4 wire fans

Will they be exposed to identical environments? Differences in air flow resistance and dirt buildup are going to cause variation. It would be a very unusual environment that is exactly symmetrical. Even if they start very closely matched the bearings will probably deteriorate at different rates.

Yes. Imagine them side by side on an infinite baffle.

But this doesn't address the electrical issue.

E.g., if I provide the same PWM *control* signal to two "same make/model" fans, will they rotate at the same rate? I.e., is the PWM talking to a controller as anything other than a "digital representation of an analog signal"?

If, for example, I drove two stepper motors with the same drive (ingoring back EMF), I would expect each to move through an identical rotation as I would be directly commutating them.

What I see of fan specifications suggests the tach is consistently designed ("two pulses per revolution" -- though that doesn't mean they are displaced by 180 degrees).

But, the PWM signal appears more of a suggestion, subject to tolerances (10%!). So, conceivably, identical fans could rotate at different speeds when driven by the same signal.

I have never come across a fan that uses stepper motors.

Yes, they will rotate at slightly different rates. > [I assume the blades and other air flow components are

I don't believe I *claimed* they were used in a fan (though many are BLDC -- which is a special case of a stepper motor, electrically commutated) rather, to illustrate a point: that the signal(s) directly controlled the ROTATION (instead of acting as a digitization of an *analog* control signal that tries to control the RATE of rotation).

So, I should "close the loop" for each fan and then drive the *setpoints* with the same signal.

If they were put into a fan they would be called synchronous motors (which what stepper motors are called by people who know what they are talking about).

Synchronous motors are driven by alternating current, as are induction motors.

Synchronous motors don't. The rotation rate always follows that of the applied magnetic field, with a phase lag that gets larger as the load gets heavier.

You need to know about the motor technology.

If they use induction motors.

We often have a temperature sensor somewhere and control the fan speeds from that. Big fans are noisy so we only want to go max when we have to. We also limit the rate of change of fan speed, to not make shocking sounds.

Air flow is entirely perverse. It doesn't do anything reasonable. Even the flow direction can be a surprise.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

There's no aviation industry. Planes would fall out of the sky at random, if anybody had been silly enough to try and invent heavier than air flight.

To John Larkin. In fact air flow is merely complicated. It goes from laminar at low air speeds (when the Reynolds number is less than 2400) to turbulent at higher speeds. There's always a region of laminar flow close to a surface.

At higher airspeeds the air flow behind a restriction to the flow can do complicated stuff - I worked on a flow rate meter that counted the number of von Karman vortices shed per second behind a bluff body.

If you want consistency, I suggest you use brushless drone motors and ESC's (electronic speed controllers). I don't know what you're trying to accomplish, but if it's stable air flow, methinks you'll do better using existing drone technology.

For "airflow performance", it's customary to use a turbine driven anemometer, thermal anemometer or Pitot tube. These work well inside a duct but badly in open air or around a tangle of airflow obstacles.

The ultimate test of air flow theory is wind tunnels.

SF State University 3D prints their own wind tunnels. Cal Poly has a hypersonic wind tunnel.

I've seen expensive card cages, with fan trays blowing air upward, where some card slots have near zero or reverse air flow.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

it's a suggestion like running the fans with a variable voltage is

The rotation is not locked to the pwm, the fan does the commutation and the pwm just limits the power resulting in lower speeds

They can slip if the torque is big enough. I have seen it.

I don't see why a regular 4 (or 3) wire fan can't give consistent performance -- if you close the loop around it. They're already designed to be used to move air.

What I was wondering was how realistic using the tacho feedback from *one* fan would be as an indicator of how N fans, driven from the same control signal, would behave.

Like driving a set of lamps from one control signal based on the sensed state of *one* of those lamps.

E.g., it is not uncommon to see a bunch of fans, side-by-side, in a blade server, seemingly operating at the same setting. *The* setting defined by *one* criteria in the box.

It seems the safer bet is to close the loop around *each* fan and then drive the setpoints for all loops as desired, counting on each controller to deliver consistent performance from the individual fan controlled.

With the tacho indicating actual rotation (and not just commutation), it should be a good enough indicator of the individual fan's operation. And, I doubt the shape of the fan blade changes over time so how the fan performs (at a given setting) should be consistent.

In the past, we've used individual sensors to monitor the airflow from the individual fans to determine faults. But, that's costly and adds the potential for another class of failures to the design.

The world is full of designers who are less competent than they think they are. They get surprised quite often.

They don't slip. They skip steps and eventually stop rotating at all.

Years ago I performed some rough-and-ready experiments on small fans to see if the airflow was consistent in practical environments. The general conclusion was that the more efficient the fan, the greater was the drop in performance when the blades acquired a thin coating of dirt and dust. We finished up using a very inefficient blade shape so as to keep the airflow constant over a wet bulb thermometer.

Whilst stepper motors are technically a type of synchronous motor, people who know what they are talking about will not call them "synchronous motors" so as to avoid confusion with sinewave-driven synchronous motors.

Or rotate backwards - or oscillate. The magnetic field has compliance and the rotor has inertia, so they will form an oscillating system at certain stepping frequencies - this can happen even it there is no load. A 'soggy' resilient coupling between the motor and the inertia of the load is a convenient way of damping this out.

The designer may choose to to mount the motor resiliently because of the noise caused by the stepping; in combination with the mass of the motor, this can also be excited into resonance by the torque reaction. Stepper motors have a lot of hidden snags.

Not to add to the confusion, but you can drive a stepper from an ac-line frequency sine wave and get a nice smoth slow rotation. Phase shift one winding with a series R-C.

I did that for a tape drive capstain once.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Interesting. But, if all the blades see the same environment, then they will all "degrade" similarly (?).

I haven't sorted out how to keep fans (and other cooling mechanisms) from being a continuous maintenance issue -- short of specifying operation in a clean room!

OTOH, automobiles suck a lot of "dirty air: into their carburetors, yet those filters don't see frequent replacement.

OToOH, residential HVAC systems seem to need monthy replaecments (our homes are dirtier than the outdoor air our vehicles breathe?)

You can also opt to drive them with pseudo microstepping drives instead (at added cost/complexity).

But most of these can be addressed in design.

We looked at a lot of rotary motors for an industrial automation project I was involved with (ages ago). The big advantage stepper motors have over synchronous and DC servo motors is they are designed to be *stopped* -- and exert significant holding torque (25Nm in our case).

Getting this sort of performance from a synchronous motor or DC servo required adding gearboxes and active brakes. And, then having to address the wind-up in the gearboxes when you wanted to change direction, as required by a governing control loop.

Where they really fall short (besides cost and drive complexity) is in higher speed "running" applications.

It's sort of the difference between driving a drift car and a bulldozer -- totally different use cases.