StarTech.com WattSmart 650 W Power Supply Review

[Hardware Secrets Team]

There has been a new article posted.

Title: StarTech.com WattSmart 650 W Power Supply Review
URL: http://www.hardwaresecrets.com/article/528

Here is a snippet:
"This 650 W power supply from StarTech.com looks like a very high-end unit, as it is bigger than traditional power supplies and uses a dual-transformer design, feature we've only seen on power supplies..."

Comments on this article are welcome.

Best regards,
Hardware Secrets Team
http://www.hardwaresecrets.com

[cypherpunks]

Gabriel, please stop repeating the ridiculous assertion that separate secondaries are better. The lowest-ripple way to build a power supply is polyphase, like all motherboard processor supplies are doing. You have separate primary switches, which operate out of phase, and you join the secondaries.

Voltage ripple comes from the current ripple in the output inductors meeting the secondary output capacitors. While the primary is switched on, the secondary output current ramps up. While the primary is off, it ramps down.

Now, imagine two such sawtooth waveforms of equal size and 50% duty cycle but opposite phase. Their sum is: zero current ripple! Which means zero voltage ripple. Now, you never achieve exactly 50% duty cycle, but the combination still greatly reduces the output ripple.

It also helps your input ripple, which is a pleasant side effect.

Now, using just one set of primary switching transistors and operating the transformers in phase is just the same as using a single bigger transformer. Whether you join the outputs together or not has no effect on the output ripple. Which is a shame.

You certainly do not get better regulation from separate secondaries; you get worse. As long as you have only one set of primary switches, only one secondary output gets to control the switching duty cycle. The other outputs follow along as best they can, but aren't in the main regulation feedback loop.

[Gabriel Torres]

Hi,

First thank you for this great message. I am always opened to learn more.

What you explain makes a lot of sense and I shall correct this in the future and use this information when I update our Anatomy of Power Supplies tutorial.

Cheers,
Gabriel.

[HPF]

Originally Posted by cypherpunks

The lowest-ripple way to build a power supply is polyphase, like all motherboard processor supplies are doing.

How important is noise/ripple? If it is within ATX spec isn't that good enough? The motherboard has it's own filtering capacitors for +3.3V, +5V and +12V and the voltage supplied to sensitive components like CPU, RAM and chipsets are, on modern MBs, are done through DC/DC converters anyway - which mean the noise/ripple from the PSU will never get to those components.

Lower noise/ripple is good for capacitor life span, but how much? 4years/10months/6days instead of 4years/10months/12days?

Seems to me that commenting noise/ripple is the new "thing" among the hardware review sites, maybe because they feel they do better reviews compared to those sites that don't use a scope. I'm not saying there is no need to measure noise/ripple, but I would like to see more soberness when commenting the results.

What I really would like to see is how well PSUs regulate in the voltage after being subjected to fast load changes, and the use of a thermal imaging cameras to reveal hot-spots inside the PSU.

[Gabriel Torres]

Hello and thanks for your comments.

Originally Posted by HPF

How important is noise/ripple?

Please read the full explanation here:
http://www.hardwaresecrets.com/article/181

If it is within ATX spec isn't that good enough? The motherboard has it's own filtering capacitors for +3.3V, +5V and +12V and the voltage supplied to sensitive components like CPU, RAM and chipsets are, on modern MBs, are done through DC/DC converters anyway - which mean the noise/ripple from the PSU will never get to those components.

If noise is too high you will transfer the noise filtering to the motherboard, which can lead to overloading the motherboard capacitors. If you are not using a motherboard with good caps you will end up with the infamous capacitor leakage problem. See more here:

http://www.badcaps.net

I also recommend you to take a look on this review:

http://www.hardwaresecrets.com/article/534

Lower noise/ripple is good for capacitor life span, but how much? 4years/10months/6days instead of 4years/10months/12days?

Temperature and ripple are the main factors for capacitor life span. The explanation is too long. For detailed information please refer to chaper 12 from "Switchmode Power Supply Handbook", by Keith Billings.

Seems to me that commenting noise/ripple is the new "thing" among the hardware review sites, maybe because they feel they do better reviews compared to those sites that don't use a scope. I'm not saying there is no need to measure noise/ripple, but I would like to see more soberness when commenting the results.

I honestly believe that all started when we published an article called "Why 99% of Power Supply Reviews are wrong":

http://www.hardwaresecrets.com/article/410

From then on reader became more exigent and other websites started to try to do something different. Since most of them don't have enough revenues to support investing on a power supply load tester (which costs around USD 3,000) several invested in a PC-based oscilloscope (which costs less than USD 300).

It is my opinion that the ripple results alone don't tell you the whole story and that is why that I continue to say that reviews that don't do load tests are wrong. Ripple is just one of the aspects one should observe.

What I really would like to see is how well PSUs regulate in the voltage after being subjected to fast load changes, and the use of a thermal imaging cameras to reveal hot-spots inside the PSU.

This would be perfect. Unfortunately we don't have the equipment to do so. Temperature is the main issue for both the caps lifespan and semiconductors ability to deliver current (de-rating).

Cheers,
Gabriel.

[HPF]

Originally Posted by Gabriel Torres

If noise is too high

Question is, what is too high? I don't think hardware review sites have the knowledge to tell. Sorry for being so blunt

Originally Posted by Gabriel Torres

you will transfer the noise filtering to the motherboard, which can lead to overloading the motherboard capacitors. If you are not using a motherboard with good caps you will end up with the infamous capacitor leakage problem.

It's true leaking capacitors has been a problem, also in PSUs (mainly the low ESR capacitors).

More here http://en.wikipedia.org/wiki/Capacitor_plague

Flawed electrolyte formulas, or flawed design, isn't good enough reason to set aside the ATX spec.

Originally Posted by Gabriel Torres

Temperature and ripple are the main factors for capacitor life span. The explanation is too long. For detailed information please refer to chaper 12 from "Switchmode Power Supply Handbook", by Keith Billings.

That's well know facts. The question is what is acceptable and what is too high/too much.

Electrolyte capacitors has evolved the last decades. Todays capacitors are better now than they were 15 years ago. If noise/ripple requirements were OK then they are most likely OK now. If not I'm sure Intel and the other chip manufactures would have called for better specs.

[Gabriel Torres]

Originally Posted by HPF

Question is, what is too high? I don't think hardware review sites have the knowledge to tell. Sorry for being so blunt

You are right and I 100% agree with you, however fortunately this isn't our case, as my background is in electronics.

Here we want to see if the power supply is within ATX spec or not, and we usually grade a power supply as a good unit if noise is 50% lower or less than what set by ATX spec.

Flawed electrolyte formulas, or flawed design, isn't good enough reason to set aside the ATX spec.

Perfect.

That's well know facts. The question is what is acceptable and what is too high/too much.

Answered above.

Cheers,
Gabriel.

[Gabriel Torres]

By the way. I just remembered about this interesting review posted at Jonny GURU that shows exactly what happens with capacitors and ripple/noise after some years:

http://www.jonnyguru.com/modules.php...=Story&reid=71

I hope this helps everybody interested in learning more about ripple and noise.

Cheers,
Gabriel.