Power Supplies

The latest review of power supplies is possible the weakest thing I’ve ever read on this site. Did I miss the part where something useful was measured? The completely nebulous loads used and the actual input and output voltage/currents are not revealed.

Instead we get a step by step how to void your warranty, and cripple your power supply, to reduce noise. Then concluded; the noisier power supply is the best.

First: REAL loads would have proved something.

Second: a clear explanation of why clean power is becoming so critical.

Third: actual measured results.

Here is my take on the why clean power is becoming so critical.

In the good olds days of TTL. A ONE was 5 volts, and a ZERO was 0 volts…in theory. But in reality anything over ~2.2V was a ONE, and anything under ~1.7V was a ZERO.

How logic works. To move a ONE you have to pump it down a pipe. When the pipe fills the pail you have moved a ONE. To move a ONE faster with the same size pipe you have to make the pail smaller. 5Volt VCC becomes 3.3Volt becomes 1.6Volts.

With the same plumbing you can fill a 1.6V pail a lot faster than a 5V pail. But to every trick there is a catch. Now your previous noise margin of 2.2V-1.7 is almost nothing.

So when you slam the power supply with a solid state load that is switching 1.6V at GHz, any fluctuations (soft power) causes the weakest transistors (nosiest) to corrupt the processes. By strait probability the weakling transistor will be in RAM, either in the caches or system.

So the most important spec for the power supply (after having enough power) is it’s damping factor. Which is never measured or stated. It’s really beyond voltage regulation because the load is dynamic and causes induction in the power leads. The better the supply, the faster it can correct for this problem. A stiffer supply drains away the induced over shot voltage.


What YOU can do: is run a utility like ASUS Probe. Monitor the core voltage. It won’t be right, but you can see the fluctuations. Now run Prime95. See how much the core voltage changes. Now look at the 12Volt supply with Prime95 running and not running. The bigger the change the closer to the limit of the power supply. This is because of the way the power supply is regulated. Only the 5V and the 3.3V are actually regulated. The other voltages are derived by ratio. Under heavy load the supply will push longer and the 12V will rise.

I hope this helps someone. The article didn’t do JACK for me.

Electrovert.

You aren't going to pull out an 18" by 18" processor board and tell me that this is what a REAL CPU looks like, are you?

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>