[time-nuts] 20221128: A question re fans and low-noise measurements / techniques

[ed breya]

Fans provide at least three kinds of noise. The audible is most annoying 
to humans, but we can adapt. The electric and magnetic noise may 
interfere with the instrument - we definitely don't want that.

If you're fixing or modifying a fan problem in a high performance, 
precision instrument, it's probably best to not stray too far from using 
the same parts and circuit design as the original. The reason is that 
you don't know - without some study and consideration - how much the fan 
interacts with or affects the rest. What you do know is that the 
original setup did its job, and the instrument (likely) met its specs.

You may be able to not only fix the problem, but also improve the 
overall performance with different fans and driving methods, but it 
takes some study to make sure.

A good thing to consider first is the location and proximity to possibly 
sensitive circuitry. If the fan is say, bolted to the back, and faces 
only power supply circuitry, and is nowhere near anything critical, then 
it's probably safe to not worry much about the magnetic fields coming 
from the motor. This allows for more options in fan choice, like AC 
versus DC.

Next is thermal - how well does it actually cool the innards? You have 
some leeway here too, since stock fans are usually chosen to work at the 
highest specified ambient temperature, and in rack mount situations. If 
your environment is more friendly, you can go with less fan air flow, 
which almost always allows for more peace and quiet. This can take a lot 
of experimenting to figure out how low you can go, and still be safe, 
cooling-wise.

Finally, you have to consider the electric power situation. Regardless 
of fan type, there will be motor ripple currents from the AC line or 
from the DC supply. These can affect the source, and also radiate EM 
fields from the wiring, so the paths may be important too - avoid the 
sensitive areas, or shield the lines properly. For AC induction motors, 
it is what it is, while for DC, lots of filtering close to the motor 
helps. You might be surprised at how much commutation current even a 
small brushless DC fan can produce. You can do some experiments to study 
it and assess the ugliness.

It may take a lot of capacitance to get acceptable ripple, especially if 
the fan supply is not isolated well from the supplies that run the 
critical circuitry. In extreme cases, a shunt regulator right at the fan 
(electrically close as possible) can make it invisible to the PS, 
leaving only the wiring between the shunt and the motor to radiate.

There's no reason nowadays to use a brushed DC or universal motor for a 
small equipment fan - leave those for hair dryers. I think virtually all 
modern small DC fans are brushless PM, and can run over quite a speed 
range. These are best for most equipment. For reduced speed, lower 
noise, try fixed lower voltages, and be sure that the motor will 
properly kick-start from off or stall (these are two different 
situations, even though the speed is zero) under all conditions.

For variable speed, analog is of course cleanest, but the power 
dissipation may become an issue, depending on fan size. PWM is more 
efficient, but adds another layer of EMC consideration. I always try to 
avoid adding SMPS/PWM circuits in low noise gear, but in many cases, the 
stuff is already powered by such, so adding a little more isn't 
necessarily bad. The trick is to figure out what problems may arise, and 
how to handle it.

One more thing about AC fans. Most of my old gear has the classic 
pancake AC fans, loud as hell. In some, I have changed to lighter gauge 
fans (the usual solution), or modified the circuits to run them slower. 
It all depends on the particular fan models. I test fans on the curve 
tracer or variac to see how they do at low or variable speeds. Some 
models are awesome, able to reliably start and run over a huge 
voltage/speed range, while others are awful, unable to start properly at 
all, then go flakey and wind up to jet engine noise level once the 
voltage is high enough. The fans that do reasonably well or great are 
marked and reserved for low speed transplants back into equipment

Ed