simple 24v fan control problem

[harp]

I am working on automated fan speed for lab psu... the fan is 24v 0.15A, and I am connected it to 820 ohm to ground, that provide minimum silent air flow on startup. It is powered around 4v at minimum point. Parallel to resistor I connected BD139, that drive a 10k NTC. Everything work like expected, but in extended test, where I isolate NTC, I have few issues, and like to ask here for help. 1. If ntc is not connected to heatsink, it go slowly to heat up and up and fan make noise than... At point "A" I measure voltage and put NTC in flame, everything is working ok till about 1.1v, and than, if pass some level it self latch and begin heating itself, pull huge current... I must power off to prevent damage. 2. I try to put 2 diode in series from "A" to ground, but it not work, not limit voltage to 1.2v... how to control max voltage on point A but not modified others resistor (it is perfectly set to what I want)?

[stj]

icrease the value of the pulldown resistor,
the ntc is shorting itself to ground!
that said, something must be smoking hot!

[petehall347]

I try to put 2 diode in series from "A" to ground, but it not work, not limit voltage to 1.2v... how to control max voltage on point A but not modified others resistor (it is perfectly set to what I want)?

would a zener work ?

[harp]

Thank you stj, you open my eyes I always think there is some kiloohm in there on ntc... I add a 220 ohm in extend to ntc to 24v, and in brief test it solve problem, not latch anymore. The ntc is 10k, and other two resistor value is determined to function at specific graph, so I cant just change value too much. I know, it is not good circuit, high voltage large current, but it is simpe enough and in practice ntc should not be over 100 deg.C... I do future test and if not satisfied, I have to add some sort of protection, with zenner, transistor... that not affect to graph characteristics... if anyone have idea, not complicated, I like to see...

[truclacicr]

Point A connects to ground via 150 ohm and 220 ohm resistors. At 3V, and allowing for a Vbe drop of 0.6V, the total current exiting that node is only about 30mA. Where do you get 1.2A? At 0.5V you should get about 3mA.

[harp]

1.2A? Really I dont know, I see that on psu display second after pass 115mA... in that time, ntc and bd139 are be hot to touch, and ntc can not to be again colded by touching like usual and fan go crazy. BD139 are later put on biger heatsink, and not overheating again...................... But ofcorse, now I have second problem, because "my simplicity"... I used ntc in metal can for testing, and it has significiant thermal mass all over the sensor. That act like small heatsink on it. Now I put flat ntc, with no thermal mass, bare sensor, and it is over heated in split of second if not strongly bond to heatsink. I dont like that, sensor should not pass current... Is there some quick fix for that, or it involve complicited schematic? I hoped that I will pass with one transistor...

[stj]

you should use pwm by linking the NTC to a 555 or something.

[truclacicr]

1.2A? Really I dont know, I see that on psu display second after pass 115mA... in that time, ntc and bd139 are be hot to touch, and ntc can not to be again colded by touching like usual and fan go crazy. BD139 are later put on biger heatsink, and not overheating again...................... But ofcorse, now I have second problem, because "my simplicity"... I used ntc in metal can for testing, and it has significiant thermal mass all over the sensor. That act like small heatsink on it. Now I put flat ntc, with no thermal mass, bare sensor, and it is over heated in split of second if not strongly bond to heatsink. I dont like that, sensor should not pass current... Is there some quick fix for that, or it involve complicited schematic? I hoped that I will pass with one transistor...

Assume that the resistance of the NTC is 0. Then 24V appears at A. Now the total current is (24V - 0.6V)/ 220 ohms + 24V / 150 ohms = 266mA.

Look at it another way. Let's split 1.2A between the two resistors. Then the power dissipation in the 150 resistor is 0.6A x 0.6A x 150 ohms = 54 watts.

[harp]

I put ntc on negative rail, and feed npn inverter. That add resistance in sensing circuit, get rid of latching and heating, but complicate circuit (wanted chart parametars) that I cant anymore follow. I manage in practice to fan start increase speed at 30 deg.c, and at 50 deg.c it is nearly fully on. I expect that all between has to be - if not linear - some sort of simple curve... but no, it jumping all over from min to max... I make some pause now to this project, and I gues that start all over with opamp control.

[stj]

i would use a cheap 8pin microcontroller with analog pins - maybe a pic12f series.
or maybe an atmel tiny85