Project: Madness (a Fan/Electronics modification)

Tealc · 23 Jul 2011 at 10:21

I've no idea what resistance is inside the transistor but now that my el-cheapo multimeter has arrived I can determine what current is flowing.

With no resistor as schematic I'm getting around 170 mA.

With a 470 ohm resistor I'm getting 18.3 mA. The voltage measured from the transistor emitter is around 10.7v from a 12.4v regulated source. So I guess the resistance of the transistor is dropping the voltage a touch.

Tealc · 23 Jul 2011 at 12:32

With some experimentation I've determined that to get slightly over 20mA I need a 390 ohm resistor attached. The voltage drop over the transistor decreases as the circuit draws more current. Is this a voltage divider effect? Not really sure how that stuff works yet.

The circuit is fully powered with no switch on my breadboard.

Tealc · 24 Jul 2011 at 01:57

Will be modding another fan (probably the translucent one to see what it looks like) and making a junction out of a terminal block. I could use crimped terminal connectors but double terminations and triple terminations are dodgy.

Waiting on 390 ohm resistors.

Could do with some 19-20 core cable too but that's hard to find.

Tealc · 24 Jul 2011 at 17:12

As you mentioned earlier putting a resistor on the LED slows down current flow and now makes the LED delay much much longer. Too long in fact, although it might be better now at slower cycle speed.

Instead of 10uF I'd need to use 1 or 2uF. This makes perfect sense as with 170mA of flow is 8 times more than the LED gets now.

Bit of a bummer really as I now have to buy 2.2 uF capacitors as well. Oh well I suppose I will have plenty of 10uF caps fir future projects. Using 1uF caps makes them a little too short to dim and 2.2 uF seems about right, better than before in fact.

I've wired up two circuits on breadboard and hooked them up to a switch and can see the difference in orange colour.

Tealc · 24 Jul 2011 at 18:07

Just to confirm it doesn't matter if I get 63v caps rather than 16v or similar as for some reason the lower voltage ones cost a lot more than the comparatively higher ones.?

Thanks.

I'm loving this tinkering. I didn't really expect it to be perfect first time out but I wasn't that far off and it does sort of work.

Tealc · 24 Jul 2011 at 20:27

And to give folks something interesting to look at here's a video.

This one has speech. I'm spoiling you lot.

Tealc · 25 Jul 2011 at 08:58

So the matrix board doesn't have copper strips. So I would need to wire wrap instead. With my shaky fingers that could be a tricky task and I have even less experience of mounting on matrix board than I do with stripboard.

I know it's ugly but I will probably stick with it on Stripboard mate. It will be mounted in a box (probably).

I was thinking of slicing the stripboard into 2 parts though.

Tealc · 26 Jul 2011 at 08:41

Might be worth grabbing some of that stuff for a future project.

There does seem to be some difference though between types. No copper. 3 hole strip. Single hole.

I find working with stripboard quite confusing so it might be a welcome change as you can plan the circuit without thinking about the long strip circuits and not worry about removing the tracks, which incidentally caused an error or two for me.

Tealc · 26 Jul 2011 at 23:38

Prepped the second fan (the semi transparent one) today by drilling 18 holes. Also cut the cables and prepped the ends ready for a terminal block and resistors I'm waiting for.

Managed to replace the 10uF caps for the 2.2uF as well. The 50v 2.2s are identical in size to the 16v 10uF ones I had fitted previously.

Hopefully the terminal blocks and resistors will come tomorrow.

A screw type terminal block (a small thing) will allow me to have several fans served off one connector if I desire such a thing. What I do know us I wont be using these Tricools or Alpenfoehn fan in my rig. I might grab some decent ones.

Tealc · 31 Jul 2011 at 10:41

The 2.2uF caps in place.

As expected I had a few things in the post over the last few days.

First up were the 390 resistors. Soldered a resistor to each of the 18 LED anode wires and then added an additional length of cable to the end to extend the overall lead to the fan. Put some heatshrink over the joint.

And the terminal block. It's a PCB mounting one with a screw type fastening for accessory wires.

I soldered the incoming wire from the controller to the PCB pin and fitted the fan wires to the screw end. It doubles over itself a bit and I'm not 100% happy with them as I thought they might be pass through terminal blocks like you get on mains.

The overall assembly with black bladed fan.

Next up the translucent Alpenfoehn fan...

Tealc · 31 Jul 2011 at 10:53

Before commiting to the drill I test ran a LED with the original LED hole powered by a 3volt lithium battery.

It's quite bright.

Having learnt about mounting 18 LEDs with the Tri-cool it was so much easier adding LEDs to the Alpenfoehn fan. Marked then drilled the holes around the circumference and added in some LEDs.

This time it was so much easier and I was able to make the job much neater than before.

And then wire it up to the controller by fitting the wires to the terminal block and power on.

Short video recorded during the day.

And shots at night time.

Video of both at night. It is not a coincidence that they both start lighting at the same point on the circumference.

And some of the Alpenfoehn fan on it's own.

It really is shocking how much noise this fan makes at it's full speed.

Overall I am pleased with the result of the second fan and it's better than the opaque.

I think I might actually go with translucent fans when building this into a system. Not the Alpenfoehn fan but maybe an Enermax TB silence.

I wonder how much energy this circuit consumes. I bet it's a fair amount.

Tealc · 7 Aug 2011 at 15:32

Now that I've finished, sort of, I thought it might be a good idea to speak about the costs involved buying the many components needed to build this.

I couldn't figure out how to copy and paste a table here, or how to import ane excel spreadsheet so a screen grab will have to do.

This list doesn't include heatshrink, the wire or solder used in the project, nor does it count the breakages, blown ICs and stuff.

LEDs cost me £4.40 for 50.

I got all but one item from the UK so I could probably have done it for half the price if I'd got everything from Hong Kong or gone for larger batches of stuff.

Tealc · 7 Aug 2011 at 19:49

It would be simple enough to connect the hdd led output to the clock input of the 4017 but not sure how you could slow the signal down to minimum 100ms. As can be seen from the videos the 4017 can accept scary fast inputs and at times hdd activity can be pretty fast.

I'm sure there are ways of slowing down and ignoring unwanted pulses but having only limited knowledge I dont have the perfect answer.

The 555 could also receive the hdd led current and only relay a pulse to the 4017 every 100ms or more. As I see it the cap if the 555 would charge a little bit on every little pulse the activity led sends. Even if they are 1ms pulses the cap would eventually charge and send a pulse down pin 3 to the 4017.

Not sure if I would want to test this with a working system though.

Funny though as I was thinking of the hdd led earlier and how it could be used for stuff.

Tealc · 7 Aug 2011 at 22:31

MAXIBOY said:
just a thought as im very impressed with this project but running lights as pairs on opposite sides on the fan might look better just an opinion

You mean two LEDs opposite each other on each fan? Or rotate one of the fans through 180 degrees? Or have on go clockwise and one go anti clockwise?

Tealc · 8 Aug 2011 at 10:56

Probably would involve totally changing the way the circuit works. It could be done though with a single 4017 decade counter.

I suppose I could hack it by joining output 1 to output 9 and output 2 to output 10 etc but I'm unsure how the joining of outputs would affect the dimming circuits.

Upon reflection though that sort of tinkering is better if the circuit were microcontroller based.

Tealc · 8 Aug 2011 at 11:30

I'm not sure that would work right as the circuit stands.

It would be simple to wire in LED 1 and LED 9 into the first active circuit on the terminal block and each LED would get 22mA of current. The trouble would start when the circuit reaches output 9 as nothing would light up. I'd then either need to stop the count at 9, which is something I didn't manage to work out on dual 4017s, or wire an additional 18 wires for the second revolution. I'd need to build on breadboard to work out how to stop reset the count.

Because it was never designed to do two LEDs at a time anything I do to the circuit now will be a hack.

Anyway here's a video of one of the fans rotated through 180 degrees so the LED light on opposing sides.

Now wondering what it would be like if the LEDs were to light clockwise on one and anti-clockwise on the other.

Tealc · 8 Aug 2011 at 20:32

Right I've tested it and if I short pin 9 to pin 15 (reset) the decade counter resets after pin 9 and starts again. Not sure why it didn't work in multisim testing but it works in real life. Will give it a go anyway just for fun and post results. Shame I didn't try this on breadboard as I could have gone for 16 LEDs which is much easier to wire onto the fan.

Tealc · 9 Aug 2011 at 16:01

Hmm no go at the moment. I've wired up a switch to pins 9 and 4017 pin 15 which seems to reset the circuit but it looks like this is actually not going to work. Pin 9 seems to stop the circuit at 8 and causes pin 9 and pin 5 to flicker. 9 is slightly less bright than 5 but there seems to be an issue there somewhere. No errors in soldering as when run normally no flickering occurs.

Shorting pin 10 to 4017 pin 15 causes the circuit to hang.

Maybe I should try pin 1 of the second 4017....

......nope that halts too and the second output gives strange behaviour too. LEDs that shouldn't be lit flicker when they aren't supposed to.

I think for the time being I'm stuck with 18 LEDs in rotation. I was thinking of getting some 8 circuit dip switches too to have dual modes but maybe the nature if the circuit doesn't allow it.

I think I might need to have a third decade counter separate from the other two which I could enable and disable via a switch. This would be simple enough with a switch on the 555 pin 3 I'd imagine and another bunch of wires on the board.

Tealc · 12 Aug 2011 at 16:02

Hmm. That's a good one. I've got almost 50 common anode RGB LEDs here already.

Sounds like an interesting project. Not quite sure how I could vary the PWM of the individual colour LEDs by using 4017s and 555s automatically but I'm sure it's possible. Maybe trimmers for each colour.

Edit: 3 hours later. Uurgh. This will be a difficult task with discrete components. I could, theoretically string 3 4017s together to trigger 9 Red, then 9 Blue, then 9 Green before the circuit starts again but fading and different colours would be tricky and use hundreds of components. Easier to go with a microcontroller I think.

More thought needed I think.

Tealc · 14 Aug 2011 at 17:25

This RGB thing has been going through my head for the last few days. I don't know why I can't just leave things alone.

So I came up with a circuit that would allow me to sequence through the individual colours Red, Green and Blue and also add mixes.

The circuit uses an additional 4017 decade counter IC, on the right, to count Tens while the first 4017, on the left, counts the units and actually drives the LEDs. This will give me up to 9 lighting effect options, more if I cascade. I've only gone up to five LEDs in this example because I couldn't be arsed to do more. :/

I then use a transistor to act as a switch from the outputs on the second 4017. I believe this is a multiplexing technique where a single anode is live when a single cathode goes to ground. Whether the way I have done this is correct or not I'm not sure. It works and that's enough for me.

Integrating dimming effect would be quite complex as I would need at least 6 transistors, if is even possible.

I've been looking at integrated RGB strip controllers such as this which would give me total control over the colour. Indeed I am already waiting for one that I will be using to modify my keyboard backlighting.

I'm not sure how I could integrate this into a sequencing circuit but should be fun trying. I could have course just have 18 LEDs lit by the controller and just have them lit all the time.

Anyway here's a video of the RGB circuit as it stands.

The video, after a time, goes Red, Green, Blue and then on the fourth a sort of Reddy/Orangey/Yellow colour which is created by driving the Red to high current and the Green element to medium current. LEDs don't dim very consistently with resistance but wiring up a variable PWM circuit would just get too complicated.

I'm grabbing some 40106 oscillator ICs which should give me a fading effect rather than this switching effect.