Baffled by Arctic Cooling Patented PWM Sharing Technology

[Syzygies]

I now have in hand some ARCTIC F PWM fans, with their patented PWM Sharing Technology. I can described exactly how they are wired, and how they are to be used. I do not understand what aspect of this design could be covered by a patent.

These fans are advertised as being able to be daisy-chained, with multiple fans controlled by a single motherboard PWM fan header. Mention is made of a PWM amplifier, which must be part of the circuit on the fan itself. All PWM fans require a circuit, to process the PWM signal. This could be a more exotic circuit than usual.

Four wires come out of the fan, as usual for a PWM fan. One of the wires, the yellow RPM monitoring wire, goes to a 3-pin fan plug. The other three wires go to a short four-wire PWM pass-through cable. These pass-through wires are thicker, able to handle the amperage of multiple fans.

Using a multimeter, I checked all possible connectivities. One can access the solder points where the wires attach to the fan chip, and the male and female fan plugs; all possible connections suspected by examining the wires is there.

One daisy chains the short PWM pass-through cables. This has the effect of splitting the power and pwm signals to all attached fans. One terminates the daisy-chain by either plugging in a conventional PWM fan at the end, or plugging in one of the 3-pin fan plugs carrying an RPM signal. Either way, exactly one fan reports its RPM back to the motherboard, which is correct.

So far, this is a completely correct and obvious solution to the exercise of wiring multiple PWM fans to a single PWM fan header, the difference being that the required custom cabling is sold connected to each fan. In comparison, there are PWM Y-splitter cable manufacturers ignorant enough to botch the necessary wiring pattern, by returning both RPM signals rather than one RPM signal. But many Y-splitter cable manufacturers do get this right.

Taking this a step further, the akasa cable is a 3-way splitter with separate power, to avoid overloading the power lines from the motherboard fan header. They correctly return one RPM signal. Others make a similar cable, and various forums propose that one should simply make this cable, once one grasps the wiring issues.

So what's the role of Arctic Cooling's PWM amplifier? A reasonable concern, that can only be answered experimentally, is how many ways a PWM signal can be shared. Wire length is not an issue here, but the PWM signal is faint, and each client fan puts a load on the PWM wire. Op amps are designed to replicate a signal without loading the source; Arctic Cooling could be taking extra pains to avoid loading the PWM signal line, such as by using an op amp. This would be a minor additional expense and a routine electronics exercise. I'm not an electrical engineer; if I understand this, many thousands of people do.

Before seeing a fan, I had somehow imagined that instead, they were regenerating a clean, full strength PWM signal at each step of the daisy chain. This can't possibly be the case, given the wiring pattern I describe.

So what am I missing, that they managed to patent here?

ARCTIC F PWM fan (PWM Sharing Technology)
akasa PWM Splitter - Smart Fan Cable

 

[alaricljs]

Try finding the actual patent.. for that matter, did they mention who owns said patent?

 

[Syzygies]

http://www.google.com/patents

I struck out, although I did recognize a few Cooler Master patents.

 

[alaricljs]

The interesting thing is that older fans on Arctic's site actually reference their own patent numbers. Makes one wonder.

 

[ncalipari]

a patent may take yeas to be registered, especially if the thing they are trying to patent is obvious

 

[alaricljs]

And it's illegal to claim a patent when it hasn't been successfully registered, hence 'patent pending'

 

[Syzygies]

This picture may be easier to follow than my description. I've plugged the RPM 3-pin cable (yellow wire) into the female PWM pass-through connector, and peeled off a sticker to expose the PWM fan circuit on the fan itself.

In this configuration, the fan could be used alone as a PWM fan, by plugging the male PWM pass-through connector into a motherboard PWM fan header.

In this configuration, the fan could instead be used as the last PWM fan in a daisy chain, by plugging the male PWM pass-through connector into the female PWM pass-through connector on the next Arctic F12 fan.

Either way, the motherboard will see the RPM signal for this fan.

In typical use, Arctic envisions that one would instead plug the CPU cooler PWM fan cable into the last female PWM pass-through connector.

 

[aigomorla]

sunbeam has a new controller thats 3.5inches wide bay device.
http://www.performance-pcs.com/catal...ducts_id=30057

It has 4 fan headers on it, because one of the headers goes directly into your pwm header on the board.

It has 3 dials with buttons on the bottom so u can control what fans get pwm, and what fans get manual.
Each channel has a total power output of 30W, or 30/12 = 2.5amps

If your going more then 2 fans, i highly recommend u go the sunbeam route, and not daisy chain off your mother board header.

Unless your board is a high enthusiast class, those headers cant support high amperage, and its a good way to kill your header quick.

 

[Zap]

Unless your board is a high enthusiast class, those headers cant support high amperage, and its a good way to kill your header quick.

The Arctic Cooling F12 fans are rated at 0.15A, not exactly drawing monster amounts of power. They are intended to be "quiet" fans with a 300-1350RPM range. Not exactly your cup 'o tea, aigomorla.

FWIW I've run two of them linked together for a while as push/pull on a Zalman CNPS10X Flex and they worked reasonably well.

 

[aigomorla]

The Arctic Cooling F12 fans are rated at 0.15A, not exactly drawing monster amounts of power. They are intended to be "quiet" fans with a 300-1350RPM range. Not exactly your cup 'o tea, aigomorla.

lol... true... but assuming they chained up another fan to it like a 4300rpm Nidac..

or is the chain up antec fan only?

 

[Syzygies]

If your going more then 2 fans, i highly recommend u go the sunbeam route, and not daisy chain off your mother board header.

Unless your board is a high enthusiast class, those headers can't support high amperage, and its a good way to kill your header quick.

I had the Sunbeam 6 Fan Controller in my new build for a bit, described in the thread Sunbeam Rheosmart 6 Fan Controller. It works in principal, but kicks on at around 10 volts. I took it back out.

The akasa PWM 30-way splitter fan cable draws separate power, getting around your objection to splitting the motherboard fan header. The issue, the entire issue, is how far one can split the PWM signal. These Arctic fans make special claims here, but I'm having a hard time deciding what's so special.

My advice to anyone building Sandy Bridge would be to run two cooler fans, using a correctly wired PWM fan Y-splitter cable off the motherboard CPU fan header. Set the BIOS for variable control. Run all the case fans at 5 volts. The bump to 12 volts makes scant difference overclocking, if one has adequate case ventilation in the first place.

This thread is more of an academic debate on what's up with Arctic Cooling fans.

 

[Syzygies]

I returned all but one fan to http://www.ultimatepccooling.com (highly recommended; great support).

There's a more fundamental problem with this Arctic design. By integrating the custom cabling into the fan, they're forcing a particular implementation of the PWM wiring pattern on users. All connections emanate from the very short pass-through cables, forcing a "star" wiring pattern, which doesn't fit my case. Not what I was thinking, ordering fans that could daisy-chain. They don't daisy-chain, they "star".

Using standard 4-wire PWM cables, leaving the custom wiring to the user, would have been much more flexible. Scythe throws in optional 4-pin power pass through cables with their fans; Arctic could throw in optional 4-pin PWM pass through cables with their fans.

Arctic could have modeled such a pass-through cable after correctly wired PWM Y-splitters, with the fatter gauge pass-through wiring carrying the RPM signal, and the thinner gauge branch wiring not carrying the RPM signal. Then one wouldn't have the separate yellow wire dangling off the fan like loose hair; it would arrive sleeved with the other wires. Either at the branch PWM header, to be ignored, or at the last pass-through PWM header, to be read.

Skip the different wire gauges (not necessary because the currents are miniscule for these gauges) and one simply has a Y-splitter PWM cable, marked to indicate the branch that carries the RPM signal.

Then, users would be free to construct any correct wiring solution, such as addressing limitations of motherboard header amperage by using the akasa cable:

akasa PWM Splitter - Smart Fan Cable

What is potentially novel about these fans are better specs for the PWM circuits on the fans, buffering the PWM signal to avoid overloading that signal with multiple fans. It would be great if more manufacturers made claims as to how many ways their circuits can successfully split such a signal.

I nevertheless stand by my original advice: One doesn't need any of this. Run two bog-standard PWM fans off the cpu fan header, using an ordinary PWM splitter cable. Run the case fans at 5 volts. One saves tremendously in complexity, and is giving up at most 2 C core temps on any Sandy Bridge overclock.

 

[alaricljs]

That star pattern is probably part of the entire method of 'PWM amplification'

 

[Syzygies]

That star pattern is probably part of the entire method of 'PWM amplification'

I don't see a smiley. Do you think wire length and network topology is actually an issue, for distributing the PWM signal? The frequency is just above audible, which is a glacial time scale for electronics.

 

[deimos3428]

This picture may be easier to follow than my description. I've plugged the RPM 3-pin cable (yellow wire) into the female PWM pass-through connector, and peeled off a sticker to expose the PWM fan circuit on the fan itself.

In this configuration, the fan could be used alone as a PWM fan, by plugging the male PWM pass-through connector into a motherboard PWM fan header.

In this configuration, the fan could instead be used as the last PWM fan in a daisy chain, by plugging the male PWM pass-through connector into the female PWM pass-through connector on the next Arctic F12 fan.

Either way, the motherboard will see the RPM signal for this fan.

In typical use, Arctic envisions that one would instead plug the CPU cooler PWM fan cable into the last female PWM pass-through connector.

I don't understand why you're plugging the 3-pin male (yellow wire) into the female PWM pass-through.

I was under the impression that it was to go on a separate 3-pin motherboard header to provide RPM info, and the female PWM pass-through connector was only used if you were daisy-chaining additional fans. Doesn't matter?

 

[alaricljs]

One fan can be cabled this way to pass its RPM to the 4pin header, any additional fans that you want RPM from would need to be cabled to other headers.

As for topology, yes topology matters seeing as how this is NOT a network it is a signal that gets attenuated by the electronics on each fan and by wire length.

 

[Syzygies]

As for topology, yes topology matters seeing as how this is NOT a network it is a signal that gets attenuated by the electronics on each fan and by wire length.

This is of course an empirical question. But USB cables can be 16', I'd be stunned if any of us has a big enough case to worry about PWM fan cable wire length.

As for topology, I did wonder if "termination" could play a role in Arctic's thinking. As in the old days, when one terminated SCSI hard drive cables, with electronics to soften signal bounce so there aren't waves traveling back and forth, reinforcing and interfering with each other.

So how could a "PWM amplifier" have an effect on the incoming signal line? It's a problem that sounds a bit like motherboard load line calibration. Motherboards have gotten a lot better at this, notice swings in current needed and respond so fast there's no voltage spiking. But the pea-sized circuit on an Arctic fan can't be doing much, I'm guessing all it can manage to do is be a very mild load.

 

[alaricljs]

Do know what empirical means?

You're comparing PWM across a ~20ga wire with an entire protocol/hardware design engineered to do its job (USB). Then you start pondering another protocol/hardware design (SCSI) and the possibility that it's similar.

PWM is simply pulses of varying duration sent down an unshielded wire. A PWM amplifier (or whatever it is they're really doing) could have a huge effect considering it's not forwarding the PWM signal, according to you and the pictures I see there is only 1 PWM wire on their controller, no pair of wires (in and out). I don't even think it's possible to amplify a signal on the same wire you're receiving the signal on. They are either using the wrong terminology to dumb it down or the marketroids are totally screwing it up.

 

[Syzygies]

Yeah, I'm a mathematician. We make broader analogies than most people are comfortable with hearing. Sometimes it works.

Clearly that one wire in (PWM signal) is sharing ground with power in. That's why there are four wires.

I was wondering about a termination mechanism analogous to SCSI, because I consider a 1 in 1000 chance of missing an idea to lack of imagination unacceptable. If I had to bet money, I'm still going with an op amp, to minimize load on the PWM line. Of course, an op amp wouldn't do squat without a ground line, too. Thankfully, there is one.

If you're curious as to the actual PWM specification, it's all here:

4-Wire Pulse Width Modulation (PWM) Controlled Fans

 

[alaricljs]

By 'in' and 'out' I was not referring to the necessary ground but to the idea of receiving the signal (via the 'in') and then fiddling with it (amplify it in this case), then sending it on (via the 'out'). This is what I would expect given that they claim: 'The patented PST (PWM Sharing Technology) is forwarding the PWM signal to all other fans inside the system.' You can't forward a signal on a line that already has the signal on it.

 

[Syzygies]

This is what I would expect given that they claim: 'The patented PST (PWM Sharing Technology) is forwarding the PWM signal to all other fans inside the system.'

Bingo! Thank you, thank you, thank you.

I fell for the same description. That's why I was stunned when I saw the actual wiring pattern.

Now, if they can forward a signal on a line that already has a signal on it, and we're both convinced that's impossible, then they're starting to meet a standard of non-obviousness, and deserve a patent. If they're merely minimizing load on the line, that's trivial, and they shouldn't be awarded a patent.

A nice mental exercise is to suppose the idea that something "impossible" is true, and try to then explain it. How would you then have your company try to do something similar, if your job depended on it?

Here, a PWM pulse is being generated somewhere on the PWM line (the motherboard header) and travels to ground through each fan PWM sensing circuit. If one of these fans on the line decided to also generate a pulse, the rest of the fans would see the sum (or the logical "or") of these signals.

So how do you sing only when someone else is already singing? If you're very quick, you could go quiet briefly to listen, every now and then, and the slower parties would never notice what's happening.

This gets much harder with more than one Arctic fan on the line, and I don't believe for an instant they're doing this. I believe they're simply minimizing load on the line. But nothing is actually "impossible". This is actually a standing cautionary joke in mathematics, the idea of someone asserting "this can't be true because I can't imagine it." We teach our students not to accept such assertions as proofs.