HardOCP and other review sites PSU reviews are wrong....

Joined
May 31, 2011
Messages
20
I am a bit disappointed with all these review sites that are doing power supply testing for the noise and ripple figure because ALL OF THEM ARE DOING IT WRONG!!!!!!!!:eek::eek::eek:

When measuring for noise and ripple of a power supply, the following should be considered:

  • Take into account common mode noises such as lighting, other power supplies, or any other source of EMI radiation or else you get the false readings!!!!!!!! Point number three will go into details on how to help with this issue.
  • Follow the standard for power supply noise and ripple test with a 20 MHZ CAPABLE OSCILLOSCOPE, the Stingray USB is oscilloscope is ABSOLUTE GARBAGE!! At 250 KHz bandwidth you might as well be driving down the road blind as a bat because you are missing all of the other noise and ripple elements of the power supply because of the cutoff limit of the oscilloscope!!! Also, be sure to specify the load current at which the noise and ripple was taken at!
  • Need to use a DIFFERENTIAL PROBE to make a PROPER measurements of noise and ripple of power supplies to help eliminate common mode interference!! Single ended probes are a POOR CHOICE. If you cannot afford a differential probe then use two single ended probes with the ground leads disconnect and terminated with 50 ohm load resistors!! You then use the math function to subtract the two waveforms to come up with the "TRUE" waveform that the supply is putting out.
  • Make sure the probe is set to X1 and not X10 when performing noise and ripple test of power supplies.
  • Finally, be sure that the load you are using is LOW NOISE. The load can put noise onto the power supply rail!! Resistors are the best choice compared to electronic load testers!


If you guys doubt me then I suggest a second source for you all here: http://www.youtube.com/watch?v=Edel3eduRj4

Hopefully, all review sites can learn from this error and will correct their testing methodologies in future review of power supplies. They will also have to revisit all power supply reviews to account for their mistakes.

They all had to do the same for video card testing ever since frame time and pacing came about from techreport guys.
 

Sith'ari

Gawd
Joined
Oct 13, 2013
Messages
573
Interesting!! Waiting to see what the answer will be from the testers (* meaning Paul Johnson, and Oklahoma Wolf in particular ;) ).
So there isn't any site that can match the criteria you mentioned above on their reviews?:(
 

teh_chem

[H]ard|Gawd
Joined
Jun 21, 2012
Messages
1,474
Electrically speaking, you may not be wrong; but practically speaking will there be a real world difference? Is the added granularity or precision beneficial among computer power supplies? Perhaps the quick-and-dirty methods you take issue with are good enough to distinguish amongst the power supplies available given the tolerances of components? I mean, maybe some of it is analogous to complaining that speedometers in cars aren't good because they don't read to the nearest tenth of a mph; it doesn't matter when all posted speeds are to the nearest fifth integer (one might argue that readings to the nearest one integer on speedometers are overly precise).

Just something to consider. No one tends to report the result on components when using a psu with 'x' amount of noise/ripple, so from one perspective, having more info on the noise characteristics may net absolutely zero practical results.
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,386
In my experience working with EMG amplification, electrical noise through circuits is a much bigger factor (100s to 1000s of magnitudes larger) than noise through radiation. [H]OCP and the rest have that covered through line conditioners.

It is not practical to have 50 different sets of resistors for each load level.

Additionally, it doesn't really matter what the absolute numbers are as long as those numbers are under ATX spec. What does matter is the numbers compared to other power supplies.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
16,731
I am a bit disappointed with all these review sites that are doing power supply testing for the noise and ripple figure because ALL OF THEM ARE DOING IT WRONG!!!!!!!!:eek::eek::eek:

When measuring for noise and ripple of a power supply, the following should be considered:

  • Take into account common mode noises such as lighting, other power supplies, or any other source of EMI radiation or else you get the false readings!!!!!!!! Point number three will go into details on how to help with this issue.
  • Follow the standard for power supply noise and ripple test with a 20 MHZ CAPABLE OSCILLOSCOPE, the Stingray USB is oscilloscope is ABSOLUTE GARBAGE!! At 250 KHz bandwidth you might as well be driving down the road blind as a bat because you are missing all of the other noise and ripple elements of the power supply because of the cutoff limit of the oscilloscope!!! Also, be sure to specify the load current at which the noise and ripple was taken at!
  • Need to use a DIFFERENTIAL PROBE to make a PROPER measurements of noise and ripple of power supplies to help eliminate common mode interference!! Single ended probes are a POOR CHOICE. If you cannot afford a differential probe then use two single ended probes with the ground leads disconnect and terminated with 50 ohm load resistors!! You then use the math function to subtract the two waveforms to come up with the "TRUE" waveform that the supply is putting out.
  • Make sure the probe is set to X1 and not X10 when performing noise and ripple test of power supplies.
  • Finally, be sure that the load you are using is LOW NOISE. The load can put noise onto the power supply rail!! Resistors are the best choice compared to electronic load testers!


If you guys doubt me then I suggest a second source for you all here: http://www.youtube.com/watch?v=Edel3eduRj4

Hopefully, all review sites can learn from this error and will correct their testing methodologies in future review of power supplies. They will also have to revisit all power supply reviews to account for their mistakes.

They all had to do the same for video card testing ever since frame time and pacing came about from techreport guys.

Actually we have been over this a million times. The oscilloscope used is simply for simplification of visualization of the output and was and is repeatedly tested against the bench Tektronics TDS460. The waveform is not 100% correct, but the amplitude is correct, if something comes up I can switch over but there is no good way to record the output. Quite simply, for ease of production and visualization the Stingray is good enough.

What you failed to understand that actually trips most people up is the requirements for actually measuring ripple/noise per the ATX12v specification in the line just under where you went bent about bandwidth:

Outputs should be bypassed at the connector with a 0.1 μF ceramic disk capacitor and a 10 μF electrolytic capacitor to simulate system loading.

That is the part that makes more of an issue and the reason why Oklahoma_Wolf and I use the SM's. They ARE configured to pass the output per the specification to be recorded.

So are all of them doing it wrong? No. Are some? Sure. Are your points valid academically on paper and without context of what has been developed, planned, tested, and used? Yes. Are your points valid in practice in context of what has been developed, planned, tested, and used? Not really.
 
Joined
May 31, 2011
Messages
20
Actually we have been over this a million times. The oscilloscope used is simply for simplification of visualization of the output and was and is repeatedly tested against the bench Tektronics TDS460. The waveform is not 100% correct, but the amplitude is correct, if something comes up I can switch over but there is no good way to record the output. Quite simply, for ease of production and visualization the Stingray is good enough.

I beg to differ, at 250 Khz bandwidth, that scope will miss the majority of the noise and ripple characteristics of the power supply. It is industry standard to have an oscilloscope set to 20 Mhz bandwidth limit when specifying ripple and noise of a power supply, but that Stingray USB scope is far below that!!

Your setup and other review sites also do not account for common mode noise. If you check the ATX specification, in addition to the capacitors, they also specify the use of a differential probe such as the Tektronix P6046 precisely for the reason of rejecting common mode noises from outside sources!!! You guys and other review sites are only using a regular scope probe that will pick up all kinds of noise from other sources in the lab. Therefore, the noise data and ripple data are all invalid.


This will be interesting to see what other EE's have to say. I've decided to link this forum to the EEVblog and I sent a message to youtube users: w2aew and mikeselectricstuff. This could be interesting discussion to come!:)

Here is the EEVblog posting: http://www.eevblog.com/forum/blog/eevblog-594-how-to-measure-power-supply-ripple-noise/

I suggest a good reading through the posts. You can even see a person posted how to measure ATX power supply noise and ripple. It requires a differential probe as I have stated.




Electrically speaking, you may not be wrong; but practically speaking will there be a real world difference? Is the added granularity or precision beneficial among computer power supplies? Perhaps the quick-and-dirty methods you take issue with are good enough to distinguish amongst the power supplies available given the tolerances of components? I mean, maybe some of it is analogous to complaining that speedometers in cars aren't good because they don't read to the nearest tenth of a mph; it doesn't matter when all posted speeds are to the nearest fifth integer (one might argue that readings to the nearest one integer on speedometers are overly precise).

Just something to consider. No one tends to report the result on components when using a psu with 'x' amount of noise/ripple, so from one perspective, having more info on the noise characteristics may net absolutely zero practical results.

Power supplies for computers follow the STANDARDIZED testing method for noise and ripple measurement that manufactures should adhere too. If these review sites do not have similar test setup, then the data is NOT IN AGREEMENT with the standards of the ATX specificiation. When devices are standardized, then the testing methodology MUST follow the standards set fourth, else the data is garbage. Now if standards did not exist for noise and ripple measurements of ATX power supplies, then your post would be valid since data can just be arbitrarily picked to meet some criteria.
 
Last edited:

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,386
Did you read the reply at all? The characteristic doesn't matter, it's the absolute values that matter. They tested it against the TDS460, which has a bandwidth of 350 mhz and the Stingray did good enough that for their testing purposes, it was fine.

The absolute numbers for staying below ATX spec and the comparative number for comparison against other power supplies are all that matter. We're not working with sensitive medical measuring equipment here, we're working with computer components that don't care what the waveform looks like as long as peak to peak values are low enough.
 
Joined
May 31, 2011
Messages
20
Did you read the reply at all? The characteristic doesn't matter, it's the absolute values that matter. They tested it against the TDS460, which has a bandwidth of 350 mhz and the Stingray did good enough that for their testing purposes, it was fine.

The absolute numbers for staying below ATX spec and the comparative number for comparison against other power supplies are all that matter. We're not working with sensitive medical measuring equipment here, we're working with computer components that don't care what the waveform looks like as long as peak to peak values are low enough.


Did they use a differential probe when they did their comparison? Nope! So you are saying that the ATX specification committee can get away with specifying a 250 Khz bandwidth scope? Maybe we should hire you as a consultant so that you can save the manufactures of power supplies a couple of thousand dollars in testing equipment when they test for noise and ripple of power supplies.:rolleyes:


It does not matter if you believe me or not. I follow all the EE's I know, and what I was taught. I don't think you can compete against a large crowd of EE's that would agree with me and not with you, or how all these review sites are doing their noise and ripple testing...

Lets see what the EEvblog has to say and what the two EE's I messaged have to say.
 

[Spectre]

[H] Admin
Staff member
Joined
Aug 29, 2004
Messages
16,731
I beg to differ, at 250 Khz bandwidth, that scope will miss the majority of the noise and ripple characteristics of the power supply. It is industry standard to have an oscilloscope set to 20 Mhz bandwidth limit when specifying ripple and noise of a power supply, but that Stingray USB scope is far below that!!

Your setup and other review sites also do not account for common mode noise. If you check the ATX specification, in addition to the capacitors, they also specify the use of a differential probe such as the Tektronix P6046 precisely for the reason of rejecting common mode noises from outside sources!!! You guys and other review sites are only using a regular scope probe that will pick up all kinds of noise from other sources in the lab. Therefore, the noise data and ripple data are all invalid.


This will be interesting to see what other EE's have to say. I've decided to link this forum to the EEVblog and I sent a message to youtube users: w2aew and mikeselectricstuff. This could be interesting discussion to.

Not particularly as it was not the first few times we went through this. Academically yes it could be an issue, in practice the limitations of waveform do not make it worthwhile to use a means were we can not easily produce an output. Would it be nice to have a better way to visualize so we wouldn't have to cross check and have this discussion again? Sure. Would you like to cut the check to cover it? I can forward you the mailing address and invoice.
 

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,386
Did they use a differential probe when they did their comparison? Nope! So you are saying that the ATX specification committee can get away with specifying a 250 Khz bandwidth scope? Maybe we should hire you as a consultant so that you can save the manufactures of power supplies a couple of thousand dollars in testing equipment when they test for noise and ripple of power supplies.:rolleyes:


It does not matter if you believe me or not. I follow all the EE's I know, and what I was taught. I don't think you can compete against a large crowd of EE's that would agree with me and not with you, or how all these review sites are doing their noise and ripple testing...

Lets see what the EEvblog has to say and what the two EE's I messaged have to say.

Are we doing a professional analysis of the power supplies, or are we doing an analysis that is good enough for the end user?

From a power supply company I would expect professional standard. From a site like [H]OCP and JonnyGuru, a review that's good enough for the end user would suffice. Needless to say, if it fails their testing, then it would have failed professional testing.

In addition, do professionals take into account the effect of heat on noise and ripple? Your EE blogs don't mention anything about that, but [H] and the rest regularly test for that in high temperature environments.
 

teh_chem

[H]ard|Gawd
Joined
Jun 21, 2012
Messages
1,474
Power supplies for computers follow the STANDARDIZED testing method for noise and ripple measurement that manufactures should adhere too. If these review sites do not have similar test setup, then the data is NOT IN AGREEMENT with the standards of the ATX specificiation. When devices are standardized, then the testing methodology MUST follow the standards set fourth, else the data is garbage. Now if standards did not exist for noise and ripple measurements of ATX power supplies, then your post would be valid since data can just be arbitrarily picked to meet some criteria.

No one is arguing that. What you are failing to prove yet are using as proof for improper testing is that there is a real-world perceivable difference between textbook standardized testing and real-world quick-and-dirty tests. As stated, academically there is nothing wrong with what you've brought up (and a lot of the issues you presented seem to have actually been addressed). Practically, can you show an example of where a quick-and-dirty testing result OK's a PSU that fails under ATX testing requirements? On a review site, not some no-name idiot.
 
Joined
May 31, 2011
Messages
20
Not particularly as it was not the first few times we went through this. Academically yes it could be an issue, in practice the limitations of waveform do not make it worthwhile to use a means were we can not easily produce an output. Would it be nice to have a better way to visualize so we wouldn't have to cross check and have this discussion again? Sure. Would you like to cut the check to cover it? I can forward you the mailing address and invoice.

I'll have a Rigol DS1102E 100MHz Digital Oscilloscope that I can send to you guys, paid for by me. This scope can store waveforms to usb and has 20 Mhz bandwidth limit function. Also, if you cannot get the differential probe that was specified or equivalent, then you can use the subtract function of the scope. The probes will need to 50 ohm terminated and 50 ohm sourced to prevent reflection noises, though. I strongly recommend getting a differential probe to get as close to factory specs of ripple and noise as possible.

Just please do not use that garbage USB O-scope... Leave that for the hobbyist on a budget.


Are we doing a professional analysis of the power supplies, or are we doing an analysis that is good enough for the end user?

From a power supply company I would expect professional standard. From a site like [H]OCP and JonnyGuru, a review that's good enough for the end user would suffice. Needless to say, if it fails their testing, then it would have failed professional testing.

In addition, do professionals take into account the effect of heat on noise and ripple? Your EE blogs don't mention anything about that, but [H] and the rest regularly test for that in high temperature environments.

I am a member of the EEvblog forum under the name AlanR. I am not Dave Jones. Why would any of us on the EEvblog forums skip any tests? Of course we would adhere to the specifications set fourth by the ATX committee.
 
Last edited:

Tsumi

[H]F Junkie
Joined
Mar 18, 2010
Messages
13,386
Well you got me there, I was having trouble finding the actual document so I was going off of Wikipedia.
 

Oklahoma Wolf

PSU Warranty Voider
Joined
Mar 22, 2005
Messages
1,450
I am a bit disappointed with all these review sites that are doing power supply testing for the noise and ripple figure because ALL OF THEM ARE DOING IT WRONG!!!!!!!!:eek::eek::eek:

All five points in the original post are non-issues with my test bench. I haven't even used a Stingray in ages. Probably wouldn't want to now that the Tektronix has totally spoiled me.

I'm personally waiting for the day someone comes to me demanding I buy a hypobaric chamber so I can test the altitude part of the ATX spec, otherwise OMG HE'S DOING IT WRONG!!!
 

jojo69

[H]F Junkie
Joined
Sep 13, 2009
Messages
10,644
WHAT???

You haven't been doing the hypobaric chamber tests???
 
Joined
May 31, 2011
Messages
20
All five points in the original post are non-issues with my test bench. I haven't even used a Stingray in ages. Probably wouldn't want to now that the Tektronix has totally spoiled me.

I'm personally waiting for the day someone comes to me demanding I buy a hypobaric chamber so I can test the altitude part of the ATX spec, otherwise OMG HE'S DOING IT WRONG!!!

Well, I got up to $73,560.89 to spend for my home electronics lab. Personally, I use the Rigol DS2102 as my scope of choice and I use the Cal Test CT3688 active differential probe when characterizing noise and ripple figures for power supplies. About $2900 of gear alone. I cannot tell you review sites what to do. And I understand that you guys may not have such large budgets of cash to spend on high end equipment. I get that....

I suppose I could start up my review site on power supply reviews with top quality gear.:cool: Competition is a good thing as it gets everyone trying to outdo the other guy. Just like how now every review site is trying to outdo the other for graphics card testing with frame pacing. It's been fun guys....I'll leave the rest to you. I've said what I said and that is all I can do.
.
 

Xinmosni

[H]ard|Gawd
Joined
Mar 12, 2012
Messages
1,974
 

Bluesun311

2[H]4U
Joined
Sep 21, 2013
Messages
2,523
I would want to do reviews, but only if no one is allowed to criticize my hard work in a way that forces me to acknowledge their criticism.

These guys do it with class: they will write a review and then discuss it openly with the public. Have they steered anyone wrong recently? Did they recommend some power supplies that actually suck? Do we really NEED to spend 60grand on equipment to measure the differences between consumer grade power supplies?

Edit: that gif is mesmerizing isn't it?
 

Hagrid

[H]F Junkie
Joined
Nov 23, 2006
Messages
9,085
I would want to do reviews, but only if no one is allowed to criticize my hard work in a way that forces me to acknowledge their criticism.

These guys do it with class: they will write a review and then discuss it openly with the public. Have they steered anyone wrong recently? Did they recommend some power supplies that actually suck? Do we really NEED to spend 60grand on equipment to measure the differences between consumer grade power supplies?

Edit: that gif is mesmerizing isn't it?

I agree, they are doing a good job.

The gif is decent. It would be mesmerizing if she had a bikini top on... :D
 

Randylahey

n00b
Joined
Oct 31, 2008
Messages
35
I'm personally waiting for the day someone comes to me demanding I buy a hypobaric chamber so I can test the altitude part of the ATX spec, otherwise OMG HE'S DOING IT WRONG!!!

Wait so the power supply i bought off you was not tested in a hypobaric chamber? I DEMAND A REFUND!!! :rolleyes: ;)
 

JimmayLong

Limp Gawd
Joined
Jan 26, 2010
Messages
333
good old book theory vs pratical experience. I'll take practical experience based on hard's testing methology.
 

stormy1

[H]ard|Gawd
Joined
Apr 3, 2008
Messages
1,053
ahum I have a degree in EE but am not a practicing EE.

low noise load?
How about measuring it in the real world with a real computer, that is the most relevant data?
Not practical to make the load the same every time so can't use it to compare supplies.
Resisters? Way too clean to represent a computer.
Electronic load testers? Reasonable compromise.

It is also obvious you have never designed a high precision power supply or you would be talking about something that you totally missed and is the most important factor.
Let me tell you a story....
There was a device that required the cleanest possible power because the signal it was trying to detect was tiny and really pushing 90s tech.
They tried many commercial supplies and all failed.
So they hired one to be custom made.
I the lab it worked perfectly the million dollars worth of test equipment said it was as perfect as it could be and tailored to the load.
When they got it in field it was no better than a $10 radio shack wall wart.
Back in it back it tested well under spec.
Back in the field it again failed.
About a month later it worked fine with no changes.
Anyone wasn't to guess why?
 

Archmage

2[H]4U
Joined
Jul 13, 2000
Messages
3,023
http://www.analog.com/static/imported-files/rarely_asked_questions/RAQ_perfect.pdf - perhaps the problem with that ideal PSU is mentioned here. It's a 1 page PDF. If not, it's still a neat page. RTFDS!

OT: Geez... I've asked for datasheets, whitepapers, specs on some fairly ordinary products. It's difficult sometimes to get a full datasheet for a halogen bulb (OSRAM usually has good paper work). LEDs are generally much better documented.
 

stormy1

[H]ard|Gawd
Joined
Apr 3, 2008
Messages
1,053
They bought power at high voltage and had an on site transformer and the site was under construction. At any given time there were 15-17 arc welders in use and the line showed huge amounts of phase shift with a lot of sags.

Which is what is something vital missing from the tests how they handle dirty power and voltage sags.
For example my power shows a lot of sags down to 100 volts to the tune of 4-5 a day. My antec green 500w when new would keep the computer up but as it got older it would take it down.
Some power supplies handle it much bettor than others which is why every computer is now on a smart ups at my place.
My industrial clients have had huge problems with white box machines and consumer class computers from dell, hp and others for the same reason.
Business class computers have had far fewer issues.

PS. despite being shown evidence to the contrary the power company for the last 10 years insists there is nothing wrong with the power in my building and will not do anything.
 

Oklahoma Wolf

PSU Warranty Voider
Joined
Mar 22, 2005
Messages
1,450
Some power supplies handle it much bettor than others which is why every computer is now on a smart ups at my place.

That's why in my lab it goes wall->power meter->PSU. No line conditioners, no UPS, no filtering. I expect them to impress me out here in the real world ;)

There's really no better way to test the really big ones, I figure. Pulling 1800 watts out of the wall can reduce line voltage by a lot, even on my 20A dedicated circuit, and the unit being tested had better deal with that if it wants to pass my testing.
 

Grandpa_01

[H]ard|DCer of the Year 2013
Joined
Jun 4, 2011
Messages
1,175
Hmmph I very rarely post anywhere but in the DC section, but after reading this I wanted to add my 2 cents here. I could care less about how accurate your test setups are, I am not building a space ship nor is this site or any of the others where I read reviews building anything along those lines. What I do build is Hotrod folding / crunching computers which are fairly demanding on power, I push my G34's to 1100+ Watts on 1200 Watt power supplies, reviewed and recommended by you guys. Guess what they work and work very well under the abusive conditions I put them through. And have been for quite some time.

If I was building something that required a better power supply such s a space shuttle or something along those lines I might care about what equipment was being used to do the testing but I am not and therefore as long as the equipment you use accurately gives you enough info for you to tell me what power supply is good and which is not, then I am happy and will continue to read your reviews and recommend to others to read your reviews.

And Electricpower go ahead and start writing reviews if you wish, I doubt I would read them though, or many others would either for that matter, I tend to shy away from arrogance as do most. And to me they would most likely be to long and drawn out and cover allot of things I would not care about. I have no doubt about what you have said but in my world it is completely unimportant maybe you should think more about what is important and what is a total wast of john Q consumers time and money.

So far the Power Supply reviews I have read at this site have been spot on for my needs in my opinion and there test equipment works just fine.
 
Joined
May 31, 2011
Messages
20
They bought power at high voltage and had an on site transformer and the site was under construction. At any given time there were 15-17 arc welders in use and the line showed huge amounts of phase shift with a lot of sags.

Which is what is something vital missing from the tests how they handle dirty power and voltage sags.
For example my power shows a lot of sags down to 100 volts to the tune of 4-5 a day. My antec green 500w when new would keep the computer up but as it got older it would take it down.
Some power supplies handle it much bettor than others which is why every computer is now on a smart ups at my place.
My industrial clients have had huge problems with white box machines and consumer class computers from dell, hp and others for the same reason.
Business class computers have had far fewer issues.

PS. despite being shown evidence to the contrary the power company for the last 10 years insists there is nothing wrong with the power in my building and will not do anything.


What did you expect with 15 to 17 arc welders running? Those are reactive components!! Usually in industrial manufacturing plants there is a capacitor bank to keep power factor as close to unity as possible. Especially when you got lots of 3-phase AC induction motors running.

I say that power supply passed the test considering the design goal did not intend the power supply to be used in an environment with poor power factor problems and heavy loading on the AC mains with nasty harmonics caused (proper EMI filtering design inside power supply would have helped with this) by the welders. Power supply designs cannot account for every possible condition. Hence, why they are spec'ed to work over a range of temperatures and altitudes, voltage fluctuations,specific frequency ranges, and electrical loading.



Originally Posted by Oklahoma Wolf

I'm personally waiting for the day someone comes to me demanding I buy a hypobaric chamber so I can test the altitude part of the ATX spec, otherwise OMG HE'S DOING IT WRONG!!!

That is not a bad idea. I can design my own little vacuum chamber to simulate low air mass at higher elevations. This could help test if the power supply can pass ATX specification for higher elevations. Not a bad idea to test this!! Thanks.:D



And Electricpower go ahead and start writing reviews if you wish, I doubt I would read them though, or many others would either for that matter, I tend to shy away from arrogance as do most. And to me they would most likely be to long and drawn out and cover allot of things I would not care about. I have no doubt about what you have said but in my world it is completely unimportant maybe you should think more about what is important and what is a total wast of john Q consumers time and money.

So far the Power Supply reviews I have read at this site have been spot on for my needs in my opinion and there test equipment works just fine.

I know the readership does not care about some of the more "technical" details. Hence, I'll need to do a pilot run and get feedback on what people want trimmed away and what people want kept or added.
 
Last edited:

stormy1

[H]ard|Gawd
Joined
Apr 3, 2008
Messages
1,053
That's why in my lab it goes wall->power meter->PSU. No line conditioners, no UPS, no filtering. I expect them to impress me out here in the real world ;)

There's really no better way to test the really big ones, I figure. Pulling 1800 watts out of the wall can reduce line voltage by a lot, even on my 20A dedicated circuit, and the unit being tested had better deal with that if it wants to pass my testing.
That is good, now if you can figure out how to simulate sharp momentary sags and see if they hold it would be even better.
But honestly no matter what you did someone would complain about it :}
There is no perfect test procedure.
 

Cerulean

[H]F Junkie
Joined
Jul 27, 2006
Messages
9,281
I wonder if in the environment where a PSU failed but very same PSU still succeeds back at factory, if putting it on an expensive line conditioner and voltage regular (or even a pure sine PFC UPS) would make a different in the environment in which it fails ...

If doing so makes the PSU work, then I think Electricpower's tests are more logical because given clean power and voltage that doesn't fluctuate wildly and isn't full of noise garbage it makes more sense. Testing for "realistic practicality" is really then an oxymoron and completely subjective. You could send that PSU to someone else whose power is cleaner and more regulated so that he can benefit from it, and give Mr. HisElectricitySux a different PSU that might handle his lesser quality electricity. Then it would be a win-win-win (except for whoever had to pay for the PSUs :\ ).

I put my behemoth workstation (not mentioned in sig -- my sig is several years outdated) on an APC voltage regular and a slightly more expensive noise cleaning and surge suppressing power strip. Pure sine? I don't have that yet. But I definitely want to protect my equipment and ensure its longevity.

But take this post with a grain of salt because what do I know? I'm not an electrician or engineer :D
 

stormy1

[H]ard|Gawd
Joined
Apr 3, 2008
Messages
1,053
Testing for "realistic practicality" is really then an oxymoron and completely subjective.
What are they testing for?
They are not testing to see if the supply is lab perfect they are testing to see if it will supply the readers computer with clean in spec power.
There will be supplies that are near perfect in the lab that fail miserably in the field. Which is why field testing is important.
I view what they should be doing as field environment testing rather than lab perfect conditions testing because that is what is most relevant to the readers.

Lab vs real world is an age old argument that first happened the very moment there were 2 engineers :}
 

Sith'ari

Gawd
Joined
Oct 13, 2013
Messages
573
Hmmph I very rarely post anywhere but in the DC section, but after reading this I wanted to add my 2 cents here........................
And Electricpower go ahead and start writing reviews if you wish, I doubt I would read them though, or many others would either for that matter, I tend to shy away from arrogance as do most. And to me they would most likely be to long and drawn out and cover allot of things I would not care about. I have no doubt about what you have said but in my world it is completely unimportant maybe you should think more about what is important and what is a total wast of john Q consumers time and money..

So you got into all this "trouble" to post again, just to make fun to someone's effort to tell us something more? If you "tend to shy away from arrogance" as you say, maybe you should start from yourself first! :mad:
I don't have electrical knowledge, and only recently i started to study more carefully how PSUs or UPSs are being constructed, but one thing is for sure:
That Electricpower specialises a lot in this area, and he spent lot of money to do something better than others do.
Maybe all this equipment is an exaggeration, for the PSU's purposes as it was stated by Paul Johnson & Oklahoma Wolf, but doesn't change the fact that that this specialization he is trying to say, only benefit will bring and personaly i never criticise someone for doing something better, even if it is not really that necessary!!
You took some "of your precious time" only to mock someone's effort and for me this is enough to describe me your character.
 

larrymoencurly

[H]ard|Gawd
Joined
Jul 18, 2002
Messages
1,635
How much would the use of a differential probe matter in real life for testing of ATX PSUs, which aren't exactly precision devices where the noise is at least 100dB below the outputs? Also I can understand common mode noise being a big problem when measuring 60 Hz sources, which is everywhere and the reason Consumer Reports used 85 Hz instead to measure noise pickup in its very recent evaluation of phonograph needles for those new-fangled vinyl records, but with ATX PSUs running at over 50 KHz, isn't noise from other sources much less of a problem?
 

stormy1

[H]ard|Gawd
Joined
Apr 3, 2008
Messages
1,053
How much would the use of a differential probe matter in real life for testing of ATX PSUs, which aren't exactly precision devices where the noise is at least 100dB below the outputs? Also I can understand common mode noise being a big problem when measuring 60 Hz sources, which is everywhere and the reason Consumer Reports used 85 Hz instead to measure noise pickup in its very recent evaluation of phonograph needles for those new-fangled vinyl records, but with ATX PSUs running at over 50 KHz, isn't noise from other sources much less of a problem?
My opinion is that if it passes without it will pass with.
If it fails then there is a chance it would pass with one if it is close.
So you might get an occasional false negative without.
Real world its not likely to make a huge difference.
If you were trying to determine what is causing the noise then that would be a very different story.
 
Top