How to make RPM Signal for 2-Wires Fan
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
How to make RPM Signal for 2-Wires Fan
As promised but I have to say this first:
Disclaimer
You have to work on inside your PC, removing your fan, connecting wires to mobo, using a solder, buying or borrow some measurement equipments (Volt meter or oscilloscope if you rich), etc...et.....
And no guarantee you will success because neither do I or nor somebody else can control the quality of work you do. So, you have been warned and I am not responsible should any damage occurs for any failure or irregularities.
This RPM circuit will also introduce some voltage drop to the fan (about 0.5 volt) so if you are using a fix 5 Vdc to the fan, your fan may not start after introducing this circuit. You will need to increase the voltage to your fan.
This circuit is also use 0Vdc as a common. So if you are using illegal method of 7 Vdc supply (using +12 Vdc and +5 Vdc line) or using fan supply with elevated common voltage, this circuit may not work unless it is modified.
OK. If you choose to proceed, here we go.
Motherboard RPM sense input
Most of the motherboard provides a pull-up resistor and need an open collector transistor connected to this input. Since the open-collector transistor (inside the fan) is turned on/off at proportional rate to the fan speed, the same pulse frequency is then created at the pull-up resistor (TTL level). A pulse counter will measures this frequency. To check if your mobo RPM sense requires an open-collector switch, measure the DC voltage to between the sense pin and common (0 Vdc - black wire). It should read about 5 Vdc. If not, yours may not be applicable to this mod.
Alternative RPM signal for 2-Wires Fan
Brushless type fan (such as axial fan on computer) will have a negative pulse during the pole switching inside the fan. Most of the fans have twice pole switch in a single rotation as you can see in the picture below. During the pole transition, the fan draws much less current compared to exciting mode, which then creates those nice negative pulses as you can see on the graphs below (take from an IC datasheet):
Our objective is then to extract this signal. The easy way to get this signal is via a sense resistor in series with the fan. The resistors size and the fan rating do matter. The higher the sense resistor, the higher the spike voltage measured, the more reliable the signal. On the other hand it also reduces the voltage to the fan for a given fan voltage supply. This is a trade-off. But it should not be a problem for SilentPCReview forum since we never want to run our fan at full speed anyway.
Here some guidelines for the sense resistors (again taken from an IC datasheet) but remember, this table is for nominal 12Vdc supply. So if your fan is running at 6 volts, you may double the value of the sense resistors.
The next step is to feed this signal to some kind of switching amplifier. This signal conditioning is needed since the spike voltage at the sense resistor is probably only 200 mV typical and even lower for down volted fan.
1st Project: A simple circuit for Fan RPM
This circuit is only used 1 transistor, 2 resistors and 1 capacitor. The circuit is below:
The R2 (edit: 10 M ohms, it was not showed on the picture for some how - sorry) bias and keeps the transistor 2N3904 in on condition. C1 blocks the DC signal from the sense resistor and passing only the AC signal (negative spike). When the negative spike reaches the base, the transistor will be cut-off and becomes high impedance and on again when the signal is 0.7 V or more at the base. Eventually, transistor will be on and off synchronously with the fan speed.
Please be aware that this circuit may not be reliably enough to switch on/off for low fan speed because the negative spike voltage is too weak to change the state of the transistors. Using FET instead of BJT probably is good idea but never tried because I have moved to the next circuit, using a comparator. If you are after a simple circuit and not too worry on low rpm accuracy, this circuit may suit you need.
2nd Project: A Less Simple Circuit for Fan RPM
The circuit is below.
An IC LM393 low power dual comparator has been used. A single chip IC is good for 2 RPM circuits since it has dual independent comparator. I have used it for CPU and PS fans in my computer, the most critical fan IMHO.
Again, the principle is no different from the 1st project but it is more reliable in low fan speed, thanks to the comparator. However, a comparator will require a reference (where do you think the name comes from?).
I use a voltage divider (R2 and VR) to get the reference. C1 (edit: 10 microFarad - not shown on the picture) capacitor is used to stabilize the voltage. You will need to adjust the VR to get the best result. I found the reference voltage (pin 2 of the IC) is best to be 60-75% of the dc voltage level on sense resistor (R1) during normal operation. It gives a good result for entire range of fan speed.
However, if you set the VR incorrectly, you will measure the noise or the harmonics of signals (very high RPM reading) or no RPM signal at all (If you do not understand, see the fan graph again, it will more make sense). So, spent a good time to tune this.
The good thing about the above setup is that the reference and fan are intentionally connected to the same voltage supply. This will maintain a proper reference level (once has been setup properly) for entire range of the fan speed with assumption that the fan is a linear load. And it has been proved to be rather linear in my experience.
R1 (sense resistor) is, again, depend on your fan rating. I have been using 5 ohms value for 200mA fan with a good result. My fan voltage is running from 6-9.5 volt, depending on the case temperature. If you are using 80 mm Panaflo (45 mA), you may need a sense resistor about 10-20 ohms depend on your fan operating voltage. Try the 10 ohms first.
As far as for comparator theory, it may take long time to explain it. I would rather skip it. You can find good information of this on the net and make it as your homework.
Well it took my 4 good hours. I do not know either this project is worthed to pursue or you just buy readily 3-wires fan. You decide. Should you proceed, OK. tell me your comment or if you have success to build this project
Regards
**Edit**
The pictures found a new home Many thanks to Fancontrol
Disclaimer
You have to work on inside your PC, removing your fan, connecting wires to mobo, using a solder, buying or borrow some measurement equipments (Volt meter or oscilloscope if you rich), etc...et.....
And no guarantee you will success because neither do I or nor somebody else can control the quality of work you do. So, you have been warned and I am not responsible should any damage occurs for any failure or irregularities.
This RPM circuit will also introduce some voltage drop to the fan (about 0.5 volt) so if you are using a fix 5 Vdc to the fan, your fan may not start after introducing this circuit. You will need to increase the voltage to your fan.
This circuit is also use 0Vdc as a common. So if you are using illegal method of 7 Vdc supply (using +12 Vdc and +5 Vdc line) or using fan supply with elevated common voltage, this circuit may not work unless it is modified.
OK. If you choose to proceed, here we go.
Motherboard RPM sense input
Most of the motherboard provides a pull-up resistor and need an open collector transistor connected to this input. Since the open-collector transistor (inside the fan) is turned on/off at proportional rate to the fan speed, the same pulse frequency is then created at the pull-up resistor (TTL level). A pulse counter will measures this frequency. To check if your mobo RPM sense requires an open-collector switch, measure the DC voltage to between the sense pin and common (0 Vdc - black wire). It should read about 5 Vdc. If not, yours may not be applicable to this mod.
Alternative RPM signal for 2-Wires Fan
Brushless type fan (such as axial fan on computer) will have a negative pulse during the pole switching inside the fan. Most of the fans have twice pole switch in a single rotation as you can see in the picture below. During the pole transition, the fan draws much less current compared to exciting mode, which then creates those nice negative pulses as you can see on the graphs below (take from an IC datasheet):
Our objective is then to extract this signal. The easy way to get this signal is via a sense resistor in series with the fan. The resistors size and the fan rating do matter. The higher the sense resistor, the higher the spike voltage measured, the more reliable the signal. On the other hand it also reduces the voltage to the fan for a given fan voltage supply. This is a trade-off. But it should not be a problem for SilentPCReview forum since we never want to run our fan at full speed anyway.
Here some guidelines for the sense resistors (again taken from an IC datasheet) but remember, this table is for nominal 12Vdc supply. So if your fan is running at 6 volts, you may double the value of the sense resistors.
The next step is to feed this signal to some kind of switching amplifier. This signal conditioning is needed since the spike voltage at the sense resistor is probably only 200 mV typical and even lower for down volted fan.
1st Project: A simple circuit for Fan RPM
This circuit is only used 1 transistor, 2 resistors and 1 capacitor. The circuit is below:
The R2 (edit: 10 M ohms, it was not showed on the picture for some how - sorry) bias and keeps the transistor 2N3904 in on condition. C1 blocks the DC signal from the sense resistor and passing only the AC signal (negative spike). When the negative spike reaches the base, the transistor will be cut-off and becomes high impedance and on again when the signal is 0.7 V or more at the base. Eventually, transistor will be on and off synchronously with the fan speed.
Please be aware that this circuit may not be reliably enough to switch on/off for low fan speed because the negative spike voltage is too weak to change the state of the transistors. Using FET instead of BJT probably is good idea but never tried because I have moved to the next circuit, using a comparator. If you are after a simple circuit and not too worry on low rpm accuracy, this circuit may suit you need.
2nd Project: A Less Simple Circuit for Fan RPM
The circuit is below.
An IC LM393 low power dual comparator has been used. A single chip IC is good for 2 RPM circuits since it has dual independent comparator. I have used it for CPU and PS fans in my computer, the most critical fan IMHO.
Again, the principle is no different from the 1st project but it is more reliable in low fan speed, thanks to the comparator. However, a comparator will require a reference (where do you think the name comes from?).
I use a voltage divider (R2 and VR) to get the reference. C1 (edit: 10 microFarad - not shown on the picture) capacitor is used to stabilize the voltage. You will need to adjust the VR to get the best result. I found the reference voltage (pin 2 of the IC) is best to be 60-75% of the dc voltage level on sense resistor (R1) during normal operation. It gives a good result for entire range of fan speed.
However, if you set the VR incorrectly, you will measure the noise or the harmonics of signals (very high RPM reading) or no RPM signal at all (If you do not understand, see the fan graph again, it will more make sense). So, spent a good time to tune this.
The good thing about the above setup is that the reference and fan are intentionally connected to the same voltage supply. This will maintain a proper reference level (once has been setup properly) for entire range of the fan speed with assumption that the fan is a linear load. And it has been proved to be rather linear in my experience.
R1 (sense resistor) is, again, depend on your fan rating. I have been using 5 ohms value for 200mA fan with a good result. My fan voltage is running from 6-9.5 volt, depending on the case temperature. If you are using 80 mm Panaflo (45 mA), you may need a sense resistor about 10-20 ohms depend on your fan operating voltage. Try the 10 ohms first.
As far as for comparator theory, it may take long time to explain it. I would rather skip it. You can find good information of this on the net and make it as your homework.
Well it took my 4 good hours. I do not know either this project is worthed to pursue or you just buy readily 3-wires fan. You decide. Should you proceed, OK. tell me your comment or if you have success to build this project
Regards
**Edit**
The pictures found a new home Many thanks to Fancontrol
Last edited by Bxwrapper on Wed Apr 16, 2003 9:02 am, edited 1 time in total.
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
That is very cool. I did some quick pricing and can get 45 of the second (comparitor) version for $4.50 each. The board is 0.7" x 0.8". It's a pcb so no ugly wires, etc. All parts are surface mount except R1 so it can easily be changed out.
You could mount it with double-sided foam tape (I love that stuff).
It has two headers, one three pin to connect to the mobo (pin-pin, or you could just solder wires to this board), one two pin to connect to the fan in and out.
As Bxwrapper pointed out, there are two comparitors in the same chip. I could look at a 2 fan version, but since the comparitor is only $0.40 I don't think you'd save much money or space.
Anyone interested? I only bring it up because we have the resources at work to do this kind of thing, and I remember how frustrating it was to go without...
You could mount it with double-sided foam tape (I love that stuff).
It has two headers, one three pin to connect to the mobo (pin-pin, or you could just solder wires to this board), one two pin to connect to the fan in and out.
As Bxwrapper pointed out, there are two comparitors in the same chip. I could look at a 2 fan version, but since the comparitor is only $0.40 I don't think you'd save much money or space.
Anyone interested? I only bring it up because we have the resources at work to do this kind of thing, and I remember how frustrating it was to go without...
-
- Site Admin
- Posts: 12285
- Joined: Sun Aug 11, 2002 3:26 pm
- Location: Vancouver, BC, Canada
- Contact:
Yes, I am very interested. 45 of the comparator version for $4.50 each -- I presume by that you mean completely assembled and ready to go if you can buy parts for 45 pcs at one time? Put me down 10 at least. Maybe more. Be great to have for 2-wire Panaflos.fancontrol wrote:That is very cool. I did some quick pricing and can get 45 of the second (comparitor) version for $4.50 each. The board is 0.7" x 0.8". It's a pcb so no ugly wires, etc. All parts are surface mount except R1 so it can easily be changed out.
You could mount it with double-sided foam tape (I love that stuff).
It has two headers, one three pin to connect to the mobo (pin-pin, or you could just solder wires to this board), one two pin to connect to the fan in and out.
As Bxwrapper pointed out, there are two comparitors in the same chip. I could look at a 2 fan version, but since the comparitor is only $0.40 I don't think you'd save much money or space.
Anyone interested? I only bring it up because we have the resources at work to do this kind of thing, and I remember how frustrating it was to go without...
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
45 is the min I can get at a decent price. My efforts to lower the number sent prices through the roof.MikeC wrote:Yes, I am very interested. 45 of the comparator version for $4.50 each -- I presume by that you mean completely assembled and ready to go if you can buy parts for 45 pcs at one time? Put me down 10 at least. Maybe more. Be great to have for 2-wire Panaflos.
Yes, completely assembled, ready to go. Note that you will have to come up with cables (I hate making cables and don't have an 'in' like I do for board assy).
35 more commitments and I'll start making them. From there it should take a week or so to get them all in the mail.
I love the smell of solder in the morning...
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
I think it is a good idea Fancontrol, it may give you better adjustment. However, it is totally your call.
If you design a drop votage of 0.5 Vdc at sense resistor with full 12 Vdc fan supply, I would make a permanent drop 0.2 Vdc on the additional resistor and having an adjustable voltage range of max. 0.3 Vdc on VR.
Good luck for your project.
If you design a drop votage of 0.5 Vdc at sense resistor with full 12 Vdc fan supply, I would make a permanent drop 0.2 Vdc on the additional resistor and having an adjustable voltage range of max. 0.3 Vdc on VR.
Good luck for your project.
Fancontrol,
Any DIY tips for creating something that'll FAKE a 3500+ RPM signal? I'm still doing it Guido style
Brice
Any DIY tips for creating something that'll FAKE a 3500+ RPM signal? I'm still doing it Guido style
Brice
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
Fanman... as you know, I'm a learning these electronics You should see my soldering techniques! Is there any way you could post a more detailed article w/ schematic, etc? I'd be happy to buy one from you.. but I'd never learn anything...fancontrol wrote: A better option, at least for a hack like me, is to use a 7406 to crank out pulses. one cap and 2 resistors gets you all you need.
Brice
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
This is an example of what I would do to make an oscillator of the second comparitor. I'd run that signal into the first comparitor and then get a drink.
If you feel like doing it yourself, be my guest. Calculating resistor/cap values for different frequencies and duty cycles is not straight forward. Send me a note if you plan on doing it and I can recommend some values for you.
Regarding the 2to3 adaptor boards, I count 29 orders. If I can get orders for 6 more I'll build them and hope I can get rid of the other 10. Hopefully I won't take a $45 bath on this. Tell your friends you want stuff and they should want it, too. BriceB, you could use one (or 10!) as the baseline for your 'tach faker' board.
If you feel like doing it yourself, be my guest. Calculating resistor/cap values for different frequencies and duty cycles is not straight forward. Send me a note if you plan on doing it and I can recommend some values for you.
Regarding the 2to3 adaptor boards, I count 29 orders. If I can get orders for 6 more I'll build them and hope I can get rid of the other 10. Hopefully I won't take a $45 bath on this. Tell your friends you want stuff and they should want it, too. BriceB, you could use one (or 10!) as the baseline for your 'tach faker' board.
-
- SPCR Reviewer
- Posts: 8636
- Joined: Sat Nov 23, 2002 6:33 am
- Location: Sunny SoCal
I'd order some but my problem is that I run most, if not all of my fans well below 2500 rpm and most MoBos won't detect rpms below approximately 1500-2200 rpm.
It's pretty easy to find your MoBo's threshold by using a Fanmate with a three-wire fan and adjusting it slower and slower. At some point the rpm will suddenly drop to "zero" even though the fan is still spinning merrily along.
It's pretty easy to find your MoBo's threshold by using a Fanmate with a three-wire fan and adjusting it slower and slower. At some point the rpm will suddenly drop to "zero" even though the fan is still spinning merrily along.
@Ralf Hutter, (name edited, my mistake)
You made a valid pont. But when you using 3-wires fan, there is possibility that in the low spinning fan, the "hall' effect is not sufficient to drive the open collector output resulting unreliable tach pulses to mobo.
The above circuit using different method. As long as the spike signal crossing the reference voltage, you will get the pulses, no matter how slow the fan running. So, the only issue here, in my opinion, is the minimum pulse mobo can detect.
On the other hand, the circuit produces 2 pulses per rotation. So, if your mobo minimum limit is 1500 pulse per minute, you will detect at least 750 RPM fan speed. My mobo (GA-7DXR) showed well below 1000 RPM (on MBM) after start-up and idle at 1300 RPM (fan voltage controlled by case temperature, 6 - 9.5 Vdc).
Regards,
You made a valid pont. But when you using 3-wires fan, there is possibility that in the low spinning fan, the "hall' effect is not sufficient to drive the open collector output resulting unreliable tach pulses to mobo.
The above circuit using different method. As long as the spike signal crossing the reference voltage, you will get the pulses, no matter how slow the fan running. So, the only issue here, in my opinion, is the minimum pulse mobo can detect.
On the other hand, the circuit produces 2 pulses per rotation. So, if your mobo minimum limit is 1500 pulse per minute, you will detect at least 750 RPM fan speed. My mobo (GA-7DXR) showed well below 1000 RPM (on MBM) after start-up and idle at 1300 RPM (fan voltage controlled by case temperature, 6 - 9.5 Vdc).
Regards,
Well, for doubter, I took a snapshot to show that the above circuit MAY give better quality than 3-wires fan. Here the picture :
I put the MBM Dashboard together with the MBM setting just to show that there is no trick for this. I also lowered the rpm alarm setting so that the picture is not distracted by the alarm.
I slowed down my PSU fan with my finger (I can still feel it !). It CAN detect down to 508 RPM (meaning 1016 pulse produced by the circuit). I did not want to hold my fan too long . And I can not guarantee your mobo will get the same result.
If I like, I will do some more precise test this weekend by manually lowering the fan voltage.
Well, I don't know why I am doing this. Probably because it is my design. That's why, I guess.
***Edit***
Re-linked the picture
I put the MBM Dashboard together with the MBM setting just to show that there is no trick for this. I also lowered the rpm alarm setting so that the picture is not distracted by the alarm.
I slowed down my PSU fan with my finger (I can still feel it !). It CAN detect down to 508 RPM (meaning 1016 pulse produced by the circuit). I did not want to hold my fan too long . And I can not guarantee your mobo will get the same result.
If I like, I will do some more precise test this weekend by manually lowering the fan voltage.
Well, I don't know why I am doing this. Probably because it is my design. That's why, I guess.
***Edit***
Re-linked the picture
Last edited by Bxwrapper on Wed Apr 16, 2003 9:05 am, edited 1 time in total.
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
I was able to tack 18 on to an existing PCB order. They should be ready to go by the 21st.MikeC wrote:Put me down for 15. I'm good for it. Now you have only 30 to go.
I can either ship them out in the order of your requests (i.e. MikeC gets the first 15, then tm gets 3, then we wait for the next order) or I can ship partials to everyone (e.g. 9 to MikeC, 5 to tm, 4 to phoner).
Let me know how you'd like it to break down. I don't expect the next 'run' will take more than a month.
-
- SPCR Reviewer
- Posts: 8636
- Joined: Sat Nov 23, 2002 6:33 am
- Location: Sunny SoCal
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
OK. It is the weekend time for me and It is time to find what is the lowest RPM my mobo (GA-7DXR) can register.Bxwrapper wrote:If I like, I will do some more precise test this weekend by manually lowering the fan voltage.
I am using NMB 3110GL-B4W-B34 fan, 220 mA @ 12 Vdc, 2-wires fan all over my rig. This is not the quite fan around but the only one available in my reach.
The lowest voltage I can go and mobo still detects the pulses is at 2.8 Vdc. The MBM showed 332 RPM. Not much air it can push at this level, however. Lower than that the MBM shows 0 RPM which is confirmed by BIOS.
Here is the picture. I assumed that you now trust me, therefore, no MBM setting is shown to save space .
I do not know other's mobos performance but I almost certain that 3-wires fans availble in the market may fail to provide reliable tach pulses if it is run on much lower voltage than its nominal.
Well, It is time for you also to find out, don't you think ?
**Edit***
Re-linked the picture
Last edited by Bxwrapper on Wed Apr 16, 2003 9:07 am, edited 1 time in total.
-
- SPCR Reviewer
- Posts: 8636
- Joined: Sat Nov 23, 2002 6:33 am
- Location: Sunny SoCal
YGEMfancontrol wrote:oh, you'll GET 15, my friend. the question was how many you want from the first batch. i'm thinking it will be 15.
then 1 each to tm, phoner, and Ralf, with the balance to follow.
please email me your mailing addresses. I expect to ship early next week. Is paypal ok with everyone?
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
I plan on building the boards today. You've got some options, and I need feedback to know how you would like the board finished out.
First, the current sense resistor varies with fan current. What value resistor would you like installed? Alternatively I could include a number of resistors and you can install them yourself.
Second, do you want headers for the fan and mobo connections? The alternative is to splice/solder this board in-line.
Third, and not as important, do you want a board that allows easy use of the second comparitor (e.g. as an oscillator)? 6 of the 18 boards have these features.
I'll mount everything but the resistor and headers and test one today. The first 18 should be in the mail by Friday, mostly waiting to get answers to these questions.
First, the current sense resistor varies with fan current. What value resistor would you like installed? Alternatively I could include a number of resistors and you can install them yourself.
Second, do you want headers for the fan and mobo connections? The alternative is to splice/solder this board in-line.
Third, and not as important, do you want a board that allows easy use of the second comparitor (e.g. as an oscillator)? 6 of the 18 boards have these features.
I'll mount everything but the resistor and headers and test one today. The first 18 should be in the mail by Friday, mostly waiting to get answers to these questions.
-
- Posts: 291
- Joined: Mon Feb 10, 2003 11:19 am
In the first post to this thread there is a table of resistor value vs. fan current. I know from experience that it's important to get that right.Phoner wrote:Sorry but most of what you ask is over my head.
At a minimum, I need to know the rated current of the fan you want to use with this converter. If there's more than one, give me the list and I'll pick the resistor that best suits your needs.
Another thing: in my haste to get these boards tacked on to an existing order there were some errors. The board will ship with a cut trace and two blue wires to make up for it. Sorry. Also, I couldn't get the boards routed out so the edges are a little rough. I'm sure you can work some dremel magic to fix this. I'll post a picture tonight.
Regarding the wire thing: I'll probably ship them with flying leads that you can crimp into whatever you're doing. That may be the easiest thing for those that are afraid of a soldering iron.
I will be using Panaflo L1A's.fancontrol wrote:I need to know the rated current of the fan you want to use with this converter.
Thanks!Regarding the wire thing: I'll probably ship them with flying leads that you can crimp into whatever you're doing. That may be the easiest thing for those that are afraid of a soldering iron.