This is a subsidiary page to the Fan Selection & Application Guide, which discusses the electrical aspects of the various fan products carried by Digi-Key. A number of general trends and patterns exist with regard to how these products are powered and how auxiliary signal inputs & outputs behave, as well as subtle variations within those patterns that are worthy of note.
Figure 1 shows the relative proportions of different interconnect styles for DC fans in stock at the time of writing. Understandably, the plain wire lead types are dominant–there’s little cause for a manufacturer to spend time & effort applying a connector before the specific connector and lead arrangement desired by a customer can be determined.
DC-input fans with the basic 2-conductor interface are pretty simple; just connect them to a suitable DC source with the proper polarity, and fan away. Avoid reversing the supply leads; modern DC fans are electronically controlled, and their direction of rotation cannot be reversed by swapping the supply leads. While many devices incorporate reverse supply protection and can tolerate such an occurrence gracefully, those that don’t are likely to suffer damage with an alacrity that correlates with the amount of current the (mis)connected supply is capable of delivering.
Figure 1. Chart of termination styles for DC fans in stock at Digi-Key at the time of writing.
A feedback or speed control feature typically adds an additional signal conductor each; common practice with small DC fans is to use the the negative power supply conductor as a shared reference for the signal conductors. The typical output signal from a small DC fan is implemented as an open-collector (or open-drain) output; the output lead from the fan is connected to the collector or drain terminal of a transistor inside the fan, which either creates a path to ground or not in order to provide an output signal.
This approach provides a high degree of flexibility, allowing the fan’s signal output to interface to common logic inputs using only a single resistor, provided that fan and logic input share the same supply (ground) reference. When using such devices:
- Select a pull-up resistor value to limit current flow through the transistor to less than the specified maximum.
- Limit pull-up voltage to less than the specified maximum. For some devices that maximum is specified as an absolute value, others also add a condition of not exceeding the fan supply voltage.
Note that while this structure is quite common, it is not universal; other devices may integrate a pull-up resistance to the fan supply or other voltage for example, which could be either a convenience or a nuisance. Check the manufacturer’s documentation for details.
Figure 2. A representative example of a typical output circuit for a small DC fan, excerpted from the ASB0312HA-AF00 datasheet. While DC fan supply leads are usually red and black in color, other lead colors may vary.
Speed sensor (Tachometer) outputs
A typical tachometer output signal from a small DC fan is a square wave that makes some integer number of cycles per rotation of the rotor; two cycles per rotation is typical though other values do occur. By measuring the frequency of this signal, the rotational speed of the fan can be measured. While such a signal might be expected to have a stable logic level under a locked-rotor condition, it’s commonly derived from rotor position and may change state if the rotor moves by some small amount, as might be caused by a fan’s attempt to auto-restart.
The excerpt from the ABS0405HHA-AF00 spec sheet in figure 3 is an example of typical tachometer signal timing. If the time (t) between adjacent signal transitions is measured in seconds, fan speed in RPM is equal to 15/t.
Locked Rotor Senor outputs
In contrast to a tachometer output, a typical locked rotor sensor output is designed to maintain a stable logic state during normal operation, and change state when a locked rotor is detected. Thus, while not useful for fan speed measurement it requires less interpretation to determine whether or not a locked rotor condition has occurred. They are often somewhat sluggish in their response, and require some amount of time following application of power or occurrence/release of a locked rotor condition to provide a valid output.
The excerpt from the Sanyo Denki cooling fan catalog in figure 4 describes a representative locked rotor behavior; note that time delays between changes in fan state and signal output are measured on a time scale of seconds.
Like signal outputs, the typical signal input to a DC fan uses the negative power supply lead as a shared voltage reference point. Most fans connect the input signal lead to the base/gate of an internal transistor, through a structure that both protects the input against static discharge and provides a default-on behavior to allow use without a control signal. Most devices are quite accommodating in terms of acceptable signal sources, and can work with logic level signals ranging anywhere from below 3 volts to the fan supply voltage or beyond, and many will also function nicely with open collector/open drain signal sources due to their internal pull-up structure. Subtle variations in these structures do exist however as illustrated in the examples in figure 5, and give rise to differing requirements, limitations, and ancillary behaviors. In particular, the potential for current flow from the fan supply into connected control circuitry should be noted; it’s the sort of subtle nuance that’s easily overlooked and which has potential to cause problems.
Figure 5. (L-R): Internal PWM input structure diagrams from the AUB0812VH-SP00, ASB0305HP-00CP4, and EFC0812DB-F00 datasheets, having nominal supply voltages of 12v, 5v, and 12v respectively.
Examples of different speed control input behaviors are shown for a 9AH0812 (figure 6) and AUB0812VH (figure 7) with the datasheets’ depictions of the internal input structure inset, captured while the input was driven using an open-collector circuit of the sort shown in figure 8. The ‘open-circuit’ voltage present at the signal input differs by nearly a factor of ten between the two, and in the case of the 9AH0812 the diagram of the input structure would appear more conceptual than literal; the diagram suggests the input feeds a bipolar transistor, while the plateau in the rising edge of the signal and maximum voltage are suggestive of a FET input.
Speed Control Techniques
Voltage speed control
Adjustment of the voltage applied to a DC fan is a means of achieving speed control or conversely, variations in supply voltage affect fan speed. The relationship between the two tends to be linear; an example of data measured from a 9AH0812P4H04 is shown in figure 9. While the rated operating voltage range for this device is 10.2~13.8VDC, at room temperature this example continued to function as supply voltage was reduced to about 3.5v, or 29% of nominal; restarting from a stopped-rotor condition required increasing supply voltage to 3.8v. Wear, changes in temperature, and other variables will affect the minimum maintainable speed in a specific circumstance;
The main benefit of this approach is a minimum of electrical noise; the inevitable blip here and there generated by the fan motor’s commutation process is still present but little else, as seen in the capture shown in figure 10. The speed adjustment range using this approach is often rather narrow if strict adherence to manufacturer’s recommended operating conditions is to be maintained, and generating the variable supply voltage is often a nuisance; to do so using linear techniques produces excess heat, (usually the thing one is trying to get rid of when using a fan) and the use of switch mode techniques generates electrical noise which runs counter to the point of using voltage-based speed control in the first place.
Figure 9. Measured fan speed vs supply voltage for a sample 80mm tubeaxial fan.
Figure 10. Waveforms from a 9AH0812P4H04 using voltage speed control. Trace 1 (yellow) shows the fan’s tachometer output, trace 2 (blue) the supply voltage, trace 3 (pink) an AC-coupled version of the supply voltage, and trace 4 (green) the supply current flowing through the fan.
PWM Speed control
Typical speed control inputs for small DC fans call for a square-wave input of 25 kHz give or take a few, with fan speed corresponding to the duty cycle of the applied square wave. Exact relationships between the duty cycle of the applied signal and resulting fan speed vary; while the relationship between the two is almost always proportional to a first approximation, some devices maintain a minimum speed when the applied duty cycle drops below some minimum value, while others will attempt to follow a speed command down to zero. Maintenance of a stable speed below roughly 20-30% of rated value is problematic for most small DC fans, and operation in that low-speed regime is not guaranteed.
In the 'scope captures in figure 11, it can be seen that the sample device used the applied PWM signal as a command for an internally-timed modulation of the supply voltage applied to the fan’s motor structure; the envelope of the current waveform retains a similar shape as in the free-running case. Note that the applied signal does not modulate the supply current directly; their frequencies differ. Also noteworthy is the amount of high-frequency noise generated on the supply network by this approach to speed control, and the effect of even a small filter capacitance in reducing it; the addition of only 4.7uF reduced ripple voltage on the supply by roughly 100mV, and supply current by about 100mA.
Figure 11. Waveforms measured from 9AH0812P4H04 using PWM speed control at 50% without (L) and with (R) a 4.7uF parallel capacitance. Trace 1(yellow) shows tachometer output, trace 2 (blue) the PWM signal input, trace 3(pink) the (ac-coupled) fan supply voltage, and trace 4 (green) the current waveform through the fan/ fan+capacitor.
DIY PWM speed control
An alternative, ‘do-it-yourself’ approach to PWM speed control of DC fans involves modulating the supply voltage applied to the fan directly at a low frequency in the neighborhood of tens of Hz, using the mechanical inertia of the fan’s rotor assembly as a filter to average the applied power and resulting in reduced fan speed. A manufacturer-provided speed control input is not required using this approach, which retains the efficiency advantages of PWM techniques compared to linear regulation of a fan’s supply voltage.
A disadvantage of this technique is that it subjects the control & commutation mechanisms in a brushless motor-based fan to ill-defined electrical stresses, and the technique tends to be frowned upon by fan manufacturers as a result. Other drawbacks include increased mechanical noise as a result of torque ripple at the PWM frequency, disrupted use of any tachometer/locked rotor output signals provided, and the need to furnish additional components. While the technique does ‘work’ in the sense that it allows fan speed to be controlled (and well enough that purpose-specific ICs are produced to facilitate this manner of use) the designer doing so assumes responsibility for any ill effect on fan reliability that results. When using this approach to speed control, limiting the rate of voltage change across the fan is suggested as a means of dealing with these disadvantages.
Figure 12. (L) A diagram of the basic DIY speed control concept and (R) an excerpt from a sample datasheet with an admonition against achieving speed control in such manner highlighted.
The majority of the AC fans sold by Digi-Key have an electrical interface consisting simply of two power supply conductors that can be connected to an AC source without regard to polarity, and a handful more (mostly those with conductive frames) employ a 3rd conductor as a protective ground. A relatively small proportion of small AC fans include control or feedback features, and because insulating them electrically from the AC supply is standard practice, devices that do incorporate them require a 4-conductor electrical interface at minimum. Many dual-voltage devices also have a 4-wire interface however, making the number of conductors with which an AC fan is equipped a less reliable proxy for inclusion of device features than in the case of DC fans.
Effect of line frequency
Most traditional AC-input fans are based on an induction motor of some flavor, and are often rated to operate from either a 50 or 60 Hz AC line frequency. Since the speed of an induction motor is closely related to the line frequency, operation from a 50 Hz source will result in a lower fan speed and a substantial ‘shrinking’ of the fan’s characteristic curve toward the origin compared to operation at 60 Hz. The newer generation of electronically-commutated AC-input fans are generally not susceptible to this effect, since they rectify the applied AC input and use it to drive what is essentially a brushless DC motor.
For traditional induction-based AC fans however, this creates a significant opportunity for confusion; a difference in line frequency causes a difference in rotational speed which causes differences in nominal air flow and static pressure ratings, and it’s not always clear (or consistent) which set of figures are being referenced. Manufacturers with good documentation will provide information for both cases; those with not-so-good documentation may rate a product for either 50 or 60 Hz operation while publishing only one set of numbers and one curve, with no clear mention of which condition applies or indication that such a thing might matter.
In the latter case, one should be mindful that nominal speed, airflow, and static pressure for AC fans with induction motors will be lower at 50 Hz than 60 Hz. It’s best to test, but if one absolutely must make an assumption about such a thing, it’s advisable to weigh the consequences of being wrong in either direction, and make the assumption on the basis of which case would prove least unfortunate.
And again–this applies to AC fans using induction motors and their derivatives only; electronically-commutated AC fans are not picky eaters as a general rule.
Figure 14. Excerpt from the SP103A-1123LST.GN datasheet, showing the shift in characteristic curves resulting from a difference in AC input frequency.