New testing methodology at Anandtech:)

[0roo0roo]

http://www.anandtech.com/casec...us/showdoc.aspx?i=3021

can't wait to see them review a whole buncha psu's. psu reviews generally suck and the last time i saw reviews using expensive equipment that actually really strained the psu's was in custom pc magazine where they used eqiupment costing thousands of dollars for a psu lineup..more than most sites can afford obviously. but they only did that once and that was like a year ago.

how much did the testing equipment cost anandtech?

 

[AWhackWhiteBoy]

I would estimate that AC source analyzer to be about $3,000 brand new and those adjustable DC load sources to be at least another $3,000 for all those modules. Geogous equipment, I read up on the AC piece and it has built in PFC which makes me a little gitty. I would love to get my hands on a used one!

The most important piece of equipement though when anazlying would be an oscilloscope, which they didn't picture in that little article. Those depending can easily cost over $15,000.

 

[John]

Originally posted by: 0roo0roo
psu reviews generally suck and the last time i saw reviews using expensive equipment that actually really strained the psu's was in custom pc magazine where they used eqiupment costing thousands of dollars for a psu lineup..more than most sites can afford obviously. but they only did that once and that was like a year ago.

I highly recommend that you visit http://www.jonnyguru.com/reviews.php?catID=1

 

[Super Nade]

Originally posted by: AWhackWhiteBoy
The most important piece of equipement though when anazlying would be an oscilloscope, which they didn't picture in that little article. Those depending can easily cost over $15,000.

More critical than the O-Scope would be the special probes to ensure that information is not lost due to bandwidth selectivity.

 

[jonnyGURU]

Originally posted by: Super Nade

Originally posted by: AWhackWhiteBoy
The most important piece of equipement though when anazlying would be an oscilloscope, which they didn't picture in that little article. Those depending can easily cost over $15,000.

More critical than the O-Scope would be the special probes to ensure that information is not lost due to bandwidth selectivity.

Yep.

I was very impressed that SKYMTL of HardwareCanucks forked over the dough for the SI-9001 differential o-scope probes for his Stingray DS1M12. All he's missing is the load tester!

 

[John]

I apologize in advance if this has already been asked and addressed elsewhere, but is there any ETA on the ripple measurements? I'm still trying to figure out whose bright idea it was to use a padded cube and chroma interface board. Obviously a lot of time and money has been invested on this high dollar equipment, and it would be nice to see articles that are "as good" as the other reputable reviewers.

 

[Super Nade]

Originally posted by: jonnyGURU

Originally posted by: Super Nade

Originally posted by: AWhackWhiteBoy
The most important piece of equipement though when anazlying would be an oscilloscope, which they didn't picture in that little article. Those depending can easily cost over $15,000.

More critical than the O-Scope would be the special probes to ensure that information is not lost due to bandwidth selectivity.

Yep.

I was very impressed that SKYMTL of HardwareCanucks forked over the dough for the SI-9001 differential o-scope probes for his Stingray DS1M12. All he's missing is the load tester!

$300 is not a bad price to pay for reliable results.

This would also do the job:-
http://www.saelig.com/miva/mer...=PSA010&Category_Code=

 

[CTho9305]

It still sounds like a seriously flawed methodology - their loads are static. It'd be much more meaningful to do the measurements on a load that simulates a system that's switching between idle and active frequently (at the resonant frequency of the supply network, for the worst-case).

 

[Super Nade]

Originally posted by: CTho9305
It still sounds like a seriously flawed methodology - their loads are static. It'd be much more meaningful to do the measurements on a load that simulates a system that's switching between idle and active frequently (at the resonant frequency of the supply network, for the worst-case).

To d that, nothing beats a real world rig. However, apart from hogs like the 4x4 AMD stunt + quad SLI, nothing even comes close to taxing today's 1kW units. That is a whole different ball-game, where reliably isolating the PSU from other components is well neigh impossible.

There is nothing wrong in using ramped up loads..

 

[jonnyGURU]

Actually, my SunMoon has a mode that tests at multiple random loads very quickly, but I find it's really impossible to gauge anything outside of "pass or fail." You can't scope it really and the voltages just bounce around 2 or 3% as would be the norm. Great for quality control, but lousy for a review.

 

[CTho9305]

Originally posted by: Super Nade

Originally posted by: CTho9305
It still sounds like a seriously flawed methodology - their loads are static. It'd be much more meaningful to do the measurements on a load that simulates a system that's switching between idle and active frequently (at the resonant frequency of the supply network, for the worst-case).

To d that, nothing beats a real world rig. However, apart from hogs like the 4x4 AMD stunt + quad SLI, nothing even comes close to taxing today's 1kW units. That is a whole different ball-game, where reliably isolating the PSU from other components is well neigh impossible.

There is nothing wrong in using ramped up loads..

Let's say we took your power supply and added inductors on its outputs. It'd handle static loads just fine, all the way up to its max rating. However, if we had, say, 100W load and quicky switched in another 100W of load, bad things would happen. If we quickly removed the extra 100W, very bad things would happen. You don't need to be at the peak limit of a power supply to get it to do bad things if it's poorly designed.