-
Save 15% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
Cooling What ever happened to Coolchip?
- Thread starter hat1324
- Start date
As patents are VERY specific, I'm pretty sure it's just the way its implemented that's patented, not the concept of reverse blowing itself.
For example, with the GTX 580 that blows that stuff off, the controller for that is probably sitting on the card itself, while the enermax fans could be the first ones to incorporate a small circuit that takes care of this thing. And that might be what they pantented: "Method for Integration of Electronic Circuit into the Frame of a Fan for use in Personal Computers implementing automated Spin Direction Reversal during Startup for Cleaning of the Fan Chassis and Dust Filters"
So mainboard manufacturers would actually be able to design their boards to implement this sort of thing to work with any fan in existence. That makes it even more curious why it's not yet widely used.
Well, H-Bridges are not terribly space efficient (the few ICs I've found that will do 12W are the size of old school DIP packages) and with the proliferation of Mini-ITX and smaller boards, it doesn't seem like a good value proposition. You could fit in another RGB LED driver or 3 in that space. Also I'm not sure if your average Super IO chip can change PWM frequencies.
Although I'm not 100% how a fan does it's management of the motor I'd assume it energizes 2, 4 or 8 electromagnets that pull the fan hub.
So in essence I think instead of the controller going: 1-2-3-4-1-2-3-4 it goes 4-3-2-1-4-3-2-1. This might be overly simplified, but wouldn't that be the easiest way to do it?
Well...the average fan's onboard controller is not a particularly smart piece of circuitry, just transistors, resistors and a Hall sensor IC:
http://pcbheaven.com/wikipages/How_PC_Fans_Work/
But notice what's in the diagram for the 3-wire fan? It's only got 2 transistors and you need 4 to run a fan both ways (making the H in H-bridge).
You are correct in that if there is enough PCB space for some extra components, a fan is perfectly capable of doing such a thing and I suspect that is MSI and Enermax's approach. However, this places a limitation on control unless the fan is PWM and both the fan and the Super IO are in agreement on what constitutes forward and backwards.
So actually, I need to make a addendum to my original statement...
For this approach to work, the fan itself needs to be 100% "dumb", just 2 wires directly attached to the motor. You can farm the H-bridge onto the fan itself probably with a control element but my point about the Super IO's limitations still stand.
Boy, did we get off topic. Apologies.
The actual running of a fan in reverse is the easy part (reverse current for a brushed DC fan, coil energisation in reversed sequence for brushless DC). The tricky part is having an external device (i.e. the motherboard, or a dedicated fan controller) tell the fan to run in reverse:
- Add an extra 5th pin to the connector for 'reverse' (proprietary, may have interference issues that prevent backward compatibility depending on board layout and nearby components)
- Use an existing pin as a reverse indicator. RPM pin could be used as an output, but could kill a fan not designed to handle voltage on that pin. Could do some weird thing with the PWM pin (e.g. negative voltage) but again could damage fans not designed to tolerate that
- Reverse 12V and GND on the motherboard. Will work right away with brushed DC fans, *shouldn't* damage brushless DC fans (a rectifier should be incorporated as a best practice) but would require a 'reversible' fan that can recognise a reversed voltage as a desire for reversed flow.
- Add an extra 5th pin to the connector for 'reverse' (proprietary, may have interference issues that prevent backward compatibility depending on board layout and nearby components)
- Use an existing pin as a reverse indicator. RPM pin could be used as an output, but could kill a fan not designed to handle voltage on that pin. Could do some weird thing with the PWM pin (e.g. negative voltage) but again could damage fans not designed to tolerate that
- Reverse 12V and GND on the motherboard. Will work right away with brushed DC fans, *shouldn't* damage brushless DC fans (a rectifier should be incorporated as a best practice) but would require a 'reversible' fan that can recognise a reversed voltage as a desire for reversed flow.
So I dug up the datasheet for the Nutovon NCT6776F Super IO chip which is very common on Asrock and Asus boards and it looks like the PWM configuration is actually like most other micro-controllers. Page 88 lists a bank of registers for configuring clock division and frequency, meaning all that's left to do is to pair the motor with the right controller. And it looks like if you have a 3 phase BLDC like on a Noctua iPPC you are in luck: looks the TI DRV8313 will do it all (although TIL you only need a 1/2 H-Bridge to drive a BLDC).
Most BLDC motors have symmetric windings so current can go either way, but as you say, it's the controllers that manufacturers pair them with to make a fan that are problematic. MOSFETs don't particularly like reversed current.
Most BLDC motors have symmetric windings so current can go either way, but as you say, it's the controllers that manufacturers pair them with to make a fan that are problematic. MOSFETs don't particularly like reversed current.
Coolchip continue to tease with a new design 'coming soon'. The packaging and colour variants make it look like Coolchip are now trying to produce a product line in-house rather than licence it to existing HSF manufacturers.
This just popped up on the S4 Mini thread: http://www.cowcotland.com/news/53501/coolchip-1u-low-profile-kinetic-cpu-cooler-retour-retour.html
Apparently this product is going to materialise soon: http://coolchiptechnologies.com/cooling-products/low-profile/
@confusis maybe request samples for review ? They are promoting that on the link above.
@confusis maybe request samples for review ? They are promoting that on the link above.
I don't have any Intel platform hardware to test it on- the thermal test system is AMD based for thermal and compatibility reasons
The Coolchip site shows some more details on the 'round' heatsink design:
In which you can see some copper below the outer fin ring, with the mounting plate sandwiched between. Could be just a copper shim, or could be a heatpipe 'wrapped' around (there's a prominent gap, which would not be necessary for just a shim).
The site also shows a second design:
Which foregoes the copper parts and folded fins for a single-piece milled aluminium outer heatsink with milled fins, that covers the entire keep-out area.
The rotating heatsink/impeller appears to be identical for both designs.
how much of the cooling of this new design is down to the impeller conducting heat and how much is due to the fixed heat sink .. It seems the original spirit of the cooler hasn't passed onto these iterations..
When (if) these heatsinks are released to production, it would be pretty easy to tell how much cooling contribution the rotating heatsink makes, either be replacing it with a plastic impeller, or just adding a washer to increase the air gap beyond the tiny distance necessary for effective conduction.
Similar threads
