How much resistance do I need?

gustavs_a · Post by **gustavs_a** » Tue Feb 17, 2004 2:16 am

The Zalman Fanmate1 still doesn't slows down the fan enough, I would like to make it 2-3 times slower (and then fine tune the speed with the Zalman controller). What resistance resistor do I need to use? Please post a formula. Thank you!

zuperdee · Post by **zuperdee** » Tue Feb 17, 2004 3:33 am

Whoa there--may I ask what kind of fan are you using? It must be a pretty high-speed fan if you still can't get it to be slow enough with a Zalman Fanmate all the way down at 5 volts.

I'd suggest that instead of trying for even more resistance, you should really consider looking into a lower speed fan. With the voltage going lower than 5 volts, I'd be concerned about that fan being able to start. Not only that, but a lower speed fan would likely be quieter than running your current fan at less than 5 volts, too.

gustavs_a · Post by **gustavs_a** » Tue Feb 17, 2004 4:16 am

I'm trying to modify the PSU, I want to bypass the PSU fan controller to fix the revving up and down issue and use the voltage from MB sysfan connector. I haven't replaced the stock 80mm SuperRed fan yet, but even with the stock fan it's much too fast. I don't know how much voltage comes from the sysfan connector, but it's obviously much higher than from PSU fan connector. A resistor could fix the problem.

zuperdee · Post by **zuperdee** » Tue Feb 17, 2004 5:19 am

gustavs_a wrote:I'm trying to modify the PSU, I want to bypass the PSU fan controller to fix the revving up and down issue and use the voltage from MB sysfan connector. I haven't replaced the stock 80mm SuperRed fan yet, but even with the stock fan it's much too fast. I don't know how much voltage comes from the sysfan connector, but it's obviously much higher than from PSU fan connector. A resistor could fix the problem.

1) What revving up and down issue?

2) SuperRed fan--I assume this is a Seasonic Super Silencer in that case?

Well, I'd suggest in this case, given what you're saying, it might still be best to try some other things.

First, just figure I might as well give a disclaimer: 1) I have never owned or hacked a Seasonic power supply before. 2) Although I know a fair amount about electronics, I am NOT an electrical engineer, so I suggest you read what I say with caution.

Second, I wonder if you have a defective unit that could be exchanged--have you considered that? The behavior you describe doesn't sound like something a good PSU should be doing.

I do think it sounds like you might have a faulty fan control circuit. If you can't return the unit, I'd say your best bet is going to be to try to bypass the circuit altogether, if you can. I don't know how one would go about designing a replacement circuit, however, so I'd say your best bet is going to be to try to see if you can find some OTHER way to control the fan speed based on the temperature of the heatsinks in the PSU, maybe by replacing the SuperRed fan with some thermistor-based fan, and mounting the thermistor to the heatsinks somehow, so the speed could be controlled linearly by the temperature. I would be careful however, to try to maintain the stock maximum airflow value if possible, cause I'm sure Seasonic must have known what they were doing in that regard. I'd hate to see you kill your power supply with too little airflow.

If you really would rather do the resistor thing in the end though, the easiest way is probably to take a look at this link: http://www.silentpcreview.com/files/misc/fancalc.zip. According to the description: "Windows utility by SPCR member Kostik computes the in-series resistor value needed to reduce fan voltage from the standard 12V to any level. Current or power rating of fan required; utility preloaded with database of fans. Version 0.4 adds RPM values & calculation, choice of 12v, 7v or 5v line voltage (default = 12v). Results displayed in a table. Great for Ohm's-Law-challenged modders!

"

gustavs_a · Post by **gustavs_a** » Tue Feb 17, 2004 5:55 am

1) A very slight change in sound pitch (from fan speeding up) when changing from 0 to 100% CPU usage. I can't hear it now, but yesterday evening when it was very quiet I could notice it. I don't think the PSU is defective, the old PSU had the same thing more pronounced.
2) Yes it's Seasonic Super Silencer 460W. I like it, it's quieter than my old Codegen 350W.

I generally don't like controlled fans, the changes in RPMs make dynamic noise that can be very annoying. The changes in RPMs are very little, I think it would be ok to use a constant RPM.

Thanks for the utility, it's just what I needed!

fancontrol · Post by **fancontrol** » Wed Feb 18, 2004 11:37 am

interesting. i ran the L1A and plotted the resistance and speed calculations vs. desired voltage.

it seems Fan Calc assumes a constant effective resistance of the fan (171.4ohm in this case), and a linear relationship between speed and voltage (S = 173 * V - 173).

then i went back to the lab and took some measurements. i'll do my best to explain without the ability to post graphs.

I used a bench power supply for flexibility. I can set the voltage for the fan (say, 9V) and the supply shows the current (say, 140mA). From that I can calculate the effective resistance of the fan (9/.14 = 64.3ohm) and the resistor value required to get there ((12-9)/.14 = 21.4ohm).

I tested two fans, one 12v w/o tach and a 24v with tach (it's all i could find lying around).

It turns out that the fan's effective resistance varies with voltage. The 12v fan, for example, goes from 57ohm @ 12v to 80ohm @ 4v. And it's not a straight line. There is, however, a linear relationship between voltage and current, but there is an offset (e.g. I = .02 V - 0.04, in the case of the 12v fan), which explains the non-linearity of resistance. So, to get resistance right--and, thus, the value of the drop resistor--you need to know two constants, not just a calculated resistance. Fan Calc falls short here.

I also used a 'scope to measure speed of the 24v fan (in free air). I've read before that speed is linear w.r.t. the square root of voltage. That was dead on. So to get this right, you also need two values. Fan Calc gets close with "Fan RPM" and "voltage drop from fan", but does not take into account the square-root thing.

Bottom line: It's a great tool because of it's pre-loaded constants and quick calculation of values you need. Just don't expect it to be dead on. But then, all the variance from fan-to-fan and use-to-use would probably cause one to hesitate about any exact numbers, anyway.

Since this post is freakin' huge already, I might as well post the data. You're on your own to do the calcs, graphs, etc.

Code: Select all

From Fan Calc, L1A fan:
volts	Rdr	rpm
11	 15.58	1727
10	 34.28	1554
 9	 57.14	1381
 8	 85.71	1209
 7	122.44	1036
 6	171.42	863
 5	240.00	690
 4	342.85	518
 3	514.28	345

2w 12v fan	
volts	current
12	0.21
11	0.185
10	0.165
 9	0.14
 8	0.12
 6	0.08
 5	0.06
 4	0.05

3w 24V fan
volts	current	rpm
20.0	0.50	2880
16.5	0.39	2490
15.0	0.35	2310
12.5	0.28	1980
10.0	0.21	1590
 7.5	0.14	1170

Michael Dooley · Post by **Michael Dooley** » Wed Feb 18, 2004 6:10 pm

That's pretty interesting fancontrol.

I used a low tech method to find the potentiometer resistance that I needed in a small project involving an AOpen H340 case and a VIA ME6000 muthaboard. I went to Radio Shack and obtained a 25K pot generally sold for volume control (I think). I put this pot in line with the feed to the PSU fan in order to control the case cooling noise and adjusted the pot so that the PSU fan was turning verrry slowly and would also resume rotation when I turned the computer off and on again.

I measured the pot's resistance and found that it was approximately 183 ohms. I bought a few (hey, they came in a baggie) 100 ohm 5 watt linear pots from an electronics supply house and installed one to replace the Radio Shack rheostat. It works like a charm.

I couldn't think my way out of a paper bag...

HammerSandwich · Post by **HammerSandwich** » Wed Feb 18, 2004 11:26 pm

fancontrol wrote:interesting. i ran the L1A and plotted the resistance and speed calculations vs. desired voltage.

YEEESSSSS - numbers to play with! Thanks for the data, FC.

fancontrol wrote:...there is an offset (e.g. I = .02 V - 0.04, in the case of the 12v fan), which explains the non-linearity of resistance.

Would you clarify that one, please?

fancontrol wrote:I've read before that speed is linear w.r.t. the square root of voltage. That was dead on.

I don't see that in your 24V data. If RPM were proportional to the root of V, then the 16.5V RPM should have been 2615.

This has long looked like a simple aerodynamic problem, but I never got a good fit before. Turns out I was missing the fan's changing resistance! Aerodynamic power goes up as the cube of speed, so I tried:

Code: Select all

New RPM = Original RPM * (New Power / Original Power)^1/3

which gives these results:

Code: Select all

3w 24V fan
volts  current  Actual   Predicted
20.0    0.50     2880    baseline
16.5    0.39     2490    2486
15.0    0.35     2310    2323
12.5    0.28     1980    2030
10.0    0.21     1590    1712
 7.5    0.14     1170    1359

This gives a satisfying match at 12.5v or more with increasing error at the lower settings. I suspect that at high speeds the aero drag overwhelms bearing friction, which becomes more important at lower speeds. That makes sense because bearing friction should be linear to RPM while aero drag is squared.

fancontrol · Post by **fancontrol** » Fri Feb 20, 2004 6:04 pm

plot V vs I. it's pretty much a line.
I = 0.286 * V - 0.764, R^2 = 0.9993

The curious part is that, unlike a resitor, the Y intercept is not zero.

and

plot (V)^.5 vs S. it's pretty much another line.
S = 990 * (V)^.5 -1535, R^2 = 0.9997

oh, and, curiously, in PWM systems the relationship between speed and duty cycle looks linear as well. I expected it to be the same as voltage (e.g. 50% == 6V). Not true.

that's all I know.

cpemma · Post by **cpemma** » Sat Feb 21, 2004 4:21 pm

I plotted voltage against speed (using MBM5 readings for rpm) for the 60mm stock Delta on my old Taisol 760 HS. Result was a roughly straight line between 7.5v and 12v, but cutting the speed axis well above zero rpm at zero volts if you extrapolate the best fit linear (though it's obviously really curving in the right direction).

I'm not sure of all the reasons, but taking a car analogy, the motor may be less efficient at high speed; I've heard mention of a "sweet spot" around 10v. As speed goes up it's working harder to overcome air and frictional resistance, and coil inductance must play a part.

A minor factor is that all fans lose a constant voltage across the transistor switch, typically around 250mV.

I wish I had something to measure RPM, I'd like to get more info on this.