Fan Monitoring Technical question

[macklin01]

Updated (and improved) experiment

Hi, Everybody!

I just refined the tests. I had originally intended to use a multimeter to test the signal by turning the fan to each of 4 positions and measuring, but because the fan requires a minimal "threshhold" speed to give a reading that is faster than my multimeter can measure a voltage, I had to settle for these cruder tests.

Here's a cut 'n' paste of the results: (where bolded, there is text from another conversation, to which this was a reply)

........................

Well, I refined the test a little bit, and the results are a little clearer. Also, there was one more result that I forgot to mention:

The fan needed to attain a sort of "threshold" speed before the LED's would flash with the RPM data. Before that, the fans would remain constantly on or off, depending upon the setup.

In the new setup, the "GND" LED indicates when the fan sensor wire is at ground, and the "+" LED indicates when the fan sensor wire is at a positive voltage.

Here's the test circuit for experiments 1 - 3:
(thank goodness for the breadboard -- easy rigging!)

Experiment 1:
fan +3V connected,
fan GND connected:

fan stalled: GND LED lit, bright
+ LED unlit
low speed: GND LED flashes (quickly) with fan signal, bright
+ LED unlit

Experiment 2:
Fan +3V disconnected,
fan GND connected:

fan stalled: GND LED unlit
+ LED unlit
low speed: GND LED flahses (more easily seen), bright
+ LED unlit

Experiment 3:
Fan +3V connected
fan GND disconnected:

fan stalled: GND LED unlit
+ LED lit, very dim
low speed: GND LED unlit
+ LED flashes, very dim

Here's the diagram for the last tests:

Experiment 4:
Rewired to supply fan +9V, common ground
(See the second circuit diagram.)
connected +9V to fan, then disconnected
+9V with GND connected and observed LED
as fan slowed to a stop.

fan at 9V speed: GND LED lit, bright,
flashes too fast to observe
+ LED unlit
fan slowing down: GND LED continues to flash
at lower speed, bright, until
fan is too slow to sustain signal
+ LED unlit
Experiment 4a:
Manual 50% PWM duty cycle, 1 second on,
1 second off:

GND LED flashes bright and is sustained during
entire cycle,
+ LED unlit


Test 3 confirms that the rpm monitoring signal occurs independent of the +ve fan voltage.

Absolutely. The tests seem to indicate a NC - GND
cycle. Furthermore, the strange results I got with the + LED in experiment 3 gave only a weak signal at best. I think it's a minor effect that won't be seen in most of our configurations. On the other hand, perhaps that weak signal would be sufficient for the motherboard to detect in a PWM scheme that modulates the GND current to the fan.


This gets me thinking about PWM and rpm monitiring, it should work if the -ve lead of the fan stays negative.

I think you're right. So long as the fan retains sufficient speed, the signal appears to come through just fine.


Unfortunately most of the PWM regulate the -ve lead of the fan


I looked at yours (and others). They do all seem to regulate the GND lead of the fan. If anybody has a PWM circuit that regulates the +V, that would be terrific!

This is pretty neat -- I think we're on to something here!


Anyone aware of any PWM that regulate the +ve fan lead (use a PnP or P channel mosfet to do the switching)

I'll keep my eyes peeled. At the end of the day, I'll probably use a potentiometer - NPN circuit for the low speed, but I'm certainly keeping an open mind to PWM, especially now that we have evidence that it would continue to supply a fan RPM signal!

-- Paul

 

[macklin01]

Design is crystallizing

Hello, all!

Thanks for all the help you've given. I thought I'd give a brief update. I've tested a prototype design on a breadboard and am ready for full implementation after the end of the month. Here are the specs:

1) 3 channels, each capable of 800 mW
2) For each channel, toggle HI (full 12V)
or Variable (controlled with low-wattage
100 kOhm potentiometer and 800 mW
NPN transistor)
3) Speed indication given by dual-color
LED's
4) Each channel turned ON/OFF by
motherboard headers.
5) Each channel returns fan RPM signal
of first fan on the channel to the
appropriate header.
6) New/Original (?) design addition: Using a
5.1V zener diode across the potentiometer
to prevent an accidental "bumping" of the
the potentiometer from turning the fans off.
Thus, all fans get a minimum of approximately
12V - 5.1V - .7V = 6.2V. In actual testing,
this minimal number is slightly higher due
to the fact that the zener is on the base
lead of the NPN.

I'm pretty happy with having come up with
#6. It was the simplest solution to a
potentially big problem.

7) Considering an in-line 2.5 A fuse for
safety's sake. Comments?
8) LED's only receive power while the fans
are receiving power. (also relay-controlled)

I'll try to post a schematic later on. But that's the basic
design, and the breadboard testing looks good. It's
also fun to watch the fans interfere with my TV
reception when I slow them down.

I'd appreciate any comments, especially regarding the fuse.
I've gotten in all my components except the solder and PCB board. (I've purposely not bought them so that I won't begin until after the end of the month. )

I'll be designing a front panel template shortly.

Thanks again, and I'd appreciate any feedback! A special thanks for all the electronics feedback earlier on! I decided to not use PWM for this circuit, under the general thoughts of the KISS principle. All of my objectives are seemingly fulfilled here.

-- Paul
__________________
-- Paul
[email protected]

 

[Diggrr]

I had thought that using a PWM would also modulate the ground potential, in effect killing the rpm signal.

Anyway, sounds good. I'd like to see a schematic, not to copy, just to see what you came up with.

Ciao

 

[macklin01]

I had thought that using a PWM would also modulate the ground potential, in effect killing the rpm signal.

Anyway, sounds good. I'd like to see a schematic, not to copy, just to see what you came up with.

Ciao

Thanks for the reply.

Well, most of the PWM circuits tend to modulate the ground lead rather than the positive lead, but if one could be devised otherwise, all the evidence I've see thus far has indicated that the fan RPM would continue to work, (or at least continue to indicate a turning fan) if the fan was spinning at a sufficient rate.

But at the end of the day, I think that the PWM is too complicated for my needs with too many potential issues.

Thanks for your interest -- I'll post a schematic tonight or tomorrow, and a picture when I'm done. I think it'll be nice. I got some 100 kOhm pots and aluminum knobs (about .5" diameter) from jameco, zeners from jameco, dual-color (red/green) LED's in a .56" diameter chrome panel mount (very slick), and some aluminum DPDT flat-lever toggles. Bought some generic 5.25" flat bay panels that I'll use as the control panel, will mount the PCB board directly on some hard drive mounting racks I have left over from an old 3.5" HDD, and basically let the panel sit with the wires attaching it to the PCB. I'll have 3-wire extensions leading out from the device to be fed to the 3 (currently) fans and headers, and I hope to have the project look pretty neat.

I'll work on that control panel schematic soon and try to have it up. Now that the design validation is done, the next hardest thing will be learning to use the Dremel on the panel without making it look terrible.

I'm curious -- what's your opinion on a fuse?

Thanks for the input!!!! -- Paul

And if I don't hear from you sooner, have a great weekend, Diggrr! (And everybody else, too.)

 

[macklin01]

Finalized schematic

Hello, Everyone!

Here's a finalized schematic for the basic circuit.

Here are some key points:

1) It is designed to support approximately 800 mA. As the typical 80mm fan draws about .12 A, several fans could be attached in parallel to the design; in such an implementation, only the first fan per channel would feed RPM information to the motherboard.

2) The zener diode sets a maximum allowed voltage drop across the resistor. Because the voltage across the resistor, plus that across the transistor, plus that across the fan equals +12V (+Vcc), limiting the drop across the resistor puts a lower bound on the fan voltage. I use this as a saftey so that the fan never stops spinning, even if the control knob is bumped.

I have found that with a zener voltage of 5.1V, the actual drop across the resistor is actually less than 5.1V, likely because the current across the resistor is comparable to that across the zener diode just prior to its avalanche current.

At any rate, I found that with .12 A draw on the circuit, the lowest voltage to the fan was approximately 7V.

3) The relay is a 12V coil (1050 Ohms resistance), so it only draw about 11 mA from the motherboard header. It can switch a maximum of 1 A; As the transistor is rated at 800 mA, it's actually limited to that lower current. But the plus side is that no heat sinks are required, and the circuit dosen't produce any perceptible heat.

4) This design allows the use of a less expensive, lower wattage variable resistor and zener diode.

5) My design has a dual-color LED. Others could easily be substituted.

6) For safety's sake, I am considering a 2.5 A, 5 x 20 mm fuse at the beginning of the circuit. Any comments????

Oh, and BTW, I have my "instrument panel" designed, measured out, and have drilled about 2/3 of the holes with my Dremel. (A new skill for me.) I'll have a scan posted soon.

All input and feedback is appreciated!

Thanks! -- Paul Macklin
[email protected]

 

[macklin01]

Feedback?

Hi, All.

Hmmm, I still haven't gotten any feedback.

I need help on a couple of items, if anybody could spare the time:

1) Fuses: good idea? bad idea? I'm thinking of a 2.5 fast-acting fuse.

2) Mounting: I have the control panel essentially finished. Where and how do you mount your circuit?

Right now, I'm thinking of mounting it on my Antec EasyUSB port.

The wires will lead from the circuit to the panel, which I'm considering attaching on top of the antec panel. Any thoughts or advice?

I would certainly appreciate any insight!

Thanks! -- Oh, and check out the "Lots of pics of my bong case " thread at overclockers.co.nz forum -- it's absolutely incredible!!!

Thanks again -- Paul

 

[Veland]

Hi!

I'll give some feedback..

First, the circuit looks very nice! Always good with several ways of doing things. A really clever solution.

So:
1) Fuse.
Not sure. It would not hurt, but I think the chance of getting a short in the fan is very small, the "normal" failure is that it stops working becuse something burns so no power goes through it. If you want to make a really safe system, this is what I propose: Use a fuse to make sure no shorts happen, but in addition make some kind of system that monitors power going thorugh the circuit. This should then give some sort of warning when no power flows (In the event of a burned fuse, or internal failure resulting in the fan not working) This would then offer "total protection". If it's a CPU fan, you could hve a relay on the power-good line to shut down the system if the fan fails.

2) Mounting
Using the Antec USB panel should work out just fine, just " measure twice, cut once" and all that. After all this work, make sure it looks good too. And when you have you dremel/drill out, thnk about other things you may need on the panel (headphones out, nic led's etc.)

Good luck, and do post pic's

 

[macklin01]

Thanks!!!!

Veland, it's good to hear from you!

And I certainly am glad to get your input!

Hi!

I'll give some feedback..

First, the circuit looks very nice! Always good with several ways of doing things. A really clever solution.

Thanks. This is very flattering.


1) Fuse.
Not sure. It would not hurt, but I think the chance of getting a short in the fan is very small, the "normal" failure is that it stops

I absolutely agree. The fuse is more for computer hardware/overload protection. I thought about the failure problem for some time. Again, I decided to go with the simplest solution -- combined software/hardware monitoring. I'll have a 2-layer solution.

The first layer is Motherboard Monitor. I'll configure it to shut down my PC if the CPU temp reaches 50C or if the CPU fan speed falls below, oh, let's say 2000 RPM.

The second layer is at the motherboard level. The BIOS can be configured to shut down the computer if the CPU reaches 60C. Alas, it can't be set any lower, but I just regard it as a second layer of protection. If all that fails, I'm probably in trouble anyway.

I know that there are some MAX... chips that can detect a fan failure, but since that's a function of the SMBus, and certainly circuits can be made to detect an event failure (e.g., an oscillation in the fan signal), but I hate to duplicate features that are already there. Maybe on a future, more ambitious project. (This is probably my first electronics project in 8-10 years.)

The idea of having a safe-guard for total project is definitely very appealing! Thanks for the ideas!

I do have a question -- in the event that the fan stops, how would the power consumption change? Increase? Decrease? I can't really say for sure.

I really like the relay idea; since some motherboards don't seem to work with MBM, (e.g., my wife's HP), having such a safeguard would be a really, really good idea!


2) Mounting
Using the Antec USB panel should work out just fine, just " measure twice, cut once" and all that. After all this work, make


Thanks for the tip. Actually, I won't be cutting the antec panel at all; I'll use the 3.5" drive mount to mount the circuit board (and possibly reserve some real estate for future projects, and lead the LED wires, etc. to another panel just above the Antec panel.

I'll try to post a picture tonight.

I absolutely agree with your comments on measuring twice, drilling once, and proceeding very slowly. The instrument panel actually went through at least three revisions, which I then tested out on index cards, before being finalized. I spent 3 days drilling out the holes -- one type of hole per night, so I wouldn't rush it, and I took time to measure and draw all items out in pencil on the panel. All the drilling is done, now, and it's ready for soldering.

The only piece I'm puzzling on now is how to attach the instrument panel to the Antec panel (directly above, flush). I'm thinking of using 2 dowels leading up from the sides. I'll try to illustrate this better when I get home tonight.


sure it looks good too. And when you have you dremel/drill out, thnk about other things you may need on the panel (headphones out, nic led's etc.)

Thanks again for the tip. Hmm, I hadn't thought about adding more feature to the front like that. I think for now, though, I'll avoid "mission creep" and just go with the basic project. I'll try to leave things in a form that can be disassembled if necessary.

Good luck, and do post pic's

Thanks!

I was also considering writing a short article for submission to overclockers.com emphasizing the often-neglected research and design process. Do you think such a piece would be worth writing?

Thanks, and have a great evening!

Well, I have my complex prelim tomorrow, and since it's four hours long and will determine my future, I suppose I should sign off.

Thanks again for all your advice and insight, today and always.

-- Paul

 

[Diggrr]

Well macklin01, good luck tomorrow.

As for the comments...well, wow.
You've gotten alot further than I could have done. I know squat about engineering circuitry beyond the occasional resistor and pot.

That looks like a great place for the board, in that Antec USB drive.

Great first run though, and I think an article would be a very nice addition to the front page. I don't think anything like this has been covered up there.

Have a great'n!

 

[macklin01]

Well macklin01, good luck tomorrow.

Thanks -- it really means a lot. Well, the results won't be back for probably a couple of weeks, but hopefully ...

Thanks for your comments.

I know just about as little. I just remember what I did here and there when I was younger, along what I've read since. These forums have been quite a help, though!

I'm looking forward to having my project complete. I'm almost completely done with the instrument panel and am ready to mount it on the Easy USB. I removed the plastic front piece, adn on the advice of the Home Depot people, used krazy glue. It turned out quite well. Not a drop of glue showing on the outside, and no glue smudges. Apparently being 8-10 years older than in my old model airplane days has helped ...

So, next step is to reassemble the Antec USB thingee, measure the drive mounting holes, drill holes into the PCB board, and then cut the board to fit. Also, I'll measure out "no-build" zones on the PCB board to the components have plenty of clearence from the mounting.


Great first run though, and I think an article would be a very nice addition to the front page. I don't think anything like this has been covered up there.

Thanks for the encouragement. I'm trying to keep notes of what I've done. Actually, this thread has helped quite a lot for that.

And I appreciate all the positive comments and encouragement I've gotten from you and everybody else!

Thanks again!

Well, here in Minneapolis, we have Thudersnow!, so I should probably get offline. (and go shovel -- no metal shovels!!!

TTYL -- Paul