-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
FEATURED Building PWM Controller for 4 wires PWM fan
- Thread starter Brutal-Force
- Start date
Guys, first: Happy new year with a lot of interresting Fan-Issues here. 
Also i enjoyed this thread alot, even when not beein' quite active here the last weeks. But my interest in that PWM-HighCFMfans issue is still alive. So i've got my latest one, the San Ace 120 9SG1212P1G06 and first connected it to my 556 circuit. There i have a loud whine noise, as you guys are talking about one year ago. (Sorry for that re-animating issue). Then i've changed C1 from 470pF to 680pF as described here, but the noise was still there. In fact, i can see and hear no difference between 470pF and 680pF, whats quite confusing to me
. Finally, i tried to adjust it by my breadboard 555x2/311-Circuit, which is designed for work with 25kHz-frequencies i thougt. But this fan is running perfect on that refering to the noise profile. There's nearly no whining at all!
But why? Is it because the circuit brings a higher output level on the PWM-rail to the fan?
Some fan pics, 'cause they were so beautiful
:
And vids
First the one one with adorable sound profile (0 to around 25 % PWM duty cycle) due to the 0-100%-Circuit:
Second the sound profile with the 556-Circuit (2 - around 30% PWM duty cycle) and that disturbing whine:
Does anyone knew this 9SG1212P1G06 one? Seems to be the same spec's insteat of another Type number.
Also i enjoyed this thread alot, even when not beein' quite active here the last weeks. But my interest in that PWM-HighCFMfans issue is still alive. So i've got my latest one, the San Ace 120 9SG1212P1G06 and first connected it to my 556 circuit. There i have a loud whine noise, as you guys are talking about one year ago. (Sorry for that re-animating issue). Then i've changed C1 from 470pF to 680pF as described here, but the noise was still there. In fact, i can see and hear no difference between 470pF and 680pF, whats quite confusing to me
. Finally, i tried to adjust it by my breadboard 555x2/311-Circuit, which is designed for work with 25kHz-frequencies i thougt. But this fan is running perfect on that refering to the noise profile. There's nearly no whining at all! But why? Is it because the circuit brings a higher output level on the PWM-rail to the fan?
Some fan pics, 'cause they were so beautiful
:
And vids
First the one one with adorable sound profile (0 to around 25 % PWM duty cycle) due to the 0-100%-Circuit:
Second the sound profile with the 556-Circuit (2 - around 30% PWM duty cycle) and that disturbing whine:
Does anyone knew this 9SG1212P1G06 one? Seems to be the same spec's insteat of another Type number.
- Joined
- Nov 29, 2010
Nothor that's amazing - I've been having that annoying whine noise with my San Ace fans too and you seem to have found the solution... but how? How indeed!
Professor bing will have to give a technical explanation
In seeing this, I'm going to build the 0-100% controller and see if it will have the same effect on my fans too. Excellent
Professor bing will have to give a technical explanation
In seeing this, I'm going to build the 0-100% controller and see if it will have the same effect on my fans too. Excellent
- Joined
- Apr 4, 2011
whoa....
another shinny PWM beast
well,
this what I've got for the new year:
a nidec, and a cigarette as a popcorn substitute to watch all those insane PWM fans movies of yours 
nothor
yes, that SG whine is so anoyying (more like HF noise to me though)...
pardon me for being curious, but have you swap the 556 with a new one?
as Lenny said,
I'll wait for professor bing
another shinny PWM beast
well,
this what I've got for the new year:
a nidec, and a cigarette as a popcorn substitute to watch all those insane PWM fans movies of yours
nothor
yes, that SG whine is so anoyying (more like HF noise to me though)...
pardon me for being curious, but have you swap the 556 with a new one?
as Lenny said,
I'll wait for professor bing
inVain, i just hoked up the SanAce to bings 0-100% Circuit, and that brought more noise comfort to me. But i can't explain that effect. It's the first real advantage of that more difficult to built curcuit now
What the difference between the 3.3A-Nidecs and yours with 2.75A?
What the difference between the 3.3A-Nidecs and yours with 2.75A?
- Joined
- Apr 4, 2011
inVain, i just hoked up the SanAce to bings 0-100% Circuit, and that brought more noise comfort to me. But i can't explain that effect. It's the first real advantage of that more difficult to built curcuit now
What the difference between the 3.3A-Nidecs and yours with 2.75A?
yeah...
I assume, the 0-100% will have it's own advantage since the frequency circuit was made of pure resistors and a capacitor network (the 2-98% was consists of pot and capacitor)
about the Nidec,
I actually got lost too
the BetaV series was discontinued by the Nidec, and I can't find the datasheet either
well...it will be more for the noise/performance when I hooked up this fan to my controller and slap it on my Ven-X latter
@Railgun, woaahh .... , that is big multi channels controller you got there !
Ok, few modifications and suggestions ...
That led will going to be toasted if you connected like that, and I assumed the reason you put the led there is to show the light intensity that is correlated with the pwm signal right ? Like really dim at low duty cycle and max brightness at max speed.
You need to change that connection and add few components like this here :
View attachment 103185
Note : For the connection at "FROM 555 OUT PIN", just get one from any 555 outputs that going to be connected to the pwm fan line and share it with the fan.
This is a really serious power consumption we're talking bout here, just becareful.
How many Amps in total for all those fans ?
I'm worry with this kind of high current in total, even though molex connector is rated or capable of 11 Amps, but that is only in theory with perfect condition and a really good molex connector. Most molex currently are those cheap made in china knock off which I doubt the capable of carrying rated current.
Suggesting to double in parallel molex connections to beef up the power connection.
Another important thing is, make sure the +12 and the Ground rails at the pcb that going to the fan have a wide pcb trace, I'd say minimal 5mm width or even wider. And once the pcb is finished, layer these tracks with another soldering lead to thicken them to ensure they're good enough to carry big current.
Finally, make sure the power supply capable of driving these mountain of industrial fans.
Yes, you are right, my mistake and thanks for pointing that out.
The IC needs ground connection to work, I will update that soon.
Also I spotted another missing connection at the kick start add on module, its the connection for the transistor's emitter pin like below pic at the arrow.
View attachment 103186
Also, since you're going to buy lots of components to build this and probably there is a chance you might have a clearance issue since you've set the max board's size, actually you can replace that transistor with the 1N4148 diode like this. Hopefully this will help abit on the board space.
View attachment 103187
, that is big multi channels controller you got there ! Ok, few modifications and suggestions ...
That led will going to be toasted if you connected like that, and I assumed the reason you put the led there is to show the light intensity that is correlated with the pwm signal right ? Like really dim at low duty cycle and max brightness at max speed.
You need to change that connection and add few components like this here :
View attachment 103185
Note : For the connection at "FROM 555 OUT PIN", just get one from any 555 outputs that going to be connected to the pwm fan line and share it with the fan.
This is a really serious power consumption we're talking bout here, just becareful.
How many Amps in total for all those fans ?
I'm worry with this kind of high current in total, even though molex connector is rated or capable of 11 Amps, but that is only in theory with perfect condition and a really good molex connector. Most molex currently are those cheap made in china knock off which I doubt the capable of carrying rated current.
Suggesting to double in parallel molex connections to beef up the power connection.
Another important thing is, make sure the +12 and the Ground rails at the pcb that going to the fan have a wide pcb trace, I'd say minimal 5mm width or even wider. And once the pcb is finished, layer these tracks with another soldering lead to thicken them to ensure they're good enough to carry big current.
Finally, make sure the power supply capable of driving these mountain of industrial fans.
Yes, you are right, my mistake and thanks for pointing that out.
The IC needs ground connection to work, I will update that soon.
Also I spotted another missing connection at the kick start add on module, its the connection for the transistor's emitter pin like below pic at the arrow.
View attachment 103186
Also, since you're going to buy lots of components to build this and probably there is a chance you might have a clearance issue since you've set the max board's size, actually you can replace that transistor with the 1N4148 diode like this. Hopefully this will help abit on the board space.
View attachment 103187
Last edited:
Guys, first: Happy new year with a lot of interresting Fan-Issues here.
Also i enjoyed this thread alot, even when not beein' quite active here the last weeks. But my interest in that PWM-HighCFMfans issue is still alive. So i've got my latest one, the San Ace 120 9SG1212P1G06 and first connected it to my 556 circuit. There i have a loud whine noise, as you guys are talking about one year ago. (Sorry for that re-animating issue). Then i've changed C1 from 470pF to 680pF as described here, but the noise was still there. In fact, i can see and hear no difference between 470pF and 680pF, whats quite confusing to me
But why? Is it because the circuit brings a higher output level on the PWM-rail to the fan?
Some fan pics, 'cause they were so beautiful
Does anyone knew this 9SG1212P1G06 one? Seems to be the same spec's insteat of another Type number.
Great pics & vids Nothor, thanks.
That metal frame Sanyo is such a beauty compared to other plastic fan isn't it ? A really fine made piece of engineering work.
Btw, your current 120mm 12volt industrial fans collections is amazing
, and suggesting to add the beast Nidec Servo G1238B12BBZP-00 into the family, a 6300rpm with 261cfm and crazy static pressure at 2.09inH2O/520Pa, I'm pretty sure you will love it. Here, just to tease you, mine -> Here. ha..ha.. About the noise, yeah, the 0-100% version by design is definitely better breed than that 2-98% version, although I'm not pretty sure how those noise was generated, we need an oscilloscope to troubleshoot that and I will do my best if time allowed (sorry, its just I'm so damn busy lately), to trouble shoot both version and share the result here.
Anyway, since you're now officially PWM fan collector, c'mon, throw those 2-98% version away, since this design is a poorman version of pwm controller, the 0-100% version is much better one and I'd say its mandatory for a pwm fan lover like you.
lol @Dave,
Told you from the beginning, again, with your fans collection there, you should build that 0-100% version by now, c'mon, its not that difficult.
- Joined
- May 7, 2011
- Location
- Cook->Kent
@Railgun, woaahh ....
Ok, few modifications and suggestions ...
That led will going to be toasted if you connected like that, and I assumed the reason you put the led there is to show the light intensity that is correlated with the pwm signal right ? Like really dim at low duty cycle and max brightness at max speed.
You need to change that connection and add few components like this here :
View attachment 103185
Note : For the connection at "FROM 555 OUT PIN", just get one from any 555 outputs that going to be connected to the pwm fan line and share it with the fan.
The goal was to find a 5v LED hence no resistor, but yes, it will change brightness based on speed. If I only use a 2.2v or whatever, the proper resistor will be in place, but I didn't think another transistor would be required. I will make the changes accordingly.
This is a really serious power consumption we're talking bout here, just becareful.
How many Amps in total for all those fans ?
I'm worry with this kind of high current in total, even though molex connector is rated or capable of 11 Amps, but that is only in theory with perfect condition and a really good molex connector. Most molex currently are those cheap made in china knock off which I doubt the capable of carrying rated current.
Suggesting to double in parallel molex connections to beef up the power connection.
Another important thing is, make sure the +12 and the Ground rails at the pcb that going to the fan have a wide pcb trace, I'd say minimal 5mm width or even wider. And once the pcb is finished, layer these tracks with another soldering lead to thicken them to ensure they're good enough to carry big current.
Finally, make sure the power supply capable of driving these mountain of industrial fans.
It's an AX1200 so as far as the PS is concerned, I think we're good. Based on what I saw from Martin's testing, they draw .22A at 12v, so that's 4.32A for the 20 120mm fans. I assume the 80mm draw a bit less, but within the limit overall. The traces are just for representation at the moment. As this is my first foray into schematics on a serious level, there's a bit of a learning curve. I've already made some changes into the design and am learning as I go to make the appropriate trace widths. My big issue is ensuring the program has the correct components/sizes so I can build the thing properly. I'm still trying to determine how to define the board size.
Thanks. I noticed I was missing that ground too and also already corrected that.
I'll make the change you indicated and replace the transistor. I suppose I could make this a two level board since I have a bit of height to play with too and split each board to three and three. It should be simple to extend the voltage rails to that other board.
Thanks again!
Last edited:
- Joined
- Jun 6, 2002
when you hook up the LED's, go with the resistor via the ground. use one that will allow LED's max rated current when your fans full speed. might even want to underrate that by about 5mA or so. depending on the quality of the LED at rated max mA spec it will possibly start getting lower and lower in brightness as time goes on faster then other known brand led's.
Last edited:
- Joined
- Jun 8, 2005
Hey guys, just found out about this thread. I am a computer engineer and deal with things like this pretty much every day.
Why use a complicated 555 circuit when you can use a more specialized chip? Something like ltc6992 would suffice with 3 resistors, 2 capacitors and the voltage divider for the controller.
Why use a complicated 555 circuit when you can use a more specialized chip? Something like ltc6992 would suffice with 3 resistors, 2 capacitors and the voltage divider for the controller.
- Joined
- May 10, 2009
Those're surface mount only and cost ~$4, plus only 20ma output currant (555 is 200ma, that's a lot of fans )
Mostly the surface mount part is an issue for most DIY people.
I'm surprised they're that expensive really, as an Attiny24/45/85 chip can do the same job (and a lot of others) and cost $2.26 or so. The attiny has to be programmed of course.
)Mostly the surface mount part is an issue for most DIY people.
I'm surprised they're that expensive really, as an Attiny24/45/85 chip can do the same job (and a lot of others) and cost $2.26 or so. The attiny has to be programmed of course.
- Joined
- Jun 8, 2005
Those're surface mount only and cost ~$4, plus only 20ma output currant (555 is 200ma, that's a lot of fans
Mostly the surface mount part is an issue for most DIY people.
Ah, true. Completely forgot that surface mount chips are not easy to mount without a board.
As for the price, wouldn't it still be cheaper due to needing less accurate external parts?
- Joined
- May 10, 2009
I've been contemplating that, and I'm not sure.
The 555 is dirt cheap itself, then you need two zeners, a pot, 2-3 resistors, and three capacitors.
The PWM generator doodad you still need a pot and two of the capacitors (power filtering), then you need a voltage divider to get 1v, then a voltage divider with the pot to turn that into the 0v-1v input the PWM chip wants.
The 555 is dirt cheap itself, then you need two zeners, a pot, 2-3 resistors, and three capacitors.
The PWM generator doodad you still need a pot and two of the capacitors (power filtering), then you need a voltage divider to get 1v, then a voltage divider with the pot to turn that into the 0v-1v input the PWM chip wants.
- Joined
- Nov 29, 2010
@bing,
Yes I will build the 0-100% but I'm very busy just now so it may take a long time. But I will start to get some parts and draw up a layout as soon as I can.
I love the 2-98% for less powerful fans like my TY-140s and the AFC1212DE... even the FFC1212DEs work extremely well on it.
I suppose the only fan that really needs the 0-100% is the San Ace, but it would be nice to have it in my rig all the time if possible!
Yes I will build the 0-100% but I'm very busy just now so it may take a long time. But I will start to get some parts and draw up a layout as soon as I can.
I love the 2-98% for less powerful fans like my TY-140s and the AFC1212DE... even the FFC1212DEs work extremely well on it.
I suppose the only fan that really needs the 0-100% is the San Ace, but it would be nice to have it in my rig all the time if possible!
@Railgun
Led is a current device rather than voltage, for example like common 5mm led, usually they're rated at 20ma max.
The reason using a transistor for that led circuit is to minimize loading the pwm signal, so its sort of stealing a very tiny current (in order of micro amps) to power the led, instead of loading it directly at the pwm signal line that might affect the pwm signal badly if its loaded too much.
Anyway, it will be great to watch the build and the result, please share it here will ya ?
@ShadowPho
I'm aware of that chip, imo its just these circuits are targetted for hobbyist and people with very minimal electronic skill & knowledge, involving a smd type component might kill the interest, and also with that small thingy is almost impossible to use in breadboarding without adding more unnecessay complexities.
Beside the 555 ic is like the universal component that is available even in Timbuktu.
Led is a current device rather than voltage, for example like common 5mm led, usually they're rated at 20ma max.
The reason using a transistor for that led circuit is to minimize loading the pwm signal, so its sort of stealing a very tiny current (in order of micro amps) to power the led, instead of loading it directly at the pwm signal line that might affect the pwm signal badly if its loaded too much.
Anyway, it will be great to watch the build and the result, please share it here will ya ?
@ShadowPho
I'm aware of that chip, imo its just these circuits are targetted for hobbyist and people with very minimal electronic skill & knowledge, involving a smd type component might kill the interest, and also with that small thingy is almost impossible to use in breadboarding without adding more unnecessay complexities.
Beside the 555 ic is like the universal component that is available even in Timbuktu.
Last edited:
- Joined
- May 10, 2009
Doable though, just requires a bit of delicate soldering:
View attachment 103264
(That character may end up controlling a fan controller at some point, it's a thermocouple amp with cold joint calibration that speaks SPI, that's an Arduino clone in the background talking to it)
View attachment 103264
(That character may end up controlling a fan controller at some point, it's a thermocouple amp with cold joint calibration that speaks SPI, that's an Arduino clone in the background talking to it)
Even its doable, for people who live outside US, buying that kind of chip for hobbyist is ridiculously expensive, for example, that chip mentioned by Railgun, if I order it from places like Digikey, they will charge me -> $4 (the chip cost) + approx. $75 (handling + s/h fee) + (some crazy import tax varies at different countries) , so it could ended up about $100 or even more, go figure.
