Flyback Diode Usage

[john21443]

Hello,

I am implementing an interface to control 4-wire fans.   These utilize constant DC for power and a separate PWM into the fan to control speed.    No external transistor is needed.   The datasheet for the IC I am using:

http://www.analog.com/media/en/technical-documentation/data-sheets/ADT7470.pdf

Figure 17 on page 21 shows use of a flyback diode (also attached below).

The attachment "schematic.png" is how I implemented this portion of the circuit (diode is D1).   J10 is connector for the fan.  A snippet of the pcb is in "pcb.png".   This will be a 2 layer board with a ground plane on the bottom layer.   The ground plane is not shown in the screenshot.

Is it OK to use the ground plane for this purpose or do I need to have a more "direct" routing of ground through the diode?

Any other feedback would be welcome.

Thanks,

John

[T3sl4co1l]

That seems fine. But why would the supply ever reverse polarity?  Is this circuit being switched on and off from some external driver??  Worst that should be needed is a few capacitors to bypass +V and that's that.

Tim

[digsys]

The most effective placement of a flyback device is directly across the coil / inductor itself. The further away the less effect it has.
If the fans have captive wires, I'd place it in parallel with the fan connectors, and stay away from ground planes all together.
Remember, that there is a HV spike right up to the diode.
As for the DC supply protection diode, a more common / effective approach is to use a zener (of appropriate voltage / power).
ie for a 5V 1A supply, use 5V6 3W. That protects from both O/V and reverse. I usually always add a polyfuse as well.

[T3sl4co1l]

Oh man.  Just looking at that datasheet now.

It was probably written by a competent digital engineer.

By "competent", I mean one that does nothing but digital design.

Just searching on "diode" turns up only six results, all of them about input conditioning -- and of those, the methods shown are inconsistent with the text, pictures, and device ratings!

Not one of ADI's most stellar products.

They've made lemons before.  They're rare, but you must always be on the lookout for them.

This looks more like a documentation failure, than necessarily anything faulty in the design of the chip.

Yeah. For the 4-wire configuration pictured, THE DIODE DOES NOTHING, can not possibly do anything, is not needed, and can be removed.  There is no discussion about the diode in the subsequent figures either (Fig. >17).  For those cases, they do show an external switch, for which the diode may be helpful.

DO make sure there's a 0.1uF ceramic and/or 10uF electrolytic, or more, somewhere between +V and GND, near the connector.

HTH,
Tim

[floobydust]

The 4-wire fan standard by Intel, does allow the tach signal to swing to 12V.
Pavouk mentions some fans have internal pull-ups to 5V.
The fan standard C4.1.1 also requires "reverse-polarity protection of the fan", so Analog Devices did that bonehead diode implementation (1N4148) 
I dislike A/D's complexity for a relatively simple control task, this IC seems overboard to me.

I would be careful and leave provisions for the extra resistors or zener clamp if a 12V swing on the tach is encountered.
The 4-wire fan standard I linked to is 2005 Rev. 1.3 so not sure if things have changed.

[john21443]

Thanks all for the insight!