0 Replies Latest reply on Nov 28, 2015 11:40 PM by nec_v20

    Sometimes tough truths have to be told

    nec_v20

      In a previous incarnation I was Senior German Engineer for Enterprise Disaster-Recovery Tech-Support.

       

      Of the over 17,000 individual cases/escalations I dealt with I failed to resolve (and I don't count kludges or workarounds as resolutions) exactly 223.

       

      This is not a boast, it is just setting a foundation.

       

      The fact of the matter though is that in over 95% of the cases I dealt with I never once had to touch our stuff. I resolved the problems anyway (except for those 223) and I didn't really care which OEM's toes I was treading on. I had Microsoft, Novell, HP, Dell, Adaptec, Quantum, Seagate along with a host of others complain to me that their customers - after I had fully resolved the problem - complained to them about the fact that they had not solved the problem even though I had proven that they were the culprits.

       

      It was no problem with my German customers, because they kept their mouths shut and were happy that everything was working. It was the English (and yes I do mean English and not British) and American customers who, after I had completely resolved their problems had nothing better to do than to get onto the people who should have gotten their problems resolved to give them a deservedly hard time about it - because let's face it, if they had had competent tech-support I would not have heard from the customers.

       

      It got so bad, that Dell sent us a letter from their legal department forbidding us in general and me in particular (I was a named party) from doing any kind of tech-support if the customer had an OEM license from Dell of our software for their backup hardware. My English colleagues understood the game and papped the customers off with some hogwash after determining that it was not our problem. My attitude was that it was our problem (my German department when we dealt with English or American customers because the English line could not handle the load) and that the buck had to stop somewhere and we were it.

       

      Did the calls from Dell customers diminish? No. Did I comply with the legal request from Dell? Of course not. When I was called to the office of the EMEA Director for Tech-Support and shown the letter, my reply was, "Dell can go and take a (Anglo-Saxon equivalent of "go forth and multiply") to themselves, and if you expect me to comply with this letter then you can go and take a (Anglo-Saxon equivalent of "go forth and multiply") to yourself as well". His reply was, "OK, but I had to show you the letter".

       

      I have innumerable "war stories" of this kind, but what I am leading up to is the following.

       

      It has been my (now over 32 years as a techie) experience that when people build a computer their last consideration is the PSU - Power Supply Unit. Personally anytime I agree to build a computer for someone the very first choice I make is with regard to a PSU - every other consideration is secondary.

       

      I am convinced - to the point of certainty - that many complaints made to the developers at AMD (especially with regard to overclocking) could be resolved if the customer was asked what PSU was in the system and if the PSU is not on an approved list then telling the person to get one that is (if they haven't irreparably damaged the hardware already). I cannot believe the amount of people I have seen who have a £250-£450 high performance graphics card running off a £20 pound "compatible" PSU.

       

      Companies who build computers also seem to consider the PSU they put into their systems as an afterthought as well.

       

      My thoughts on the matter are summed up in the following:

       

      Power Supply Unit: PSU or IED?

       

       

      Why is there such a difference in price between various Power Supply Units which purport to do the same thing?

       

      The secret is known as "cost down".

       

      What happens is that a design for an absolutely perfect PSU is created, then there is a round of eliminating components to save cost. If that works then another round of elimination is started and tested until there is a design which no longer works - one then goes back a step to the absolute bare minimum which did work.

       

      So that's it? Not by a long shot. Now that there is a bare minimum of components which will work the next phase of "cost down" is put into effect. In this round the high quality components (such as capacitors) are replaced by lower quality ones. This is then tested and if it still works then those lower quality components will be replaced by even worse quality components. This continues until the PSU once again fails and then the process is taken one step back to the design which did still work.

       

      So that's it then rock bottom has been reached?

       

      Not quite; now that the design has been minimised and the components have been "cost downed" the focus is on the safety of the device itself. Do we really need all of them? Of course not, so various safety components are eliminated or replaced by cheaper alternatives until they can technically say that they are in compliance with safety standards without actually lying outright.

       

      How can you tell a PSU from a potential IED (Improvised Explosive Device)? Simply by picking it up. If you pick up a PSU and it feels like all it needs is to be sealed and have some helium added to it to make it fly away then you are dealing with an "AL-Q Taliban Special" (AQTS). Another indicator would be the presence of a switch on the back of the PSU around the power switch where you can choose between 110 volts and 230 volts. If one looks at the label on the PSU and it has an entry for "-5V" on it then it is an IED candidate. If you look at the Amps with regard to the various voltages and the "+3.3V" and "+5V" are higher than the "+12V" then again you are dealing with an AQTS".

       

      Last but not least there is of course the price. If a deal looks too good to be true then in the case of a PSU it most certainly is.

       

      What are the typical "features" of an "Al-Q Taliban Special" IED? One of the main "features" is that this type of device will generally use your hardware as a fuse to defend itself against detonating. It is only logical (well in the minds of the designers) that the PSU has to be defended against damage by eliminating the cause of the threat, namely the motherboard and/or some other offending parts of your hardware.

       

      If you are lucky then all that will happen is that the AQTS will just not turn on. Not so lucky if if it makes large "bang" noise and/or issues magic smoke. Unlucky if the "bang" noise and/or magic smoke results in the PSU catching on fire, and you can see the results of something like that here: http://www.theregister.co.uk/2012/11...vs_reg_reader/

       

      Of course the worst luck is if it catches fire and you are not there (or asleep) to put it out before you lose your home and/or your own life and life of your loved ones.

       

      Another "feature" of the AQTS is that although it is rated at a certain wattage it will never attain that performance in real use. Generally one of these devices will be "good" if they can actually put out half of the rated wattage. One thing you can be certain of however, if you buy a "700 Watt" IED then you will NEVER, EVER, be able to run 700 Watt worth of hardware from it - not even close.

       

      One should not expect any kind of efficiency from an AQTS and the only "80+" energy rating is the writing on the box. What this means is that within a year (if it lasts that long) the AQTS will have ended up costing you more in wasted electricity than buying a proper PSU in the first place.

       

      The most trivial of the "features" of the AQTS is the absolute dearth of connection cables issuing from it. Not only will there be a lack of those cables but the ones that are there will also be much too short.

       

      The cheap price of the AQTS is also offset by its short lifespan. If it lasts for more than a year of normal use before it dies (and hopefully does nothing else) then the person who bought it can consider themselves lucky.

       

      So what can you expect from a PSU that costs a bit more money? Safety features for one (for both the PSU and the attached hardware), they will also perform to their rated wattage and also have a priority for the "+12V" rail over the other two. Really high quality PSUs (like the Corsair AX860i) will happily run ABOVE their rated wattage. For instance the AX860i has been tested and showed no signs of wear running at almost 1,000 Watts: http://www.kitguru.net/components/po...60i-review/11/ If anything goes wrong with a real PSU then it will quietly shut down, that is it will go out with a whisper rather than a bang. You can also expect the PSU to last and do its job for years.

       

      In conclusion next to pouring a bucket of water over your running computer the PSU has the potential to do the most damage to your system. If you have a neighbour downstairs who is considering building his own computer and you see him eyeing advertisements for an "Al-Q Taliban Special", offer to give him an extra $30 or so towards buying a decent PSU - it might just save your life.

      The developers at AMD would be doing themselves, and their tech-support - never mind the most important people of all, the customer - a favour if they created a blacklist of PSU's which would basically void any kind of support.

       

      Excessive ripple (electrical "noise") from a PSU can and does result in hardware such as graphics cards or CPUs either failing or being destroyed. There is nothing the developers at AMD can do about this.

       

      AMD should be forthright about this and come out publicly by stating which PSUs are, and above all which PSUs are not, commensurate to be used with their hardware. If this steps on the toes of "strategic partners" who toss a PSU into their systems as an afterthought then so be it. It has to be made clear to people that "compatible" just does not mean the same thing as, "It will work".

       

      I have written to retailers asking them to take PSUs which I know to be substandard off their list - and some have done so.

       

      As developers you should be pushing your company to take a stand against substandard PSU's being used in conjunction with your products.