Choosing Filter Capacitors for DC Motor Controller

It does not make a big difference.

Statistically, using 2 instead of 1 will give you more reliability, but usually practical considerations such as availability and cost are the ones taken into account.

It won't matter either way. Electrically, two capacitors might give you lower ESR (Equivalent Series Resistance) for better filtering performance, but most brush-type DC motors don't care anyways. Packaging might dictate two smaller capacitors that fit in a lower profile space. Manufacturing cost might dictate one capacitor to reduce touch labour. Supplier availability could favour one or two capacitors. Only you can really answer these questions.

But really, it is all moot anyways.

The real question is the purpose that you think the capacitors serve. If your goal is to keep ripple from affecting the motor speed, then you don't need them since the mechanical inertia of the belt and rollers will dampen any electronic ripple.

Maybe acoustic noise? But usually the belt and belt drive system is a lot noisier than the laminations of the motor.

Whether you use 1 or 2 capacitors of the same total capacitance should make no difference.

By the way, don't forget the reverse biased diode in parallel with the motor. Otherwise, the motor's inductance will give you some nasty voltage spikes when you chop the input current.

https://electronicsforu.com/electronics-projects/hardware-diy/speed-control-dc-motor-using-pwm

There is some difference, this is data for BHC Aerovox 200VDC caps, ESR milliohm. Sorry, CR4 is bad at keeping neat spacings in tables - it takes out one's tabs & spaces.

μF, ESR[100Hz], Ripple Current[100Hz, 10kHz]

1500, 90, 6.7 amp, 9.1 amp

3300, 48, 11 amp, 13 amp

2 x 1500 has slightly lower ESR, distinctly higher ripple current compared 3300. Also the bigger caps (by case size) have longer life - 11000h for 1500, 18000h for 3300 @ rated 85'C with rated voltage & ripple. Losses for 3300@13amp ~I²r = 169 x 0.048 = 8.1 watts

The ripple currents are at 85'C ambient, x 2.1 @ 50'C.

Do not forget you have bridge pulsing away at 120 Hz & FET @ 10 kHz, losses probably additive.

If you do not use caps, ripple fed back into mains could interfere with others. impedance of 3300 at 10 kHz is 40 mohm, x 20 amp = 0.8V. What it will be without cap?? There is no assurance of what mains impedance will be at 10 kHz (high impedance could make behaviour of FET/motor unpredictable), but CISPR give values of 50-200 ohm for standard test "line simulation" networks to feed equipment under test [honestly, this is a matter of standardisation - surveys find values of ohms to kilohms fluctuating wildly with frequency from 10 kHz up].

A brief look at EMC regulations found VDE required max 10 mV @ 10 kHz emission to mains from 250W battery charger effective 1982. Schaffner example shows it required filters in both input & output leads to meet EMC emission standard.

Thank you. I will use just one capacitor, 2200uF/250V.

Now I'm working on the PCB prototype. It's my first time designing a PCB which includes high and low power devices together. I'm analysing the traces width/thickness for the high power side. I'm using Diptrace. For now I've just placed the components. I can create another post on this for further help.

I will just add that without the reservoir caps the noise fed back into supply could interfere with your own AC PSU/MCU - PSU & controller much closer to noise than other users of mains. Possible you get trouble even with filter caps.

I get your point. I've been looking into others DC motor treadmills controllers and they always include one or two filter capacitors. I always try to do a research on the topic I'm working and end up getting confused as I find opposite answers every time. I have even got a bit discouraged while researching, I've seen people who tell:

"Stop what you're doing, don't deal with mains"

"You can buy a cheap DC motor controller on Ebay"

"Why doing what is already invented?

But as for my viewpoint, I'm just trying to develop my own controller with educational purposes. I don't mind if it's invented or not. I don't mind I have to pay way more to develop something it's already invented and can be bought at low price. That's not my priority. With time and experience in electronics design, then I can try to do something new.

Back to technical stuff, I was wondering if I rectify the mains and filter it, and I get a peak voltage of 170V. But my DC motor is 125V rated, and many others are rated at 90VDC. That way, I would have to use a very limited duty cycle range, right?

The 125 and 90 v dc motors are typically used with phase controlled SCR bridge.

approx 110vac x .85 = 94 volts DC is a phase controlled SCR bridge output.

Yeah, well, why play football or tennis? It has all been done before!! And those cricket & baseball bats & balls are dangerous!!

According to my calculation 115VAC rectified will give 160V peak, with 120Hz ripple which falls to about 130V with 2200 μF reservoir & 10 amp load, mean value say 145VDC.

The speed of the motor will depend on mean rotor voltage [permanent magnet like a shunt motor with fixed excitation].

Suppose you want a speed of 2500 rev/min from motor 125V/5000 rev/min nominal. That will require 62V - the duty cycle δ must be ...

δ = 62/145 = 0.43 = t_on/(t_on + t_off)

With (t_on + t_off) = 100 μs, the on time would be 43 μs.

This would be modified according to torque loading and voltage drop in motor resistance. To get 5000 rev/min, 125V mean would have to be applied to motor (mean supply about 130V @ 20 amp load, ripple down to 100V); δ ~ 1.0.

500 rev/min would need δ ~ 12.5/170 = 0.07 - even less at small load since IR drop in motor at full amps may be 10% say, δ = 0.07 x 0.9= 0.066.

That is not a limited duty cycle range.

Note that the ripple frequency is 120Hz, while 3600 rev/min is 60 Hz - You will probably get problems with speed control close to this where motor wants to lock into ripple frequency.

without a circuit sketch one can only blow smoke.

I already have the complete schematic of the circuit. Just that I haven't upload it for this question.