86% Efficient AC-DC ATX supply - Page 2

ColdPhreze · 11-23-2004, 01:33 AM

MrPerfectionest, yes, this is from my professional experience working with power supplies over the years. When you push any device that creates heat to the manufacture's stated limits, components break down at a much faster rate than normal. Power production vs. component breakdown is an exponential curve. At 150 watts, an OPUS is being strained, any more and severe strains occur. That 150watts is it's maximum continous power, which is going to generate more heat, and stress certain components to their maximums. This will *always* lead to a decreased lifetime. If you keep below the maximums, you will prolong the lifetimes of any heat-creating product.

Let me put it this way: Consider an automobile which is designed and rated to operate safely at 6000rpm, and red-lines at 6500rpm. This means it is ok to rev to 6000 rpm, and even up to 6500rpm occasionally. However, does it mean it is ok to constantly run it at 6000rpm? NO! You will blow that engine in a short amount ot time, even if you change the oil and other fluids every day.

I stated that blanket rule, then also stated that you have to look at every rail separatly... Thus driving home that one must consider both total power and individual rails.

Opus isn't lying, their product does output a constant 150 watts. The only thing that wasn't considered or told to those purchasing it, is that it's life expectancy is reduced at that wattage.

Consumers should be better informed... but that is kinda the point of this forum

ColdPhreze

choyak · 11-23-2004, 01:42 AM

At my work we make AC-DC-AC converter systems rated more than 60 kilowatts and they are near 90% efficient. The AC-DC is a PFC and the DC-AC is an inverter using IGBTs. I see no reason that DC-DC is at least 90%, depending on the topology. The weirdness to circumvent though is it is either dual stage or flyback. This is necessary because the DC-DC needs to both buck and boost.

MrPerfectionest · 11-23-2004, 12:22 PM

Quote:

MrPerfectionest, yes, this is from my professional experience working with power supplies over the years. When you push any device that creates heat to the manufacture's stated limits, components break down at a much faster rate than normal. Power production vs. component breakdown is an exponential curve. At 150 watts, an OPUS is being strained, any more and severe strains occur. That 150watts is it's maximum continous power, which is going to generate more heat, and stress certain components to their maximums. This will *always* lead to a decreased lifetime. If you keep below the maximums, you will prolong the lifetimes of any heat-creating product.

I know what heat does to components.

I am talking about specifications plain and simple.

Opus states that their device will deliver:
+3.3V: 10A max, 13A pk, 50mV p-p max 2%
+5 V: 10 A max, 13A pk 50mV p-p max 2%
+12 V: 5 A max, 6.5 A pk 50mV p-p max 2%
-12 V: 0.5 A max, 1.2 A pk, 150mV p-p max 10%
5 V standby: 1 A max, 1.2 A pk 50mV p-p max 2%

from -20C to 70C with a MTBF of 150,000 hrs. They also provide a limited 3 yr warranty.

This is their specification and I will live by this. If my application stays within those ratings then I fully expect to be able to turn my opus on and have it run for 17 yrs. This may not be realistic (and I would be interested to see how they arrived at that number) but Opus states they will warrany it for 3 years. So at the least I can expect that or else they replace the product, plain and simple. Thats all there is to it.

Quote:

Let me put it this way: Consider an automobile which is designed and rated to operate safely at 6000rpm, and red-lines at 6500rpm. This means it is ok to rev to 6000 rpm, and even up to 6500rpm occasionally. However, does it mean it is ok to constantly run it at 6000rpm? NO! You will blow that engine in a short amount ot time, even if you change the oil and other fluids every day.

Your analogy doesn't work at all because when a mfg. produces a car the 'specification' (I use the term loosely here) is 'normal' driving conditions. Normal driving conditions aren't 6000rpms 24hrs a day.

Ricky327 · 11-23-2004, 12:38 PM

What is the "normal" condition for the OPUS?

rushnrockt · 11-23-2004, 12:53 PM

Hey there MrPerfectionest,
Although one would think that the numbers given by a PS manufacturer for each rail should add up, it is not quite the case. You probably noticed that for example regular cheapy PSUs out there rated at 380W cannot compete with say an Antec 300W. The reason is that the cheapy PSU's looked at each rail separate and added up the numbers to get their power rating. In reality, with all rails loaded at the same time, power supply is strained to keep them going, hence the '80%-90%' rule mentioned previously. I am an electrical engineer (granted, only beginning) and as far as I can see, any rating that a product has is quite idealized, and since OPUS doesn't say what kind of load they use, its definitely a great idea to be on the safe side.

ColdPhreze · 11-23-2004, 01:30 PM

MrPerfectionest, I agree with rushnrockt, what are "normal" operating conditions for an opus? I am SURE that the norms are not to constantly pull the maximum continous currents for 3 years, so my analogy holds true. And, any idiot knows that specs are idealized, and that you cannot expect to constantly pull their max continous rated values for long. It's called marketing.

Also, why do you ALWAYS argue with me? You seem to be making a habit of it.

ColdPhreze

MrPerfectionest · 11-23-2004, 03:00 PM

Quote:

Also, why do you ALWAYS argue with me? You seem to be making a habit of it.

Because you know just enough to be dangerous. Someone that doesn't have a clue will read what you write and think it is gospel. Then in the face of irrefutable facts you will cling to your original arguments or morph your statements into something that you claim wasn't "totally incorrect" (reference the thread about "car frames").

Quote:

I am SURE that the norms are not to constantly pull the maximum continous currents for 3 years, so my analogy holds true

This is what I mean, you have chose to focus on my word choice of 'normal' and then tried to change what the meaning of 'normal' is the way I used it. You state that normal isn't 3 years of continuous operation. Of course that isn't what a normal computer user would consider normal every day use but you completely took my useage out of context.

I am talking about specifications. You do understand what a specification means in the industry right? I take normal use to be anything that falls within what Opus says the device will do. Opus states it will deliver the loads I posted above, within the operating temperatures they state, with the input voltages they state, and it will do it for 150,000 hours on average before failure. If their product doesn't do that then their specification is lying and as a consumer you have plenty of reason to complain.

There is no reason at all to think semiconductor based devices operating WELL within their specified ranges can't run continuously given that they stay in their operating ranges and capabilities.

Quote:

You probably noticed that for example regular cheapy PSUs out there rated at 380W cannot compete with say an Antec 300W. The reason is that the cheapy PSU's looked at each rail separate and added up the numbers to get their power rating. In reality, with all rails loaded at the same time, power supply is strained to keep them going, hence the '80%-90%' rule mentioned previously.

I am fully aware of this. The car stereo makers have been doing this years and will use all the tricks in the book to inflate their specifications. This is nothing new. But, we aren't talking about commodity AC-DC power supplies here. The commodity market is an entirely different animal than the niche the Opus is fulfilling. In the commodity market it is marketing. Corners are cut, lots of tricks are pulled to make that .50c-$1 margin that they are making on their product. They know enough people will be buying their products that won't be operating near the limits and they know most educated consumers that are looking to push the limit aren't going to buy their product. So they avoid a potential quality problem this way. Within this market you will find companies that produce respected supplies that will do what they say they will do, and you pay for it too.

The Opus is not in a market where they can produce a product that won't do what it will claim. They will lose business quickly if their product doesn't perform. They even go as far as offering a 3 year warranty on their product, and yet you question whether it will do what it says it will do???

Quote:

I am an electrical engineer (granted, only beginning) and as far as I can see, any rating that a product has is quite idealized, and since OPUS doesn't say what kind of load they use, its definitely a great idea to be on the safe side.

As am I with plenty of years in the industry.

Quote:

any idiot knows that specs are idealized, and that you cannot expect to constantly pull their max continous rated values for long. It's called marketing

If Intel or Sun or HP or any of quite a few other companies came back to my company and asked why our devices broke and we told them this my multibillion dollar company would be out of business very quickly. What we put on the spec is what our device will do, and we are even IN the commodity market.

ColdPhreze · 11-23-2004, 04:19 PM

MrProfectionist, it seems you are the one who is taking the dangerous grounds, suggesting to the audience here that they stress their psu to it's maximum ratings, while I suggest not to. You stated "If my application stays within those ratings then I fully expect to be able to turn my opus on and have it run for 17 yrs. This may not be realistic...", so who is being dangerous with their knowledge?

I'm not getting into it with you about car frames again. My local automotive mechanics agree with me, I don't care if you do. Who knows more about frames, you or a mechanic? Don't answer that, it's retorical.

I do understand about specifications, I just don't believe OPUS's are completely true. I beleive they have fallen victim to marketing ploys, and they will still remain as a company because it's the only decent DC-DC with a sdc. There is no competition, so that niche market you're talking about will still buy their product even when it is problematic because there is no better, and because so many people diss using an inverter like it's bad or something, while the rest of us using inverters attempt to warn (and when they won't listen we laugh at) those overloading an OPUS.

Actually, normal computer use for me and most of the people I know is 24/7 always on. Of course, that is different in a car computer, and I did not here make the distinction. But I didn't need to, because you stated that bit about 17 years and using it at max rated continous values, which suggests you will be leaving it on 24/7.

Opus doesn't specify their "normal" use, so really this argument is for naught, except that IT IS WISE for the consumer to err and be on the safe side. In the very least, DON'T overload it.

As for Intel or Sun or HP or any other large company... Those circumstances are different, just as you state this niche market is. Most retail consumers don't know enough to test a psu to verify if they do what they are rated, so companies can strech the specs wthout any major reprocussions. Commercial consumers, however, will put it to the test.

If you had as much electronic engineering experience as you profess to, you shouldn't be arguing with me about erring on the side of caution - it is a good engineering practice. You should always take the manufacturer's worse case specs, and make them doublly worse to assure that they will operate properly in your application. If you need more current, it's not like OPUS is the only thing out there. You can always use a SDC and Inverter, you won't have any problem if you buy an inverter of 400watts rms or more.

ColdPhreze

rushnrockt · 11-23-2004, 04:38 PM

May be (here is a brand new idea) instead of going after fellow enthusiasts with idealistic assumptions, you should go out there and find out what exactly OPUS believes its normal conditions are. I am not into taking sides, but ColdPhreze was correct about the "80-90%" rule for power supply loading. I've been around Computer modding longer than I've been an EE, and if you were to check out power user recomendations, the load they put on the supply is always less than the theoretical number, since you can never predict what kind of a power surge you might have at some instant. You can call that insurance.
Now if you really want to prove everyone wrong, get one of those OPUS power supplies, set it up in the lab that I am sure your company has, and load up each one of the rails. To make it more interesting, switch loads on and off at random, and see the power spikes and dips, my guess is that at full 150W load, you will find some discrepencies from the numbers provided by manufacturer.
By the way, this is not from engineering classes, but business clases, when you are the only player in the market (niche or not) you get to say whatever you want, its not like non-existing competition would say anything about it

P.S.
What I think OPUS should do... is have a person checking on all these posts from time to time and actually have "official" input on the matter. I think it would help their business out if they were here trying to clear out whatever misconceptions we as users might have.

MrPerfectionest · 11-23-2004, 10:15 PM

Quote:

MrProfectionist, it seems you are the one who is taking the dangerous grounds, suggesting to the audience here that they stress their psu to it's maximum ratings, while I suggest not to. You stated "If my application stays within those ratings then I fully expect to be able to turn my opus on and have it run for 17 yrs. This may not be realistic...", so who is being dangerous with their knowledge?

I had to type that post twice and forgot a bit of information the second time. The original post said that even though Opus states 150,000hrs MTBF that I was curious about how they arrived at that number but at the very least they give a 3 year warranty (I did state that) and I would expect to be able to run the device this long at least.

But the post I did make added the claim about semiconductors operating in their safe regions will last forever.

You say this:

Quote:

I do understand about specifications, I just don't believe OPUS's are completely true

then say this:

Quote:

Opus doesn't specify their "normal" use, so really this argument is for naught, except that IT IS WISE for the consumer to err and be on the safe side. In the very least, DON'T overload it.

which clearly shows you have no clue what a specification is. Please get this concept iinto your head. A specification is the law when we are talking about things of this nature. From dicitonary.com:

Quote:

specifications A detailed, exact statement of particulars, especially a statement prescribing materials, dimensions, and quality of work for something to be built, installed, or manufactured.

I know very well what what goes into dc-dc converters and though I haven't given a full evaluation of the design, a glance at the components is enough to have a good idea that it will do what it says it will do. It isn't hard to acheive the design they have (I have some designs of my own that outdo theirs that aren't overly complex) and if they don't meet the ripple specification (wow there is that word again) set forth by the ATX standards committee then there is a good chance whatever is attached to the Opus simply won't run at the specs they say.

Quote:

I just don't believe OPUS's are completely true. I beleive they have fallen victim to marketing ploys

If I worked for Opus I would be insulted by this comment. What gives you the basis to make this statement? People attaching multiple hard drives, lcd screens (these three alone probably enough to put us over the 5A rating on the 12V line) and other devices causing the Opus to break is certainly not a basis to make this claim. What answer do you have to the many multiples of people that have the Opus and are totally satisfied with them?

Quote:

they will still remain as a company because it's the only decent DC-DC with a sdc. There is no competition, so that niche market you're talking about will still buy their product even when it is problematic because there is no better, and because so many people diss using an inverter like it's bad or something, while the rest of us using inverters attempt to warn (and when they won't listen we laugh at) those overloading an OPUS.

You have a somewhat valid point about them being the only supplier but if their product is as poor as you claim then plenty of people would already be off the bandwagon and buying into other solutions (mainly a seperate supply and sdc) which are almost as feasible to implement and not too much more expensive.

Quote:

If you had as much electronic engineering experience as you profess to,

I'll assure you it is as much as I profess and probably more.

Quote:

you shouldn't be arguing with me about erring on the side of caution - it is a good engineering practice.

Finally you hit on something correct. The biggest problem is your whole experience (btw, have you ever owned an Opus) is based on AC-DC converters from the desktop world and you use this as a basis alone for the claims you are making against the Opus. Had you any real experience in the industry you would realize that your 80-90% claim (the cause of this whole discussion) in regards to the Opus is ridiculous.
My job everyday is to make sure that our customers don't get failing devices. To do this we test guard band into the product. This way we get rid of all the devices that are borderline on our spec. and only ship those devices that fall well on the good side.

This can be easily seen in the peak ratings that Opus puts on each individual line. This is the guardband they are building into the product. They also limit the temperature range to 70C. Probably every component in the opus has a rating of at least 85c, some will have 100c and a few will even have a 125c rating. Since it is a switching supply it will be low in heat by nature. At 90% efficiency it is wasting 15W of heat at full power. At 70c (their max spec) 15W of heat distributed across multiple regulators isn't enough given proper heatsinking (which the Opus has) to raise the internal temps of any of the devices near high enough to become critical.

Quote:

I've been around Computer modding longer than I've been an EE, and if you were to check out power user recomendations, the load they put on the supply is always less than the theoretical number, since you can never predict what kind of a power surge you might have at some instant.

Again, this statement is based off of your AC-DC supply experience but it is pretty much taken care of with the peak rating given to us by Opus. Besides, if we buy components (hard drives, screens, etc) and add all their max ratings and keep the total less than the Opus spec then we are fine. The hard drive won't pull more than it claims it will or will you assume that hard drive mfgs are lying about their specs also? This dc-dc supply probably runs in the 200khz range or more for all the regulators and is plenty fast enough to handle the transients thrown at by any device connected to it.

Quote:

By the way, this is not from engineering classes, but business clases, when you are the only player in the market (niche or not) you get to say whatever you want, its not like non-existing competition would say anything about it

There is existing competition and again, If I were an Opus employee I would be bothered by both of your assumptions that they are trying to be purposefully misleading.

YOU CAN'T ARGUE WITH A THREE YEAR WARRANTY OR THE FACT THAT TONS OF PEOPLE ARE HAPPY WITH THEIR PRODUCT AND THEIR SERVICE IN THE EVENT OF A FAILURE.

How much more does it take?

I would love to entertain a well thought out post with good reasoning to support your claim.

ColdPhreze · 11-23-2004, 10:47 PM

MrPerfectionist,

I have a HUGH responce to all of that, but once again, I tire of arguing with you. It is like talking to my 2 year old. You are too thick-headed to fully understand what I say, and too hot-headed to reason with.

Once again, I will refrain from replying to you again, as this is going nowhere. You obviously have your opinion, we all know this. I know you are simply trying to live up to your name, but lets face it, you aren't perfect.

Goodbye

auto_robotics · 11-23-2004, 11:15 PM

Yikes. This thread reminds me of some of the politics forums I frequent.

Anyways, I'll leave the power supply spec details to your expert judgment. I'm an electrical engineer too, but for me that largely means a concentration in control theory and digital signal processing.

Mr.Perfectionist: You asked what I was using power for a while back? Short answer is I'm a bored engineer who likes projects on the side of work. I've developed a few algorithms in C to find lane markings on a highway. I want to hook up a camera to my computer with a frame grabber card, do some image processing and develop a lane departure warning system like on some luxury cars. I could go with DSPs (and may well later), but a nice simple way to crunch some numbers is through a decently fast P4.

Next step from there is using GPS and actuators on the steering column and pedals to do waypoint navigation (another thing I've already developed code for). It works in simulation! Haha, famous last words of the engineer.

MrPerfectionest · 11-23-2004, 11:36 PM

Its amazing really.

I systematically tore down your claim with extremely good factual based reasoning, yet you refuse to admit that you are incorrect. And I am the hard head?

Here, because I enjoy this and am bored, we can walk through it and please if I am wrong somewhere along the way point it out.

Quote:

As a rule-of-thumb, you will severly shorten the life of a PSU if you draw more than 80-90% of it's stated capacity. Following that rule-of-thumb, you can safely draw 120-135 watts from an OPUS (these a approximate, but are a good practice when designing a PC). Also, remember that you could be pulling more current on one rail than it is designed for, and much less on another, still be within 120-135 watts, but the PSU won't last becuase that one rail being stressed will die.

So you make the statement about only 120-135W as a general power consumption, but you confounded your own claim with the statement about stressing one rail. The second statement was correct and could have stood on its own.

My guess is you will attempt to claim (this is your change the argument/claim strategy) that by the 80-90% safety factor you will protect all rails from being overstressed. This too can readily be shown to not work in all situations.

I asked for more support of your claim and addressed the part about a general power consumption number.

Quote:

MrPerfectionest, yes, this is from my professional experience working with power supplies over the years

Your experience is based on AC-DC power supplies which the Opus is not and exist in an application and market entirely different (which I explained a couple of times).

Quote:

When you push any device that creates heat to the manufacture's stated limits, components break down at a much faster rate than normal. Power production vs. component breakdown is an exponential curve

While the spirit of this statement is ok (that more heat can cause more failures), it is based on other devices that may or may not be similar similar to the Opus in their built in robustness. Yes the curve gets exponential near the edge. But the only component you consider is the Opus itself and you go by a single 150W number as given by the manufacturer. What you don't understand is the technology that is actually in the Opus and your blanket statement about component failure can't be readily applied to the 'system' that is the Opus.

Quote:

I stated that blanket rule, then also stated that you have to look at every rail separatly... Thus driving home that one must consider both total power and individual rails.

You don't have to consider total power if you consider the individual rails alone, this concept alone nullifies your original 80-90% claim (among other things).

Quote:

Opus isn't lying, their product does output a constant 150 watts. The only thing that wasn't considered or told to those purchasing it, is that it's life expectancy is reduced at that wattage

Its amazing how you know this information and can make this claim despite Opus saying the MTBF (thats 'mean time before failure' in case you didn't know) is 150,000hrs. Even if they fudge this number by a factor of 10 (pretty absurd don't you think?) the device can still be expected to live for almost 2 years at their ratings. On top of all this they give you a three year warranty.

I gave tons of other good reasoning supporting why I think your statements are completely and utterly baseless. In general your failure to respond is enough for me and hopefully others reading this post to realize you were wrong.

The point of all this that as long as a user of the 150W understand what his power draw is and stays just under the current specificationf or each rail (generally speaking it is probably even ok to go a little outside of it) they can expect their device to last quite a long time.

Quote:

I know you are simply trying to live up to your name, but lets face it, you aren't perfect

As if your failure to respond isn't enough, your resorting to name calling certainly is.

What is most funny is the fact that my name is spelled incorrectly on purpose, it is simply an ironic name. You missed that one too.

rushnrockt · 11-23-2004, 11:41 PM

Alright Mr, you like to quote, lets quote (but do read to the end please)