Generate Fan RPM Signal W/O Fan

In the mean time, I am planning to use the system by splitting the yellow tach line to the rest of the fan connections except I read somewhere that there isn't enough power in the tach line and it needs to have a separate circuit to divide it. Do you know how to do this? Any suggestions?
Thanks
 
Last edited:
In the mean time, I am planning to use the system by splitting the yellow tach line to the rest of the fan connections except I read somewhere that there isn't enough power in the tach line and it needs to have a separate circuit to divide it. Do you know how to do this? Any suggestions?
Thanks

I can't delete this repetition post. I thought I lost my previous post, apparently it went to page 3
 
Last edited:
Off the top of my head, I can think of about 5 ways to multiply that tach line. An OR gate, AND gate, PNP transistor + a few NPNs, an Op-amp, a buffer IC...

How many fans are we talking here? Six? I'd suggest a hex buffer IC. Tie pins 2,4,6,8,10, 12, and 14 to GND, tie pin 16 to 12V, and take your source signal and connect it to pins 1 and 15. The remaining 6 pins you connect to the various RPM pins on the mother board.

If I were in your shoes, though, I'd get an oscilloscope and probe both the blue and yellow wires when the fans are plugged in to see what the server is actually doing.
 
Off the top of my head, I can think of about 5 ways to multiply that tach line. An OR gate, AND gate, PNP transistor + a few NPNs, an Op-amp, a buffer IC...

How many fans are we talking here? Six? I'd suggest a hex buffer IC. Tie pins 2,4,6,8,10, 12, and 14 to GND, tie pin 16 to 12V, and take your source signal and connect it to pins 1 and 15. The remaining 6 pins you connect to the various RPM pins on the mother board.

If I were in your shoes, though, I'd get an oscilloscope and probe both the blue and yellow wires when the fans are plugged in to see what the server is actually doing.


Thanks Mohonri
I must get an oscilloscope and see those wires yes. I just checked the voltage on the blue line and is 2.55V, and on the yellow line it is 0.01V. I need me an oscilloscope.


Thanks for the circuit and IC link.
 
Ok. I didn't get the Oscilloscope from the ad I ended up using the alternative.
I used ArduinoScope (https://github.com/iroth/simplearduinoscope) and I have some pictures.
First, when the pc is turned on, the fan goes at full rpm and the yellow and blue lines show these images:
Blue
blue_full_rpm.jpg

Yellow
yellow_full_rpm.jpg


then, the fan slows down and the oscilloscope shows different output:
Blue
blue_reduced_rpm.jpg

Yellow
Yellow_reduced_rpm.jpg


kind of rough idea, the blue is PWM. The yellow line wave, to me almost looks like a cardiac ecg! LOL. Any suggestion on how to proceed with this?

Thanks!
 
...
How many fans are we talking here? Six? I'd suggest a hex buffer IC. Tie pins 2,4,6,8,10, 12, and 14 to GND, tie pin 16 to 12V, and take your source signal and connect it to pins 1 and 15. The remaining 6 pins you connect to the various RPM pins on the mother board.
...

From the Mouser page: Supply Voltage - Max: 6 V;
If you want to use something on 12V, I would recommend to look in the 4000 series, something like the CD4010BE.
 
From the Mouser page: Supply Voltage - Max: 6 V;
If you want to use something on 12V, I would recommend to look in the 4000 series, something like the CD4010BE.

My bad--thanks for the correction.

@andrewkluz--I can't tell where the zero point is on the oscope, but it looks like the blue wire carries a speed signal, just like the yellow wire in a traditional PC fan. At high speed, the waveform has a much higher frequency. The Yellow wire would appear to be a signal from the PC to the fan, telling it how fast to go.

One other thing I just noticed. Fans give two pulses per revolution on the sense line. You'll need to recalculate for that :)
 
My bad--thanks for the correction.

@andrewkluz--I can't tell where the zero point is on the oscope, but it looks like the blue wire carries a speed signal, just like the yellow wire in a traditional PC fan. At high speed, the waveform has a much higher frequency. The Yellow wire would appear to be a signal from the PC to the fan, telling it how fast to go.

One other thing I just noticed. Fans give two pulses per revolution on the sense line. You'll need to recalculate for that :)

Hello Mohonri

Thank you! I was just thinking of that now when I came to read your reply and it just makes sense that the blue must be the tachometer signal. I will test again.
 
Mohonri ... Thank you very much!!

All I had to do was change the yellow to blue on the simulator I made a week ago and it works perfectly!
What a shame ... the standard was perfectly reversed and fooled me. I am sure there will be some one who will come across a DL585 or DL580 fan connector issue and this will be a perfect example.

Thank you all that helped!!
 
Woohoo! I'm glad we (finally!) got it figured out. Do you have a build log somewhere for that water cooled beastie?
 
Look up the datasheet for a fan, like a Delta used in HP servers. http://media.digikey.com/pdf/Data Sheets/Delta PDFs/PFC1212DE-F00.pdf

Look at the FG signal. You may need more current, and I'm not sure if it triggers multiple times per rev (it probably does) -- a tiny Arduino may be a better fit here since you can adjust the parameters easier, just put a transistor on one of the PWM outputs so you can pull up to +12V.

I PWM control these fans with an Arduino (PFC 1212DE), they're backwards from normal fans in terms of PWM values. I can try testing the tach wire and seeing what kind of output it gives.
 
Oh I posted only after reading the first two pages -- but the tach being reversed with the PWM being reversed vs standard PC fans is not exactly surprising :)
 
Woohoo! I'm glad we (finally!) got it figured out. Do you have a build log somewhere for that water cooled beastie?

Glad we did!
No not yet. I take pictures in the process (some of the major steps at least) but I didn't build a log. I probably would fit at homeservers.com.


@FrankD400: Thank you very much for your input! The current available for each connection has to be at least 3.5A since each fan (Nidec BETAV VA450DC) for each connection is also rated at that Ampere. Did I understand you?

Next challenge, silence the screaming PSU fans. Water cooling or changing the fans.

Regards
 
Hey
I know its old topic but...

Would any1 be willing to make me 4 units of fake PWM fan signal that can replace this fan > PFR0812XHE - 9A14

Closest data sheets I could find > http://www.mouser.com/ds/2/632/PFR0812XHE-515953.pdf
http://partner.delta-corp.com/Products/FANUploads/Specification/PFR0812XHE-SP00-REF..pdf

Need 4 units, or 1 if I can connect 4 fans to 1 slot. I guess to replace server fans for some quiet 120mms... Will might need more if this works out... ?

Just name ur price :- )

Regards
Dariusz
 
Hi everyone and Mohonri !

I realize this is an older thread, however, the solution seems to be valid still for what I'm trying to do. I have a Netgear GS728TP Managed Switch that sits on my desk, that utilizes 2 12V 40x40x20mm Fans which each operate at .18A, and 9000 RPMs and connect via their own 3-pin header pin.

They're loud as heck, and I need something quieter...

Interestingly enough, there's several threads out there on "replacing these fans" for similar switches, but in fact they're for older switches that don't monitor the RPM. Trust me... i've looked.

I bought 2 Noctua NF-A4x20 FLX fans, which operate at 5000 RPMs normally, but can even be downed as low as 3700 rpms via a set of resistors. When connected, and the switch initially plugged in, the fans spin up momentarily, and then stop and the switch sees it as a fan error and cuts off power to them. I'm assuming that it's because it's looking for an input of 9000 RPMs, tries to spin them up only to find that they reach 5000 RPMs and then die. Also, I assume that since I'm not "eliminating" the fan, I would simply splice in the 12V & Ground of the circuit to the 12V and Ground of the Noctua fan and not use the PWM signal from the fan, but the circuit.

Based on what was described here, I believe this circuit would work, however.... I'm confused on the "exact" parts I need. I like the POT approach, but could someone provide me with a complete "parts list" to build this? Part numbers or direct links would be super helpful to help me build this. My ears are screaming over here :)
 
The LM555 you can pick from this list: https://www.mouser.com/Semiconducto...ducts/_/N-4k3ax?P=1z0z63x&Keyword=555&FS=True
(they all work the same)

EDIT: A previous version of this post had bad math. This should be better:
For 9000 RPM, you'll need a signal that will strobe at twice that rate (there are two pulses per rotation on the PWM line), so 18,000 pulses per minute, 36,000 half-pulses per minute, or 600 half pulses per second.

The datasheet says that the length of the half pulse will be 0.693 * R * C. You want that to be 1.66mS. A bit of algebra leaves you with 0.00166/.693 = R*C = .0024
If you use a .1uF capacitor (a very common size), you'll want a 24kOhm resistor.
For your capacitor, anything off the first page of this search will do: https://www.mouser.com/Passive-Comp...yzvvqx?P=1z0wrj5Z1z0z819Z1z0z7l5&Ns=Pricing|0
Your 24kOhm resistor you can select from these: https://www.mouser.com/Passive-Comp...yzvvqx?P=1yzeg1fZ1z0z819Z1z0z7l5&Ns=Pricing|0
 
Last edited:
The LM555 you can pick from this list: https://www.mouser.com/Semiconducto...ducts/_/N-4k3ax?P=1z0z63x&Keyword=555&FS=True
(they all work the same)
For 9000 RPM, you'll need a signal that will strobe at twice that rate (there are two pulses per rotation on the PWM line), so 18,000 pulses per minute, 36,000 half-pulses per minute, or 10 half pulses per second.

The datasheet says that the length of the half pulse will be 0.693 * R * C. You want that to be 0.1. A bit of algebra leaves you with 0.1/.693 = R*C = .0693
If you use a .1uF capacitor (a very common size), you'll want a 693kOhm resistor. Now, that's kind of high, so let's use a 1uF capacitor and a 69.3kOhm resistor (or something close to it). 68k is a common size. It'll give a fake RPM slightly higher, but should still work.
For your capacitor, anything off the first page of this search will do: https://www.mouser.com/Passive-Comp...yzvvqx?P=1z0wrj5Z1z0z819Z1z0z7l5&Ns=Pricing|0
Your 68kOhm resistor you can select from these: https://www.mouser.com/Passive-Comp...=1yzekisZ1z0z819Z1z0z7l5Z1z0z5h6&Ns=Pricing|0

Thanks Mohonri - So in this case, I won't need a Potentiometer in the circuit at all, it will basically be your schematic you laid out and without the pot?
 
correct--the fixed resistor would take the place of the potentiometer
 
Oh my goodness, you're right! usafltg - you'll want to check out the updated version of my post above. Apologies for the mistake!

So, I did the correct resistor as described above... but I can't get the circuit to work with the fan. The Fan still powers on briefly, then shuts off. As I am using a REAL fan, and not eliminating it completely, just replacing it with a much quieter version, my diagram is a little different. Also the connections are a bit different on my board. Pin 1 is Power, Pin 2 is the sensing wire, Pin 3 is ground. The original fan is a 3 wire 12VHD, the specifications of which can be found here --> https://www.digikey.com/product-detail/en/delta-electronics/EFB0412VHD-R00/603-1401-ND/2560525

Looking at the data sheet of the original fan, it indicates that the middle wire is "RD" which I am guessing is Rotation Detector vs frequency generator. Not sure if that plays a difference here or not.

The new fan specifications can be found here --> https://noctua.at/en/nf-a4x20-flx

The wiring i'm trying to use is as follows. Not sure why this isn't working...

2020-03-07_22-26-23.png
 
Last edited:
From the datasheet, it almost looks like the fan treats that third line differently from how a "normal" fan does. I may be misinterpreting the datasheet, but it kind of looks like the fan holds that line low as long as the fan is spinning, rather than outputting a square wave to indicate its speed. You don't happen to have an oscilloscope handy, do you? Or even a multimeter? If you hook up the original noisy fan, and probe the voltage at the "pwm" pin, what voltage do you see?
 
It looks to me like the green wire is connected to the wrong pin on the left side. It looks like it's connected to pin 3 when it should be connected to pin 2.
 
From the datasheet, it almost looks like the fan treats that third line differently from how a "normal" fan does. I may be misinterpreting the datasheet, but it kind of looks like the fan holds that line low as long as the fan is spinning, rather than outputting a square wave to indicate its speed. You don't happen to have an oscilloscope handy, do you? Or even a multimeter? If you hook up the original noisy fan, and probe the voltage at the "pwm" pin, what voltage do you see?

I have a FLUKE but no oscilloscope. Let me probe the PWM line on voltage while they are going.

Also, I did have the green wire on the left side on pin 3, but that's because there's a resistor between 2 and 3, I figured it did not matter on which side of resistor the green was, but I'll test in position 2 as well.
 
I have a FLUKE but no oscilloscope. Let me probe the PWM line on voltage while they are going.

Also, I did have the green wire on the left side on pin 3, but that's because there's a resistor between 2 and 3, I figured it did not matter on which side of resistor the green was, but I'll test in position 2 as well.

Mohonri

I ran the test while it's running. Not sure if you want me to try it immediately at start-up as well.

When I have the fluke grounded to the Chassis The pins read as follows:

Pin 1 RED (VCC) = 11.7 Volts
Pin 2 BLUE (assumed PWM) 5.72 Volts steady
Pin 3 BLACK (Ground) = 5.86 Volts

IMG_2266.jpg
 
I have a FLUKE but no oscilloscope. Let me probe the PWM line on voltage while they are going.

Also, I did have the green wire on the left side on pin 3, but that's because there's a resistor between 2 and 3, I figured it did not matter on which side of resistor the green was, but I'll test in position 2 as well.

It doesn't matter for the resistor itself, but in the circuit it does. In the diagram you posted, the resistor is affecting the pin 3, the way you wired it, it's affecting pin 2.

If i remember 555 circuits right, you're basically setting it up so the output is feeding the threshold pin at a higher voltage than the trigger, so the circuit timer never fires again after the first tick.
 
It doesn't matter for the resistor itself, but in the circuit it does. In the diagram you posted, the resistor is affecting the pin 3, the way you wired it, it's affecting pin 2.

If i remember 555 circuits right, you're basically setting it up so the output is feeding the threshold pin at a higher voltage than the trigger, so the circuit timer never fires again after the first tick.

ryan_975 well, I gave it the ol college try, and that didn't work either ;) Worth a shot.
 
ryan's right that it's important where the green wire goes--because there's current passing through the resistor, the voltage at the two ends won't be the same.

In your photo, it appears that you're using the multimeter wrong--the black probe isn't contacting anything. You need to put the black probe on the black wire to the fan, and the red probe to the blue wire. Chassis ground may or may not be a good place to use as a reference.
 
ryan's right that it's important where the green wire goes--because there's current passing through the resistor, the voltage at the two ends won't be the same.

In your photo, it appears that you're using the multimeter wrong--the black probe isn't contacting anything. You need to put the black probe on the black wire to the fan, and the red probe to the blue wire. Chassis ground may or may not be a good place to use as a reference.

Mohonri

I tried it both ways, but only gave reference to grounding it at the chassis level the first time.

If I attach the negative to the black cable. Then the blue wire fluctuates... between -0.135 and -0.15

Also if you keep black attached, then red measures at 6.36
 
So the blue line is basically sitting at 0, at least when the multimeter is sampling it. And 6V on the red line is about half what I would expect. There's something sort of unusual going on here.
 
So the blue line is basically sitting at 0, at least when the multimeter is sampling it. And 6V on the red line is about half what I would expect. There's something sort of unusual going on here.

That's why I didn't trust the numbers given when I placed the multimeter measurement on the black line. if I ground it to the chassis (although it looked as though the black lead was not touching anything, the chassis itself is grounded) it shows an expected voltage of 12V. and the blue wire showed a voltage of half of that.
 
That's why I didn't trust the numbers given when I placed the multimeter measurement on the black line. if I ground it to the chassis (although it looked as though the black lead was not touching anything, the chassis itself is grounded) it shows an expected voltage of 12V. and the blue wire showed a voltage of half of that.

Well... in case you wanted something even MORE odd.... as I was testing, suddenly the voltage of both vans kicked up and there was a noticeable audible increase in decibal levels of the fans, plus the error light came on even though both fans were spanning. Tred restarting the unit, and light still indicates fan failure despite them both running at full speed. So..... I said, what the heck, wonder if I try plugging in the new fans now without the circuit what will happen. Sure enough..... they started working.

I could care less about the fan light blinking. The thing works. No idea what just happened. but both new fans are working, and they're silent. My ears can rest.
 
how much time and effort have you put into making a $9 part?

A lot but the circuit never worked to begin with, nor does it still. To Mohonri 's point, the board / fan doesn't behave like a normal fan, and there's some weird voltages going on. When i looked at the original schematic just as he did, the PWM didn't look the same as what was being described here. I don't think a pre-fabricated circuit would have been useful here either. I still don't really know what the culprit was here.... I did order an oscilloscope just because I needed to get one anyways.

Also something else to throw in, I had 2 sets of aftermarket fans I was trying here, 6V and 12V because of the weirdness going on which I had tested before I even went down the path of trying to replace the fans. Both gave the same problem. If anyone cares to try and figure this out.... Lookup the Netgear ProSafe GS728TP Version 2. Has to be the exact one. I could only find information about different switches which use different fans and connections.... not this specifc one, and I could never find enough information on the technical specifications of the pins that power / monitor the 2 fans.
 
Back
Top