I'm trying to understand how PSUs / transformers work together

Discussion in 'Power Supplies' started by Cov, May 14, 2009.

  1. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007
    I'm gonna give the following statements and you tell me whether I'm right or wrong please:


    The number of ampers indicate only the potential of a transformer.

    Meaning: if I connect only a low power consumer to a high power PSU / transformer set, it will consume only the power at the consumer's rate and not more.

    If the PSU has only low power consumers connected, it draws less current from the transformer. The PSU as well as the transformer stay well under the full load limit, and every component works fine and stays cool or warm.

    If the PSU has to power 3 HDDs, a DVD-R, a quad CPU and a discrete high end graphics card, it demands as much current as it needs from the transformer / or as much current as the transformer can deliver.

    >>> If the transformer cannot deliver the current requested, it runs at 100% which lets the components run just about, if at all ... and the transformer gets very hot and can get damaged.

    >>> If the transformer can deliver the full requested capacity of the PSU (i.e. 8A), then the components and the transformer run fine, but the PSU runs at 100% and gets too hot and can get damaged.

    Is that about right, what I think how it works ?


    If yes, then an overpowered transformer (i.e. 12.5A) cannot damage an underpowered PSU (i.e. 8A), as long as the voltage matches (i.e. 12V), because the PSU is not able to draw more than the maximum number of ampers it can handle anyway (i.e. 8A).

    Or am I thinking wrong and the PSU, which is layed out to handle up to 8A, can in fact be damaged by connecting consumers drawing 8A or more ... which makes the PSU draw more current from an overpowered transformer ... following a damage of the PSU.

    Can you confirm or dispute ?

    Thanks




    PS: the reason why I started this thread, is that I have chosen >>> this <<< PSU and >>> this <<< transformer.

    As you can see, the PSU can handle up to 8A while the transformer can handle up to 12.5A

    I am now confused whether the overpowered transformer can damage the PSU und the connected consumers.
     
  2. Zero82z

    Zero82z [H]ard as it Gets

    Messages:
    27,897
    Joined:
    Jan 20, 2004
    Yes, although your usage of the term potential is not really correct.
    I'm not familiar with the design of the picoPSU, but it may have some form of over-current protection which will shut it down before it exceeds its rated capacity for power delivery. If it does not have OCP, it will keep supplying as much power as the devices connected to it need, although if the draw becomes too high it will cause the PSU or the transformer to fail, which will obviously destroy the part in question as well as possibly causing harm to other components.
    In theory, the PSU should never get too hot or become damaged while supplying a load that it is rated for, even if it is being maxed out completely.
    Yes.
    See above. If more than the PSU's rated capacity is drawn from it and it lacks OCP, it will still supply as much as it can, unless it is loaded to the point where it fails. The PSU will not fail due to drawing too much current from the transformer per se; the issue is that it will attempt to supply more power than it is capable of and that will harm its components.
     
  3. MrF

    MrF Gawd

    Messages:
    707
    Joined:
    Feb 18, 2009
    That is not a transformer. It could contain a transformer as well as a rectifier and a voltage regulator.
    Anyway, the stated Ampere rating is what it can deliver.


    That is correct as long as you do not forget the efficiency. So, it will consume a little bit more which turns into heat.


    Yes


    OK


    If you load it more than it can deliver, its output voltage will drop. Yes, it will get hot.


    I don't think you are looking at this right. The 8A figure is what the PSU needs. But, your components may need more.
    See next.


    If the components overload your PSU, the PSU voltage will drop. I would worry about the CPU and RAM etc.



    I think where you should start from is your components. You need to calculate exactly how much total power your components will need.
    Since the spec says that the input current of the PSU should never go over 8A, its input power cannot go over 8x12=96W.
    Taking stated efficiency into account, the PSU output power will be limited to 86W.
    Now, give it some margin.

    You also need to check each individual rail since each rail has a specific rating.
     
    Last edited: May 15, 2009
  4. NichoTL

    NichoTL Gawd

    Messages:
    772
    Joined:
    Apr 15, 2008
    The 8A figure is the max load on the 12V output line of the PSU. Nothing else. Which means 96W on the 12V rail. It has nothing to do with the amount of current flowing between the transformer and the PSU.

    The PSU is rated at being able to output 150W, which means that to produce that much power it will require (from the transformer) at least 150W DC (the PSU is not 100% efficient), which would be more than 12.5A that the transformer can output. I haven't seen the PSU efficiency so it would be difficult to calculate what kind of transformer would cover ALL the PSU needs (175W? 200W?).
    That being said, in most modern computers the power drawn on the other lines (3.3 and 5) is fairly small, so what you need to make sure is that the power ON THE 12V RAIL will not go above 96W (120W peak). Or less, even, to be on the safe side. Again, without the efficiency of the PSU, it's hard to calculate what is a safe limit (because the transformer won't sustain the PSU's 150W peak).
     
  5. Elledan

    Elledan [H]ard|DCer of the Month - April 2010

    Messages:
    15,885
    Joined:
    Oct 18, 2001
    I read the product page the OP linked to in his other thread, and the PicoPSU (150 Watt version) mentions that the 12V DC input on it (some wires with a connector) can only take 8 A max, while 12.5 A would be required for 150 W. A 4-pin connector would be required for the other 4.5 A according to the page.

    To me this just smells like poor design/too thin cables on the side of the PicoPSU. This is the real issue, not what the PicoPSU (beyond this sloppy input stage) or transformer do.
     
  6. NichoTL

    NichoTL Gawd

    Messages:
    772
    Joined:
    Apr 15, 2008
    I think the 8A max in the text refers to the 3.3V and 5V rails, even though that's not clear. There's no way a 150W DC to DC PSU would be limited to 8A current on a 12V input line.
     
  7. MrF

    MrF Gawd

    Messages:
    707
    Joined:
    Feb 18, 2009
    It is not. Its default connector is!!!



    Precautions for operating this DC-DC converter:

    -For fanless operation de-rate the output of the 3.3 and 5V rails by ~35% or
    ensure PSU surface temperature should not exceed 65C, whichever comes first.

    -Combined and sustained output should not exceed 65% of total power or or
    ensure PSU surface temperature should not exceed 65C, whichever comes first.

    -Input current should not exceed 8A. For current higher loads, we suggest using
    a 2x2 mini-FIT JR as an input connector.


    -Peak load for individual rails should not exceed 60 seconds.

    -For long life operation, PSU surface temperature should not exceed 65C.



    Edit: That is from the manual, a PDF file on the first link in the original post. Here is a direct link to it:
    http://www.cartft.com/support_db/support_files/PicoPSU-150-XT-Manual_EN.pdf

    Edit 2: This is poor marketing (specmanship). It is quite misleading.
     
    Last edited: May 15, 2009
  8. MrF

    MrF Gawd

    Messages:
    707
    Joined:
    Feb 18, 2009
    Regardless of what PSU you use, this is how I think you should analyze it.

    Find out how much power you will need from each rail (12V, 5V, 3.3V).
    Give it some margin to be safe. For example, multiply them by 1.2 to give yourself 20% margin.

    Make sure the PSU can provide the required power on each rail. If not, do not do it. You could damageg your components as well as the PSU.

    Use the efficiency for each rail to calculate how much input power the PSU will need for each rail.

    Add the three required input powers (in Watts).
    You can divide that by 12 to see how much current it will need.
    If what you call transformer cannot deliver this, do not do it. It can damage the components, the PSU and itself.
     
  9. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007
    How about capacity ?

    I have read that all PSUs are supposed to come with ...

    OVP = Over Voltage Protection / SCP = Short Circuit Protection / and OCP = Over Current Protection.

    In the specifications to the Pico PSU 150-XT it says:

    I presume that is OVP ? But no mentioning of SCP or OCP there.

    The worst that can happen is that the consumer(s) are not supplied with enough current ?

    So the PSU can get destroyed ... and the consumers possibly as well ? ok.

    If the PSU draws too much current than its rated for, how does it damage the consumers before it dies ?? I would have thought the designers had built in some kind of fuse to protect the other components.
     
  10. Zero82z

    Zero82z [H]ard as it Gets

    Messages:
    27,897
    Joined:
    Jan 20, 2004
    Yes, that would be more accurate.
    Not all PSUs have it. However, quality units do.
    That is OCP, not OVP. OVP is only when the voltages become too high, but since the voltages are regulated, overloading the PSU will not cause the voltages to exceed specification (unless it's a poorly-regulated PSU).

    According to that quote, the PSU should have protection for excessively high loads. However, I wouldn't suggest that you test that out.
    No, the worst that could happen is that the PSU explodes and takes your hardware out with it.
    Correct.
    If the PSU fails, unpredictable things can happen. There are measures in place to prevent the attached hardware from being damaged, but things can still happen.
     
  11. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007
    Something like that ?

    1. The transformer
    ... uses a transformer (primary & secondary winding) to reduce the 100 - 240V AC into 12V AC
    ... the rectifier diodes convert the 12V AC into 12V DC
    ... electrolytic capacitors compensate for varying load conditions, smoothing the output waveform and assists in limiting the current demand

    2. The PSU
    ... should actually be called power converter and not power supply
    ... lowers the 12V DC into 3.3V and 5V

    Efficiency
    ... is the ratio between the power being pulled from the wall and the power being delivered to the PC.
    Usually the PSU achieves its highest efficiency when delivering between 40% and 60% of its maximum capacity.

    The output voltage will drop as well ? I thought only the current will drop.

    I would have thought all that can happen was an undersupply of the requested current.
    The PSU gets over heated and can get damaged.

    Does the PSU voltage drop below its own specifications and if so, why ?
    Where does the cause of the consumer's damage come from ?
    Does the PSU deliver far too much current before it dies ?

    That's a good idea ! ok, let's see ...


    The CPU E8400

    ... has a TDP (Thermal Design Power) of 65W. I'm not sure if the following statement is correct:
    CPUs for desktop computers typically use more power than any other component, except perhaps recent technology video cards.
    The steady trend in CPU power supplies over the past decade has been towards using lower voltages and having considerably higher currents.

    How can a 150W PSU / transformer set deliver enough power for the consumption shown below I wonder ?

    [​IMG]


    According to Everest, the CPU core voltage is 1.10V at 2 GHz on my E8400. Speedstep does only affect the frequency and not the voltage, right ?

    [​IMG]
    source


    The HDD

    It consumes 10W in average and up to 25 at start, so I have been told.


    The motherboard

    Not conclusive


    The RAM

    Not conclusive, but this sounds somewhat reasonable:

    620mW per 128 MB = 4.96W per GB = 19.84W for 2 x 2GB

    Is there a difference in consumption between RAM at idle and load ?

    By the way, my RAM modules take 1.80V and the CPU takes 1.10V so where do they get these voltages from ?
    I thought there are only 12V / 5V and 3.3V being offered by the PSU.
    Or is there seperate converter on the motherboad, where the 12V 4-Pin connector is ?


    So, Power = Voltage times Current ..... 12V times 8 Ampere = 96W going down to 86W.
    How can 86W be enough for CPU / HDD / motherboard and RAM at load ?
    Why is the CPU rated 150W if only little more than half is left ?

    Misc


    I don't know whether these components are worth to mention in the calculation:

    • Power / HDD & Fan LEDs.
    • Case fans
    • USB devices
    That one ? ...

    [​IMG]
     
  12. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007


    All I have to worry about is ...
    • that the voltage of the transformer (12V) is equal to that of the PSU
    • that the current / wattage of both is sufficient for the consumers
    • that the load is located at around 50% of the transformer for the best efficiency
    • that the transformer / PSU set does not get too hot by overloading
    Is that correct ?

    I don't get why the 12.5A matter since the PSU cannot handle more than 8A anyway.

    This one ? ...

    [​IMG]


    It's all confusing to me though. The correlations between PSU / transformer / efficiency / rails / idle / load / input requirement / ect ...
     
  13. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007
    Aaahh ... so that means: No matter what, the 150W PSU cannot supply more then 96W due to the thickness of the cables.
    But why would it take a 4-pin connector to upgrade to 12.5A ? Isn't it possible to replace the stock cables with thicker ones while keeping the 2-pin DC-Jack (female) 2.5 * 5.5 * 10 mm ?

    I agree 100%.
     
  14. MrF

    MrF Gawd

    Messages:
    707
    Joined:
    Feb 18, 2009
    The first block should be called an adapter, not a transformer.

    Every block has an efficiency. The PSU has an efficiency. So does the adapter.

    To understand why the output voltage drops if you overload a voltage source, you should read up on "Output Resistance".
    You should also read up on "Voltage Regulators" and see how the effective output resistance shoots up when you overload it.

    If you have a 12V source, and you place a resistor as a load at its output, the current through the resistor will be 12/R.
    So, if the resistance is R=10 Ohms, the current will be 12/10=1.2A.
    The source's job is to keep its output voltage constant at 12V.

    Now, if this source is rated to provide only 8A, this means that you can reduce the size of that resistor down to 1.5 Ohms and it will continue to keep its output voltage at 12V.
    But, if you reduce the resistance any further, the output ripple will probably increase and the voltage will eventually drop. That is why the output current cannot go higher than 8A. If it kept the output voltage at 12V, the resistor current would keep going up.


    If your CPU loads your PSU more than it can handle, the 12V voltage will drop or/and its output ripple will increase. Either way, you could be damaging the CPU.
     
  15. Elledan

    Elledan [H]ard|DCer of the Month - April 2010

    Messages:
    15,885
    Joined:
    Oct 18, 2001
    I have no idea what the design motivations behind the PSU are, but unless the PCB traces for the input power or so are too thin as well, then yes, swapping out the wiring may allow you to put the full 12.5 A through it.
     
  16. blivit4

    blivit4 Limp Gawd

    Messages:
    154
    Joined:
    May 10, 2007
    The power supplies I've seen described above are linear: transformer - rectifier - filter - voltage regulator. However, all PCs use switched mode power supplies. They work different. Also keep in mind that not all the voltages in a PC are supplied by the PS, but derived locally on the MB and most new video cards.

    Most end users should just treat the PS as a Black Box: you know what goes in and what comes out, but don't know, or care, what is happening inside it.
     
  17. MrF

    MrF Gawd

    Messages:
    707
    Joined:
    Feb 18, 2009
    Who said it can?
     
  18. NichoTL

    NichoTL Gawd

    Messages:
    772
    Joined:
    Apr 15, 2008
    It looked on the pictures like the 2*2 mini-Fit JR is already on the input wire. On it is plugged a standard power jack (which seems to be the one limited to 8A). So to provide up to 12.5A, you must bypass that jack adapter...
     
  19. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007
    Our fellow [H]ard Forum member drgnfang told us in posting No 11 of this thread, that he runs the M350 case with ...

    CPU = E5200 with themperatures idle in the 30s and load in the 50s
    motherboars = Zotac i630
    RAM = 2 x 2 GB
    HDD = 5400 rpm
    PSU = Pico PSU 150W with 80W 6.6A brick !!


    If that works fine, then researching won't bring me any closer to the answer.
    I will be getting the parts soon and of course, let you know the outcome.



    Thanks to ....
    • Zero82z
    • MrF
    • NichoTL
    • Elledan
    • blivit4
     
  20. Zero82z

    Zero82z [H]ard as it Gets

    Messages:
    27,897
    Joined:
    Jan 20, 2004
    You mentioned that you will be using a high-end discrete video card however, which would certainly not fit within the capacity of the 150W PicoPSU and the adapter you are considering.
     
  21. MrF

    MrF Gawd

    Messages:
    707
    Joined:
    Feb 18, 2009
    If a voltage source is overloaded, its output noise could increase beyond specification, or its output ripple could increase beyond specification, or it could become unstable, or its temperature could increase beyond safe limits, or its output voltage could decrease below specification, or its expected life could become shorter than specified, or any combination of those.

    There is no general rule that can tell you which will happen for any design. Only the designer or manufacturer can tell you that. Or if one has access to the design, could try to reverse engineer it.

    I am sorry about my previous over-simplistic comment about this.
     
  22. Cov

    Cov Gawd

    Messages:
    917
    Joined:
    Dec 3, 2007
    [​IMG][


    I got the answer now: