A loading resistor for picky PSUs?

PSUs: The source of DC power for all components in the PC & often a big noise source.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee, Devonavar

Post Reply
notareal
Posts: 67
Joined: Mon Jul 11, 2005 9:03 am
Location: Finland

A loading resistor for picky PSUs?

Post by notareal » Tue Mar 14, 2006 5:36 pm

MikeC wrote: The simple solution is to add loading resistors within the PSU to guarantee the min load. Done simply, this will cost a couple percent in efficiency, maybe an additional 10-15W power draw all the time. (I believe this is the approach taken with the Antec Smart Power 2 series, which explains its relatively lower efficiency compared to the NeoHE from the same make.) The more sophisticated PSUs are apparently going to current-sensitive loading that only applies the resistance when the current is too low.
Tired of picky NeoHE? Perhaps we should design a nice little loading device that can be connected to a molex connector. It should turn off after the PC has power on.

Maybe some DIY fan controller that will turn fans on full speed for a moment and then slow fans down. Any better ideas?

TomZ
Posts: 386
Joined: Tue Feb 07, 2006 2:59 pm

Post by TomZ » Tue Mar 14, 2006 6:09 pm

For me, I think it is more hassle than it is worth. I would rather expend some efforts to RMA the PSU and/or replace it with one of a better design that doesn't suffer from this problem. There is no reason the PSU has to have this limitation.

GoranS
Posts: 18
Joined: Fri Jan 23, 2004 2:03 pm
Location: Ronneby, Sweden

Re: A loading resistor for picky PSUs?

Post by GoranS » Wed Mar 15, 2006 12:47 am

notareal wrote: Tired of picky NeoHE? Perhaps we should design a nice little loading device that can be connected to a molex connector. It should turn off after the PC has power on.
Won't the PS shut down or "act strange" if the load drops below the "1 amp" limmit a short while after startup?

Adding a Zalman fanmate might do the trick? If I’m not mistaken it uses voltage division to control the fan speed. The current the fan uses needs to pass through this variable resistor as well and it gets warm or even hot => power is drawn. ;) I’m not sure if the power drain on the +12V rail will be equivalent to the fans rated power consumption at full speed, but that would be my first guess. :D

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Wed Mar 15, 2006 1:37 am

The max power draw of any VR is generally no more than than the load it's controlling, so I don't think a low power fan is going to be enough, w/or w/o a VR.

Earlier today, I experimented with resistors on the 12V line to see how low the load could be before one of our afflicted PSUs would stop turning on. With 1A on the 12V line (12 ohm resistor), this particular PSU would always turn on. Dropping the load down to 0.46A still turned it on. Less than 0.4A and it would not turn on. At least for this particular high efficiency PSU, a 24 ohm resistor across any 12V output leads was enough to always allow it to boot. The power dissipated in the 24ohm resistor was 6W. It got fairly warm.

In a real system, removing the dummy load after boot should have no effect; the 12V power draw would always be high enough from the other components once the system is running. If you have a low power system and put it in sleep mode, you might have trouble though...

Also, in a real system it might be possible to lower the dummy load further, perhaps to 0.3A (3.6W)... because even if the motherboard has a turn on delay for the 12V current to the CPU, there should be some draw even from the HDD so the total 12V current could be >0.46A.

I will experiment more with the early NeoHE PSUs on hand & some of the 12V-delayed mobos to see how much of a dummy load is need on the 12V line for consistent turn on. For a typical minimalist system (1 HDD, 1 GHz RAM, modest vidcard, 60W CPU) I am guessing that it will be no more than 0.5A -- 24 ohms.

notareal
Posts: 67
Joined: Mon Jul 11, 2005 9:03 am
Location: Finland

Re: A loading resistor for picky PSUs?

Post by notareal » Wed Mar 15, 2006 1:53 am

GoranS wrote: Won't the PS shut down or "act strange" if the load drops below the "1 amp" limmit a short while after startup?
Tested this with my NeoHE; after PC managed to power on, I simply disconnected those spare devices I been using to help power on and NeoHE still stayed on, reboots ok. But if I turn it off, it will not power on, unless the additional load is added again. So removing the load after power on is not a issue.

Edit: Did not test the sleep mode!

GoranS
Posts: 18
Joined: Fri Jan 23, 2004 2:03 pm
Location: Ronneby, Sweden

Post by GoranS » Wed Mar 15, 2006 4:14 am

MikeC wrote: Dropping the load down to 0.46A still turned it on. Less than 0.4A and it would not turn on. At least for this particular high efficiency PSU, a 24 ohm resistor across any 12V output leads was enough to always allow it to boot.
So it would not be completely unreasonable for a single SATA II barracuda to put enough load on a +12V rail? My particular 250 GB drive is rated to draw 7,2 watts in idle mode. (I assume that is from the +12V line as it is the only one mentioned in the specs.) That would draw 0,6 amps from the +12V rail. Adding a low power fan + fanmate to the same rail could perhaps add another 0,1 - 0,3 amps and that rail should be in business. :)
... At least in theory. :roll:

I saw a post somewhere in the forums that questioned how much current a hard drive draw right from the start. Do they really draw less in the startup moment? One way of reasoning would suggest they could draw more current instead. That would be because they accelerate from a standstill up to full rotational speed, and that should require some extra power. But maybe there is more to it?

jaganath
Posts: 5085
Joined: Tue Sep 20, 2005 6:55 am
Location: UK

Post by jaganath » Wed Mar 15, 2006 4:30 am

I saw a post somewhere in the forums that questioned how much current a hard drive draw right from the start. Do they really draw less in the startup moment? One way of reasoning would suggest they could draw more current instead. That would be because they accelerate from a standstill up to full rotational speed, and that should require some extra power.
Here is the manual for the Seagate 40GB 7200.7 PATA; current profile at startup is on page 9:

ST340014A manual

On the 12V rail there is a clear surge power draw at startup (as you would expect) to about 1.6A and then drops down to about 0.5A after 6 seconds; the difference between startup and operating current on the 5V line is less marked.

Devonavar
SPCR Reviewer
Posts: 1850
Joined: Sun Sep 21, 2003 11:23 am
Location: Vancouver, BC, Canada

Post by Devonavar » Wed Mar 15, 2006 7:06 am

GoranS wrote:So it would not be completely unreasonable for a single SATA II barracuda to put enough load on a +12V rail? My particular 250 GB drive is rated to draw 7,2 watts in idle mode. (I assume that is from the +12V line as it is the only one mentioned in the specs.) That would draw 0,6 amps from the +12V rail. Adding a low power fan + fanmate to the same rail could perhaps add another 0,1 - 0,3 amps and that rail should be in business. :)
... At least in theory. :roll:

I saw a post somewhere in the forums that questioned how much current a hard drive draw right from the start. Do they really draw less in the startup moment? One way of reasoning would suggest they could draw more current instead. That would be because they accelerate from a standstill up to full rotational speed, and that should require some extra power. But maybe there is more to it?
In theory, you should be correct. Most hard drives draw in excess of 1A on the +12V line when they spin up and then settle down to ~0.3-0.4A at idle. By that time, though, the processor should be active and the power draw from the HDD shouldn't matter.

However, if that was actually the case, none of the power supplies out there should have problems at all. The issue seems to become more complicated where hard drives are involved.

For the PSU that MikeC tested, we found that if wouldn't power up even with a total of three IDE drives connected, which should have drawn well in excess of 1A.

My hypothesis is that, for a crucial split second when the power supply is first turned on, hard drives do not draw any power from +12V. Perhaps it needs to initialize the disc controller before it spins up the disc. In any case, the power supply in question would not start up with only a hard drive connected.

GoranS
Posts: 18
Joined: Fri Jan 23, 2004 2:03 pm
Location: Ronneby, Sweden

Post by GoranS » Wed Mar 15, 2006 9:01 am

Devonavar wrote:For the PSU that MikeC tested, we found that if wouldn't power up even with a total of three IDE drives connected, which should have drawn well in excess of 1A.

My hypothesis is that, for a crucial split second when the power supply is first turned on, hard drives do not draw any power from +12V. Perhaps it needs to initialize the disc controller before it spins up the disc. In any case, the power supply in question would not start up with only a hard drive connected.
Is there any reason to expect it to be any different for SATA II drives? I saw a post somewhere suggesting that it might be. (Someone said that one or possibly two PATA drives did not make the NeoHE work but one SATA drive did.)

How about fans... Do they draw their peak current when they accelerate? Would that help?
...But I guess any help from that would be just about voided if a fanmate was used to keep the fan at a noiseless level. :(

Devonavar
SPCR Reviewer
Posts: 1850
Joined: Sun Sep 21, 2003 11:23 am
Location: Vancouver, BC, Canada

Post by Devonavar » Wed Mar 15, 2006 9:52 am

There is anecdotal evidence that SATA drives draw more power than PATA drives, but from what I've seen the increase is mainly on the +5V line. I don't know that it would help. Also, there are a few SATA drives that will not spin up at all unless there is an active disc controller connected at the other end (mainly Hitachis). Since the disc controller is on the motherboard and the motherboard isn't drawing enough power, these drives may be likely to cause more problems. Also, SATA drives that support staggered spin-up may also cause problems.

I don't think there's any hard and fast rules for hard drives. Some may work, others may not. I'd like to be able to prove one way or the other, but I don't think there's enough evidence out there...

pony-tail
Posts: 488
Joined: Sat Aug 23, 2003 4:39 pm
Location: Brisbane AU

Post by pony-tail » Wed Mar 15, 2006 1:35 pm

It's a bit late for me but I have had this problem (with an Asus A8N-SLI premium) The problem does not seem to be confined to the Antec PSUs but to a few others as well - But they all seem to be High end (expensive ones ) .A $50au Omni (abrand I had not heard of) booted my machine without issue - yet a $250au Phantom will not .
Why have they setup these PSUs so as to need a load on the 12v line just to boot , where the older ATX PSUs do not ?
I know that the time out on the Asus board has to do with compatabilty , but it seems like the two camps are not talking to each other and the consumer pays.
I wound up getting a Silverstone 650 watter because that was the only good quality PSU they had in the shop that would boot my machine . The resistor ,if it works in all cases is a good cheap fix , but I have to ask - should we have to do this just to get the thing to work ?
Even my soloution is not that good , the noise drives me nuts but my computer works (very fast and completely stable ) and all my random reboots , no powerup issues are gone .
I hope they fix this in there next generation - without introducing some new catastrophic issue . so I can just sell this thing on and upgrade - most likely minus the SLI For me it just was not worth the bother.
Great work yet again - MikeC

notareal
Posts: 67
Joined: Mon Jul 11, 2005 9:03 am
Location: Finland

Post by notareal » Thu Mar 23, 2006 2:45 am

Here's a simple circuit I used to solve the power on problem of the NeoHE. Thanks to cmcquistion's Terminal Block controller and Falstad's Circuit Simulator.

Code: Select all

 notareal's simple diode fan controller with a dummy load aka the NeoHE FIX.

     D1        D2        D3        D4        D5        D6    
 +---|>|---+---|>|---+---|>|---+---|>|---+---|>|---+---|>|---+---------+---------+
 |                                                           |         |         |
 |   C1                                                      |         |         |
 |  +| |       R1                                            |         |         |
 +---  |---+--/\/\/--+                                       |         |         |
 |   | |   |   18    |                                       |         |         |
 | 10000u  |         |                                      FAN1      FAN2      FAN3
 |   R3    |   R2    |   D7                                  |         |         |
 +--/\/\/--+--/\/\/--+---|>|---+                             |         |         |
 |   200       18              |                             |         |         |
 |                             |                             |         |         |
 |                             |                             |         |         |
+12                           ///                           ///       ///       ///


D1...D7 = diodes 1N4xxx (drops voltage to 7.2 V)
R1...R2 = resistors 18 ohms / 20 W
R3      = resistor 2 x 100 ohms / 2W (I will replace this with 470 ohms + led)
C1      = electrolytic capacitor 10000 uF / 16 V
FAN1    = Antec TriCool (Waiting for the Ninja)
FAN2    = D14SL-12
FAN3    = D90SL-12

I just used some resistors and capasitor that I had available. This is not a optimised setup.

Code: Select all

 This maight be even better, as it will give a kick start to fans and thus alows rather low voltage in fans (more diodes).

     D1        D2        D3        D4        D5   ...  Dn    
 +---|>|---+---|>|---+---|>|---+---|>|---+---|>|---+---|>|---+
 |                                                           |
 |             Dn+1                                          |
 |         +---|>|---+---------------------------------------+---------+---------+
 |         |                                                 |         |         |
 |   C1    |                                                 |         |         |
 |  +| |   |   R1                                            |         |         |
 +---  |---+--/\/\/--+                                       |         |         |
 |   | |   |   18    |                                       |         |         |
 | 10000u  |         |                                      FAN1      FAN2      FAN3
 |   R3    |   R2    |   D7                                  |         |         |
 +--/\/\/--+--/\/\/--+---|>|---+                             |         |         |
 |   200       18              |                             |         |         |
 |                             |                             |         |         |
 |                             |                             |         |         |
+12                           ///                           ///       ///       ///


D1...Dn = diodes 1N4xxx (one diode will drop voltage 0.75-0.8V Add enough for your need)
R1...R2 = resistors 18 ohms / 20 W
R3      = resistor 2 x 100 ohms / 2W (I will replace this with 470 ohms + led)
C1      = electrolytic capacitor 10000 uF / 16 V
FAN1    = Antec TriCool (Waiting for the Ninja)
FAN2    = D14SL-12
FAN3    = D90SL-12

NOTE! This is not tested, just run at the simulator.

deCarnac
Posts: 3
Joined: Sun Mar 19, 2006 3:42 pm
Location: Norrköping, Sweden

Post by deCarnac » Fri Mar 24, 2006 1:39 pm

Usually I would have gone with a neat solution as posted above (I have a NeoHE that is not working yet). But I have no suitably large capacitors or high-power resistors, and they are a bit difficult and expensive to buy where I am. What do you think of this fairly cost-effective modder's alternative?:

Code: Select all

deCarnac's low-budget NeoHE fix (not tested yet)

             /                     
+12V -------o  o---------+
           switch        | 
         (optional)     (X) lamp
                         |  12V 10W
GND ---------------------+

Should draw about 0.8A.
Lamp may be a small halogen lamp. It could be used to illuminate the desk or PC chassis (or just be tucked away in there). Unlike a fan, it's noiseless. Either way, be aware of its high temperature!

MikeC
Site Admin
Posts: 12285
Joined: Sun Aug 11, 2002 3:26 pm
Location: Vancouver, BC, Canada
Contact:

Post by MikeC » Fri Mar 24, 2006 1:52 pm

deCarnac -- it would work. The light could even be useful atop the desk. ;)

qviri
Posts: 2465
Joined: Tue May 24, 2005 8:22 pm
Location: Berlin
Contact:

Post by qviri » Fri Mar 24, 2006 3:11 pm

Maybe I should get a P150 after all and run my router off the PSU :D

GoranS
Posts: 18
Joined: Fri Jan 23, 2004 2:03 pm
Location: Ronneby, Sweden

Post by GoranS » Thu Apr 13, 2006 1:37 am

MikeC wrote: Earlier today, I experimented with resistors on the 12V line to see how low the load could be before one of our afflicted PSUs would stop turning on. With 1A on the 12V line (12 ohm resistor), this particular PSU would always turn on. Dropping the load down to 0.46A still turned it on. Less than 0.4A and it would not turn on. At least for this particular high efficiency PSU, a 24 ohm resistor across any 12V output leads was enough to always allow it to boot. The power dissipated in the 24ohm resistor was 6W. It got fairly warm.
I got around to installing a power resistor in my system. Now I can keep my fans at the lowest speed even at startup. :D But I have to say, “fairly warm” was no exaggeration. ;) In non moving air my resistor runs at about 65°C.

I used a 20 ohm resistor that can take up to 30 watts, max temp. 150°C. I mounted it on a cooler with twice the cooling capacity the resistor requires. Then I mounted the cooler and resistor on the outside directly in front of the 120 mm case fan so that it will get some cooling. I placed it dead center on the fan grill to obstruct the airflow as little as possible. In this position the temperature drops to on average 58°C (the temperature was measured on the cooler, both close to the resistor and as far away as possible. There was a 2-3 degrees difference).

Now to my question. Is the temperature I measured reasonable? A 20 ohm resistor should generate 7,2 watts on a 12 V line. Is this really enough to bring a big cooler up to around 60°C? Or did I muck something up?

Resistor: http://www.elfa.se/elfa-bin/setpage.pl? ... 012526.htm
Cooler: http://www.elfa.se/elfa-bin/setpage.pl? ... 245192.htm

Note: A cooling paste was used between resistor and cooler. Don't try without.

notareal
Posts: 67
Joined: Mon Jul 11, 2005 9:03 am
Location: Finland

Post by notareal » Thu Apr 13, 2006 3:35 am

GoranS wrote: I got around to installing a power resistor in my system. Now I can keep my fans at the lowest speed even at startup. :D But I have to say, “fairly warm” was no exaggeration. ;) In non moving air my resistor runs at about 65°C.
Sounds correct, the surface of the resistor is quite small. I used two much larger wire resistors. If I where you I would build someting like this to cut power use of the dummy resistor. It should give enough time to power on (~500 ms) and also discharge the capacitor after the power is turned off.

Image

or like here.

GoranS
Posts: 18
Joined: Fri Jan 23, 2004 2:03 pm
Location: Ronneby, Sweden

Post by GoranS » Fri Apr 14, 2006 2:36 am

Yes, that would bee a neat solution. :) But the way I see it the system I have draws to little power for me to turn of the dummy load. :( ...Why on earth did I get that silly thought in my head then, you might ask.

Well, before the resistor was mounted I used the speed of the fans to regulate the power drain on the system. This worked fine for booting, I just cranked up the case fan and powered up the system. Then I reduced the fan speed to a minimum. To my surprise it was on several occasions not possible to turn the computer off with the fans on low speed!?!? :shock: Windows would start shutting down and right when you would expect the computer to actually power of, it would just sit there with the stupid “windows is shutting down” sign. As I figure it the system reduces the power drain at shutdown which brings it below the PSU limit for switching off and the system simply stays on. :? ... In a very unexiting mode I might add. ;)

As for the small surface of the resistor it should be countered by the quite large cooler it's attached to. Or do you mean to tell me that your wire resistor has more surface area than the resistor cooler I use?!?! :shock: ...... ;)

notareal
Posts: 67
Joined: Mon Jul 11, 2005 9:03 am
Location: Finland

Post by notareal » Sat Apr 15, 2006 2:25 am

GoranS wrote:As for the small surface of the resistor it should be countered by the quite large cooler it's attached to. Or do you mean to tell me that your wire resistor has more surface area than the resistor cooler I use?!?! :shock: ...... ;)
Those two resistor I used have total surface area ~18 cm2 or greater as the surface is not smooth. I think they are also called as cement resistors. No heatsink needed. The larger one has similar size (10W100 ohm) rating). Total cost 1.60 €.

Image

Post Reply