Fan Monitoring Technical question

[macklin01]

Hello!

I've been batting this around in the overclockers.com.au forum for awhile, and I thought I'd pose the same question here.

Basically, I'm designing a fanbus which allows the motherboard to turn the fans ON/OFF and retain its RPM monitoring capabilities. I know how to do the former with a relay, but the latter ...

Originally, I had planned to just connect the RPM signal wire from the fans to the motherboard headers. This will certainly work when the fan's GND is connected to GND, but if doing the 7V mod, the GND signal now gets +5V, which is the same, as far as I understand, as what's used in the pullup resistor on the motherboard's RPM circuit. AFAIK, ordinarily, the RPM monitoring circuits on the motherboard expect an open collector (at GND) signal, so the wire gives a signal that varies from high (+ voltage) to low (app'x 0 to .4V). My original, nieve design of a direct hookup, under a 7V mod, would at worst damage the motherboard header and at best probably give spurious readings.

Does anybody have any ideas on getting a 7V mod to work with the fan RPM readings? Has anybody used a 7V mod with a fan RPM/power splitter with success?

Right now, the solutions I've heard are to always connect the fan's GND to GND and vary the positive voltage with either voltage regulator ICs, a rheobus, or the like. But nobody has come up with a solution for when the fan's GND signal is at an elevated potential, such as +5V.

Thanks! -- Paul

PS: Here's the thread at the aussie site:

http://forums.overclockers.com.au/showthread.php?s=&threadid=45034

 

[Iron Hawk]

i dont think the seven volt mod will work with the rpm signal. that would be like puting 5 volts into your mobos ground, giving a 7volt mod to your entire mobo. NOT A GOOD IDEA. you will need some sort of voltage regulator/resistor thingy to lower the speed of your fan.

 

[macklin01]

thanks


i dont think the seven volt mod will work with the rpm signal. that would be like puting 5 volts into your mobos ground, giving a 7volt mod to your entire mobo


Iron Hawk,

Thanks for the reply.

I don't think that the 5V would damage the case, per say; as I understand it (I'll test with a multimeter soon), the fan signal swings in a square wave between high and low, the high usually at whatever's supplied the fan (e.g., 12V), and the low near ground, around, say, .4V. 5V would be within the expected range.

I'm planning on testing with a multimeter soon, but that's what I've been able to gather from NLX and ATX FanM documentation on the web.

At any rate, it won't provide the proper signal, so I'm not willing to plug it in with the current design.

I'll probably wind up using a voltage regulator circuit at the end of the day, but I still like the idea of trying to use the output from the fan signal to run, say, a transistor which would send the "normal," expected values back to the motherboard.

Any thoughts on such a design?

Thanks again for the response! -- Paul

Also, and ideas on where I could find precise specs on how the motherboard header and fan RPM signal communicate?

Thanks again !! -- Paul

 

[Diggrr]

I haven't looked at a fan's circuit board lately, but I'm thinking the only way to have the best of both worlds is to modify the circuit of the fan.
If one side of the hall sensor (tiny transitor looking thingy) on the fan circuit goes direct to ground, you may be able to de-solder that leg of the sensor and provide a seperate ground wire for it. (tricky soldering, but doable).
This would allow the sensor itself to see normal -5v potential, and the fan's power ground for the windings to be at +5v ....basically splitting up the two systems that are usually tied together by the ground bus on the circuit board.

Sounds like simplicity itself, but you'll have to check the circuit. Like I said, I haven't looked in a while. Hope it'll work for ya!

 

[macklin01]

thanks

Diggrr,

That is a fantastic icon! (Okay, I know that there must be a different term for that ...)

Thanks for your response.

I haven't looked at a fan's circuit board lately, but I'm thinking the only way to have the best of both worlds is to modify the circuit of the fan.

I hadn't really thought about modifying the fans. That's a good idea. The only catch is that I was hoping to create a fan bus that is as nearly universal in compatability as possible. It should allow for running unmodified fans.

But if I were just designing for specific fans, I'd definitely give it a look -- it's a terrific idea!

Thanks -- Paul

Sounds like simplicity itself, but you'll have to check the circuit.
Well, I'm a bit newer to this than you are.

Thanks again for the great ideas -- I love these forums!

-- Paul

 

[Diggrr]

I had another thought about my suggestion, I think that hall sensor also times the pulses to the power the windings...that makes the fan brushless. I'll have to look into a fan I have to see if it can be separated as I suggested.

I'll let you know unless someone else chimes in.....ciao


Oh, and Welcome to the forums!

 

[macklin01]

I had another thought about my suggestion, I think that hall sensor also times the pulses to the power the windings...that makes the fan brushless.

Ah, yes! Thanks. I found a description at

http://www.comairrotron.com/Engineering/MonitorFanPerf.htm

and

http://www.nmbtech.com/engineering_section_nw2.html

I know that the signal uses an open collector, if that helps. I'm just trying to get a feel for how the signal voltage varies, and where current flows from and to in the course of the cycle.
I'll let you know unless someone else chimes in.....ciao


Oh, and Welcome to the forums!

Well, thank you very much! It's great to be here, and I'm learning a lot very quickly. I'll look forward to showing my final design. Thanks again for your help!

-- Paul

 

[macklin01]

Iron Hawk, Diggrr,

Thanks again for all your input! I just about have my design finalized, and when I have time (probably at the end of April after my complex analysis prelim), I'll put it together and give pictures.

I decided to use a spdt to toggle between a full 12V and a simple NPN whose level is determined by a 100 kOhm potentiometer. I'll leave room for an on/off theromostat control later.

I bought an extra fan to play with and plan on running it at 3V on a breadboard and testing the signals on the signal wire: I'll stop the fan at various locations and compare the outputs with a multimeter.

I do have a couple of finishing touches to work out, and I'd appreciate any advice:

1) Using a capacitor which can start the fan turning no matter how low the initial voltage.

2) Building a timer circuit which can toggle between different RPM fan outputs and feed them to the motherboard header.

Again, thanks for all the help!! If you have any ideas on this polishing, I'd appreciate it very much!!

-- Paul
[email protected]

 

[Veland]

Since we're talking about fan RPM..

I've got an old Geforce256 from Asus with it's own software for card voltages, temperature and fan rpm. But I'm using another HSF running from a separate PSU. This works perfectly, and gives very good temps. The problem? The monitoring software can shut down or reduce the speed on the card if it's overheating, but for the moment it always jumps to front since it doesn't see a fan rpm!

So what do I want? A circuit or ideas for designing a circuit that simulates an RPM signal.

..or just some good links..

 

[macklin01]

Since we're talking about fan RPM..

I've got an old Geforce256 from Asus with it's own software for card voltages, temperature and fan rpm. But I'm using another HSF running from a separate PSU. This works perfectly, and gives very good temps. The problem? The monitoring software can shut down or reduce the speed on the card if it's overheating, but for the moment it always jumps to front since it doesn't see a fan rpm!

So what do I want? A circuit or ideas for designing a circuit that simulates an RPM signal.

..or just some good links..

Have you tried a splitting cable that routes the fan's RPM wire to your graphics card and the power to wherever it's being powered?

What's the voltage of the fan, compared to the original fan?

-- Paul

 

[Veland]

The second PSU is a 5 channel homemade unit that gives me voltages from 1.25V-16V. And since it's completely separate it has "its own ground" so routing a wire to the card would not work.

At least I don't think so, but I have no formal training in electronics, just as a hobby for a couple of years..

 

[macklin01]

Hi, Veland

The second PSU is a 5 channel homemade unit that gives me voltages from 1.25V-16V. And since

Wow, that's really neat! Do you have schematics and/or pics anywhere? I'm really interested in
that.

it's completely separate it has "its own ground" so routing a wire to the card would not work.

Hmmm, I think you're right. It may or may not work, but I think it's a good idea to isolate the two systems. This is very similar to the problem I posted earlier: A fan runs with a voltage supply which has a "ground" different than that expected by the header. How do you transmit the fan signal to the header?

I've been thinking about it. I think you could try to rig up some sort of op amp or solid-state relay circuit.

I was going to post a rough idea of using a relay to do the job, but I highly, highly doubt a relay would keep up with a few thousand switches per minute, at least not for very long.

But I think some sort of switching would be in order. Hmm, but the trick will be isolating the two circuits.

At least I don't think so, but I have no formal training in electronics, just as a hobby for a couple of years..

Ahh, you're speaking to another hobbiest. Maybe if we put our heads together we can solve the problem!

Well, ttyl -- Paul

 

[Veland]

After looking through my "Big book of components" I see some possibilitys.

1. Electronic SPDT switch (I've got no experience with these, and they're pretty expensive)

2. Semiconductor relay (Should work, but I don't know how to "feed" then, as the only info is amount of mA needed to drive them)

3. Optoswitch (Probably the best solution, but input voltage should be pretty stable, so it would probably need a voltage regulating circuit)

So, what remains?
To know exactly what the fan sends out (digital or analog waveform, min and max voltage) and what the fan header accepts.

Let's take a hypotetical situation:
The fan gives a 12V pulse on each rotation (12V->ground)
You run using the 7V trick, so it send 12V->5V
Between sense and 5V is a voltage regulator giving appropriate voltage for the optoswitch
The other "side" of the optoswitch is connected to sense and 12V

As I have no knowledge of the inner working of fans and sensing, some assumptions may be wrong, but you get the point?


We'll continue the work, we'll figure it out..

Happy soldering!

PS! I WILL make a tutorial on the 5 channel PSU, but I'm buried in schoolwork right now.. Taking a masters degree is a lot of work..

 

[macklin01]

Hey!

After looking through my "Big book of components" I see some possibilitys.

1. Electronic SPDT switch (I've got no experience with these, and they're pretty expensive)

Okay, so that's out.

2. Semiconductor relay (Should work, but I don't know how to "feed" then, as the only info is amount of mA needed to drive them) [/B]

I think basically we could hook one lead up to the fan signal, and the other to the ground of the fan circuit to run the "semiconductor coil" of the relay.

I think I will attach my rough, stupid idea for a relay switch.


3. Optoswitch (Probably the best solution, but input voltage should be pretty stable, so it would probably need a voltage regulating circuit)

I'm not sure what an optoswitch is; sorry!


So, what remains?
To know exactly what the fan sends out (digital or analog waveform, min and max voltage) and what the fan header accepts.

I've been studying this a bit. As far as I know, the fan sends out an open collector signal. I thing that means that the motherboard, with an open transistor collector and +5 or +12V (probably +5) /pullup, has its collector hooked up to the fan signal. Fan signal gives a square wave +V volts or ground. When +V (usually 12), it's higher than the +5 for the circuit on the MOBO, and when ground, it's lower, and so the MOBO gets its open collector turned on and off.

I'm not positive, though. I'm not sure when I'll have time, but I had planned on taking out a multimeter and testing voltages on a spare fan I have. I'll run it at 3V, so it's not turning, and keep rotating the fan through different positions while observing the voltages. Hopefully, we should see a square wave between +3V and 0 volts. At the +7V mod, we'd see a signal between +12V and +5V, and neither is below the +5V of the mobo circuit, so no signal. (Constant on or off, I believe.)

Does any of this sound right? It sounds like your a better hobby electrical engineer than I am. (I'm just starting up; last did it as a kid in the 80's)


Let's take a hypotetical situation:
The fan gives a 12V pulse on each rotation (12V->ground)
You run using the 7V trick, so it send 12V->5V
Between sense and 5V is a voltage regulator giving appropriate voltage for the optoswitch
The other "side" of the optoswitch is connected to sense and 12V

As I have no knowledge of the inner working of fans and sensing, some assumptions may be wrong, but you get the point?

I think such a voltage regulator could work. But weren't you trying to isolate the circuits? I guess it wouldn't be such a big deal. Hmmm. Sorry, I'm still trying to get a handle on the inner workings of the fan; I've told you about all I've been able to find.


We'll continue the work, we'll figure it out..


I think so!


Happy soldering!


After the end of the month, probably. I have my complex analysis prelim before then (industrial and applied mathematics program). And what's your masters work in?


PS! I WILL make a tutorial on the 5 channel PSU, but I'm buried in schoolwork right now.. Taking a masters degree is a lot of work..

Terrific! Good luck on your work! I'm not sure if I'll get to test out the fan this weekend -- my wife's mom is visiting for the weekend, and I really should be studying conformal mappings this weekend.

Well, I'll try to keep up with the forum over the weekend, but in the even that I can't, have a great weekend!

Where are you a masters student? U of Minnesota, here.

ttyl -- Paul

 

[Veland]

The schematics look about right. And I suddenly remembered seeing a circuit converting SPDIF -0.5/0.5V to TTL 0-5V, so it could possibly be useful. Take a look at this

Oh, I'm working on my masters at Agder University College here in Norway. Well, the actual work on my masters is next year, now it's all about how to do it..

Sorry for the topic drift right here, but I just got to show the CD digital -to- optical converter i made of some old junk lying around. Here:

 

[macklin01]

nice!

Hi, Veland

Well, the weekend's been hectic, and I haven't had a chance to do any testing whatsoever. I'm glad to hear that the circuit sounded workable. I suppose that an op-amp type of circuit might work, too, but I'm a lot less experienced at this than you are. Also, I'm pretty sure that the fan signal is an open collector signal. Does that mean that the signal oscillates (in a square wave) between plus +V and 0V or does it just oscillate between grounded and no connection?

It seems strange to me, the more I think of it, that a fan would provide +12V signal at any time, as the current draw from the external circuit (motherboard) would likely be comparable to the current of the fan. That doesn't make sense to me from a design standpoint.

It's so hard to catch up on what I didn't learn when I was 12.

The schematics look about right. And I suddenly remembered seeing a circuit converting SPDIF -0.5/0.5V to TTL 0-5V, so it could possibly be useful. Take a look at this

Oh, I'm working on my masters at Agder University College here in Norway. Well, the actual work on my masters is next year, now it's all about how to do it..

Cool! I went to your website -- some pretty sharp-looking stuff there! And you're a business student? Your grasp of electronics is much better than mine!

Sorry for the topic drift right here, but I just got to show the CD digital -to- optical converter i made of some old junk lying around. Here:

No problem -- it's a very sharp, neat looking package. Nicely done!

Some other fun projects I've considered, after my fan bus:

1) Adding an FM radio receiver to the from bezel of my PC, and feeding signal directly to the sound card, and draw power source directly from teh PSU.

2) Adding an FM transmitter to transmit MP3 audio to all the radios in the house.

Well, I've gotta' run. Have a nice Sunday!

-- Paul

 

[Veland]

I found some more info, this time directly related to my wish on simulating fan RPM. But in this discussion some interesting facts are found. The most important is that the SENSE signal is NC - GROUND signal. So it's either "floating" or connected to ground.

By the way, I know little to nothing about complex circuits, I still think of electricity as water flowing in tubes, meeting resistance, valves and such stuff. But after some pondering, I usually figure it out

Then, other stuff! My masters is in Information Systems, with a focus on things like communication, collaboration, IT enabled business transformation and such.

I like the radio ideas, I've always been fascinated på radio signals. Maybe you could build a RF remote for the computer? I built a normal IR receiver, but I'd like a RF receiver working from the 5Vstb power so I could remotely switch on the computer!
All the plans up in my head...

Take a look at my IR receiver as well..

 

[macklin01]

I found some more info, this time directly related to my wish on simulating fan RPM. But in this discussion some interesting facts are found. The most important is that the SENSE signal is NC - GROUND signal. So it's either "floating" or connected to ground.

Ahh, that's what I thought! I thought it would be strange if they'd have a 12V - (motherboard voltage) current flowing through the sense wire. Thanks!!! Do you have a link on where I can read more about it?

Well, I've gotta' run, so thanks for the response, but I'll write more soon! This is such a fun thread!

-- Paul


By the way, I know little to nothing about complex circuits, I still think of electricity as water flowing in tubes, meeting resistance, valves and such stuff. But after some pondering, I usually figure it out

Then, other stuff! My masters is in Information Systems, with a focus on things like communication, collaboration, IT enabled business transformation and such.

I like the radio ideas, I've always been fascinated på radio signals. Maybe you could build a RF remote for the computer? I built a normal IR receiver, but I'd like a RF receiver working from the 5Vstb power so I could remotely switch on the computer!
All the plans up in my head...

Take a look at my IR receiver as well..

 

[macklin01]

Hi, Velund. I'm back!

I found some more info, this time directly related to my wish on simulating fan RPM. But in this discussion some interesting facts are found. The most important is that the SENSE signal is NC - GROUND signal. So it's either "floating" or connected to ground.

Again, ahh, that's terrific!


By the way, I know little to nothing about complex circuits, I still think of electricity as water flowing in tubes, meeting resistance,

I think that's a very way of looking at it. And you certainly manage to do good things with it! That was a great compact IR receiver! Do you have a serial port on the front of your case, then? I seem to remember that the serial port can supply a voltage through some of its pins. Is that true?


Then, other stuff! My masters is in Information Systems, with a focus on things like communication, collaboration, IT enabled business transformation and such.

It sounds like a rewarding career option. Enough tech to be fun, and enough business to make a living. Actually, I've heard that that can be a good niche to be in. Cool.

I'm in industrial and applied math, here. I mostly focus on numerical analysis, and my thesis topic will be numerical simulation of tumor growth, with some modeling of angiogenesis.

But in academia-land, I'm forced to concentrate on passing my written prelim exams in algebra (oh, yes, verrrry useful to writing code ) and complex analysis, so I haven't had time to do research in ages.


I like the radio ideas, I've always been fascinated på radio signals.

Thanks!! You know, I think I actually saw a decently cheap kit for sale via order from radio shack. I think it could be pretty easily modified to interact within the computer. Might be more expensive than it needs to be, though.


Maybe you could build a RF remote for the computer?

Now THAT would be great! You could start up your computer when you hop off the bus. Maybe another remote function to start it dialing the internet!

I think the components would be do-able. I'm not sure how one would go about writing the software, however. And would it interact with the PCI slots through an interrupt? Through a serial port?

What software do you use for your IR remote? I think that's absolutely awesome!

Take a look at my IR receiver as well..

Incredibly compact and neat!

Well, I should go. Thanks for the fun ideas. I hope we can continue this thread and get a better understanding on teh fan RPM signal. I think we've got it, though.

If it really is NC (no charge / no electrical path) - ground, then I think we could basically hook it up to another open collector signal. If it is now a +5V -- NC square wave, we could use it, with some high resistance to drop the voltage (after all, we don't want a huge amount of current flowing out the fan signal), to run an inverter of some sort and generate just the open collector signal the motherboard is expecting. I haven't had time to write out a circuit, but do you think this sort of idea would work?

Thanks again -- this has been a great thread!

-- Paul

 

[macklin01]

Hi, Veland.

I ran some tests on the fan I had running around. First, I'll post a simple circuit picture (at the end).

The fan used was a standard 12V fan (coolermaster) with 7V startup. I did all tests at 3V so that the fan would be stalled and I could hand-crank the fan to view the results.

Basic circuit was fan connected to +3V and GND, and the signal wire passing through a resistor and LED (from +3V to the fan signal).

Stalled will always refer to the fan with just the 3V supplied, and low-med. speed will refer to hand-spun. I didn't have a chance to test with a multimeter, but I will soon. These are preliminary estimates, and they should hopefully fit into the information you found.

test 1:
fan: +3V connected
GND connected
LED: as indicated

stalled: LED always lit
low/med: LED flutters with fan speed

test 2:
fan: +3V connected
GND disconnected
LED: as indicated

stalled: LED unlit
low/med: LED unlit

test 3:
fan: +3V disconnected
GND connected
LED: as indicated

stalled: LED unlit
low/med: LED flahses

test 4:
fan: +3V connected
GND disconnected
LED: polarity reversed, fan signal through resistor
and LED, then to GND

stalled: LED lit
low/med: LED flutters

ALL other configurations with LED reversed: always unlit

It's late, so I have no conclusions per say. The first three tests seem to indicate what we expected. I'm not sure about #4, though. But, I'm pretty tired. Talk to you later -- Paul