Multi-channel fan controller for several PWM/Non-PWM fans Vs a decent MB?

[LennyRhys]

My guess is that if you go with resistors; you're going to need some mighty big ones so that they don't just burn off... and using those still leaves us with the fact that all that voltage that they are not sending to the fan is converted to heat that you're goint to be putting into the case... It kind of defeats the purpose of high powered fans if you're just going to be moving hot air around.

All good advice. I use wirewound resistors on all my non-PWM fans and they are great on the less power-hungry fans, but on the Delta the resistors get HAWT, so I tend not to use them for extended periods of time. I used a 10W 5.6R to knock the Delta EHE down to approximately 5v, and it did a great job but also melted through the plastic encolsure that housed the resistor and toggle switch This is where PWM trumps other methods of speed control: highly efficient, and no extraneous heat.

I got some hands on figures. i dont do math .
delta 1amp 12v 12W fan, requires a 4.8V drop to get to 1/2 power (not speed)
resister has .5A at 4.8V = 2.4W <--waste heat
fan itself only using 6W

for 1/3rd power
resister has about 7.3V drop at .33A = 2.4W <--waste heat
Fan using only 4W
all figures are approximate.

I found a great fan resistance calculator which I bookmarked for convenience:

http://www.blackfiveservices.co.uk/fanspeedcalc.php?

 

[Martinm210]

Something else you need to think about regarding thermally throttling the fan is CPU core dependant vs water.

I haven't tried either myself, but "Think" it would be a much nicer and smoother ramp up/down using water. Under core control (MB), the water will slowly rise until it either hits the limit under low loads OR...it will hit every time you do something slightly intensive. So it would very much probably ramp up and down rather abruptly.

However...using water temperature...because of water's natural specific heat storage benefits, you would essentially give your ramp up/down a natural capacitor of sorts and smooth out that ramping problem.

You would simply have to set the water temperature to a correlating acceptable core temperature.

For this reason, I think ultimately you'll need to look beyond motherboard control if you are very particular about noise.

I only say this because my work computer has a PWM fan on it, and I personally find that ramp/up down every time I do something intensive to be a bit annoying. Rather than just spin up partially, it' pretty much pegs high rpm then back down to low in very short order. Perhaps this could be smoothed out with more precision PWM settings (Work computer locked down by our big brother IT folks), but I just haven't been happy myself at least with that rig. It's definitely better than listening to a fan full bore all the time, but the ramping up and down so quickly is a bit disappointing.

The solution to this will of coarse require a fan controller, one that has the ability to control fans via some sort of temperature probe you can attach or insert into the fluid stream somehow.

As far as you're discussion on PWM vs voltage control of the fans...either is fine. The only real advantage I see with 4 pin PWM fans is "Some" have a little lower RPM floor, but both if designed optimally are equal on the noise front. If anything, my preference sways toward voltage regulation, it's less testy and has shown to have smoother sound quality in some fans.

Hope this helps..

I haven't decided quite what I'm going to try. I have my son's pc that's fairly minimal, it just has a single MCR320 mounted in the top with one case fan. In his rig, I may go ahead and try a set of four Arctic Cooling F12s and give PWM a try, I just don't think it's quite the control I want. I've got a single loop though, so it would probably work ok with some careful PWM settings. I just think controlling by water temp would be much more ideal for water cooling.
Cheers!
Martin

 

[Seebs]

Man... I love this site. Only at OCF can we take something so simple as controlling the speed at which a fan spins and turn it into the next C.E.R.N project!!!


That ocean vs glass of super heated water reminds me of the "What's heavier: A ton of bricks or a ton of feathers" curve ball that our physics professor threw at us back in my High School days.

@Martinm210
If that PWM fan at work is ramping up and down is not because of the nature of PWM, but most likely because the motherboard header is not doing the controlling in a smooth curve and doing it in 10% increments instead (most motherboards do it like that)... Hence the want/need for a smooth PWM controller like the ones being built here... Not exactly what the OP is looking for since he has stated he wants the entire thing to be automated, but a good starting point... Now if someone could figure out a way of taking the PWM signal from the MoBo headers and smoothing it to a nice curve; then we could probably build a fully internal controller that takes the MoBo's PWM signal, makes it nice and smooth and then feeds that signal to the fan.

@m0r7if3r

Amen brother!!!!

 

[Martinm210]

Yeah, the more I think about it...the more I'm personally drawn to something like the mcubed Tbalancer bigNG/mini or AC poweradjust controller. Not quite sure which is better, but they both seem to have tools to tweak things via software.

Or...very basic like the mcubed fanamp would probably do it too without breaking the bank and could do it all analog.

 

[Brutal-Force]

I have built this controller (see my link) and have been running a San Ace for 6-8 months.

The problem with running this off a motherboard is a) the current draw is immense and will quickly burn out a fan header and b) motherboard manufactuerers set up the PWM signal in 10% increments from 0-100%.

A) is bad, no matter how you slice it, but you can get by just rewiring the Power to the PSU and the Signal to the board.

B) is not so simple. While yes you can effectively jump in 10% increments, this works for lower spead fans because you incremently just 10% of the rated fan speed. 1000RPM fan jump 100 RPMS at a time, 2500 RPMS jump 250 RPMs. But a 6000 RPM fan jumps ~600+/- RPMs which is akin to switching a Hairdryer with 5 different speed settings up and down till it drives you insane.

If your looking for a quiet system, then just go with a slower fan. If your looking for maximum cooling, then you will need a high speed fan. IF you do this, you need to make a PWM Controller. There are no official manufacturers (at least that I could find 6-8 Months ago) and if they are, they are surely more than $100.

We can discuss semantics/physics all day and night and until you actually do it, you can not talk about it. Theory and practice are two different things, and up until now there is probably only a handful of us who have practical knowledge. I am not an engineer, but let me tell you, building this was fun, but it was also a *****. I learned a lot of valuable hands on from Bing. Give it a try if this is what you want.

The simple answer to your question is:

a) No, there are no multi channel controllers to run PWM fans unless you build one.
b) If you try to just undervolt a high speed PWM fan, you will be plauged with Clicking and whining and you will regret it, not to mention you will loose performance.
c) The Motherboard option should only be used if you are going with a fan under 2000 RPMs. Even when I used a 2500 RPM fan, it was annoying.
d) Non-PWM is fine for a multi set up. Go a Multi-fan controller for the case fans and a PWM for the CPU and then build your own controller.

Just do it!

 

[hokiealumnus]

First off - Martin, welcome to OCF. It's about time you got here and it is truly an honor sir.

Now, to the topic at hand. While I don't think a dedicated controller is imperative, as complicated as this is getting, I think it's probably a good idea. It will a) keep your wiring cleaner, b) allow you to see the RPM and fine-tune fans as you want and c) save you a lot of headache separating the fan's controller & voltage wires, running them all over the place to your headers (see a)).

What I don't get yet - why are you getting PWM fans? If you're getting Scythe GTs, they come in three-pin models. As a matter of fact, Sidewinder only has three pin models.

For the reasons above, I'd recommend giving up a bay to the most attractive fan controller to you and using voltage-controlled fans. If you're going for under-stated and don't like flashy, bright lights, the Lamptron Touch -reviewed here- is even completely blank (except for the power indicator in the lower right corner) until you power it up. Just make sure you get quality, with good wattage ratings. Watch out for those that look good but are only rated 7W/channel. You seem attuned enough to realize that already though.

EDIT - If you really wanted to avoid a bay controller, still get the voltage-controlled fans and you could go with a GUI controlled T-balancer bigNG that Martin mentioned.

PWM seems to be the biggest obstacle. If you can avoid that altogether, your options expand quite a bit.

EDIT II - Cleaned up your thread for you.

 

[jalyst]

Apologies for the ridiculously delayed response again, crazy time of year.

I had posted to get an interest check, but there were no bites, so I decided against producing them. Initially your looking at a 50-100 dollar cost for a "not so pretty" controller. And its pretty much beta, so you take what you get, which is pretty expensive. The bright side is that one should be capable of running multiple fans. I can also make it pretty small. Again, price and looks make it a not so desirable device, not to mention temp controlled is not even in the equation. this should be considered extreme cooling based off the price alone.

Thanks for the offer,
But running-off my MB will give me temp activated fan control, as well as voltage/RPM feedback, for 2 grps of fans & 2x individual fans.

If I can't have this functionality on a discrete* then I'm giving-up too much.
The only thing YTBD, is if my 2x PWM headers are of sufficient quality to lower these 25mm 5400RPM fans, to their lowest rated speed of 1000RPM.

*I'd prefer that it interfaces with the MB to attain this, but if it's all integrated into the controller & that's the only option on the market, then I'm happy to settle.

Chipset/vrm/mem don't need active cooling for overclocking, unless you want to break some hardware records on hwbot; and chassis fans in that area are usually extracting air, not actively cooling. I built my controller for benching, and when I say benching I mean voltages of 1.65v on the CPU and QPI (forget PWM fans and copper heatsinks...most people using these voltages use dry ice or liquid nitrogen LOL).

Def. not going to bother with a level of OC'ing that requires dry ice or LN!
Not sure what sort of voltages I'll be able to push on my core i3-530 & H20-220 Edge WC kit, but I'm definitely not going for any benching records!

Why not consider passively cooling your system, especially if it is an HTPC and it's going to be underclocked most of the time? I tried passively cooling my CPU @ 4GHz when I got the copper TRUE, and I was amazed at how well it held up...

I prefer active, because there'll be times when I want to 'ramp-things-up' a bit.
But it would be fun to dabble in passive, so thanks for the suggestion!

To be honest I would not risk plugging high powered fans to the MoBo header for anything other than the PWM signal and RPM sensor... I would not dare draw the power those things want from the MoBo header from fear of frying the damn thing. I think that with fans rated at anything higher than 1A (12W) you should be looking at powering them from the PSU... Remember that it's not just the 12W, those things will draw almost double their rated amperage at startup and that spike might be enough to send your MoBo crying to a dark corner.

Yeah I realise all that, think you may have missed the point of that Qn.
Plz lemme know if you'd like me to clarify

You found the non PWM from a US retailer? And they have them in stock?

Nope didn't say that, sorry for any confusion.

If you want to go all out; take a look at these: G1238B12BBZP-00
They are double the price of your GT 5400 rpm... But you get 261CFM and the highest static pressure that one can find on a 38mm thick fan. The drawback to these is that they work on their own PWM frequency (500Hz to 5KHz) whereas MoBos usually work on the 16KHz or 25KHz frequency; so for that fan you are definitely looking at building your own custom solution.

LOL, yeah I saw them mentioned in another thread, that's getting too hard-core for me.
Pretty sure I'd be very happy with 2x 25mm PWM 5400RPM Nidec's!

You could contact support for the MoBo manufacturer and ask them what PWM frequency they run their MoBos at. The general consensus is that PWM frequencies are either 16KHz or 25KHz.

Yeah def. going to get onto that now, will scour my manual 1st of course.
It sounds like it's only going to be 16Khz or 25Khz, thanks for the tips!

That could be a case of that particular fan model being calibrated to run at 16KHz and the MoBo header running at 25KHz... In that case, the lowest duty cycle from the MoBo header would not be low enough to make the fan slow down all the way down to 1000rpm.

Interesting, something I'll have to look-out for once I'm looking for 5400ropm Nidec's to pair with my MB, thanks!

 

[jalyst]

with PWM the fan itself is designed to not drop below 20% speed, that is why he isnt getting full control of it. the intel spec stuff that the fan makers are following somewhat, says that.
that way it always starts up, even if the fan controller croaks, and they do and have.

Well that's great then isn't it???
Because that means I should be able lower the 5400RPM PWM fans I'm considering to 1080RPM (20% speed)!

which is sort of strange, because one of the best purposes for PWM is keeping the voltage the SAME, and sending it in chopped up to the motor. by doing that many motors will start way slower than they will with voltage control.

??

if all you want is to have the RPM on the board, the power from the PSU and to cool them out when desired, a "SPDT Center off switch" and 2 resisters will easily give each fan 3 speeds, quiet , normal and blow me away. its cheap easy leaves most stuff the same, and i only ever need 3 speeds anyway. but i would have to tune those speeds carefully.
that would be basically ignoring the PWM and just voltage controlling, or tuning it even further so that any PWM could operate still at the resisted voltage/power. then you put the switches on a faceplate where there is shared space because a hard drive doesnt need all that space, and it stupid easy simple, nothing can burn out over time, and it gets it done.

Not sure I get this, why not just make use of the PWM capability of my MB?
Assuming it's 25Khz, and the fans I'm looking at are too, couldn't I expect to get them down to 20% (1080RPM)?
Of course I wouldn't power them from my PWM headers, both would be plugged directly into my PSU (X-650).

I don't know that it has to do with following specs... PWM fans are normally in the 16KHz or 25KHz frequencies... Motherboard manufacturers generally make their PWM controllers work on the 25KHz frequency (most common one)... The reason some fans will not slow down below 10% or 20% on certain motherboards is not due to the fact that they are not to spec, but to the fact that the motherboard manufacturers configure the headers to go to no less than said 10% or 20% duty cycle.

I see, thanks.

As for controlling PWM fans bypassing the PWM circuitry... True; you can do it by just voltage regulation... But why would you spend the extra money on PWM fans if you're just going to use them as regular 3-pin fans?

Agreed, I have the capacity for 2 PWM fans, just not 100% certain yet whether I can lower them to approx. 1000RPM.
The two remaining 3-pin headers can be used for: (1) up-to-6x AP-14's (2) up-to-4x AP-15's.

Another drawback of this approach is the fact that we are talking about high powered fans here... the weakest fan mentioned on here pulls 1.14 amps at 12 volts... that makes this fan a ~14W fan... The most powerful fan on this thread draws nearly 50W or power.... My guess is that if you go with resistors; you're going to need some mighty big ones so that they don't just burn off... and using those still leaves us with the fact that all that voltage that they are not sending to the fan is converted to heat that you're goint to be putting into the case... It kind of defeats the purpose of high powered fans if you're just going to be moving hot air around.

Just seems unnecessary to me based on the solution I have with my MB.
And doesn't seem like it'll meet my requirements for a discrete controller going forward.

see what they have done by making this complicated, it is many times more simple.
If you build your Own PWM , there is no requirement to talk to the curcuit on the fan, and , if you overbuild it it wont fail.

So that means if I plug PWM fans into the PWM header of my MB, I can bypass the circuit of the fan because there's already a PWM circuit on the MB?
Does the same apply for discrete controllers with one or two channels that use a PWM circuit?

OR
you can talk to the controller on the fan, with PWM pulses. but then you wont be able to control a non-pwm fan, so why would you build a controller that only works for some things?

But there are controllers out there with at least one channel that uses a PWM circuit.
Or are they all basically duds, why is that exactly?
If I had to build my own, I wouldn't want to ditch PWM just for the sake of compatibility between PWM and non-PWM fans.

I'd want at least 4-channels;
2x that are LVC and between them can accommodate up-to-10x AP-15/14's.
2x that are PWM and can supply the kind of juice required by the two 5400RPM's I'm looking at.

But I wouldn't build it, because as mentioned earlier in the thread.
I want much more functionality than just this from a discrete controller.
So it'd be far too difficult to engineer all that on my own!

the curcuit to do it, isnt that different, one mosfet difference, and full control at the controller, of ANY fan. not just a PWM fan, and every PWm fan too reguardless of thier communication needs.
So if you build a PWM controller as we have seen, you can put the one last component in and control anything, not just a fan that expects some specific pattern of pulses. even control your LEDs with it.

I see the sense in what you're trying to say...
But I have 'wants' from a controller that go beyond accommodating for PWM/non-PWM fans (besides, my MB's headers basically do that*), & my MB comes closest to catering for them as-of-now.

*although LT I'd prefer more 'channels', so I can have more fans on their own channel, providing rpm/voltage feedback or being independently controlled etc.

LOL...
This thread is turning into something so much complicated than it really needs to be. The OP wants a way or running high powered fans via PWM signal... To be quite honest; I'd just wire the PWM and RPM sensor wires to the MoBo header and feed the +12V and GND wires straight from the PSU...
That way you would have full control over the fans via BIOS/SW while still being able to run the high wattage monster fans shown here. No real need to put anything in between the MoBo and the Fan that way.... Yes; you'd be limited to whatever duty cycle the MoBo header is set up to, but any PWM control for free would still be better than now PWM control at all.

Yeah it sure did get way-off tangent there for a while, LOL
Agreed, for my intended use my MB is the best way to go, for now.
I just hope my MB can lower the 5400 Nidec's to approx 1000RPM.
No "biggy" if it can't, I'll just commence my search for a discrete controller.
LT I'm hoping to find one that gives me more power/flexibility than my MB.
Ideally it'd work in tandem w/my MB's EFI & Win7/Linux, but that's pipe-dream stuff at this time

All good advice. I use wirewound resistors on all my non-PWM fans and they are great on the less power-hungry fans, but on the Delta the resistors get HAWT, so I tend not to use them for extended periods of time. I used a 10W 5.6R to knock the Delta EHE down to approximately 5v, and it did a great job but also melted through the plastic encolsure that housed the resistor and toggle switch This is where PWM trumps other methods of speed control: highly efficient, and no extraneous heat.

Yeah I prefer the idea of having separate circuits for the different types of fans.
Got PWM fans, get the most out of them on PWM circuits (esp. for the power-hungry fans).
Got a bunch of 3-pins, get the most out of them on LVC circuits etc.

 

[jalyst]

Something else you need to think about regarding thermally throttling the fan is CPU core dependant vs water.
I haven't tried either myself, but "Think" it would be a much nicer and smoother ramp up/down using water.
...using water temperature...because of water's natural specific heat storage benefits, you would essentially give your ramp up/down a natural capacitor of sorts and smooth out that ramping problem.
You would simply have to set the water temperature to a correlating acceptable core temperature.
For this reason, I think ultimately you'll need to look beyond motherboard control if you are very particular about noise.
I only say this because my work computer has a PWM fan on it, and I personally find that ramp/up down every time I do something intensive to be a bit annoying. Rather than just spin up partially, it' pretty much pegs high rpm then back down to low in very short order. Perhaps this could be smoothed out with more precision PWM settings (Work computer locked down by our big brother IT folks), but I just haven't been happy myself at least with that rig. It's definitely better than listening to a fan full bore all the time, but the ramping up and down so quickly is a bit disappointing.
The solution to this will of coarse require a fan controller, one that has the ability to control fans via some sort of temperature probe you can attach or insert into the fluid stream somehow.

This is an interesting suggestion, thanks.
Could the water become an unreliable source, if in time too much of it evaporates?

As pointed-out further down in the thread...
This jagged ramping up/down may simply be due to poor PWM config, or a fan with a top RPM that's a bit too high for the 10% increments.
But it still sounds like an interesting 'supplement' to the usual core measurement that's done, so thanks!

As far as you're discussion on PWM vs voltage control of the fans...either is fine. The only real advantage I see with 4 pin PWM fans is "Some" have a little lower RPM floor, but both if designed optimally are equal on the noise front. If anything, my preference sways toward voltage regulation, it's less testy and has shown to have smoother sound quality in some fans.

Isn't PWM also much more efficient?
Interesting point RE 'smoother' sound quality for linear voltage control, quite surprising.

I haven't decided quite what I'm going to try. I have my son's pc that's fairly minimal, it just has a single MCR320 mounted in the top with one case fan. In his rig, I may go ahead and try a set of four Arctic Cooling F12s and give PWM a try, I just don't think it's quite the control I want. I've got a single loop though, so it would probably work ok with some careful PWM settings. I just think controlling by water temp would be much more ideal for water cooling.

I love how you say a single MCR320 is fairly minimal, in reality it's not compared to the avg. Joe's PC!

Now if someone could figure out a way of taking the PWM signal from the MoBo headers and smoothing it to a nice curve; then we could probably build a fully internal controller that takes the MoBo's PWM signal, makes it nice and smooth and then feeds that signal to the fan.

That would be so sweet,
A device that could take the signal from both headers, pass it to 2 PWM circuits of a controller, but in a more 'smoothed-out' fashion

Yeah, the more I think about it...the more I'm personally drawn to something like the mcubed Tbalancer bigNG/mini or AC poweradjust controller. Not quite sure which is better, but they both seem to have tools to tweak things via software.

Yeah, alas they're best options I've seen so far

The problem with running this off a motherboard is <SNIP> b) motherboard manufactuerers set up the PWM signal in 10% increments from 0-100%.
<SNIP>
B) is not so simple. While yes you can effectively jump in 10% increments, this works for lower spead fans because you incremently just 10% of the rated fan speed. 1000RPM fan jump 100 RPMS at a time, 2500 RPMS jump 250 RPMs. But a 6000 RPM fan jumps ~600+/- RPMs which is akin to switching a Hairdryer with 5 different speed settings up and down till it drives you insane.

This is one of the most critical points made so far...
It may mean having these 5400RPM's coming-off my MB's PWM controller is impractical

If your looking for a quiet system, then just go with a slower fan.

I may have to consider that if this 10% increment you speak-of holds true for my MB.
Especially if my MB is not able to lower these fans to 20% of their top-speed (1080RPM).
I guess I should 'bite-the-bullet', order, & find-out how bearable/annoying these 10% increments are on a 5400 PWM Nidec.
Maybe my MB's special... doubtful!

If your looking for maximum cooling, then you will need a high speed fan. IF you do this, you need to make a PWM Controller. There are no official manufacturers (at least that I could find 6-8 Months ago) and if they are, they are surely more than $100.

Assuming the 10% increments apply to me (they prolly do), & assuming they're just bearable (the fans will usually be at their lowest RPM).
Do you still think it's imperative to build or buy a discrete controller?

I don't mind paying more than $100 if a controller with 2x decent PWM channels exists, but it has to meet all my other basic requirements
Also I don't think anyone's explained (apologies if they have) what the problem is with pre-builts out there, that have several LVC & one PWM circuit.

b) If you try to just undervolt a high speed PWM fan, you will be plauged with Clicking and whining and you will regret it, not to mention you will loose performance.

Thanks for pointing this out, this conflicts with what one of the other posters was saying earlier in the thread.

c) The Motherboard option should only be used if you are going with a fan under 2000 RPMs. Even when I used a 2500 RPM fan, it was annoying.

Hmm, you're probably right, still tempted to try one though
Especially given I still don't know the specifics of the PWM control implemented for my MB, will find out ASAP!

Now, to the topic at hand. While I don't think a dedicated controller is imperative, as complicated as this is getting, I think it's probably a good idea. It will a) keep your wiring cleaner, b) allow you to see the RPM and fine-tune fans as you want and c) save you a lot of headache separating the fan's controller & voltage wires, running them all over the place to your headers (see a)).

There's not too many controllers that bring anything equivalent to a MB+BIOS/SW wrt (b) (at least for my needs).
But I'm sure once everything's set-up & there's wires everywhere, that it'd be nice to have one for (c)!

And I like controllers because of the larger no. of channels (headers), many with the ability to provide more juice than a typical MB header.
But those multitude of channels aren't much good if RPM/voltage can't be captured, & speed can't be adjusted via temp.
If this can't be done in concert with a MB/BIOS/OS, then all this should at-least be integrated into the controller itself.
e.g.
-LCD for RPM/voltage/temp feedback...
-multiple & accurate temp probes...
-12 channels (unlikely) for 'each' fan, so that rpm/voltage can be displayed independently, & fans can be auto-adjusted independently.
-ability set temp thresholds to manage fans/fan-group speeds automatically.

I think there's a few out there that come close to these requirements...
Alas I don't think there's any w/2x PWM circuits that have the right qualities (not sure what I'm looking for), & enough juice to drive 5400RPM fans.

What I don't get yet - why are you getting PWM fans? If you're getting Scythe GTs, they come in three-pin models. As a matter of fact, Sidewinder only has three pin models.

Mainly because of the qualities of PWM driven fans, greater efficiency/control etc.
But if it's too tricky to obtain the 5400RPOM 25mm Nidec's from Japan, then I may well end-up ordering 1x 5400RPM Scythe GT.

For the reasons above, I'd recommend giving up a bay to the most attractive fan controller to you and using voltage-controlled fans. If you're going for under-stated and don't like flashy, bright lights, the Lamptron Touch -reviewed here- is even completely blank (except for the power indicator in the lower right corner) until you power it up. Just make sure you get quality, with good wattage ratings. Watch out for those that look good but are only rated 7W/channel. You seem attuned enough to realize that already though.

I'm not in any rush to move to a discrete controller, if in say 6mth there isn't one that can accommodate "all" my requirements.
Then I may well end-up getting one that does 3-pin fans only...
I'd rather not give up a bay if I can afford it, which is partly why I'd prefer a controller that interfaces with the MB/BIOS/OS.

EDIT - If you really wanted to avoid a bay controller, still get the voltage-controlled fans and you could go with a GUI controlled T-balancer bigNG that Martin mentioned.

Yeah I think this satisfies more of my requirements (I need to re-list them!) than anything else I've seen yet.
I doubt it'll get much better any time soon , but I'll be happy with my MB for some time yet.

PWM seems to be the biggest obstacle. If you can avoid that altogether, your options expand quite a bit.
EDIT II - Cleaned up your thread for you.

Yeah I'm starting to think PWM is causing more complication than it's worth.
Just need to determine a few things & if they don't pan out, then I may move toward 3-pin 5400RPM 25mm Nidec's (GT) instead.
Thanks for that!

 

[GIXXERGUY6]

I want to strap some of those fans to my daughter and see if I can get her to hover around the house! JESUS!!!!!

 

[hokiealumnus]

Re: PWM efficiency - Without hard evidence, I can't say for certain; but if you think logically, they are both pulling the same amperage at a given speed, they're just controlled in a different manner. I couldn't imagine any gain would be significant even if there was one.

Frankly, I don't think your silver bullet exists. The only thing even close is the T-Balancer bigNG mentioned previously.

I'm also somewhat concerned about your fan choices. You say this is going into an HTPC which will occasionally be used for gaming, but mostly will be at stock or under-clocked. Presumably, due to the concern with controlling them, this will be in a room that you want it quiet most of the time. Largely, high amperage fans are not quiet, no matter the speed. If they're on, you can hear them. They're quieter and for some are tolerable.

I have three deltas and two panaflos on radiators in my main rig. They're all high power (3,400 and 2,750 RPM, respectively), yet are a couple thousand RPM under what you're considering. I have them running about 5V and can still hear it across the room. It's nothing compared to when they're turned up for sure, but they let you know they exist. I'm fine with that b/c the computer is in our office. If it were in my living room, it would be a problem.

I fear you are overestimating how much noise you will reduce when controlling a 5400RPM fan. You can control it, but you will definitely still hear them. Motors that strong still make noise even if running at very low speeds.

 

[PigeonStriker]

Hi, haven't been to this forum in 7 years, but I felt compelled to reply to this thread.

I read the front page article on the 555 timer controller, and the thread behind it, and this thread. I didn't know what the 4th wire is for, but now I do. Good stuff.

Anyways, a while back I made my own fan controller to PWM control a bunch of 3 and 2-pin fans. After fixing a few bugs, it's been working just fine for years. I learned one lesson:
-I had 3 fans in my case and (naturally) wanted one fan per channel. Since the microcontroller I had lying around only had 2 hardware PWM channels, I ended up implementing PWM in software. The final result is only 3-5 usable speed settings depending on the specific fan's characteristics. Needless to say, this is less than ideal.

Commenting on some of the rest of the things in this thread...
-'Smoothing out' a PWM control signal is a bit hard since it's a square wave. Simple linear analog filtering will just round out the corners. A microcontroller could easily do it though, and probably with some thinking an analog circuit could do it too.

-Your wishlist... is quite ambitious. It's not impossible though, but it would take some time. To do all of your wants, you'll need some kind of processor. On the bright side, any half decent 8-bit microcontroller could do it.

-Interfacing with the BIOS is pretty much impossible (I think?). Interfacing with the OS / computer is doable via USB / serial / parallel, but your OS would need to be running a program to communicate with the fan controller.

-Reading a fan's RPM should be doable.

-Reading temperature is easy depending what kind of signal the temperature sensor gives.

-Automatic closed-loop control of the fans is doable given temperature input and PWM output.

-Automatic closed-loop control of the fan to a specific rpm range should be doable too.

-Manual tweaking of PWM frequency / duty cycle to avoid unpleasant noise would be easy too.

-User input via buttons / switches / dials, and output to a 7-segment display / LCD is easy.

-8-12 channels should be fine, because microcontrollers generally don't have a lot of PWM outputs built-in (re: lesson learned above), so some kind of dedicated PWM chip would be needed, and the microcontroller would just send it commands.
->Actually after some quick searching, there are microcontrollers with enough PWM outputs, but it might be hard to find one with a convenient package.

-Staggered startup / limiting inrush current / high startup duty cycle -> super low running duty cycle is all easily doable.

-Driving high-powered fans just depends what components are used for the output circuit. For my fan controller, the output was just a power nmos with the microcontroller's output connected to the gate and a few other things for safety.

Price estimate... $50-100 assuming nothing gets broken in assembly and depending on the quality of components, the amount of protection stuff included, and the number of output channels.

 

[jalyst]

@PigeonStriker

Apologies for my delayed response, this has been lowered dramatically on my ToDo list.
Too many other more pressing concerns....

It sounds like you're saying all of my needs/wants are do-able only if I'm prepared to fab. everything.
Alas short of a detailed "plan of approach"*, there's no way I have the time/skills to do it all myself

I need something pre-built which comes close to satisfying everything, & with some "franken-modding" is perfect.

Any ideas/suggestions?
I won't be able to focus on all this for some time yet, but just trying to keep the torch burning.

*maybe there's an avid arduino user out there that's outlined it all?

 

[jalyst]

@All prior posters in this thread...
I don't suppose you could add replies to my responses to you?
#27, #28, & #29

Re: PWM efficiency - Without hard evidence, I can't say for certain; but if you think logically, they are both pulling the same amperage at a given speed, they're just controlled in a different manner. I couldn't imagine any gain would be significant even if there was one.

Yeah I've ordered 1x AP-31, along with 4x AP-15.
I'll prolly still try & obtain 1x PWM equivalent of the AP-31.
If only to compare the two LT...

Already have 2x AP-14, I'll prolly order 2->4 more if the fans that come with my case are crap.

Frankly, I don't think your silver bullet exists. The only thing even close is the T-Balancer bigNG mentioned previously.

This is what I'm thinking of doing, see it and subsequent posts.
I'll make do with my MB initially though, just to learn what I can & can't do!

I'm also somewhat concerned about your fan choices. You say this is going into an HTPC which will occasionally be used for gaming, but mostly will be at stock or under-clocked. Presumably, due to the concern with controlling them, this will be in a room that you want it quiet most of the time. Largely, high amperage fans are not quiet, no matter the speed. If they're on, you can hear them. They're quieter and for some are tolerable.
I have three deltas and two panaflos on radiators in my main rig. They're all high power (3,400 and 2,750 RPM, respectively), yet are a couple thousand RPM under what you're considering. I have them running about 5V and can still hear it across the room. It's nothing compared to when they're turned up for sure, but they let you know they exist. I'm fine with that b/c the computer is in our office. If it were in my living room, it would be a problem.

I realise all this don't worry
The majority of the time I'll have 2-4 AP-15 on the rad + maybe some shrouds.
If I like the way the AP-31 (or it's PWM cousin) works, I may order a few more for the times I want to 'go crazy' with OC's & just for fun

I fear you are overestimating how much noise you will reduce when controlling a 5400RPM fan. You can control it, but you will definitely still hear them. Motors that strong still make noise even if running at very low speeds.

Yeah based on what I've heard from people in this thread...
I now realise I prolly won't have 'fine grain' control via the MB*, & may not even be able to lower it to 1080RPM**.
And even if I can lower it that much, I realise it won't necessarily be the same as an AP-15 lowered to that RPM.

As for the non-PWM equivalent...
I won't be able control it's speed at all via the MB, all the more reason to start thinking about discrete controller options!
Yikes 4am....

Thank again so much everyone & good-night!
*yet to be substantiated
**also yet to be substantiated

 

[hokiealumnus]

I did, I'm just sorry to say I don't have any more to add. It seems like the choices are out there for you, you just have to make a decision at this point. Sorry, I don't have enough electronics knowledge in this field to assist with any frankenmodding.

 

[jalyst]

@hokiealumnus
Thank-you for taking the time to respond, it is very much appreciated.

Which of these "choices" you speak of, stand out to you as my best option?
Could you possibly address just these two points, where you can?

This is what I'm thinking of doing, see it and subsequent posts.
I'll make do with my MB initially though, just to learn what I can & can't do!

Yeah based on what I've heard from people in this thread...
I now realise I prolly won't have 'fine grain' control via the MB*, & may not even be able to lower it to 1080RPM**.
And even if I can lower it that much, I realise it won't necessarily be the same as an AP-15 lowered to that RPM.

As for the non-PWM equivalent...
I won't be able control it's speed at all via the MB, all the more reason to start thinking about discrete controller options!

*yet to be substantiated any idea?
**also yet to be substantiated ditto?

 

[hokiealumnus]

In your situation, if you want to mix PWM and non-PWM fans, here's what I'd use:

• bigNG + miniNG for the voltage-controlled fans
• MOLEX + PWM-control & RPM wires to the MB combination for the PWM fans

That's the only thing (seemingly) that's going to give you easily attainable, automated control of both types of fans with off-the-shelf components.

You'll just have to keep your fingers crossed the MB controls the PWM fans well enough to tame them. IF it's possible (I know too little about the xNG to say yes or no), I'd go all voltage-controlled fans and daisy-chain whatever bigNG/miniNG combination is necessary to control all of the fans.

 

[PigeonStriker]

@PigeonStriker
It sounds like you're saying all of my needs/wants are do-able only if I'm prepared to fab. everything.

Yes. I'm not on the up-and-up for the off-shelf-hardware out there, but I think meeting every last thing on your wishlist perfectly is too niche that anyone would make a retail product for it. But if someone does, I'd buy it.

Alas short of a detailed "plan of approach"*, there's no way I have the time/skills to do it all myself

If this project ever becomes a super high priority for you, let me know.

 

[jalyst]

In your situation, if you want to mix PWM and non-PWM fans, here's what I'd use:

• bigNG + miniNG for the voltage-controlled fans
• MOLEX + PWM-control & RPM wires to the MB combination for the PWM fans

That's the only thing (seemingly) that's going to give you easily attainable, automated control of both types of fans with off-the-shelf components.

You'll just have to keep your fingers crossed the MB controls the PWM fans well enough to tame them. IF it's possible (I know too little about the xNG to say yes or no), I'd go all voltage-controlled fans and daisy-chain whatever bigNG/miniNG combination is necessary to control all of the fans.

Thanks, this is the way I've been thinking of going for some time (1st mentioned in my other thread).
Was just curious to know whether you think it's solid, or if there's better way to go.
Thanks for affirming that you also feel it's the best, of a limited set of options....

 

[bing]

Yes. I'm not on the up-and-up for the off-shelf-hardware out there, but I think meeting every last thing on your wishlist perfectly is too niche that anyone would make a retail product for it. But if someone does, I'd buy it.

Jalyst, this is an excellent response and also is the real answer for your quest.

What you dream of is a very specialized custom made circuit that I believe that it can not be made by just some sketching the circuit at the paper then build it in one or two days even for an senior EE. It is a really a serious electronic project.

I guess you just have to live with the current options that already mentioned previously.
Unless you can find an EE to design and build it from the scratch just for you.