fan emulator based on 555 - questions

[magnetic256]

Hey guys,

Recently I've been undertaking a quick project to build a 'fan emulator' (using a 555 timer to simulate a tachometer signal). Using this schematic here:
https://www.techidiots.net/notes/fake-fan-sensor

Here were the parts I purchased:
https://www.digikey.ca/product-detail/en/texas-instruments/TLC555CP/296-1857-5-ND/277502
https://www.digikey.ca/product-detail/en/vishay-sfernice/T93YA502KT20/T93YA-5.0K-ND/1587722

Top view:

Bottom view, with magenta highlighting to explain what I was connecting with the solder blotches (yeah, not the best but I was sort of lazy-- I probably made my FX-951 feel ashamed it had to solder that):

In a nutshell it doesn't seem to be generating a signal (although it could just be the computer I was using isn't suitable since it's a bit odd and has PWM and TACH swapped-- even swapping the signals I still wasn't getting anything). I'll try a different computer tomorrow but just wanted to ensure I had all of the ducks in a row.

In a nutshell my questions are:
A) did I purchase the correct flavour of 555?
B) is there anything wrong with the vishay 5K pot that's not suitable for this?
C) Does pin 3 of the 555 (according to the circuit diagram) get soldered to the MIDDLE pin of the 5K pot, and pin 2 of the 555 to either of the *outside* pins of the 5K pot?


Cheers,

PS: I've been meaning to join the EEVBlog forums for awhile, guess this finally makes a good reason to do so

[alsetalokin4017]

I can't tell from your photos if you've wired it together properly, but I've built the circuit on my breadboard and it does work. You do understand that it acts as a current sink, right? You won't get much "signal" from the Pin 7 output, as the circuit pulls Pin 7 low for about half the time and leaves it floating for the rest of the period. So to see if your circuit is working you can connect one end of a 1k ohm resistor to the positive supply, the anode of an LED to the other end of the resistor, and the cathode of the LED to Pin 7, then scope the Pin 7 and you should see the signal. The LED will be flashing so fast you probably will just see it as continually on, but if you make the capacitor larger (for testing purposes) like 10 or 100 uF you should be able to see the LED flash.

To answer your specific questions:
Yes, the 555 version you bought should work.
No, there is nothing wrong with the specific trimpot you've selected. It should work fine.
Yes, the Pin3 output should go to the wiper of the pot, which is _usually_ the center pin.

[magnetic256]

I can't tell from your photos if you've wired it together properly, but I've built the circuit on my breadboard and it does work. You do understand that it acts as a current sink, right? You won't get much "signal" from the Pin 7 output, as the circuit pulls Pin 7 low for about half the time and leaves it floating for the rest of the period. So to see if your circuit is working you can connect one end of a 1k ohm resistor to the positive supply, the anode of an LED to the other end of the resistor, and the cathode of the LED to Pin 7, then scope the Pin 7 and you should see the signal. The LED will be flashing so fast you probably will just see it as continually on, but if you make the capacitor larger (for testing purposes) like 10 or 100 uF you should be able to see the LED flash.

To answer your specific questions:
Yes, the 555 version you bought should work.
No, there is nothing wrong with the specific trimpot you've selected. It should work fine.
Yes, the Pin3 output should go to the wiper of the pot, which is _usually_ the center pin.

Thanks for the help,

Well, it appears that the second adapter I made started working and spoofed an 8000 RPM tachometer signal (computer picked it up fine). However, after adjusting the trimmer to a lower value it stopped working after a few turns. I have a feeling the vishay trimmers are the culprit (I tested the connections with my multimeter and there was nothing wrong, and the circuit is fine too). So I've ordered a variety of different trimmers to try, we'll see if the next round is successful. I'm going to redesign the layout a bit to make reworking components easier.

[magnetic256]

After troubleshooting and building a lot of prototypes, I found the issue. It has to do with how the circuit is built around the capacitor.

I've tried a variety of different capacitors and the only one (that I had on hand) which would work for longer than a few seconds was a 68uF vintage Sanyo OSCON. Since it's ceramic instead of electrolytic, the ESR is much lower.

Basically what ends happening is that I have to adjust the pot (dropping the amount of resistance) and then quickly increase it-- as this is done the capacitor undergoes a transient period-- and it's ONLY during this transient period does the 555 timer output an RPM value. The more extreme the resistor is adjusted, the shorter the transient period lasts; thus resulting in an RPM output that only lasts for a few seconds. So for instance adjusting the 68uF OSCON at 5000 RPM, it will only output for like 3 seconds. Whereas if I adjust it to 400 RPM, it's more akin to 5 minutes.

Is there any way to reliably build this circuit where it's not reliant on a transient period? The only workaround that I could think of is oscillating the 5K pot thus allowing the capacitor to perpetually be in its transient period... but that could be unreliable.

Or perhaps the signal being outputted from the 555 could be simulated with something like a teensy?

As for why this circuit worked for everyone else, my only guess is that maybe a low-quality 1uF capacitor was used where it perpetually stays in a transient period and never fully charges (like a TEAPO or G-Luxon or something). I don't have any of those on hand, only Rubycon / Chemicon / Vishay etc.

[alsetalokin4017]

Ah.... no. I don't think you have identified the problem in your circuit.

I happen to still have the circuit built up on a breadboard. I've tested the circuit with the 1 uF value of  tantalum, poly-film, ceramic, and several different brands of electrolytic capacitors. It works through the full range of the 5K pot adjustment, from about 115 Hz to about 50 kHz. It continues to work as long as I supply it with power and it does not require any "fiddling" to start up.
Some capacitors do give a better shaped signal than others; right now I have it working nicely with a poly-film type and this seems to give the best "square" pulses with minimal over/undershoot.

You do understand what I said in my last post about the thing working as a current sink, right? Did you try the LED indicator system that I suggested? Do you have an oscilloscope?

I would suspect that you have one or more bad solder joints on your pad-per-hole construction and that's why it isn't working properly for you. Also, in your photo of the board it looks like you might have a short between Pins 6 and 5 of the 555.

Did you breadboard the circuit first on a solderless breadboard, and test it as I suggested with an LED and also an oscilloscope?

[alsetalokin4017]

The thought just occurred to me: if the only way you are testing your circuit is by plugging it into the fan connector of the motherboard -- maybe the motherboard is trying to regulate the fan speed by reducing/shutting off the power to the "fan". You could check this by connecting an LED+resistor between the Red and Black wires, or use your DMM to check the voltage on the Red lead wrt the Black lead.

By the way.... with a 68 uF capacitor and the 5 k pot, the circuit works from about 1.5 or 2 Hz to a maximum of about 850-950 Hz depending on applied voltage. I don't know what your fan controller will make of this slow speed range, but I'll bet it will be trying to speed the "fan" up by providing it with constant power, more or less.

[magnetic256]

Ah.... no. I don't think you have identified the problem in your circuit.

I happen to still have the circuit built up on a breadboard. I've tested the circuit with the 1 uF value of  tantalum, poly-film, ceramic, and several different brands of electrolytic capacitors. It works through the full range of the 5K pot adjustment, from about 115 Hz to about 50 kHz. It continues to work as long as I supply it with power and it does not require any "fiddling" to start up.
Some capacitors do give a better shaped signal than others; right now I have it working nicely with a poly-film type and this seems to give the best "square" pulses with minimal over/undershoot.

You do understand what I said in my last post about the thing working as a current sink, right? Did you try the LED indicator system that I suggested? Do you have an oscilloscope?

I would suspect that you have one or more bad solder joints on your pad-per-hole construction and that's why it isn't working properly for you. Also, in your photo of the board it looks like you might have a short between Pins 6 and 5 of the 555.

Did you breadboard the circuit first on a solderless breadboard, and test it as I suggested with an LED and also an oscilloscope?

I've actually made mulitple instances of this circuit and tested with my multimeter, there are no shorts; It's built precisely to that schematic.

The only variable that affects (anything) is the capacitor after many iterations of playing around with this thing.

The thought just occurred to me: if the only way you are testing your circuit is by plugging it into the fan connector of the motherboard -- maybe the motherboard is trying to regulate the fan speed by reducing/shutting off the power to the "fan". You could check this by connecting an LED+resistor between the Red and Black wires, or use your DMM to check the voltage on the Red lead wrt the Black lead.

By the way.... with a 68 uF capacitor and the 5 k pot, the circuit works from about 1.5 or 2 Hz to a maximum of about 850-950 Hz depending on applied voltage. I don't know what your fan controller will make of this slow speed range, but I'll bet it will be trying to speed the "fan" up by providing it with constant power, more or less.

Motherboards only regulate fans with PWM not with the tachometer signal, as we're omitting the PWM pin in this circuit design the motherboard has no control over it. However, I went ahead and stole 12v and ground from the main power supply, and it the same behaviour ensued which is expected.

I think it actually makes more sense to just pull the power and ground off of a molex connector instead of the fan connector (I can just feed the fan connectors the tachometer signal to keep things simple; and it's possible to just split it to give multiple connectors the same signal, thus eliminating the need for multiple adapters).

---

I've hooked the circuit up again (using the 68 uF OSCON), and it looks like its stable at 132 RPM running for over 20 minutes. However, that's pretty useless for my needs :p
Tried adjusting it beyond that and it just cannot go beyond that. Using the 1uF electrolytic capacitor results in a value that cannot be held.

[alsetalokin4017]

Well, you've got to be asking yourself why the circuit works for everybody who has built and tested it, except you.

Perhaps there is something about the pulse width (it isn't perfectly symmetrical 50-50 Hi-Lo throughout the frequency range) that your computer doesn't like.

The frequency range of this circuit is set by the combination of the capacitor value and the potentiometer value. Whether or not this action is what your computer wants to see is a different issue altogether.

But wait--- 
I've just done some further testing on my breadboard and I found that the 555 does heat up unacceptably at the top end of the frequency range when using a 1 uF cap and a 12 volt supply. Maybe the problems you are seeing have to do with chip heating.  With a lower supply voltage like 5V, and using a 1 uF capacitor and a 5k pot, the circuit is perfectly stable in doing _what it actually does_. Since it operates by alternately floating and pulling pin 7 low, it should still work for your tachometer input if you use a 5V supply to the 555 rather than the 12v supply, I think.

I'd suggest you try using 5V supply and see if that improves the performance with the 1uF or 2.2 uF cap.

[Ian.M]

TLC555CP is CMOS, unlike the original bipolar NE555, ao you *MUST* *NOT* leave any inputs floating. 

Your problem is probably the floating Reset pin (4).  Link it to Vdd (8 ), for the 555 to operate normally.

Also the datasheet recommends 0.1uF decoupling cap from Ctrl (5) to Gnd (1), and you should probably also have one from Vdd to Gnd.

Finally, electrolytic timing caps aren't the best for 555 circuits - its usually better to use a smaller non-electrolytic cap and increase the timing resistor(s) to compensate to get the required frequency or period.  If you must use one, the marked -ve side *MUST* go to Gnd.

[alsetalokin4017]

Excellent points. I was using a NE555N for my breadboard testing since that is what was used in the original circuit linked by magnetic256. But replacing it on my breadboard with the CMOS TLC555P indeed causes instability in the frequency of the output signal. Connecting Pin 4 to Vdd solved the problem and produces a very stable signal. I didn't bother to include any decoupling caps on my breadboard.
This version is also much less susceptible to frequency changes as the supplied voltage changes, nor does it heat up with 12v supply, as far as I can tell.
So maybe the simple fix of connecting Pin 4 to Vdd is all that is needed to solve magnetic256's issue.

[magnetic256]

TLC555CP is CMOS, unlike the original bipolar NE555, ao you *MUST* *NOT* leave any inputs floating. 

Your problem is probably the floating Reset pin (4).  Link it to Vdd (8 ), for the 555 to operate normally.

Also the datasheet recommends 0.1uF decoupling cap from Ctrl (5) to Gnd (1), and you should probably also have one from Vdd to Gnd.

Finally, electrolytic timing caps aren't the best for 555 circuits - its usually better to use a smaller non-electrolytic cap and increase the timing resistor(s) to compensate to get the required frequency or period.  If you must use one, the marked -ve side *MUST* go to Gnd.

Gahhhh!!!! That was it!!!!

I should have clued into that possibility (especially since the old vintage SIGNETICS 555 was used by the other guys), and I know Dave mentioned it in a 555 video.

Well thanks, it's working as intended now.


Cheers