Or alternatively.. isn't it just easier to use a 4-wire PC fan and just drive the PWM directly to the (usually) blue wire? Problem solved ??

I think what would get a lot of interest here is if you offered not just the boards, but a plug and play add-on for those who are intimidated by the thought of soldering anything (though they would still need to solder headers).  I think a lot of people would like to order something that is ready to plug into the board at one end, and plug into a fan at the other.  Maybe include the headers and a 40mm fan for a complete kit.

++1

I too would be interested in one of these as a complete unit for future addon to my new SD3.

K

Zarni I agree-  I think as the community expands, more people are going to be looking to add peripherals that they would like to control via common interface  (gcode, naturally).

I²C would be a prime candidate to allow commands&data to cross from PC through printer to any number* of slave devices.  If so-inclined, SD could capitalize on this idea by just soldering in a jack to the I²C pins and modifying firmware to support a custom gcode command to send data to node ID, and another gcode command to receive data from a node ID.  This would allow designers to share or sell  all sorts of custom sensors and/or drivers  that anyone can just plug-in  (as if it were a USB device). Fantastic!

* standard I²C is limited to 128 device addresses.  you probably wouldn't connect 128 peripherals to your printer, but designers would have to be cautious not to reuse the same address someone else did-  otherwise you couldn't attach both even if they were the only 2 peripherals you wanted to use

there's a good side and a bad side to the idea that you treat printer parts as peripherals and give them addresses...

to start with, so long as you agree an addressing scheme, then you don't really have an issue with things having the same address. (split the 128 addresses up into device classes, (e.g. actuators, limit devices, cooling devices, temperature sensing devices etc))

eventually your imagination takes you to a place where everything is just plugging into a machine.
so when your machine turns on it does a basic systems test and goes to a discovery mode, and you have parts that reply with information.
for example on a standard solidoodle, you'd get replies from xyz and e motors. xyz end stops, perhaps they'd be more clever and have a learned idea of how big the axis etc were...
so when a printer turns on it knows it has 3 axis, 150mm long and a single colour print head.
(like how when my PC turns on it knows it's got a 3 button scroll mouse connected, or if I put a different mouse on (say 5 button) then new functions are opened up)
When you stick an SD card in your panelolu that's 180mm square in two colours, the printer can just tell you that it can't print that.

basically, if you start hanging things off a proper bus and putting addressable devices in like they were peripherals then the possibilities are pretty endless.

In addition, if you were going to use a 1 wire bi-directional bus.
it's make sense that whilst you were throwing away all the conventional printer wisdom that you also threw away the conventional controller boards and made one with a CAN bus.

The reason for this is that this bus rather than being master/slave oriented is multi-master with a way of giving preference to devices based on address that aids conflict resolution.

so (for example) in your car,
You may use stalk controls to increase the stereo volume.
if your car wants to tell you that the brakes are faulty then the brakes win because they are addressed lower.

To put that in a printer setting, control messages from the panelolu, saying that you want to increase/decrease speed would outrank (and be given preference to) messages regarding cooling.

But the trouble is, when you start sticking Uc's into everything you suddenly increase the cost.

As for this being the future of 3d printing.
Personally, I think that the ease brought by embedding chips in everything, and adding loads of user easy. (so just add or takeaway a fan etc) far outweighs the few pence spend on doing it.

However as free as the Rep-Rap movement (that most commercial makers are taking their direction from) goes, Adrian Bowman said that he did not like the idea of adding chips to everything. and given that RepRap (as a brand) is controlled by him, it seems unlikely that any standard could ever be endorsed...

Any chance you would be willing to combine an inverter with your driver?  I have been experimenting with an air pump for an aquarium and have got great results due to the highly directed nature of the airline tube, however the only reasonably priced air pumps with enough cfm are the AC ones. I'd love to have that gcode controlled if there is a way... I've come up with quite a few examples online, but it seems like it would take a pretty bulky board to combine an inverter and driver.

No inverter-  wattage too high for the SD.  And with external supply-  why bother stepping down only to step-up..
I DO want to experiment with aquarium pump & airline tubing, since it should be easier to use tubing rather than mount a fan directly. I already experimented with using a fan + tubing but alas there is zilch-balls for air pressure.  Instead, I would employ a TRIAC to chop up the 60hz mains voltage to dampen the aquarium pump (google for zero-cross dimmer circuit)-  but this is still speculative whether or not it will work according to the pump's design  (tho it should, if timed properly upon the zero-cross).  As well, the cost for such a kit would be higher-  not only do you need to buy an aquarium pump but the AC chopping framework will require some optoisolation, an MCU, and extra connectors/protection for dealing with mains AC.  Long story short-  maybe in the future

Technically there is only 1 signal wire (PWM)-  but to make the story complete you need a common reference (GND).  Further, since the circuit doesn't supply it's own power so you need +12V, as well.  So there are 3 inputs  (+12v, Gnd, PWM),  and 2 outputs  (Fan+, Fan-).

I have not tested the circuit using MOSFETs.  It would probably work fine, but there is some legitimate reason for choosing oldschool trannys. The design for under-volt protection assumes the ~0.7v saturation voltage present with an an NPN transistor.  Second, you'll get better output regulation if you use the suggested PNP transistor for the buck supply-  it might be discomforting because the PNP transistor can get rather warm during operation at 40-60% fan speeds, but this is simply because the transistor is being largely driven in "active amplification mode"  rather than the pure on/off you might get with a MOSFET-  the output is smoother. It's not warm enough to require a heatsink.  In any case, a typical 12v 40mm fan can draw upwards of 2 watts so do NOT use typical 3906 and 3904 NPN/PNP transistors  (the tiny half-circle shaped ones)-  get power transistors in the TO-220 package, like seen in the picture.  Regardless, though, it should still work fine with MOSFETs  (just make sure you use an N-channel type in place of the NPN,  and a P-channel type in place of the PNP)

4-pin input jack only has 3 pins used

I'm just surprised this thread hasn't gotten hardly any attention

I check SoliForum every day, and I KNOW that this thread was not visible as "new post" yesterday.  This has happened quite a bit recently.  Puzzling