EP45-UD3 power distribution opinions.

[kick]

System is as in my sig.
Take it for granted that I wont be changing my psu,tho that would be the easiest solution.
My q only concerns the 12v distribution.
Im not worried about net posts that say the psu is a pos,its worked fine for me for 3 years.
psu specs
psu mechanics:
After some investigation recently.
The 12v1 rail supplies every component/output that needs 12v except the cpu
The 12v2 rail supplies the atx 4/8 pin cpu connector (cpu only)

The rails are seperate.ie the outputs originate from different points on the cb at least.

the 12v rail amperage ratings are shown as 18a for either.
Its slightly disturbing to see 12v1 rated as 15a in one listing on the spec sheet.
That would mean a max 12v output of 180 w for everything except the cpu.

New gx card has a theoretical max power consumption of 225w tho in practice it wont reach that.
How high will it go.not sure but at maximum overclocks total test systems(from reviews) power consumption reaches around 400w.sli ones over 600w,but thas not relevant to me.

The only psu test I know of that will stress both the gx card and cpu AND give ripple/noise + voltage readouts is from OCCT.

Ive reduced my number of fans to increase the headroom on the gx side but i would like to have them back in.

Normally the psu stays within limits but with hi overclock on gx and fans running full speed+ lots of heat, ripple/noise come close to the 5% limit on 12v output and v drops to 11.58v(also within limits but i feel I can get some unused power off the 12v2 rail).
3.3v and 5v outputs are rock solid ,low noise/ripple.

Im sure with a cpu overclock there is more than 95w going thru the 12v2 supply,but I estimate(from reviews again that its less than 140 even at 4ghz).

Getting to the q soon ..
=============================
I want to pull some of the power off the 12v2 rail
There are 2 options
1.use the 12v2 power to power some of the system fans and add back in some of the ones taken out.
2.Connect the 12v2 rail to the second 6pin pcie gx card input.the lower power draw one(furthest from the edge of the gx card).
Yeh nvid recommends that both additional 6 pin connectors to gx cards come off the same rail but many psus use different rails for the 6 or 6+2 pin connectors if they have more than 1 gx connector.

So which option would be best?

[yuu]

Re: EP45-UD3 power distribution opinions.

No power supply that cheap can afford separate lines, you can pull all the 12v power from either 12v1 or 12v2, they are one. The cpu is around 100 watts priming, Gpu is 150 watts in games, the rest is around 50 watts tops.

[profJim]

Re: EP45-UD3 power distribution opinions.

I vote for option #3, where you buy an inexpensive 12 volt, 3 amp unit to power all your fans.

I don't recall reading any psu reviews where the psu OCP (over current protection) was set at the max rated amperage for the 12 volt rails, it's often 10 - 20% higher than the maximum rating for each rail.

[kick]

Re: EP45-UD3 power distribution opinions.

Hey PJ
.no actually some of those links you posted,led me to reviews of psu's cutting out at the amperage limit on rails.
Anyway Im probably "micturitioning against the prevailing airflow"
Short of buying an ammeter and hooking it in to the cpu power socket ( I dont fancy the idea of a seperate 12v psu to fans lol option 3).
Altho there are some "combine 2 crappy pc psu's" articles out there.

Ill butcher some cables and connectors or wooh buy some( maybe even grab some vrm gx card heatsinks !) and try both options out and let this post sink into obscurity.
8:15

[profJim]

Re: EP45-UD3 power distribution opinions.

Just to be clear, my option #3 is to use a low amperage 12 volt "unit" that would power a printer, scanner or other peripherals.
Many of these units only provide 1 amp. I have a 12 volt Canon unit that is about 4" x 2" x 2" and it only cost 99 cents at a thrift store.
I wouldn't have any qualms using one of these units to provide 12 volts to my fans.

Some cheapo power supplies don't implement OCP and will burn or explode when overloaded.
Many middle to top tier units can sustain 10 - 20% over their rated power output and still meet ATX specifications for voltage regulation and noise/ripple limits.

[kick]

Re: EP45-UD3 power distribution opinions.

yeh pj i understand.
I just dont fancy having a seperate transformer's wires running around inside the pc .
Ive no reason to suspect the psu of blowing up.
The rail limitations are a concern . Im merely looking for opinions on using the excess power on 12v2 of which i feel there is more than 12v1.

If the psu 12v output was classed as 33+ amp and well designed single rail. it wouldnt matter.
Total (stated) power capacity is fine.Ill get some connections and self test and benchmark. To see if there is any difference

[profJim]

Re: EP45-UD3 power distribution opinions.

I look forward to seeing your results.
It will be interesting to see if you can free up and use some "trapped" power from one of the under-utilized rails to power your fans.

**edit** Is Jeantech the OEM for your psu? If not, you might find a psu review for another 500 watt psu that uses a very similar build (i.e. same pcb and components).
This might provide information that details the maximum 12 volt power output. Some dual rail units don't implement OCP protection and you really have a single rail setup that appears to meet the older psu limit of 20 amps per rail.

[kick]

Re: EP45-UD3 power distribution opinions.

psu origin:
I cant say for sure .
there are no certification numbers(Sold in Europe,so there dont need to be altho it does have pfc).
From hardware secrets ,there are no "clues" as to manufacturer.
No green insulating tape,no "name stamped on cb or any other indicator.
It is of east asian origin..gee what a suprise.
from hand written (tested checked ok) notes inside psu.

Sirtec is the only alternate manufacturer Ive seen listed for Jeantech but there is no definite conformation.There is no review I can find for the exact psu,tho a lot must have been sold thru "pc world" (computer store chain).
The 12v supply is definitely "split" into 2 seperate outputs unlike some units Ive seen reviewed that say they are but arent.

Anyway rather than ramble on Ill connect up the fans first (even tho i will lose some speed readouts).
That should transfer about 20w total between rails (-10 then +10) ..oooh impressive! ,not really but it should be enough to "bench" whether rail loading makes any difference.
As I say everything is working and it may just be the extra gx card power draw pushing total power draw and temps toward the psu overall limit .Ill test some variations next week

edit : For those who are interested

There is a lot of internecine psu activity in the far east. Sirtec has been manufacturing psu's for a lot of companies since the beginning of the century antec , thermaltake.coolmax and a load of others.
Sirtec seems to have 2 offshoots /allied companies (for psu) sort of the R+D companies(Jeantech and High power) but they are so low profile in "the west" they seem to be more aimed at selling their tech"look what we can do" to other psu sellers and oem manufacturers of pc's than in direct selling.
Jeantech has been scaled down(for psu) with High power pumped up since about 2006.High power now mentions jeantech in advertising as more of an adjunct.
Anyway the similarities across the 2 R+D ranges(as I call them) and to some extent Sirtec are very high.
Things like the ez extract molex connectors and the psu readout on the back of the power supply .
example

There are a load of variations starting from about 450w consumption limit upwards to 1200w 4, 12v rails 5, 12v rails power limits between rails. rails certified at 13a + 17a and total pwer consumption between those not to exceed a certain wattage .
then 3 other 12v rails one of which is certified at 25a and 2 at 18a. etc etc.
As I said tho ill go on the theory that my 12v1 rail is limited to 180 w and connect up to thwe 12v2 rail ,to check the action.At present Im checking what games I could get the price up to allow free delivery on some decent connectors.
Guaranteed "superior " psu's to my presnt one involve at least $100 outlay.

[profJim]

Re: EP45-UD3 power distribution opinions.

An easy fix is to go with the Silverstone ST-1500 watt psu with:

• 8 x 12-volt rails (25 amps for each rail, 30 amps peak maximum per rail)
• 4 of the 12 volt rails are dedicated for video cards
• 80 PLUS Silver rated

If you don't like the nearly $400 (USD) price, check out the 18 Newegg power supplies (80 Plus Silver or Gold rated) with single 12-volt rails and 1000+ watts
price range: $185 - $300

[Psycho101]

Re: EP45-UD3 power distribution opinions.

messing about with PSU wiring, shifting cables to different rails etc is a really bad idea. I wouldn't recommend anyone mess about with how a PSU is wired up unless they are highly electrically qualified. That includes simply splicing wires. Like you said, you don't actually know for sure what your PSU is doing in terms of power distribution, OCP etc, so why risk fcuking with it?

As for OCCT's readouts, they're pretty worthless IMO. The only way to get semi accurate ripple measurements is using an oscilloscope in conjunction with some pretty specialist equipment. Who knows how accurate OCCT is? After all it's reporting software, not hardware. When testing a PSU for example for review, a lot of very expensive equipment is needed to load the PSU, record power draw etc. All you can tell form OCCT is that there may be a little ripple and some voltage drop. Forget "5% ripple" as that's got as much chance of being precise as I have of winning the lottery.

If the PSU isn't up to the job, sell it and buy one that is. That's the only solution, other than buying a specialist 12V only GPU 5.25" bay PSU for the card.

500W is a little on the restrictive side any way, so maybe ebay it and get something between 650W and 850W and with a single 12V rail. That way you can use the whole output of the PSU without having to deal with pointless seperate rails that have zero benefit anyway on modern PSU circuitry.

[profJim]

Re: EP45-UD3 power distribution opinions.

^^^^ I agree!!!
You've had your current inexpensive psu for several years.
Why take a chance that the psu might fry your current (or future) hardware.

Ya, I know, it's easy for me to spend your money ;)

[Tyreman]

Re: EP45-UD3 power distribution opinions.

messing about with PSU wiring, shifting cables to different rails etc is a really bad idea. I wouldn't recommend anyone mess about with how a PSU is wired up unless they are highly electrically qualified. That includes simply splicing wires. Like you said, you don't actually know for sure what your PSU is doing in terms of power distribution, OCP etc, so why risk fcuking with it?

As for OCCT's readouts, they're pretty worthless IMO. The only way to get semi accurate ripple measurements is using an oscilloscope in conjunction with some pretty specialist equipment. Who knows how accurate OCCT is? After all it's reporting software, not hardware. When testing a PSU for example for review, a lot of very expensive equipment is needed to load the PSU, record power draw etc. All you can tell form OCCT is that there may be a little ripple and some voltage drop. Forget "5% ripple" as that's got as much chance of being precise as I have of winning the lottery.

If the PSU isn't up to the job, sell it and buy one that is. That's the only solution, other than buying a specialist 12V only GPU 5.25" bay PSU for the card.

500W is a little on the restrictive side any way, so maybe ebay it and get something between 650W and 850W and with a single 12V rail. That way you can use the whole output of the PSU without having to deal with pointless seperate rails that have zero benefit anyway on modern PSU circuitry.

Having been an electronics major in secondary school I would concour completely
often connections on PS's are not what one expects and then as far as OCCT's ripple it a very dubious showing
I personally give it no credence other than program puffing
a good 'scope and few other bench requirements are in order to ascertain ripple
Also the circuits quality itself and its parts determine low ripple

[kick]

Re: EP45-UD3 power distribution opinions.

Movin a little off the q here.
Thats why the original post was so long. nm

My psu is still (for now) at least, ok There arre no failures at the maximum stress im applying.
If I had a definite single rail psu it wouldnt matter as long as it was well constructed.
I mean I wouldnt be too happy about 400 w going thru a single supply point on the psu cb .
On hard testing there is a tendency to head towards the atx psu limits the difference is the new gx card.Its power management is better.So at lower load its probably working better than my old one.
If i wasnt "clocking" the new gx card also,it would be np.In theory it isnt even as it is ,but with everything except the cpu limited to 180 w maximum on 12v :

Taking into consideration :
1. I merely want some opinions on the 12v2 supply
(a) should I connect it to the gx card and which of the 2 connectors?
(b) shouild I connect it purely to as many fans as possible max rated power draw in the region of 10w.They would be running at full speed the whole time so power variations to the cpu shouldnt be a problem.
(c) maybe even connect it to the main gx card 6 pin input.
2.From the info supplied above the 12v2 supply has more headroom than the 12v1
3.After nearly 3 years the psu has reached its maximum degredation.ie could be down to 450-475 maximum output.It is practically dust free since the last cleaning tho.
4.Psu's are more expensive in Europe
5.If..I stress" if" changing psu i would buy a $40 equivalent 750w one i have been looking at(no it doesnt have a 5 year guarantee) but it does have all the connectors i require.
The connectors I need are on the wiring loom.if the psu is terrible i can cannibalise them.

Im working on the maximum overclocked requirement for my setup is 400 w(including gx card overclock)
If it was sli with 2 of the same gx card it could be 600w+
i7 920/930 overclocked with 2 gx cards sli overclocked 700+ etc etc.

Ill post how it goes,any improvement on heat/overc;ock limits etc sometime in the next week or so.

[wazza300]

[Psycho101]

Re: EP45-UD3 power distribution opinions.

You're going to gain nothing in the way of overclocking headroom or stability by doing this. In fact the risks that you're going to fry something are greater than the chance of anything positive coming of this.

Ask yourself a few questions before you start splicing wires, re-soldering or what ever it is you're planning to do. I say "what ever it is you're planning" because you really haven't worded clearly what you're actually planning to physically do to your PSU, and the more I read your posts the less clear it becomes.

Question 1 Do you either a) have access to circuit diagrams of your EXACT model of PSU originating from the OEM (and the capacity to read and understand them) or b) have found a write up by a competent person who has done something similar to your EXACT model of PSU?

Question 2: Can you afford to replace ALL your hardware (worst case) if you should screw up?

If the answer to EITHER of those is "No" then leave the PSU alone.

No offence intended, I'm just going to be perfectly honest here...

After nearly 3 years the psu has reached its maximum degredation.

A PSU never stops degrading. Where you got the "after 3 years degradation peaks" thing from I don't know, but it's completely wrong. A PSU NEVER stops degrading. Electrolyte inside capacitors constantly degrades day by day, week by week, year by year until the capacitor fails completely. there is no "maximum degradation".

From the info supplied above the 12v2 supply has more headroom than the 12v1

That information is no where near enough to tell you what the PSU is actually doing in terms of power distribution. You do not know at what point the OCP is set to kick in. One would hope that there is a sensible +/-10%-15% on them, especially considering how restrictive the rail containing the PCIE connectors is set up. For all you know, the PSU may actually employ intelligent OCP and will be able to supply more current on the 12V2 rail as long as load on 12V1 is light.

" if" changing psu i would buy a $40 equivalent 750w one i have been looking at

That 750W PSU you've been looking at is cr ap. But how do I know that? You haven't even linked it! Simple. You WILL NOT get a quality 750W PSU for $40. You'll struggle to get a decent 350W-400W supply for that amount of money. Yes, it might have all the connectors you need, but that's because connectors are just cheap bits of wire, even sh1t PSU's have lots of connections. That PSU will either be a prehistoric design with p155 poor efficiency and stability or it will be made with extremely cheap materials.

I've seen PSU's priced like that many many times. There are tell tale signs. Does the manufacturer tell you at what intake temperature the PSU is rated? If not, don't buy it. If it stated tested at 25c, don't buy it. A PSU should be able to easily supply its max output wattage while sucking in air at 40c+. After all, what use is a 750W PSU at 25c intake temp, when the ambient case temp of an average PC is usually over 30c? Also take a look at the amperage/wattage rating for the +5V rail. If there seems to be an unusually large amount of power allocated to it, then it's definitely a prehistoric model, more suited to powering a P3/early P4 or AMD Athlon/Athlon XP based PC.

I can appreciate the want/need to tinker and play with things but a PSU rewally is not something you should be tinkering with in any way. Don't be tight. Buy a decent larger PSU if you feel you need it. If you don't or you haven't experienced any problems then leave the thing alone. PLaying around and re-wiring it's about as good an idea as wiring your nipples to the mains.