Hi,

I have a 12V DC fan drawing 18 amps.
I want a 2 second ramp for the fan startup. My DC supply system doesn't like abrupt amp increases you see.
Can someone draw a simple circuit to do this with component values calculated? It's a high temp (say 105 degree C) application.

Note that this is not the same as a normal current limiting DC motor startup -- those applications limit amps in the first few millisconds of startup. I'm talking instead about a really slow startup in the order of 2000ms.

Thanks!

Paul

 

How often would the slow start circuit be engaged?
What is your skill set for assembling a small electronic circuit?
Is this a one-off or a production item?
Does the fan have to be at 100% in 2 seconds, or could it be at 83% at 2 seconds, 95% at 3 seconds, 99% at 5 seconds?

PWM (pulse-width modulation) is a popular option for delivering a variable voltage to a fan without dissipating a large amount of heat in the pass device. But in your case I'd go with a simple ramp up of either a P-channel power MOSFET in the +12 or an N-channel in the ground, something that goes to almost zero impedance once the fan is going full speed. An R-C ramp with a 1 second time constant could be tuned to give the exponential ramp detailed above. MOSFETs are not nearly as repeatable and predictable as bipolar transistors when it comes to exact turn on performance, but they have exceptionally low loss when fully on. If you need something more tightly controlled, a linear voltage or current ramp can be done by adding an opamp and sense resistor. Either way, not a big problem to solve.

What country are you in, what kinds of parts do you have access to, is this for a MIL or other rugged environment?

ak

 

Here is a circuit you could adapt using a high pass transistor for higher current, there are several examples out there on the web.
Max

 

How often...

hi AnalogKid,

the fan would be engaged, say, once a minute. it should last for years.
it is a one-off private affair. it's is a 105 degree C environment.
a precise linear ramp is not required. something that asymptotes toward close-to 100% is fine -- just as you describe.
I'll be ordering all the parts online from any of the popular electrical parts site.
i have done plenty of sodering -- so no problem there. i happen to have an old cathode osciliscope + auto-ranging multimeter.

Here is...

hi Max,

cool thanks. i see this circuit has a 20V input. My input is 12V output is 12V.

is there a circuit for this with the correct values?

 

Max, your circuit uses a LM317. His fan draws 18 amps. (The LM317 is only good for 2 amps.)




Edit: Oops, sorry, I didn't catch the mention of a pass transistor. @MaxHeadRoom

 

Max, your circuit uses a LM317. His fan draws 18 amps. (The LM317 is only good for 2 amps.)

Note that Max mentioned using a pass transistor for the higher current.

ak

 

You are turning a 216 watt fan on and off once a minute? Gotta ask why. Or did you mean it is on for one minute at a time? If so, how many fan cycles per hour/day/whatever?

Here is a minimalist approach. LT Spice thinks it will work. At 18 amps, every milliohm is 1/3 W, so use a very fat FET. Note that fan current probably will be very low until Vout reaches 6 V or so, then rapidly ramp up to approx, 6 to 9 A and track the output voltage to full current at 12 V. All fans are different, so you'll have to check yours.

ak

 

hay that was fast only 4 components -- cool. ramp is perfect.

the 3 second lag to start is not ideal :-/
how many amps can Si4427DY handle?
i don't know what the draw/impedance of the fan is at low speed, but the fan does have a 40 Amp fuse, so i think something that can handle >40 amps is best to be on the safe side. sorry, i should have mentioned that.

the fan is thermostat controlled.

i don't mind spending $50 on parts if that's what it takes.

 

The FET is something I pulled out of the LTS library to illustrate the method. Something that can dissipate 10 W and has Rdson under 0;01 ohm is a starting point. Digi-Key's parametric search engine is excellent for this.

A more complex circuit would start the ramp immediately. Opamp integrator, etc. Again, is high side control required, or can we use a N-channel FET to GND?

ak

 

Your ramp gradient and flattening-off is fine as you have it.

Can't you somehow make it so V(gate) starts at 11V? I assume this will pull the red line to the left by approximately 3 seconds.

Just add a 10meg across C1. No?

 

well downloaded this. really easy to use. managed to duplicate your curve. ak you are really making me do far to much work here ;-) LOL

 

the circuit below seems to work fine. got it to ramp immediately as you can see from the screenshot.

i searched on digikey and found a mosfet i like for only $2: FQPF47P06

there are actually a number that would do, i want something with a hole so i can screw it into the aluminum panel next to the fan.

problem is there is no correlation between the selectable mosfets in ltsplice and those on digikey.

how do i put something on digikey into ltsplice?

this is all taking far to many hours

 

the circuit below seems to work fine. got it to ramp immediately as you can see from the screenshot.

Need to back up a bit. I took a slightly different path but got basically the same result - the first cycle. then I extended the simulation for a 2nd on/off cycle and the fan comes on immediately. The problem is that there is no path to discharge the timing capacitor. And things probably will get worse when the simulation is updated to have a power source that goes to a high impedance when off rather than a 0 ohm path to GND. I'll try it with your version and see. More later.

ak

 

Gopher - Paul was talking about his hours, not mine.

ak

 

Put a 12 V car battery across your DC supply.

 

Here is a high power motor controller. The IRF 1405 MOSFET can handle 168 A. Make sure the 2,200 uF 50 V electrolyte is low ESR. Everything else should be polypropylene or polyester 100v, and normal electrolytic but preferably tantalum. All resistors metal.
How you use this circuit, well ?
Switch a resistor, wound from heavy gauge copper wire, in series until the fan comes up to speed, then full throttle. Your DC supply should suffer the initial inrush of current.
Or, modify your power supply.
Bypass the over current sense circuit.
The base of Q3 is tied to 5 v through a 1 k resistor, and driven through 56 Ω directly from a PIC port.

 

no

the circuit i need should be around 5 or 10 components max.

simple enough to soder the leggy things to each other and not need a circuit board

then pot the thing into a 1 inch cube with 2 input wires and 2 output wires.

gonna try simulate maxheadroom's design on ltspice

my apologies --- I didn't realize this would tax all you geniuses so heavily. thought getting a list of components would be a simple task.

thanks ak for showing me ltspice

thanks max -- will try your design

bye all and good luck

 

Trust me.
Soldering three legged things without a circuit board, or a jig of some kind, where bits can be soldered and un-soldered easily, will end up in grief and a lot of throwing your arms in the air. You may never finish the circuit due to pure frustration.
It's important to set you work area out specifically for the task of building this circuit, and treat it like your baby.

 

pot the thing into a 1 inch cube with 2 input wires and 2 output wires.

Don't pot the FET (unless you like firework displays). It dissipates an average 18W or so in the first 5 secs.

 

no
the circuit i need should be around 5 or 10 components max.
simple enough to soder the leggy things to each other and not need a circuit board
then pot the thing into a 1 inch cube with 2 input wires and 2 output wires.
gonna try simulate maxheadroom's design on ltspice
my apologies --- I didn't realize this would tax all you geniuses so heavily. thought getting a list of components would be a simple task.

If you had been more forthcoming with the physical requirements and your expectations early on, you would have gotten much better responses. Your expectations indicate a lack of experience with these kinds of power levels. You are trying to control over 300 W. That will not happen in 1 cubic inch, even if all you want is a saturated switch. If the fan were a simple resistor, peak power dissipation in the control device would be over 50 W. In a +105C ambient there is very little thermal headroom to dissipate that energy before the device reaches its max operating temperature.

ak