Possible to customize Power Supply for external use ? 5

[mikeguava]

Hi,

I am trying to build a mini PC which has a very high power consumption ( ~400 - 450 Watts)
I cannot find any standard PSU that meet my size requirments and I am considering custumizing a Power Supply to have it sit externally - by lengthening all the cables ( 20 Pin etc. ). This would help the size of the unit as well as in terms of heat.

Question: Would the extending the cable length cause any problems - I am talking about 3m / 9feet in total length.

Thanks in advance for any input.

Mike

 

[jurgen36]

I did read all the msg's and can not understan the P4 3.8???
Is this a mistake? I do not believe Intel makes them at all they tried 3.6 and then gave up and stopped with 3.4 Ghz. I only wonder? Regards


Jurgen

 

[Frank4d]

14 ga is overkill. You would be using something capable of 20 amps in open ait to a connector that is rated at 5a.

Current carrying capacity isn't the issue in this situation. Rather it is voltage drop due to wire size, wire length, and load current. As I mention above, a 5 volt 7 amp load connected to 10 feet of 18 gauge wire will result in 4.59 volts at the load.

 

[mikeguava]

mainegeek wrote: "Any ROM drives going in this case? Just curious as to how
the bay(s) are configured for air flow and component clearance. I'm assuming vertically side mounted."

I am actually installing a Pioneer DVR-K04La slot drive dual layer dvd - writer . This unit is usually used in laptops and only measures 5" by 5". Coupled with a mini ide to regular IDE converter it hopefully will work like a charm not using too much energy and not not producing too much heat
A regular size rom drive would not fit in the case.

Figuring out the placement of components is actually a bit of a challenge in terms of thermal considerations. Im thinking of putting the drives below the motherboard sealing the drive compartemnt off from the motherboard/Videocard compartement. I would also have a fan in that bottom compartment hoping to push the heat out from a bottom opening

jurgen36 wrote:"I did read all the msg's and can not understan the P4 3.8???"
Nope - they actually are selling 3.8GHZ P4s. Not too many OEM are using them yet, but Dell for example uses them in their XPS Gen 4 lines. Most online stores carry them now.

Ed all your input has be tremendous. I need to buy you some drinks next time that I am i Atlanta
I am glad that I will be able to use thinner cable. This helps out a lot.

About the PSU with Pots - I read that the Foxtron 500 and 530 Watts PSUs have adjustable pots - right now I have 2 seasonics PSU - but didn't dare to open the box voiding the warranty and finding out that they don't have the pots...I probably will return them and go for one of the Foxron units.
Once the cables are made - I will check the voltages ( before and after) and if there is a significant difference will I open the box and play with the pots.

Cheers


Mike

 

[mikeguava]

Quote:
14 ga is overkill. You would be using something capable of 20 amps in open ait to a connector that is rated at 5a.

Current carrying capacity isn't the issue in this situation. Rather it is voltage drop due to wire size, wire length, and load current. As I mention above, a 5 volt 7 amp load connected to 10 feet of 18 gauge wire will result in 4.59 volts at the load.

Frank do you think it would be ok to use an 18AWG and justraise the voltage output from the PSU via the pots?

A a couple of drinks beer are coming your way too!

Cheers

Mike

 

[edfair]

Agree about the voltage drop being a potential problem. But I am one that holds to optimism with power cable or wires and would have to have the VOM at the load measuring what the real applied voltage ended as before I would believe it. My concern is the 7 amp figure. Possibly overstated but who knows. But the option is to jumper drive connections at the computer end and effectively have parallel 18 ga feeding 2 drives if you have such. Or even paralling all the 5 and 12 v and associated 0v returns for the drive connections in the box.

Ed Fair
Give the wrong symptoms, get the wrong solutions.

 

[Frank4d]

Frank do you think it would be ok to use an 18AWG and justraise the voltage output from the PSU via the pots?

It is not a good idea... unless you are absolutely certain that you won't unplug something without first turning the pot down. The load would be reduced causing the voltage to rise which could fry your MOBO.

 

[Frank4d]

But the option is to jumper drive connections at the computer end and effectively have parallel 18 ga feeding 2 drives if you have such. Or even paralling all the 5 and 12 v and associated 0v returns for the drive connections in the box.

Good idea. Paralleling the wires this way would have the desired effect if done at the computer end.

 

[CobolKid]

As long as you increased the length of wire for all the outputs by the same length, then adjusting the regulator would theoretically work, however you will run into two distinct problems.

First, p.c. power supplies generally incorporate "output clamps" for protection. These typically consist of a silicone-controlled rectifier (SCR) connected across the positive and negative of each particular voltage, which are fired by over-voltage detectors (and usually crowbar detectors too.) Their purpose? To protect the p.c. from meltdown should the rectifiers, a regulator circuit or a power fet in the supply fail. In the realm of solid-state electronics, fuses and breakers are incredibly slow acting, while SCR’s are quite fast. SCR’s are used because once an SCR fires, it will not stop conducting from anode to cathode until the control signal is removed and its anode voltage drops to zero. This means that a nanosecond over-volt or a one-cycle crowbar will fire the SCR, which will then conduct, pulling down the output that it is protecting well before damage can occur, and then hold it down until the supplies fuse melts, or the user unplugs it. In some supplies, the clamps are adjustable, in most they are fixed.

If you increase wire length by nine feet, you will probably see a voltage drop that is around one-volt at twelve volts, even with high quality larger gauge wire. If you crank a voltage up, you will be narrowing the differential between its over-volt clamp and the additional ripple that will occur at the supply will have the effect of narrowing the tolerance of the voltage’s crowbar clamp. Even if you can isolate the clamps adjustment, you still should not change it without the proper test equipment and knowledge; if you adjust it incorrectly, you risk a meltdown.

Second; typically, not all the voltages in a p.c. power supply are adjustable; often you’ll find one or two that use another as a reference. Considering the load caused voltage variance that would be present at the nine foot end, I’d have some doubt that you could accurately re-adjust all voltages to remain withan an acceptable range.


Steve

 

[mikeguava]

Hi,

another important input - thanks Steve.

I actually had found this PSU for example

http://www.newegg.com/app/ViewProductDesc.asp?description=17-104-152&depa=0

which has adjustable voltage pots on the outside - to adjust voltages on the power rails.

I now realize the new problem with th SCRs. But I feel that I need to try nevertheless to at least minimally adjust the rails - hoping to get the system stable. If the voltage was to drop really one volt - which I hope to avoid by using heavier gauge wire - I think I'd need to take this step.


If I was to check the voltage drop with a multimeter - does the PSU have to be under full load then in order to adjust it properly? I was hoping to keep all the expensive parts clear of the motherboard and components and just test the PSU voltages on an old POS mobo without really anything attached to it

Cheers

Mike

 

[CobolKid]

The supply should be under it’s intended operating conditions (which should also be a somewhat stable load) to achieve best results when adjusting. You can initially adjust it under close conditions, however be sure to check all adjustable output levels once it is under its intended operating conditions. If you intend to power up the supply without connecting it to a computer, be sure to read the supplies label first, many power supplies list MINIMUM current loads for the different voltages. It is highly recommended you load ALL the outputs to a level that exceeds their given minimum loads, but does not exceed their given maximum loads, before powering it up. Although I've not used one myself, several inexpensive power supply test boxes are available that are suppose to do this, check out some of the on-line “wholesale” hardware sites for details. If you decide to obtain one, be sure it meets your particular supplies loading requirements. You will still need to be wary of the scenarios described in my last reply, in addition;

As to the wire gauge "thing," you will not eliminate the voltage loss of a direct current run; you will only moderate it and its subordinate effects to some degree. Yes, the value of parallelled resistances is the average of the value of each separate resistance; in this situation, you rapidly reach the point where a large gauge addition makes very little difference. That’s the math, i.e.; the "nature of the beast."

With D.C. the load is resistive, where with A.C. the load is inductive. Resistive == voltage, inductive == current. Voltage is determined by the source, Current (load) is determined by the demand. (Although in general, a demand that exceeds a source's capability will cause a voltage drop.) Furthermore, the load (demand) determines the actual effect of both a resistor and an inductor. With a resistor, no load equals no voltage drop, as you increase load the voltage drop increases too. With an inductor; as you increase load, the current loss caused by the inductor itself increases. Elimination of the resistive and inductive losses encountered in D.C. and A.C. power transmission has another name, it’s called superconductivity.

What would be best? To locate a supply that has remote sense lines for the different major voltages. This is common with (other than pc) much of the electronic equipment of the world. In this scenario, each voltage has a +, a -, a sense + & a sense - ; you connect the sense lines to the point at which voltage regulation for the particular voltage is desired. Note that the sense wires themselves suffer negligible voltage drop because the sense circuit causes very little load, it is a “signal.” But why go to such an extreme? As the actual voltage drop incurred in D.C. power transmission is determined by both the distance (wire length) and load (demand); the further away from the regulation point one gets, the more the voltage will vary for a given change in load. Computer’s demands (loads) tend to vary substantially, constantly and quite rapidly. The result? Lots of hf, mf & lf noise at the machine as the power runs grow in length. A dvdm won’t reveal this, it requires a reasonably fast oscilloscope to see. You may not suffer obvious difficulties; any damage may be far subtler, in fact, it may not rear its ugly head for some time.

I’ve seen high end p.c. supplies with remote sense, it’s been awhile and I don’t run one because they were quite expensive, however I do run multiple supplies on both my workstation (680 watt & 500 watt) and its exact duplicate, due primarily to the number and types of connected peripherals AND the randomly occurring problems I encountered when I first assembled them. Took a bit more than most would imagined to do.



FYI; Edison’s first power systems were D.C., if you were next door to the supply your lights were bright, but within a few blocks they were dim & yellow. One mile away there was no glow at all. Lights flickered and changed intensity as the total load (from all connected) on the transmission wires changed. It was actually Tesla who “invented” Alternating Current (A.C.,) and the shunt wound (rotating opposed magnetic field) A.C. motor, which revolutionized the “whole power thing


Steve

 

[Frank4d]

If I was to check the voltage drop with a multimeter - does the PSU have to be under full load then in order to adjust it properly? I was hoping to keep all the expensive parts clear of the motherboard and components and just test the PSU voltages on an old POS mobo without really anything attached to it

The +5 volt output needs to be loaded to at least 5 amps or it will not regulate properly... which will cause the +5, +12, and -12 outputs to be low. When you plug real loads in, these outpus will rise. Best to plug in the old POS MOBO and then measure the outputs.