Help with splitting a 12V PC fan tacho signal

Thread Starter

joe brick

Joined May 18, 2013
21
I modified #12's nice simple circuit from post #12 (spooky) to light the LED when the fan pulse stops;
Very cool, THE RB! Thank you very much! (I think I'm beginning to understand what you meant in your earlier post :D)

A couple of questions about this circuit if I may:

(1) Is there a simple way to modify this circuit to work with 12V rather than 5V?

(2) Is it possible to connect a 12V buzzer to it as well? and if so, how? in parallel with the LED?

(3) At mouser.com it says that the MPSA14 is obsolete. does it matter? is there a newer equivalent that will work in this case?

Thanks again for this, man, much appreciated!

Joe
 

#12

Joined Nov 30, 2010
18,210
1) It is already designed to work with 5 volts. Double the resistor to use 12 volts. 270 becomes 560 ohms.
2)You will probably need to add a current amplifier to get a buzzer to work. The buzzer would be instead of the LED in this circuit, but it probably won't work because it needs more current than an LED. Still, you are welcome to try. You might luck into a high performance transistor that would run a buzzer.
3) The MPSA14 is just the first transistor that I found that is a low power Darlington. Any low power Darlngton transistor should work to run an LED. Much more than that and you'll have to add another stage.

Now, let's see if I'm less scatter brained than yesterday:rolleyes:
This is what I'm really thinking about:
 

Attachments

Last edited:

Thread Starter

joe brick

Joined May 18, 2013
21
1) It is already designed to work with 5 volts. Double the resistor to use 12 volts. 270 becomes 560 ohms.
2)You will probably need to add a current amplifier to get a buzzer to work. The buzzer would be instead of the LED in this circuit, but it probably won't work because it needs more current than an LED. Still, you are welcome to try. You might luck into a high performance transistor that would run a buzzer.
3) The MPSA14 is just the first transistor that I found that is a low power Darlington. Any low power Darlngton transistor should work to run an LED. Much more than that and you'll have to add another stage.

Now, let's see if I'm less scatter brained than yesterday:rolleyes:
This is what I'm really thinking about:
Hey #12, many thanks for the explanation!

and nice carpentry skills :)

So just the 270 changes to 560 for this circuit to work with 12Vdc? the two other resistors and the capacitor stay the same?

Also, as this is intended to be a failure alarm, I'm afraid both the LED and the buzzer are an absolute must (I'd like to know if there's a problem even if I'm not sitting at the PC).

I don't mind adding another stage (more components I assume?), if you could please show me how to do it :D

thanks again for the help in this,

Joe
 

THE_RB

Joined Feb 11, 2008
5,438
All you need to get the LED and buzzer working from the same circuit, is to put the buzzer between the transistor collector and the 12v supply.

Then the LED and it's series resistor also go between the collector and the 12 supply.

Then the transistor can turn both on together. :)
 

#12

Joined Nov 30, 2010
18,210
Try it. The worst that can happen is that the buzzer makes no noise or a a silly noise and you add a transistor later.
 

Thread Starter

joe brick

Joined May 18, 2013
21
All you need to get the LED and buzzer working from the same circuit, is to put the buzzer between the transistor collector and the 12v supply.

Then the LED and it's series resistor also go between the collector and the 12 supply.

Then the transistor can turn both on together. :)
Try it. The worst that can happen is that the buzzer makes no noise or a a silly noise and you add a transistor later.
cool, thank you very much guys! will try it out and report back with the results:)
 

Thread Starter

joe brick

Joined May 18, 2013
21
Hi Everyone,

First of all, thanks again to all who took the time to read and reposed to my post. Your suggestions & advice have been very helpful!

On my end, after getting the parts and spending two weekends breadboarding & testing, it’s finally time to share the results & write up a progress report :)

Before diving in, I just wanted to say that in what follows, I’ve tried to be as clear & detailed as possible (especially since I’ve encountered some difficulties along the way and need further assistance), so this post has turned out a bit long, but please bear with me.

Also figured a quick recap might be useful at this point. In a nutshell, I’m trying to put together a circuit that is capable of duplicating the tach signal of a 3pin PC fan (typically the yellow wire), so that this signal could then be sent to two different locations simultaneously: (1) to a motherboard header (for RPM reading); and, (2) to a (tach-based) fan failure alarm.

Hence, the complete circuit consists of two main parts: a tach signal duplicator & a fan failure alarm based on the said tach signal.

After spending a lot of time testing various configurations, here’s the initial version of the circuit I ended up with. Having tested it, it seems to be working alright - that is, if the tach signal is interrupted, the LED lights up and the buzzer starts making a continuous (& very annoying) beeeeeeep…

Nevertheless, it would be great if you could take a close look at this circuit and see if I’ve made any errors or otherwise missed something important (see notes & clarifications below):

CIRCUIT_DIAGRAM_1.1.png


A few points about the above circuit (in the hope that you can actually see it):

(1) R3 has been change from 470R to 100R (see The RB’s original design in post 21). This was done because with 470R, when the fan was powered at a lower voltage (anything under 5.5V), then the Alarm LED would start blinking even though the tach signal was running fine & the LED was supposed to be off at that point.

(2) The latching (on/off) switch is there to prevent the alarm from going off in cases where there’s no fan connected to the circuit (no fan = no tach signal = alarm on). On the other hand, the momentary switch gives the opportunity to make sure that the LED & buzzer are indeed working during normal operation of the circuit (it’s simply a testing button).

(3) As noted in the schematics, the fan is connected to a voltage-based fan controller that is capable of delivering between 2.5-12Vdc (the controller also has a kick-start feature so the fan doesn’t stall at start-up even if the voltage is set to a very low level).
With this initial working version of the circuit in hand, I’ve tried to take it to the next level & this is where I run into a serious difficulty.

My idea was a very simple one: I wanted to modify the sound coming from the buzzer when the alarm goes off from a continuous beeeeep to an intermittent beep-beep-beep-beep (similar to the sound trucks & vans make when backing-up).

Having researched different ways to achieve this, in the end I decided to go with a 555-based design that I picked up from the net. Built independently, the design works perfectly, creating precisely the sound and frequency I was looking for.

However, when I tried to incorporate it into the main circuit, I run into a specific problem I can’t figure out how to solve (despite making numerous attempts).

Here is the modified circuit which is supposed to work exactly as the first, except for the particular sound that buzzer makes (explanation of the problem below):

CIRCUIT_DIAGRAM_1.2.png


The problem is this: THE LED IS CONSTANTLY BLINKING when the alarm is supposed to be off (the fan is connected & the tach signal is going into the circuit as it should). Moreover, it seems to be blinking at the exact pulse frequency which is coming out from the 555 timer... :eek:

I've been racking my brain as to why this is happening and I can’t understand it… :confused: given that the LED’s anode has a permanent connection to the 12V line, it has to be sinking into something, but what?

It’s worth noting in this context that the alarm mechanism of this modified circuit appears to be working exactly as it should, that is, if the fan’s tach signal is interrupted (or the momentary switch is pressed), then the LED lights up in a continuous manner and the buzzer gives a strong beep-beep-beep… only when everything is working normally & the alarm is supposed to be off, the LED keeps blinking.

I’ve tried using diodes, different transistors, and all sorts of other things, but nothing stops the LED from constantly blinking without disrupting some other part of alarm’s operation.

Could anyone please help me fix this design to stop the LED blinking problem and have the alarm mechanism work as is intended?

Also, any advice/correction/comment on how to improve this circuit in general (& perhaps reduce the number of parts) would be very welcome :)

Many thanks in advance,

Joe
 
Last edited:

Thread Starter

joe brick

Joined May 18, 2013
21
Having received nothing in response to my last post (& question), I thought I'd bump it.

Is there no one here who can help me with this?
 

THE_RB

Joined Feb 11, 2008
5,438
Your fan tach fail detect looks like our original circuit where the tach pulse keeps the darlingon transistor turned off. That looks ok, but I would use a much larger base-emitter cap than your 10uF cap, I would use maybe 470uF.

Then you check that when the tach pulse is running that the darlington B-E voltage waveform is low, it's highest points should be under 1v.

If that is so, then the darlington is properly OFF when the tach pulse is running.

I don't like your method of making the 555 modulate the piezo. When the 555 output pin goes low, it turns on the PNP transistor and it will flow some small amount of current through the piezo, and through the E-B of the PNP into pin 3 of the 555 timer. That is enough current to make the piezo emit some sound.

If you swap the PNP to the other side of the piezo, then current through PNP E-B will not go through the piezo.
 

Thread Starter

joe brick

Joined May 18, 2013
21
Hey THE_RB, thank you very much for the great suggestions & detailed clarifications! Very much appreciated! :)


As a result, we've come a step closer, but we're not just there yet as I'll do my best to explain below.

Your fan tach fail detect looks like our original circuit where the tach pulse keeps the darlingon transistor turned off. That looks ok, but I would use a much larger base-emitter cap than your 10uF cap, I would use maybe 470uF.
Following your suggestion, I changed the 10uF cap to a 470uF, but that only made the alarm go off after a relatively long time from the moment the tach signal was cut off.

After checking out a few other values, I settled for 47uF which is bigger than the original 10uF, but still makes the alarm go off rather quickly (as I'd like it to do).

Is there a specific reason why a bigger cap would be preferable here that I'm not aware of? (from what I saw so far, changing cap values didn't make any other difference apart from the said time difference).


Then you check that when the tach pulse is running that the darlington B-E voltage waveform is low, it's highest points should be under 1v.

If that is so, then the darlington is properly OFF when the tach pulse is running.
Checked this, and it looks fine, I think. Stayed constant at 0.65V.

Btw, I've been trying to figure out how is it possible that the Darlington (which is an NPN) is kept off when the tach pulse is running and "on" when it stops... :confused: shouldn't an NPN stay "open" until voltage/current is applied? or is it the reversed 1N4148 that does the trick by manipulating the current somehow? I'd be very grateful if you could clarify how this works as I've been scratching my head (and googling everywhere), but couldn't find an answer to this mystery.


I don't like your method of making the 555 modulate the piezo. When the 555 output pin goes low, it turns on the PNP transistor and it will flow some small amount of current through the piezo, and through the E-B of the PNP into pin 3 of the 555 timer. That is enough current to make the piezo emit some sound.

If you swap the PNP to the other side of the piezo, then current through PNP E-B will not go through the piezo.
Swapped places between the PNP and the piezo and indeed the LED stopped blinking when the alarm is off! :D

Make me very happy :) but like I said at the beginning, we're not quite there yet as I discovered another hitch...

More specifically, when the fan is powered with the full 12V, the circuit works great. Tach signal goes in and everything is working smoothly. When the Tach signal is disrupted, the alarm and buzzer go off as they should (with the strong intermittent beeping like I wanted).

However, if the voltage that feeds the fan is lowered, at a certain point (from about 10V downward), the buzzer stops emitting sound properly. Instead, it goes down significantly in volume and starts making rapid (and much quitter) clicking sounds.

I've tried different combinations to sort this out (including replacing the PNP with a MOSFET) but to no avail. Given that the alarm part of the circuit is based on grounding the LED & buzzer (rather than supplying them with voltage), I can't see why changing the voltage to the fan should affect the working of the buzzer...

It's crucial that the fan alarm will function regardless of the fan voltage, otherwise it would be useless if it prevents me from being able to control the speed of the fan.

Can you (or any of the other electrical experts out there) figure out what might be causing this strange phenomenon? and how might it be prevented?

thanks again for all the help,

Joe

p.s. is there a way of changing the title of this thread? and, if so, how can anyone tell me how could this be done? (seems appropriate with the shift in focus from tach duplication to fan alarm).
 

THE_RB

Joined Feb 11, 2008
5,438
...
After checking out a few other values, I settled for 47uF which is bigger than the original 10uF, but still makes the alarm go off rather quickly (as I'd like it to do).

Is there a specific reason why a bigger cap would be preferable here that I'm not aware of? (from what I saw so far, changing cap values didn't make any other difference apart from the said time difference).
...
Once the cap is sufficiently large it won't make much difference. But if the cap is too small the voltage there will have ripple, and the high points of the ripple may cause false triggering. The size of the cap depends on the frequency of the tach pulse, which I did not know. You're the one on the front lines, so if 47uF works great then that's done. :)



...
Checked this, and it looks fine, I think. Stayed constant at 0.65V.

Btw, I've been trying to figure out how is it possible that the Darlington (which is an NPN) is kept off when the tach pulse is running and "on" when it stops... :confused: shouldn't an NPN stay "open" until voltage/current is applied?
... I'd be very grateful if you could clarify how this works ...
I think you already got it. The 100k resistor will eventually charge the cap, and when >1v will turn the darlington on and make the fault alarm. But the pulse (which is active low) will cause the cap to be discharged through the diode, so the cap can never charge >1v when the pulse is occurring.

...
Swapped places between the PNP and the piezo and indeed the LED stopped blinking when the alarm is off! :D ...
Excellent. :)

... I discovered another hitch...

More specifically, when the fan is powered with the full 12V, the circuit works great.
...
However, if the voltage that feeds the fan is lowered, at a certain point (from about 10V downward), the buzzer stops emitting sound properly. Instead, it goes down significantly in volume and starts making rapid (and much quitter) clicking sounds.
...
I think this one is straightforward. The 555 timer and buzzer need 12v. All the time. Instead of running them from the fan voltage which might vary down to 10v, just run them from the fixed 12v supply inside the PC.

I thought you were already doing that, based on your schematic which shows a "12v" label for their supply.

Anyway, you are on the home stretch now. :)
 

Thread Starter

joe brick

Joined May 18, 2013
21
Thank you very much, THE_RB!

However, regarding the voltage lowering problem...

I think this one is straightforward. The 555 timer and buzzer need 12v. All the time. Instead of running them from the fan voltage which might vary down to 10v, just run them from the fixed 12v supply inside the PC.

I thought you were already doing that, based on your schematic which shows a "12v" label for their supply.
I wish it was as straight forward as that... but in fact the buzzer and LED are already connected to the 12V of the power supply (not to the fan voltage) exactly as shown in the diagram.

That's precisely what's so confusing about this :confused:

My (very uneducated) guess is that maybe this problem has something to do with the frequency? unfortunately, I have no way of measuring this as I don't have the right equipment (and can't afford to buy it at the moment).

Attached is the latest version of the circuit. It includes the previous changes (swapping places between the PNP and Buzzer, change to 47uF), as well as a simplified configuration for hooking-up the 555 as a timer for the buzzer.

help?
 

Attachments

THE_RB

Joined Feb 11, 2008
5,438
Hmm, ok if you think it's the fan pulse frequency that's going to be easier to diagnose with a 'scope.

Without a scope you have to evaluate the different circuit modules.

If the fan pulse is good, the darlington is OFF, and its base will be well under 1v, and there is zero power to the piezo so it makes NO sound.

Are you saying as the fan voltage is lowered to 10v, it still spins fine but the piezo STARTs making sound? And the sound is "clicking sounds"?

Please provide more info as to why the piezo is running (is the fan stalled at 12? and 10v?) and what all the voltages are. :)
 

Thread Starter

joe brick

Joined May 18, 2013
21
Hmm, ok if you think it's the fan pulse frequency that's going to be easier to diagnose with a 'scope.

Without a scope you have to evaluate the different circuit modules.

If the fan pulse is good, the darlington is OFF, and its base will be well under 1v, and there is zero power to the piezo so it makes NO sound.

Are you saying as the fan voltage is lowered to 10v, it still spins fine but the piezo STARTs making sound? And the sound is "clicking sounds"?

Please provide more info as to why the piezo is running (is the fan stalled at 12? and 10v?) and what all the voltages are. :)
ok, got it (& thank you very much THE_RB for bearing with me on this :))

Before going into explaining the problematic situation, saying a few words about the setup & testing method might be helpful in clarifying things.

SETUP

The fan's + and - pins are connected to a voltage based fan controller that feeds it between 3-12Vdc.

the fan's tach signal pin is connected by a wire to the circuit discussed here (first into the "tach duplicator" part and then into the "alarm" part, all as shown in the schematics attached to the previous post).

Btw, I made sure that the voltage based fan controller and the "duplicator/alarm" circuit have a common ground.

TESTING METHOD

To test the functioning of the alarm circuit, I simply connected/disconnected the wire that goes out from fan's tach pin and into the "duplicator/alarm" circuit.

Hence, the fan never actually stalled during testing. It was always rotating according to the voltage it was fed (3-12V), and the only difference was whether the tach signal was physically going into the alarm circuit or not.

(The reason I used this method is that, in real life, it is possible that the fan will receive power but still not function properly due to some mechanical problem - the only way to know this is according to the existence/non-existence of the tach signal.)


TWO SCENRIOS

Two scenarios emerged from the testing - one good and one not so much.

Scenario (1) - The fan is powered at 11-12V.

No problem whatsoever. That is to say:

When the wire that carries the tach signal from the fan is connected to the alarm circuit, the alarm itself stays OFF (no sound from piezo and LED is off).

When the said tach wire is disconnected, the alarm turns ON (piezo emits a strong beep-beep-beep sound & the LED shines continuously).

At this point, if the tach wire is re-connected, then the alarm turns back OFF and all is well :)

Bottom line, when the fan is power with 11-12V, the alarm mechanism works exactly as it should.


Scenario (2) When the fan is powered at 3-10.5V, things start well but then something goes wrong.

More specifically:

When the wire that carries the tach signal from the fan is connected to the alarm circuit, the alarm itself stays OFF (no sound from piezo and LED is off).

So far, so good.

However,

When the said tach wire is disconnected, the LED turns on as it should, but the piezo doesn't emit the strong beep-beep-beep sound as before. Instead, it makes these faint & very rapid clicking noises.

Bottom line, if the fan is powered by anything between 3-10.5, when the alarm turns ON, the piezo doesn't emit the strong beeping sound it's supposed to.

Like I said in the previous post, this is especially baffling as the piezo is powered directly by the 12V power line of the power supply and therefore isn't supposed to be affected by the voltage that feeds the fan.

I hope this makes the exact problem clearer.

However, if more info and/or explanations are needed please let me know.

Thanks again for all the help :)

Joe
 

THE_RB

Joined Feb 11, 2008
5,438
Thanks for taking the time to post a complete description. :)

OK, if the 12v remains good to the 555 circuit at all times, the issue must be with something coming from the tach duplicator circuit.

Your tach duplicator circuit also says it is running from 12v, is this correct? Or are you running that from the fan PSU?

IF the dup cct is running from fixed 12v, and you disconnect the FAN_TACH connector, T1 turns on, T3 turns off, and the alarm should work perfectly.

You can test that, with the fan voltage low (3v to 10v range), disconnect the FAN_TACH wire, and measure that T1 is turned on >0.6v at its base, that T3 is off (about 0.1v at its base) and T4 darlington shouldbe hard on with about 1v on its base.

And those voltages shuold also be identical at any time the FAN_TACH wire is disconnected, even if fan PSU voltage is 12v.
 

Thread Starter

joe brick

Joined May 18, 2013
21
Very much appreciated, THE_RB! :)

Can't tell you how great it is to have someone to tackle this with rather than having to go at it alone! I’m very grateful for your help!

Back to the issue at hand, I started by taking the circuit apart & rebuilding it just to make sure I've got all the connections right as shown in the latest diagram (REV_2).

After that, I've done some more testing & been getting exactly the same results as before (including the buzzer sound problem), so I think it's safe to assume the problem isn't a mistaken or bad connection somewhere.

Nevertheless, I think we are slowly zooming in on the problem (but I have a different theory about the source as I'll explain below).

Before that though, here are some more testing results & info:

You can test that, with the fan voltage low (3v to 10v range), disconnect the FAN_TACH wire, and measure that T1 is turned on >0.6v at its base, that T3 is off (about 0.1v at its base) and T4 darlington shouldbe hard on with about 1v on its base.
T1, T3 & T4 Voltages

Checked the 3 transistors as instructed and seems to me that the results are exactly what you said they should be:

T1 (BC547)

Vbase 0.44-0.45V (Alarm OFF, Tach wire Connected)

Vbase 0.81-0.82V (Alarm ON, Tach wire Disconnected)

T3 (BC547)

Vbase 0.44-0.45V (Alarm OFF, Tach wire Connected)

Vbase 0.01-0.02V (Alarm ON, Tach wire Disconnected)

T4 (MPSA14, Darlington)

Vbase 0.53-0.54V (Alarm OFF, Tach wire Connected)

Vbase 1.27-1.28V (Alarm ON, Tach wire Disconnected)

(all measurements were made against the circuit's common ground)

Looks ok, right?


And those voltages shuold also be identical at any time the FAN_TACH wire is disconnected, even if fan PSU voltage is 12v.
Yep, checked the above transistors (in both alarm ON & alarm OFF) while the fan was powered with different voltages (3-12V) and same values as above every time :)


T4 Voltages

I've also checked the Vbase voltages of T4 (the BC327) & this leads me to thinking that the real culprit is actually the 555 circuit...

But first the hard data:

T4 (BC327)

Vbase ~10.3-10.9V (Alarm OFF, Tach wire Connected)

Vbase ~10.3-10.9V (Alarm ON, Tach wire Disconnected)

(again, measurements were made against the circuit's common ground)

That doesn't look right, does it?

The Vbase of this transistor constantly changes (modulates?) in small "jumps" between these end-values (~10.3 and 10.9V), presumably due to the pulsating signal it's getting from the 555.

Btw, I've tried replacing T4 with a P-Channel MOSFET (MTP50P03HDL), as well as a BC547 (I know it's NPN, but figured it's worth trying), but none of them worked properly (that is, similar problems with the buzzer's clicking noises).


Voltage to FAN & Buzzer sound

I've taken a closer look at how changes in the voltage of the fan affects the sound the buzzer (I should perhaps note that voltage to the fan is completely separate from the rest of the circuit, except for the common ground, so changing this voltage doesn't change the voltage to any other component in the circuit).

Here are some of the results (notice the “wave-like" behavior):

At 5.25V – beeping (i.e. not just “clicking” or making “tik-tik” noises) in the correct rhythm and pitch, but volume is much lower than it’s supposed to be.

At 6.55V and at 10.5V – beeping the correct rhythm, but pitch is much higher & sound much lower than supposed to be.

At 7.60V and at 10.0V – initial short beep, followed by faint clicking (i.e. no beeps after the first, just rapid “tik-tik-tik…”).

Between 8.50-9.50V – only faint and rapid clicking sounds.

At 11.50V – buzzer alarm works fine – strong “beep-beep-beep…”.


I've got no idea whether this info is helpful in any way, but perhaps you could make sense of it.



555 Pulse - The Real Culprit?

I've also taken some measurements of the pulse coming out of the 555 timer in the current setup (as shown in the diagram).

The signal itself (measured at PIN 3 of the 555, against the circuit common ground), swings fairly rapidly between ~1.76V - ~11.5V.

The voltmeter shows these values more or less in quick succession: 1.76, 2.54, 4.11, 5.20, 7.45, 9.28... (it might be a continuous change, but showing in discrete "jumps" due to the sampling delay of the voltmeter itself).

Btw, these values on PIN 3 remain constant regardless of changes in the voltage that feeds the fan (3-11.5V).


Present Situation

From all the tinkering & testing I've done with the circuit so far, it looks to me like the most problematic part is the 555 timer.

I would gladly swap it with something different that hopefully won't mess up the rest of the circuit like the current one seems to.


While the signal from the 555 seems to fluctuate up and down in a wave like form, what I really need is a repetitive HIGH-LOW toggle that will continuously switch the T4 (the BC327) on and off in relative rapid succession, thereby creating the continuous "beep-beep-beep" sound when the alarm turns on.


Is it possible to build a circuit like this with the 555 (or perhaps with other components)?

I'd really like to try this so any suggestion would be most appreciate (a diagram with the components marked out would be also very helpful in this context).

Many thanks,

Joe
 
Last edited:

THE_RB

Joined Feb 11, 2008
5,438
OK, first, you said the BC327 is T4, it's actually T5. No drama. :)

It looks like the BC327 may not be turning off perfectly, so to fix that, put a 1k or 2k2 resistor across its base and emitter (B-E).

However that may not fix this weird fault symptom.

Seriously I think you have a miswire. The buzzer beeping seems to match the fan voltage, almost exactly like it was powered from the fan voltage. No you've said the buzzer and 555 are powered from fixed 12v, but your symptoms match what we would expect if the buzzer was powered from the reduced fan voltage.

One test you can do is to disconnect R3 so the darlington T4 is disconnected from all the fan and tacho circuit. Then the alarm will be fixed hard on.

At that point, there shoudl be no influence of the fan or tach cct on the alarm sound. So check with different fan speeds etc.

If the sound is still changing, you must have a circuit miswire of some type.
 
Top