Continue to Site

Eng-Tips is the largest engineering community on the Internet

Intelligent Work Forums for Engineering Professionals

  • Congratulations IDS on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!

Defeat an open collector sense line?

Status
Not open for further replies.

PeterEFry

Computer
Feb 2, 2007
2
Cheesy question:
I have a network appliance with a 1U form factor power supply with a couple bad fans -- they rotate, but with excessive noise and vibration (it seems to be their normal mode, but it's hard to tell without samples for comparison). I would like to replace them -- in fact I have new (better in every way) fans installed -- but I received a rude awakening: The PS uses fans with a locked-rotor sensor, not a tachometer.

Grrr. I'd like to defeat the sensor input.

According to the original fans' data sheet, they supply an open collector (NPN, emitter -> ground); judging by my tests, this seems to be accurate. I would expect a 5V (most likely) or 12V pull-up for the shutoff logic, with a 1k or 2k resistor. Now, I'm a straight-software-logic kind of guy, so my first impulse is to simply ground the pins, but the possibility of excessive current flow worries me. If I'm right, I get a few mA, but if I'm wrong... So here I am. What would y'all recommend? Remember, I want a constant "on" (pulled low) state, and I have to cram the solution into a tiny, already-too-full box. I also would rather not disassemble (i.e. destroy) the original fans (they're not readily available, and a replacement PS is $750), so reverse-engineering their internals is a bit problematic. Same with the PS itself -- naturally, it's all glued together.

I'm going to make a few more checks, but I'm not sure if they'll provide me any useful data: Resistance between my sensor pins and 5V & 12V, just to see if I have a clear path to the pull-up resistor; perhaps current flow using the old fans.

Thanks.

Peter E. Fry
 
Replies continue below

Recommended for you

Can you hook a voltmeter between to the sense wire and the chassis ground to get some idea of what the running installed signal is?

Keith Cress
Flamin Systems, Inc.-
 
Ground the sense wire with a 100mA fuse.
Looks like a 1/8W resistor, solders in like a resistor. Costs a buck.



Mike Halloran
Pembroke Pines, FL, USA
 
Any open collector should be less than a couple mA. Use a 100 ohm resistor to ground. Cheap and even connecting it wrong wouldn't damage anything. Post back with the result. I haven't run into any of these three wire fans yet. I love these defective power supplies for project and would like to know what this third wire actually detects.
 
Good points about the fuse and the resistor -- thanks. I have some 250mA fuses lying around for another project, but when I think about it, the 100 Ohm resistor yields a maximum 120mA on an open 12V line, which is the most I can have. So I threw it in, and got .3mA. Perfect. My ground points aren't perfect, so I may use an 82 Ohm resistor instead -- I'll test it again once I've built the wiring. Funny -- it looks like an AC signal. Eh -- WTH, it works, and I'm back in business. Thanks again.

Most of your three-wire fans have a tachometer -- a square wave out driven off of a Hall-effect sensor with a transistor. Comair Rotron has some good docs. 2 pulses/rev standard. The original fans in this PS have a locked-rotor alarm wire instead -- a special-order option. I'd recommend grabbing the datasheets for your particular fans. In my case, I looked after the fact, and I had to rip all of my stuff apart and rebuild it -- a personal peeve. The info wasn't hard to find -- a Google search got me:
...which has a part number key and datasheets. Not all manufacturers have this data handy for OEM parts -- another peeve. Me, I'm a believer in COTS (but not C[insert acronym for excrement here]OTS -- parts have to perform).

In my experience, the tach implementations vary and aren't terribly reliable -- no standards. But hey.
 
There are two possibilities:
1.) Error = 0 No error =1
2.) error = 1 No error =0

The first is the most likely. Try to disconnect the sense line.

If this doesn't work case 2.) applies - so GND the line.
No resistor or fuse is necessary.


Plesae read FAQ240-1032
My WEB: <
 
Status
Not open for further replies.

Part and Inventory Search

Sponsor