yeah -- a lot of people swore off Taiwanese caps as a result.

_________________
Sail forth! steer for the deep waters only!
Reckless, O soul, exploring, I with thee, and thou with me;
For we are bound where mariner has not yet dared to go,
And we will risk the ship, ourselves and all.

I'm surprised electrolytic capacitors are still in use at all. They possess a very fundamental flaw that is catastrophic to a device:



An electrolytic capacitor is highly polarized. The dielectric inside will break down if the direction of the electric field is reversed, meaning the above video happens if current ever flows in the reverse direction.

_________________
Buckle your pants or they might fall down.

Heh, I like how the first one explodes and some gal asks "Are you exploding capacitors?" and then when she gets answered in the affirmative, asks the follow up question "Are you taking safety precautions?"

_________________