2-pin fan voltage controller from 3V PWM/DAC, advice?

[fivesixzero]

I'm pretty new to analog design and power circuits so any advice here would be appreciated.

Goal is to control voltage of a 2-pin 12V rated (4.6-15V capable) BLDC fan based on an input 3V PWM or 0-3V DAC signal. Should be able to handle up to at least 300mA current draw.

Using a FET switch would work fine for limiting current with PWM, I think? But I'd like to use voltage here instead, if possible. I know I could just get a 4-pin PWM capable fan but its not always possible to find a PWM fan in the CFM/pressure range I'm looking for.

While fixing up an HP 6644A power supply recently I noticed that they basically do this exact thing in that design.



I created a simple-ish Falstad simulation to learn more about how that circuit works and see if it'll work with the PWM/DAC input I'm expecting to get from the EMC2101 controller I'm prototyping with, which can be programmed to do either PWM or a 0-3V DAC output on its "FAN" pin.

https://tinyurl.com/yzwsryxo

I won't have the ability to test this idea on a breadboard for a bit so I'm hoping to get a sanity check here. Back at the bench I have a few op amps to try out (not the original MC34701 though) as well as a TIP122 and a few other FETs and darlingtons to play with, so component selection is flexible.

So is that circuit I'm playing with a sane choice? Or am I way out in left field and better off just doing the 'easy' thing with a FET switch circuit?

[David Hess]

That HP design is basically what I would suggest.  An operational amplifier drives a power pass transistor within its feedback loop with appropriate frequency compensation to handle an output capacitor.

Besides the frequency compensation, the tricky part is arranging for a 0 to 3 volt control signal to be offset to produce a higher minimum output voltage; this requires some type of voltage or current reference which in the HP is provided by the regulated power supply voltage.  Some BLDC fans will fail to start if their input voltage is raised gradually.

[oPossum]

An old thread with a few circuits that I built and characterized... https://www.eevblog.com/forum/projects/pc-motherboard-pwm-4-pins-to-3-pins-digital-converter/msg24284/#msg24284

[fivesixzero]

That HP design is basically what I would suggest.  An operational amplifier drives a power pass transistor within its feedback loop with appropriate frequency compensation to handle an output capacitor.

Besides the frequency compensation, the tricky part is arranging for a 0 to 3 volt control signal to be offset to produce a higher minimum output voltage; this requires some type of voltage or current reference which in the HP is provided by the regulated power supply voltage.  Some BLDC fans will fail to start if their input voltage is raised gradually.

Thanks for the sanity check. Glad to know I wasn't totally off the mark.

Getting the offset right was a bit of a challenge, but it looks like it just comes down to tuning the resistor values on the left side (input from DAC/PWM) and the right side (feedback) to find a good balance. With the HP circuit this just takes a few iterations and can result in a pretty nice 4.6V to 12.6V spread.

I'm still foggy on a few things in that circuit though. Specifically the function of the two zeners - VR771 / 9.1V, which is actually 10V in the real circuit and VR770 / 7.5V. Its hard to tell from the simulations what particular purpose those serve or whether they're actually required.

[fivesixzero]

An old thread with a few circuits that I built and characterized... https://www.eevblog.com/forum/projects/pc-motherboard-pwm-4-pins-to-3-pins-digital-converter/msg24284/#msg24284

Nice! Cool stuff, I'm definitely going to spend some time looking over that thread in the coming days. Thank you for sharing.

[David Hess]

I'm still foggy on a few things in that circuit though. Specifically the function of the two zeners - VR771 / 9.1V, which is actually 10V in the real circuit and VR770 / 7.5V. Its hard to tell from the simulations what particular purpose those serve or whether they're actually required.

VR770 acts to level shift the output of the operational amplifier up by 7.5 volts because its +15 volt supply voltage does not allow it to drive emitter follower Q770 directly.

VR771 limits the output voltage but I do not know why that is necessary.

[mariush]

Have  you considered using a led driver IC as a fan controller?
There's buck led drivers which accept a pwm signal or a dc voltage level to set the current to a percentage of the maximum configured through a resistor.

example https://www.diodes.com/assets/Datasheets/PAM2861.pdf

vset :

• Multi-function On/Off and brightness control pin:
• Leave floating for normal operation.
• Drive to voltage below 0.2V to turn off output current
• Drive with DC voltage (0.3V < VSET< 2.5V) to adjust output current from 12% to 100% of IOUTnom
• Drive with PWM signal from open-collector or open-drain transistor, to adjust output current. Adjustment range    1% to 100% of IOUTnom for f < 500Hz
• Connect a capacitor from this pin to ground to increase soft-start time.    (Default soft-start time = 0.1ms. Additional soft-start time is approx.1.5ms/1nF)

[David Hess]

A lot of integrated variable voltage regulators could be made to work.  For linear ones, the limiting factor is the dropout voltage but a single external transistor can get around this; the regulator then becomes a combined voltage reference and error amplifier.

A more universal part like the 723 would be interesting and I have seen this done.

[fivesixzero]

I'm still foggy on a few things in that circuit though. Specifically the function of the two zeners - VR771 / 9.1V, which is actually 10V in the real circuit and VR770 / 7.5V. Its hard to tell from the simulations what particular purpose those serve or whether they're actually required.

VR770 acts to level shift the output of the operational amplifier up by 7.5 volts because its +15 volt supply voltage does not allow it to drive emitter follower Q770 directly.

VR771 limits the output voltage but I do not know why that is necessary.

Thanks for that breakdown. I'm trying to learn as much as possible while implementing this, largely so that I don't end up with 'cargo cult' mentality where I just use pre-made circuits without understanding them. This circuit has been a good challenge since I have enough knowledge to grok the basics of how it works but there are enough 'new' things (like these zeners and the darlington setup) that I've got plenty to chew on.

Have  you considered using a led driver IC as a fan controller?

I had started by looking at a few ready-made fan and LED driver/controller ICs but I found that I don't (yet) know enough about that space to zero in on precisely what I'd need from one of those. I figured designing a circuit from jellybean parts would be useful exercise and its already paying off pretty well in that regard.

That said, I appreciate the suggestion. I hadn't seen that particular chip yet and its already got me thinking about simpler 'v2' ideas using ICs like these.

[David Hess]

VR770 acts to level shift the output of the operational amplifier up by 7.5 volts because its +15 volt supply voltage does not allow it to drive emitter follower Q770 directly.

VR771 limits the output voltage but I do not know why that is necessary.

Thanks for that breakdown. I'm trying to learn as much as possible while implementing this, largely so that I don't end up with 'cargo cult' mentality where I just use pre-made circuits without understanding them. This circuit has been a good challenge since I have enough knowledge to grok the basics of how it works but there are enough 'new' things (like these zeners and the darlington setup) that I've got plenty to chew on.

The way the schematic is laid out makes it more difficult to grasp.  I would redraw it for analysis.

There are several ways to make that sort of level shift allowing a lower voltage amplifier to control a higher voltage load including using a constant current through a resistor.  The advantage of using a fixed voltage like that is that no gain is added within the feedback loop which aids stability.  If the level shifting element is bypassed with a capacitor, then it works even at high frequencies.  Of course without gain, the maximum change in output voltage is limited to the output range of the operational amplifier.

VR771 might be there to prevent applying too high of a voltage to the fan.  With it in place, the maximum voltage at the transistor emitter is about 15.4 volts.  Without it, the operational amplifier output could potentially be pulled up to at least +15.6 volts under fault conditions making the emitter voltage 21 volts which might damage the fan.

D605 prevents the base-emitter junction from being reversed with current supplied by C772 if the operational amplifier output falls quickly.  I am not sure this is really needed to protect the power transistor but it does not hurt.

Update: R778 also supplies some of the fan power and would also supply current through the reversed base-emitter junction when the operational amplifier output is low.  Diode D605 blocks this.

I really like that notation in the schematic around input filter capacitors C602 and C603.  It shows single point connections to the capacitors to prevent return current loops on the ground side and to preserve low ESR.  The Tektronix PS503A power supply is constructed to do that also as shown below.