Why have a +600 Watt PSU in ITX Factor?

[System Error Message]

2 hours ago, Quaker said:

Lol! I can't help laughing at this.

 

1. Just because a power supply is rated at 1Kw does not guarantee that the components in it are any better "quality" than those in a 500watt supply. Many of the components are exactly the same - it's mostly just the actual current carrying components that are more highly rated.

2. While it's true that the 1Kw PSU would only be using 40% of it's capacity, at 400watts it's producing roughly the same amount of heat as a 500watt supply. The only difference in thermals would be the difference in efficiency between the two supplies when running at 400watts. It's quite possible that a 1Kw supply could actually use more power at 400 watts output, than a 500watt. depending upon the overall efficiency and design of the particular supply.

3. Differences in the temperature of any particular components does not negate the fact that you would be essentially creating the same amount of actual "heat". The heat doesn't disappear because you measure a lower temp - it's simply more diffused*.

4. At 400watts output an 80+ rated power supply would draw roughly 500 watts from the wall. The only difference would be the "actual" efficiency of the particular power supply at the 400watt level, which would only vary by a few percentage points. As above, the 1Kwatt could be 85% efficient at that output while some 500watt supply could be 86% or whatever.

5. The life of the capacitors would be determined by the quality of the capacitors and the heat/temperature of the capacitors themselves. In a higher rated power supply the capacitors would generally be "larger" - that is, have more microfarads - but they could be the same voltage and "quality" as in a lower wattage power supply. It's the voltage (and the power supplies ability to not exceed it), quality of construction, and localized temperature that most affects longevity.

 

* this also applies to CPUs and GPUs as well. When you lower the temperature of a CPU/GPU by increasing the cooling, you still have the same amount of heat being produced - it's just not localized at the chip as much. The same quantity of heat still goes into the surrounding atmosphere however - it doesn't magically disappear.

go and do some voltage regulation and run simulations too. Components could have the same volt ratings but the amp ratings will differ. If the PSU has a much much higher capacity it could also have components in parallel providing extra redundancy.  The most important bit of the PSU, the switching bit is what produces heat. The transistors and switches will defer a lot between PSUs of different wattages. They dont use the same components, only the same design, for example PSU of 500W would have a much smaller heatsink and a transistor rated only for a particular amount of amps while a 1KW PSU will have a much bigger heatsink and a miuch heavier duty transistor which given the same power would heat the transistor at the 400W PSU much more than in the 1KW PSU. Electricity has nothing to do with heat, heat comes from the resistance and resistors heat up. So components that arent design to take that much heat operating at a higher % of capacity will be at a higher temperature than a  component operating at lower capacity as theres less resistance.

 

Electrolytic capacitors are commonly used in power applications. unlike solid caps used in signaling or on boards they are stubborn and like to hold their charge. The higher the load the more the capacitor is charged and discharged. This process does wear out the capacitor after a while but the less the amount drained from the capacitor each time the longer the capacitor will wear out. This applies to electrolytic capacitors which you will find in every power delivery application as to use solid state capacitors of the same capacity will be very expensive (basically those rectangular tiny caps you see around CPUs, boards, etc).

 

Ask on EEVblog forums for example, you will see that i am not wrong. Most here dont have a clue about the component level workings of electronics and what different things do. Ofcourse the quality of the components matter but theres a reason why a PSU is 500W and another is 1KW even if from the same PSU line or brand.

[H0R53]

On 8/13/2017 at 4:16 PM, Minibois said:

'cuz it's possible.

 

But uhhh, I got a 750W PSU while I would be just fine with a 550W PSU. I did this because that means my PSU can run fanless at all times (even at max load). Not sure if these SFX (I assume SFX?) PSU's can run without spinning the fan, but that is what I'm just gonna guess.

 

On 8/13/2017 at 4:37 PM, manikyath said:

because some people have very weird fetishes when it comes to overclocking, PSU headroom, and/or quiet operation.

 

On 8/13/2017 at 4:52 PM, Stefan Payne said:

Because there is no sense in making a 550W Titanium PSU

And because you can.

My system uses 480W under full load and I have a 600W PSU - my system is the quietest thing in my apartment and I have two small refrigerators, a laptop, another tower, my NAS, my media PC, and an air conditioning - my PC is quieter under load than my laptop is at idle.

[manikyath]

Just now, H0R53 said:

My system uses 480W under full load and I have a 600W PSU - my system is the quietest thing in my apartment and I have two small refrigerators, a laptop, another tower, my NAS, my media PC, and an air conditioning - my PC is quieter under load than my laptop is at idle.

my main rig struggles to break the 300W barrier, 750W power supply, because fanless operation

my server... can barely touch 100W and it has a 650W 80+ titanium power supply

[H0R53]

4 minutes ago, manikyath said:

my main rig struggles to break the 300W barrier, 750W power supply, because fanless operation

my server... can barely touch 100W and it has a 650W 80+ titanium power supply

If I didn't have, oh idk, a 1050Ti with a custom BIOS and a 600Mhz overclock, five hard disks, and a quad core Xeon as well as light-up everything I'm sure I'd be using 100-200W less. I could easily get by with a 450W.

[Quaker]

21 hours ago, System Error Message said:

. Electricity has nothing to do with heat, heat comes from the resistance and resistors heat up.

Actually electricity (in electronic circuits) has everything to do with heat. It's the consumption of electricity that produces the heat. What did you think causes the heat in a resistor - magical pixies?

(And, of course, it's not just resistors that have "resistance" and/or generate heat.)

 

The 400watts that the power supply would be delivering (in this hypothetical case) is the power consumed by the components - CPU, motherboard, GPU, etc - and would not be the source of the heat produced within the power supply.

The heat produced within the PSU is the result of the power consumed within the PSU itself in order to deliver that 400watts.  In this hypothetical case, an 80+ power supply would be drawing 500 watts out of the wall in order to deliver 400watts output. It's that extra 100watts of power consumed that generates the heat within the PSU.

Whether the 80+ power supply is capable of delivering 1Kwatt or only 500watts, at 400watts output there would be 100watts of wasted power. This wasted power is essentially turned into the same quantity of heat.

 

The components in a 550 watt power supply could be just as capable of withstanding this heat as the components in a 1000watt supply.

 

The level of noise produced by a power supply would be a function mostly of the cooling system. A 1Kwatt power supply generally has, as you said, larger heatsinks, etc, and so can often deliver lower levels of power without even turning on the fan, but, it is, of course also possible to build a lower wattage power supply with larger heaksinks so it's fan doesn't run as often either. There are even some passively cooled power supplies that have no fan at all.

Generally speaking, it comes down to a price/performance issue. It's possible to build power supplies with all manor of cooling systems that develop more or less noise based upon the internal components and the quality, size, etc, of the fan (if any).

[bowrilla]

I can pretty easily build a system with ~500–550W of max power consumption even in ITX format. A 600W psu wouldn't be enough in my books. Sure, it could do the job, but it's not ideal. Better efficiency and as was pointed out before probably better noise level. And keep in mind: if you want to OC you want some headroom.

[Corsair Blake]

24 minutes ago, bowrilla said:

I can pretty easily build a system with ~500–550W of max power consumption even in ITX format. A 600W psu wouldn't be enough in my books. Sure, it could do the job, but it's not ideal. Better efficiency and as was pointed out before probably better noise level. And keep in mind: if you want to OC you want some headroom.

Mind showing a part list? I'm curious to see what a 550W ITX rig would look like.

 

600W is perfectly fine and ideal for a mini-ITX system.

[bowrilla]

27 minutes ago, Corsair Blake said:

Mind showing a part list? I'm curious to see what a 550W ITX rig would look like.

 

600W is perfectly fine and ideal for a mini-ITX system.

Even though "easily" was an exaggeration since you'd need to pick the extreme editions, but still possible with off the shelf parts. That case is also made to support a full custom loop and since those 519W are without any cooler, just add a proper pump and maybe 3 or 4 fans. Then you'll end up with about 550W. I'm pretty sure there will be a few Mini-ITX X299 boards coming. I have my doubts about Threadripper and Mini-ITX though. Of course, this kind of rig would be an absolute niche build, i.e. for mobile workstations or an extreme mobile gaming and content creation rig. A 7700k with proper OC will also break the 500W mark together with Vega 64 and a proper OC. 3x 120x120 fans to consume some power as well.

 

PCPartPicker part list / Price breakdown by merchant

CPU: Intel - Core i7-6950X 3.0GHz 10-Core Processor  ($1559.79 @ SuperBiiz) 
Motherboard: ASRock - X99E-ITX/ac Mini ITX LGA2011-3 Narrow Motherboard  ($199.99 @ Newegg) 
Memory: G.Skill - Trident Z 32GB (2 x 16GB) DDR4-3200 Memory  ($356.99 @ Newegg) 
Storage: Intel - Pro 6000p 512GB M.2-2280 Solid State Drive  ($199.99 @ SuperBiiz) 
Storage: Samsung - 850 EVO-Series 1TB 2.5" Solid State Drive  ($338.88 @ OutletPC) 
Video Card: AMD - Vega Frontier Edition 16GB Frontier Edition Video Card  ($999.00 @ B&H) 
Case: Silverstone - RVZ01B Mini ITX Desktop Case  ($84.99 @ B&H) 
Total: $3739.63
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2017-08-16 13:00 EDT-0400

 

Of course, with a GTX 1080 TI you'd be a bit better off but 450W can be reached as well. OC your CPU and you'll get near that 500W mark.

 

Don't get me wrong, for like 99% of Mini-ITX cases a 600W psu is perfectly fine. I have your SF600 in my case and it's doing a fine job. 

 

[System Error Message]

3 hours ago, Quaker said:

Actually electricity (in electronic circuits) has everything to do with heat. It's the consumption of electricity that produces the heat. What did you think causes the heat in a resistor - magical pixies?

(And, of course, it's not just resistors that have "resistance" and/or generate heat.)

 

The 400watts that the power supply would be delivering (in this hypothetical case) is the power consumed by the components - CPU, motherboard, GPU, etc - and would not be the source of the heat produced within the power supply.

The heat produced within the PSU is the result of the power consumed within the PSU itself in order to deliver that 400watts.  In this hypothetical case, an 80+ power supply would be drawing 500 watts out of the wall in order to deliver 400watts output. It's that extra 100watts of power consumed that generates the heat within the PSU.

Whether the 80+ power supply is capable of delivering 1Kwatt or only 500watts, at 400watts output there would be 100watts of wasted power. This wasted power is essentially turned into the same quantity of heat.

 

The components in a 550 watt power supply could be just as capable of withstanding this heat as the components in a 1000watt supply.

 

The level of noise produced by a power supply would be a function mostly of the cooling system. A 1Kwatt power supply generally has, as you said, larger heatsinks, etc, and so can often deliver lower levels of power without even turning on the fan, but, it is, of course also possible to build a lower wattage power supply with larger heaksinks so it's fan doesn't run as often either. There are even some passively cooled power supplies that have no fan at all.

Generally speaking, it comes down to a price/performance issue. It's possible to build power supplies with all manor of cooling systems that develop more or less noise based upon the internal components and the quality, size, etc, of the fan (if any).

than why do efficiency curves differ given a particular load? This would mean less heat is generated at higher efficiency given the same loads between PSUs.

[rockon5622]

Clearly we need 1500w sff.

I7 3970x with this motherboard

 

 

And 2 of these

 

4 way crossfire with 290x

All on itx err e-itx

[turkey3_scratch]

On 8/15/2017 at 11:18 AM, Quaker said:

5. The life of the capacitors would be determined by the quality of the capacitors and the heat/temperature of the capacitors themselves. In a higher rated power supply the capacitors would generally be "larger" - that is, have more microfarads - but they could be the same voltage and "quality" as in a lower wattage power supply. It's the voltage (and the power supplies ability to not exceed it), quality of construction, and localized temperature that most affects longevity.

Not just heat/temperature, but ripple also. How much ripple is on a particular capacitors is like one of the absolutely vital key factors to its lifetime. That's not something you or I really know either.

[Imbellis]

56 minutes ago, rockon5622 said:

Clearly we need 1500w sff.

I7 3970x with this motherboard

Holy crap. 4 Dimm slots on itx.

[Stefan Payne]

2 hours ago, Imbellis said:

Holy crap. 4 Dimm slots on itx.

Well, that ain't no ITX; that's DTX.

[Imbellis]

29 minutes ago, Stefan Payne said:

Well, that ain't no ITX; that's DTX.

Eh. Same thing - small board.

 

But really not.

[rockon5622]

26 minutes ago, Imbellis said:

Eh. Same thing - small board.

 

But really not.

E-itx

[Quaker]

On 8/16/2017 at 3:45 PM, System Error Message said:

than why do efficiency curves differ given a particular load? This would mean less heat is generated at higher efficiency given the same loads between PSUs.

Efficiency curves differ because circuit design, components, etc, varies between different models/brands.

 

Yes, less heat is generated at higher efficiency because less power is wasted internally. (duh)

However, there is no reason to assume that a particular 1Kwatt PSU would be more efficient at a particular load.

A 1Kwatt PSU that is 85% efficient at 400watts output would produce the same amount of heat as a 500watt PSU that is 85% efficient at 400watts output. Note - this is the total quantity of heat, not the temperature of any particular part.

 

[System Error Message]

2 hours ago, Quaker said:

Efficiency curves differ because circuit design, components, etc, varies between different models/brands.

 

Yes, less heat is generated at higher efficiency because less power is wasted internally. (duh)

However, there is no reason to assume that a particular 1Kwatt PSU would be more efficient at a particular load.

A 1Kwatt PSU that is 85% efficient at 400watts output would produce the same amount of heat as a 500watt PSU that is 85% efficient at 400watts output. Note - this is the total quantity of heat, not the temperature of any particular part.

 

you argue theory i argue practice. Tomaeto tomato.

[Stefan Payne]

On 16.8.2017 at 8:45 PM, System Error Message said:

than why do efficiency curves differ given a particular load? This would mean less heat is generated at higher efficiency given the same loads between PSUs.

Because of Design Choices, Transformer, quirks in the design, switching frequency and so on.

There are Controllers that switch between PWM and FM Mode operation. That 

Champion Micro CM6900 is such a chip. Depending 

 

15 hours ago, Quaker said:

Efficiency curves differ because circuit design, components, etc, varies between different models/brands.

 

Yes, less heat is generated at higher efficiency because less power is wasted internally. (duh)

However, there is no reason to assume that a particular 1Kwatt PSU would be more efficient at a particular load.

A 1Kwatt PSU that is 85% efficient at 400watts output would produce the same amount of heat as a 500watt PSU that is 85% efficient at 400watts output. Note - this is the total quantity of heat, not the temperature of any particular part.

 

That is absolutely correct.

I don't see any way to disprove your argument.

 

13 hours ago, System Error Message said:

you argue theory i argue practice. Tomaeto tomato.

Mate, you lost the argument, admit it and stop trying to avoid the topic.

He is absolutely correct. And it's just utter nonsense to argue about efficiency of two different PSU without knowing the efficiency curve (measured, in at least 10% steps).

 

Because there are units out there that are utter garbage at lower loads - like Huntkey X7-1200W and others.

And there are units out there that are very efficient at lower loads and can compete even with Platinum units or beat them at certain loads. Cooler Master V-Series (550-750W) is such an example. 

For example:

https://www.techpowerup.com/reviews/CoolerMaster/V550/6.html

https://www.techpowerup.com/reviews/FSP/HGX550/6.html

https://www.techpowerup.com/reviews/Cougar/GX-S550/6.html

https://www.techpowerup.com/reviews/FSP/Twins-500/7.html

 

All are 80plus Gold, all have different efficiency curves...