voltage controlled variable resistor?

A

Autochthon

Limp Gawd
Joined
Apr 28, 2005
Messages
427
Ok I'm tired of manually adjusting my cpu fan but want to keep my high currrent fans. Is there a variable resistor or ciruit that can replace my 2.5K pot? basically it needs to vary from 0 to 2.5k ohms as voltage goes from 0V to 12V.
thanks
 
thanks good stuff and I've actually built and use a couple of those circuits. But I was hoping for something I could plug into my existing controller that I could control from the voltage output wire on the motherboard.
 
there's digital pots... but you might need some extra circuitry to make it work since i'm not exactly sure what signal will be controlling it
 
My thought was that the voltage output line of my MB fan cpu fan control header sent a voltage ranging between 0v and 12v and can be linked through software and adjusted according to MB temperature sensors. I am currently manually setting this using a 2.5 k pot on a circuit like one on the website "theshadow27" linked to. If I could replace the 2.5 k pot with an equivalent device that would vary resistance with input voltage. I could then adjust my high current fan through standard software.

I'll look into digital pots thanks
 
It would take a little figuring, but a FET could serve your purposes quite nicely. You'd have to figure out some sort of feedback scheme to make it work, but it could quite possibly be simpler in the end than using a microcontroller to drive digital pots and whatnot all.

SOmething to look into. I'd do some of the figuring for ya, but time is a little short at present;)

good luck eh.
 
well, if he's using a voltage between 0 and 12 volts to control it, i'd put an ADC in line with a digital pot and call it a day.

or possibly some type of opamp circuit could acheive what he wants...
 
plot said:
i'd put an ADC in line with a digital pot and call it a day.

or possibly some type of opamp circuit could acheive what he wants...
Click to expand...

I looked at digital pots a bit and I see there are 256 step 2.5K ohm versions that would work fine if I can come up with a way to simply convert the voltage into a DC signal. All of the circuits I came across seemed fairly complicated. Any recommendations for the ADC?

As for Fets a brief glance tells me I haven't any idea how I start to use one to convert voltage to resistance. Basically it looked as though you could use them as voltage controlled switches so I can imagine it is possible.

thanks for replies I'll read up a bit more on ADC and fets

here is the best I've found so far but it seems to need a clock input signal.

This ADC in conjunction with a digital pot may work. Would I need to adjust the 0 - 12V input to work within the range of this ADC? I've never used one of these is Vref my input signal? or is it Vdd? Would I put a 12V signal to Vref give me a 0 to 12V range for Vdd? or is that backwards?

What you think?
 
How much current is "high current"? I think it would definitely be possible to rig up an op-amp circuit to do it.
 
My current circuits can handle 5A, but If I just hook up the 1 fan it's 1A. I don't know if that 1A is startup current or not. My present circuits with heaviest loads have 1A, 1A and 2A

This Digital Pot might work? I'm guessing SPI is serial port interface.
 
Autochthon said:
My current circuits can handle 5A, but If I just hook up the 1 fan it's 1A. I don't know if that 1A is startup current or not. My present circuits with heaviest loads have 1A, 1A and 2A

This Digital Pot might work?
Click to expand...


Holy crap! That's gotta be a hefty hefty fan.

Edit: I don't think that'll work for you either. It lists the maximum voltage at the resistor terminals as Vdd, and 2.7V < Vdd < 5.5V.

I'm thinking a opamp+FET combination may be in order here. I'm really hungover but I'll try to whip something up.
 
vantec tornado is rated 1A, the two sunon 120mm are .5 A each and the delta 80mm are .86A each I have 2 of those and a .36A panaflo on one circuit
 
fat-tony said:
Holy crap! That's gotta be a hefty hefty fan.

Edit: I don't think that'll work for you either. It lists the maximum voltage at the resistor terminals as Vdd, and 2.7V < Vdd < 5.5V.

I'm thinking a opamp+FET combination may be in order here. I'm really hungover but I'll try to whip something up.
Click to expand...

I only have a vague idea as to what plot had in mind. But would the voltage at the Dpot matter? isn't it just taking a digital signal from the ADC? the ADC seems to be able to handle voltage from 0 to Vref as long as both chips are powered by a 5V source I wouldn't have thought it mattered that Vref is up to 12V?
 
Autochthon said:
I only have a vague idea as to what plot had in mind. But would the voltage at the Dpot matter? isn't it just taking a digital signal from the ADC? the ADC seems to be able to handle voltage from 0 to Vref as long as both chips are powered by a 5V source I wouldn't have thought it mattered that Vref is up to 12V?
Click to expand...


The problem I see there is with the digital pot. I did just re-read some things and realized a few things I was missing, but my problem still stands :).

In the circuit in the page that theshadow27 linked to, you've got your pot down at the bottom there (they had a thermistor, same deal). The problem I saw was the actual voltage across it. In the datasheet for the AD5162, they list the maximum voltage ACROSS the actual resistor as Vdd. I didn't go and actually calculate the currents and voltages going through though, since your configuration will probably be different and I'm lazy :D.

However, if you just need to track the voltage put out by the mobo's header, I think I have a pretty simple circuit that will work for you, but want some feedback from other people first to make sure I'm not entirely out to lunch. I haven't picked specific parts, since I really have no "practical" experience with MOSFET selection (all theory baby :)).

So yeah, my warning stands: this circuit will probably work. I simulated it, and it looks OK. It might (probably will) need some caps in it to filter out noise In The Real World. The lab's closed today or I'd go build it and check to make sure I'm not entirely out to lunch here.

EDIT5: Removed circuit because it was wrong.

Edit: Oh yeah... sorry that it looks like ass. I hate circuitmaker, but it's all I had installed that would simulate. Make sure to ignore that part number. It has no grounding in reality. It was chosen at random.

Edit2: Also, that Vin is actually where the voltage comes from the mobo. It's just at 5V in the picture because I was testing it at different values.

Edit3: AHA! Another thing!

The motherboard is probably doing PWM already. If you can get ahold of an oscilloscope and check what's coming out of the board you could tell for sure. I realize that could be a tricky thing for a lot of people to get their hands on though.

If the board IS doing PWM, then you could drop the entire opamp and just use the FET as a switch.

Edit4: Hmmm. Once I replaced the opamp with a "non-ideal" one I found that it didn't work as well :). I'll drink some more water to try to clear my head and keep playin.
 
I should have been clearer this is the nearer the actual circuits I use though I have LM350's instead of 317. I'll look over your circuits thanks for the effort :)

I bet your right about the PWM from the MB. Though that makes me wonder where the heat from the circuit is being dissipated.
 
Autochthon said:
I should have been clearer this is the nearer the actual circuits I use though I have LM350's instead of 317. I'll look over your circuits thanks for the effort :)

I bet your right about the PWM from the MB
Click to expand...


I just realized that that circuit is entirely broken :).
 
Well... in my hung over state, I got sidetracked by all of the FET talk and completely forgot about the simple solution.

If you can a) tolerate a 0.7V drop between the motherboard voltage level and the output voltage level, and b) deal with about 160mA of draw from the MB power header then I think I have a solution for you :)

simpleanswer.png


I ran that all through the simulator, and I believe that it's entirely sound. If you short-circuit the fan, you may blow your mobo header or burn out the transistor, but if you keep it only hooked to the fans, you should be fine.

Edit: Oh yeah, that's actually the transistor to use too, and it will need a heatsink.

On digikey, that transistor is $2CDN.

Edit2: I've been thinking about the heatsink some more. What are you using for your LM350 right now for a heatsink?
 
fat-tony said:
I ran that all through the simulator, and I believe that it's entirely sound. If you short-circuit the fan, you may blow your mobo header or burn out the transistor, but if you keep it only hooked to the fans, you should be fine.

Edit: Oh yeah, that's actually the transistor to use too, and it will need a heatsink.

On digikey, that transistor is $2CDN.

Edit2: I've been thinking about the heatsink some more. What are you using for your LM350 right now for a heatsink?
Click to expand...

thanks again just keep drinking those fluids I need you of sound mind ;)

I'm using a HS I salvaged out of an old PSU for the one that I have clamped down to 3V as it gets warm. The other one only drops down to 6V so it is just an using an old pentium HS. They have to be isolated from ground because the LM350 case is Vout.

So what's the range on R1 in your circuit? could a safety be rigged for the MB header maybe with one of those diodes you hate so?

edit: Just took a look at that transistor. Is it just an power amp? If so, I wouldn't need 15A. I bet 2 or 3A would be plenty
 
Autochthon said:
thanks again just keep drinking those fluids I need you of sound mind ;)

I'm using a HS I salvaged out of an old PSU for the one that I have clamped down to 3V as it gets warm. The other one only drops down to 6V so it is just an using an old pentium HS. They have to be isolated from ground because the LM350 case is Vout.

So what's the range on R1 in your circuit? could a safety be rigged for the MB header maybe with one of those diodes you hate so?

edit: Just took a look at that transistor. Is it just an power amp? If so, I wouldn't need 15A. I bet 2 or 3A would be plenty
Click to expand...

R1 is actually the fan itself. I used a 3ohm to simulate at roughly 4 amps full bore, and a 2ohm on paper to simulate at 6 amps (at 12V). Using the 3ohm in CircuitMaker, I got a maximum load for the motherboard at .16A, which I figure is probably acceptable.

As far as the transistor goes, yes it's a power transistor. I chose it simply because it was relatively cheap and seemed to be a "standard" part. I can have a peak after work and see if there would be any problem using a smaller one. I'm not sure if it will result in much of a cost savings though.

Edit: Ahhh yes, and the heatsink. At 2ohm, I have a power dissipation of 18W, which seems like it could be a lot. As that resistance increases though, the power drops considerably. That's a considerably larger load than you'll be using though, so I think you should be fine. For a 1A (12ohm) load, you should be looking at a peak Pd = 3W. That will probably still require a heatsink, but not much of one.
 
ok sounds good I'll go track the transistor (or it's smaller equivalent) down and build a test circuit. I'm not concerned about the transistor blowing but would sure like to protect the motherboard header. I don't know enough know how to limit current draw on the mb output leg but it seems it would be fairly simple.

edit: you know I think I have a half dozen Tip31c around I think they're only rated at 3A 40V would they work?
 
Autochthon said:
ok sounds good I'll go track the transistor (or it's smaller equivalent) down and build a test circuit. I'm not concerned about the transistor blowing but would sure like to protect the motherboard header. I don't know enough know how to limit current draw on the mb output leg but it seems it would be fairly simple.
Click to expand...

if you can handle a .7v drop between the motherboard header and the transistor, stick a diode in there, it will protect the motherboard.

or, find a germanium diode, it's only a .3volt drop i think.
 
plot said:
if you can handle a .7v drop between the motherboard header and the transistor, stick a diode in there, it will protect the motherboard.

or, find a germanium diode, it's only a .3volt drop i think.
Click to expand...

How would that protect the motherboard? I'm not saying that there isn't a way to protect the mobo with a diode, but how would you install this to add further protection than what the BE junction already provides?

My concern with the motherboard is that there is no real current-limiting involved here. I don't think adding a diode will provide any further limiting...
 
i see, so you think the mobo is gonna source to much and fry itself?
 
Autochthon said:
edit: you know I think I have a half dozen Tip31c around I think they're only rated at 3A 40V would they work?
Click to expand...


Well, I think they could work. As long as you're careful about not overloading the thing and make sure they're well heatsinked. The Tip31C has pretty low gain it looks like. If you're driving 1A worth of fans, you can expect to be drawing around 40mA from the motherboard. That's not too bad.

I'm pretty sure that the piece of paper I had worked out on last night put the max power dissipation in the transistor around 18W. The data sheet for the tip31c says it can do 40W as long as you keep the case temperature at 25'C, which could be challenging.
 
plot said:
i see, so you think the mobo is gonna source to much and fry itself?
Click to expand...

From what I understand, people usually hook fans up to molexes instead of motherboard headers for that exact reason. I've never blown out a mobo fan header before, but it seems to be a prevailing concern.
 
fat-tony said:
From what I understand, people usually hook fans up to molexes instead of motherboard headers for that exact reason. I've never blown out a mobo fan header before, but it seems to be a prevailing concern.
Click to expand...

That is certainly my concern. If the fan manufacturer goes to the trouble to wire an extra molex connector, and leave out the two MB header power wires, that leads me to believe they don't want me pulling that much current off the header.

fat-tony said:
My concern with the motherboard is that there is no real current-limiting involved here. I don't think adding a diode will provide any further limiting...
Click to expand...

yep that's how i read your worry. I was thinking something like a fuse rather than a diode.

The only fan i'd use on this circuit is the 1A tornado. So it sounds like you think it will work? No capacitors needed? wire length from transistor to header and transistor to fan shouldn't even be 12 inches each.

I'm fairly certain temps at the location of the transistor shouldn't exceed 30C I have a temp probe laying against the side of my CPU/HS base that seldom goes above 30C at full load
 
Autochthon said:
That is certainly my concern. If the fan manufacturer goes to the trouble to wire an extra molex connector, and leave out the two MB header power wires, that leads me to believe they don't want me pulling that much current off the header.



yep that's how i read your worry. I was thinking something like a fuse rather than a diode.

The only fan i'd use on this circuit is the 1A tornado. So it sounds like you think it will work? No capacitors needed? wire length from transistor to header and transistor to fan shouldn't even be 12 inches each.

I'm fairly certain temps at the location of the transistor shouldn't exceed 30C I have a temp probe laying against the side of my CPU/HS base that seldom goes above 30C at full load
Click to expand...

As I said above, at 1A, you're not looking at very much current from the mobo header. I think it should work. I won't promise. It's been a while since I've actually made anything except for on a piece of paper (haven't had a lab for a year or so).

At 1A, it looks like your maximum power dissipation through the transistor will be about 3W. The datasheet specifies a maximum of 2W for an ambient temperature of 25'C, or 40W for a case temperature of 25'C. From what I understand, that implies that you will have to heatsink it to keep the case temperature lower.

If you're interested, here's how I figured that out:

Vout = Vin - 0.7V
Vce = 12V - Vout = 12V + 0.7V - Vin = 12.7V - Vin
Ic =(Vin - 0.7V)/Rfan

Ptransistor = Vce*Ic
= (12.7V - Vin)*(Vin-0.7V)/Rfan
= (-Vin^2 + 13.4Vin - 8.89)/Rfan

To find the maximum, we take the derivative wrt Vin

d Ptransistor/d Vin = (-2*Vin + 13.4)/Rfan = 0
=> Ptransistor is at a maximum when Vin = 6.7V

Plug it back in:

[email protected] = (-(6.7)^2 + 13.4(6.7) - 8.89) / Rfan
= 36V^2 / Rfan

Now, if the fan draws 1A at 12V, by ohm's law it must have a resistance of 12ohms (V=IR).

[email protected] = 36V^2 / 12ohm = 3W
 
i think i might have an idea... one of the two settups where you have resistors connecting the base and the collector, then put in a current limiting resistor. i'll work out the math tomorrow night and see if my idea holds any water
 
Hmmm well I wired it all up and the circuit seemed ok. The emitter varied voltage as I changed base voltage I hooked it up and it's not working :(. The fan runs fine but it is running max speed. I fear that I have to complete the ground to the header on the motherboard for the voltage to vary. Speedfan says it should be running at 20% at my current temperatures but it isn't. Is there any problems if I run a lead from ground on the header to ground on the fan? Will it run the 1 amp of current through that lead? can I put a resistor on it to limit the current?
 
Autochthon said:
Hmmm well I wired it all up and the circuit seemed ok. The emitter varied voltage as I changed base voltage I hooked it up and it's not working :(. The fan runs fine but it is running max speed. I fear that I have to complete the ground to the header on the motherboard for the voltage to vary. Speedfan says it should be running at 20% at my current temperatures but it isn't. Is there any problems if I run a lead from ground on the header to ground on the fan? Will it run the 1 amp of current through that lead? can I put a resistor on it to limit the current?
Click to expand...

Hmmm... There *should* be a common ground throughout the computer. Try measuring the voltage between the base and the ground.
 
plot said:
i think i might have an idea... one of the two settups where you have resistors connecting the base and the collector, then put in a current limiting resistor. i'll work out the math tomorrow night and see if my idea holds any water
Click to expand...

That definitely has potential IF you use a high gain transistor. Another option would be to have an op-amp voltage follower. That should reduce the input current to almost nothing.

The high gain transistor though I can imagine would typically only come as a Darlington pair, which means ~1.4V drop across the two bases. Most op-amps can't hit their rails either. I think in this situation we'd probably want to be able to get as close to the rails as possible so that in a "needs serious cooling" situation we're not limited too much by the driver circuit.
 
fat-tony said:
Hmmm... There *should* be a common ground throughout the computer. Try measuring the voltage between the base and the ground.
Click to expand...

I measure mb header ground with power ground and had negligible resistance maybe 0.5 ohms. It measures 12.25V then starts to taper of as the fan controll software kicks in. You think the transistor could be bad? I have a tester what should it read? I have 2 transistors one reads 44 the other 38 the units are a mystery to me.

edit: ok nevermind it is working now. Don't ask me why the only thing I can imagine is I had the the 3pin connector I use to plug into the transistor shifted over 1 pin. it was in an awkward spot and I couldn't see it so it is possible.

thanks bunches, I don't understand why someone doesn't sell fans with something like this. it's simple, compact and useful. I just epoxied a small HS with transistor right onto the body of the fan.
 
Autochthon said:
I have 2 transistors one reads 44 the other 38 the units are a mystery to me.

snip

thanks bunches, I don't understand why someone doesn't sell fans with something like this. it's simple, compact and useful. I just epoxied a small HS with transistor right onto the body of the fan.
Click to expand...

The 44 and 38 would be the current gain through the transistor. Glad to hear you got it working!

Just curiousity, how warm does the transistor get?
 
fat-tony said:
Just curiousity, how warm does the transistor get?
Click to expand...
I'm using a 1/3 or 40% of an old 486 HS and the rpms are about 1500 which is about 30% of max. The temperature is cool to touch so less than body temp. Where I have it situated one of my 120mm blow holes blows on it so it get good airflow even at low speeds.

edit update on temp: at 3200rpm which should be about 2/3 of max the HS was warm to touch I'd guess maybe 40C (with no fan on it)
 
I could use a bit more help if you game, Another odd issue popped up today that has me concerned. I made a second circuit to power my blowhole (2) 120mm (.5 A each) and a secondary 90mm (.14A). These run off a secondary power supply that is tied to ground with my primary. I started up my system and the fans powered up and slowed as they should the 120's dropped below their starup voltage and stopped, the 90 didn't, BUT I realized the circuit that powered those 3 fans hadn't been turned on yet which meant they had to be drawing the power thru the base leg of the transistor or something else very odd was happening. Either way it's got me concerned even though it appears to work. I'm nervous about testing. I'm afraid i'll blow the MB headers.

edit: up for 2 weeks or so now and all is working fine.
 
You must log in or register to reply here.
Back
Top