1
\$\begingroup\$

I'm trying to suppress EMF coming from a brushed DC motor that I am running. I've tried using 0.1uF capacitors across the motor leads as discussed here: https://www.pololu.com/docs/0J15/9.

It works great for reducing noise, but it makes the motor respond very differently to the same PWM power supply. Specifically, I'm running an L298 at 30kHz.

Before, I was getting my intended minimum RPM at around 50% duty cycle. The motor was also running very steadily with constant power.

With the capacitors, I am getting a much higher RPM at a lower duty cycle and I'm having trouble running the motor at a sufficiently RPM. Also, at a constant PWM, the motor will start up and gradually slow down a little bit.

Is there a better way to control noise? I can understand vaguely why a high pass filter might have issues with a high frequency PWM - but the motor lead capacitor approach and PWM both appear to be the standard in implementing small, brushed DC motors.

\$\endgroup\$
3
  • 2
    \$\begingroup\$ L298: bad idea. L298 at higher switching speads: really bad idea. \$\endgroup\$ Jul 30, 2021 at 21:30
  • 1
    \$\begingroup\$ I'm not sure why you wouldn't use an L298 - it is an older part, and it is bipolar, so it doesn't really have "on resistance." If you run your circuit as shown in Figure 6 in st.com/resource/en/datasheet/l298.pdf, keep the traces short and fat, you will be about as good as you can be in a PWM design. Pay attention to use a non-inductive, low ESR capacitor between power and ground and good diodes. As you have learned, capacitance on the motor leads and PWM don't mix. The relatively low switching speed won't hurt when you are trying to eliminate noise. Good luck! \$\endgroup\$ Jul 30, 2021 at 22:04
  • \$\begingroup\$ Which motor are you using? \$\endgroup\$ Jul 31, 2021 at 2:55

4 Answers 4

2
\$\begingroup\$

You plainly shouldn't be using a capacitor of that size at the motor with PWM. The capacitor is literally meant to smooth away current changes, and your PWM is trying to rapidly change the current. Using both makes no sense.

You'll want a flyback diode, and input-side capacitors to your switching transistors.

Also, you really want something less bad than the L298: it's really an obsolete part; it's high on-resistance means you're wasting a lot of power in the IC that would otherwise be available to the motor, and that you have to get rid of as heat. Atop of that, it's slow and doesn't integrate any good protection circuitry.

\$\endgroup\$
1
  • \$\begingroup\$ That makes sense. Are there any good guides for things like the "flyback diode, and input-side capacitors"? I chose the L298 because it's common, and I don't really need great efficiency. It seems like the TB9051FTG might be a good alternative. \$\endgroup\$
    – Celongar
    Jul 30, 2021 at 22:02
0
\$\begingroup\$

I agree that you should try smaller caps from drain to source of the FET's. Use the smallest ones that give the desired reduction in noise.

You can also use shielded cable for the phase wires. Terminate the shield to DC bus ground on the PCB and stator or motor housing on the motor. If you can't manage a fully shielded cable, you can add a ground wire running from PCB to motor housing or stator assembly (terminate the ground wire the same way as the shield).

You can also use a ferrite on the phase wires. All the conductors, including shield and ground wire if any should go inside the ferrite.

\$\endgroup\$
0
\$\begingroup\$

I would never use a cap directly across hard switched PWM. You could blow up the caps or the PWM switch. Now that you have done this and found that your noise has gone down it would be good to keep the cap. You can do this by placing inductance between the PWM drive and the motor with the 100n. The inductance does not need to be large because it only needs to divorce the 100n. Remember that the motor inductance is in mH and this smoothes the current. This added inductance can be in uH. Try 47 uH with a current rating suitable for the motor.

\$\endgroup\$
0
\$\begingroup\$

30 kHz is too high for L298, I would say 5 kHz is the sweet spot.

\$\endgroup\$
5
  • 1
    \$\begingroup\$ Can you elaborate and expand your answer? \$\endgroup\$
    – Voltage Spike
    Aug 1, 2021 at 16:11
  • \$\begingroup\$ Adm Kuznetsov - Hi, As my fellow moderator has kindly commented, your post is very short and it's unclear how it answers the question at the top of the page. This means both of your answers so far have not been of the expected depth nor specific enough to be well-received here. From the help center I linked on your previous (non-)answer, here is some guidance about writing good answers. Please also look around the site at the sort of answers getting lots of upvotes & also read here about why low-rep users cannot write comments. \$\endgroup\$
    – SamGibson
    Aug 1, 2021 at 16:24
  • \$\begingroup\$ [continued] You can Edit your answer to improve it. Thanks. \$\endgroup\$
    – SamGibson
    Aug 1, 2021 at 16:24
  • \$\begingroup\$ An L298 is fine at 30kHz. \$\endgroup\$ Aug 1, 2021 at 16:48
  • \$\begingroup\$ I used L298 a long time ago, maybe there were some improvement or newer type since then. The problem was that below 30% PWM signal, the output start current ( say, from 25% to 30% ) was not proportional at higher frequency. My application was sensitive to this kind of unstable output current. With 5kHz frequency, at the lower end of PWM , the output is more stable. \$\endgroup\$
    – user293074
    Aug 2, 2021 at 9:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.