ChielScape

if you do, make sure to test that with 180+W, to make sure it doesnt melt. (hence the ratings)

__________________

oliverw92

By melt do you mean the plug?

__________________

ChielScape

the cables, plug, whichever runs out of luck first.

__________________

oliverw92

http://en.wikipedia.org/wiki/Molex_c...tion_System.29

Each pin can take 11A, therefore 132 watts on each pin. The transistors on our circuitboard could take up to 5A if you had a perfect heatsink i.e. even better than a megahalems on each one. Realistically, if you put a ram sink on each transistor, you are looking at 2A per channel max. Thats 24 watts. Even so, most fans do not use that amount of power. You could run 6 fans at 2A each off of our circuit, or 8 at at 1.4A each. Those would still be some pretty powerful fans.

At the moment we are sorting out PWM noise by using a capacitor to reduce the pwm slew, in laymans terms, not stopping it instantly, removing the clicking. If that doesn't work, my idea is to split the PWM signal coming out and send them both to transistors, but one of them goes through a NOT gate first, inverting the signal. The transistor this one is connected to will be allowing a low voltage, say 1-2v, to the fan. The other will be allowing 12v. This means the fan will never turn off. I might breadboard this and see if it will work.

__________________

ChielScape

my T-balancer has a software ajustable PWM frequency, and that does a perfect job at making the fans sound nicely.

also, how's a waterblock for those FETs sound?
what's their efficiency and max operating temp? (noob question though: is the max amp rating based purely on a realistic operating temperature achievable, or something else? like electromigration, etc)

__________________

oliverw92

The max amperage of 5A is based on if you had a infinite cooling source, i.e. if you can remove all the heat made. you could probs get to 4A if you watercooled it. I do not know the exact numbers, i will have a chat to brian (he designed the board). So yeah the 5A is probably something like electromigration. 2A is the realistic limit do to temperature however. LN2 on a fan controller would be a tad excessive .

This board will obviously allow you to adjust the PWM frequency using software, however as you probably know, if you want a large fan control range, depending on the fan, when you get low down you get into horrible noises.

__________________

ChielScape

technically, you always remove all the heat you generate, as that is what causes the temps to stabalise. question then is, does watercooling keep the temp below the max when operating at 5A, since i assume thats the amperage at which the max temp is spec'd.

and about the noise, it doesnt happen when you tune the frequency right.
I have Ultra Kaze 3000's on PWM. those fans are said to have a real bad clicking noise even undervolted by a normal controller. i use them on PWM and i dont get any clicking at all. (neither on analog, so people are just spreading lies, but that aside) so from my experience it works great, even going down all the way to 5%

i cant say that about other fans obviously, but im pretty confident though.

__________________

oliverw92

For my minime project i have made myself a fan controller using PWM. It is different to the PhantomLED one but uses same principles. I tried a Scythe S-Flex on mine and i get horrible clicking. The only time i dont is if i use a tiny 100pF capacitor as the timing capacitor. This cap controls the range (i.e. how low down you can drop the PWM frequency). However with that tiny cap i got barely any range, i could just about hear the change in pitch of the fan. At 0.1uf, the original, there was clicking, but it was bearable, and mounted with rubber washers would be no problem at all. Any cap inbetween the two would result in the fan emitting a horrible high pitched tone. However i tried another fan on there, a 60mm 3krpm fan and the 0.1uF works perfectly, whilst anything else gives a little bit of noise. It really does depend on the fan and how it is made, how much current/voltage is needed to start it etc. I'm not certain, but i would be pretty sure if you put a scythe s-flex on the T-Balancer, you would get clicking at low speeds.

As i have said before, my solution is to alternate between the 12v and a lower voltage, you could use caps to smooth the transition. This would remove the 'digital' signal effect of the transistor switching on and off.

__________________

ChielScape

perhaps you could turn the PWM signal into an analog one, basically making the effective voltage the real voltage. you still have the high power capacity advantage PWM has over a rheo, but you wont have any clicking whatsoever. you do lose the advantage to start below starting voltage though.

__________________

oliverw92

The pwm signal is a miniscule current and voltage. So you would need a transformer. Also, the PWM signal is binary i.e. on or off. If i turned it into an analog signal by transforming it into the required 12v, it would still be on and off. You could slow the slew of the on/off by using a capacitor, but this wouldn't help much.

__________________