My understanding is that within the audible frequencies - we don't like the annoying sound for starters and for higher frequencies the ripple is less (for current in the motor). Disadvantage is that faster switching in the MOSFETs make then hotter. One more thing - does operating at a low a frequency PWM frequency in a control loop introduce phase-lag?
Which frequency for PWM in motor control?
Started by ●October 26, 2016
Reply by ●October 26, 20162016-10-26
On 26.10.16 08.38, gyansorova@gmail.com wrote:> My understanding is that within the audible frequencies - we don't like the annoying sound for starters and for higher frequencies the ripple is less (for current in the motor). Disadvantage is that faster switching in the MOSFETs make then hotter.You should not care too much about audible frequencies in context of power dissipation in MOSFETs. MOSFETs switch within several nanoseconds if you want. With the highest audible frequencies you are far away from this. But increasing frequencies and switching speed also causes an increase of electromagnetic emission. This might cause serious problems with compliance.> One more thing - does operating at a low a frequency PWM frequency in a control loop introduce phase-lag?Not really. Usually the inertia is the dominant factor followed by the integration in the motor coils and/or the output filter. If you choose the PWM frequency that low that this is no longer true the motor would stutter. So choosing the appropriate switching frequency is mainly an engineering task. And furthermore in some cases FM provides better results than PWM because it avoids to switch the power on for very short times which itself contains very high frequencies, even if the repetition rate is low. Just take an FFT of your output to see the resulting frequency spectrum. With respect to audibility you should also be aware of subharmonics. Non-linearities and coupled resonances (including that of the control loop) may cause fractions of the PWM frequency to come into play. So even with 20 kHz PWM you might create audible results at 10 kHz or 6.7 kHz or even lower. Marcel
Reply by ●October 26, 20162016-10-26
On Wednesday, October 26, 2016 at 10:00:12 PM UTC+13, Marcel Mueller wrote:> On 26.10.16 08.38, gyansorova@gmail.com wrote: > > My understanding is that within the audible frequencies - we don't like the annoying sound for starters and for higher frequencies the ripple is less (for current in the motor). Disadvantage is that faster switching in the MOSFETs make then hotter. > > You should not care too much about audible frequencies in context of > power dissipation in MOSFETs. MOSFETs switch within several nanoseconds > if you want. With the highest audible frequencies you are far away from > this. But increasing frequencies and switching speed also causes an > increase of electromagnetic emission. This might cause serious problems > with compliance. > > > One more thing - does operating at a low a frequency PWM frequency in a control loop introduce phase-lag? > > Not really. Usually the inertia is the dominant factor followed by the > integration in the motor coils and/or the output filter. If you choose > the PWM frequency that low that this is no longer true the motor would > stutter. > > So choosing the appropriate switching frequency is mainly an engineering > task. And furthermore in some cases FM provides better results than PWM > because it avoids to switch the power on for very short times which > itself contains very high frequencies, even if the repetition rate is > low. Just take an FFT of your output to see the resulting frequency > spectrum. > > > With respect to audibility you should also be aware of subharmonics. > Non-linearities and coupled resonances (including that of the control > loop) may cause fractions of the PWM frequency to come into play. So > even with 20 kHz PWM you might create audible results at 10 kHz or 6.7 > kHz or even lower. > > > MarcelI've not heard of FM being used for motor control. What sort of FM? How to get dc with FM - what carrier?
Reply by ●October 26, 20162016-10-26
On Wed, 26 Oct 2016 09:28:49 -0700, gyansorova wrote:> On Wednesday, October 26, 2016 at 10:00:12 PM UTC+13, Marcel Mueller > wrote: >> On 26.10.16 08.38, gyansorova@gmail.com wrote: >> > My understanding is that within the audible frequencies - we don't >> > like the annoying sound for starters and for higher frequencies the >> > ripple is less (for current in the motor). Disadvantage is that >> > faster switching in the MOSFETs make then hotter. >> >> You should not care too much about audible frequencies in context of >> power dissipation in MOSFETs. MOSFETs switch within several nanoseconds >> if you want. With the highest audible frequencies you are far away from >> this. But increasing frequencies and switching speed also causes an >> increase of electromagnetic emission. This might cause serious problems >> with compliance. >> >> > One more thing - does operating at a low a frequency PWM frequency >> > in a control loop introduce phase-lag? >> >> Not really. Usually the inertia is the dominant factor followed by the >> integration in the motor coils and/or the output filter. If you choose >> the PWM frequency that low that this is no longer true the motor would >> stutter. >> >> So choosing the appropriate switching frequency is mainly an >> engineering task. And furthermore in some cases FM provides better >> results than PWM because it avoids to switch the power on for very >> short times which itself contains very high frequencies, even if the >> repetition rate is low. Just take an FFT of your output to see the >> resulting frequency spectrum. >> >> >> With respect to audibility you should also be aware of subharmonics. >> Non-linearities and coupled resonances (including that of the control >> loop) may cause fractions of the PWM frequency to come into play. So >> even with 20 kHz PWM you might create audible results at 10 kHz or 6.7 >> kHz or even lower. >> >> >> Marcel > > I've not heard of FM being used for motor control. What sort of FM? How > to get dc with FM - what carrier?I think he's referring to constant on-time, variable off-time control. -- www.wescottdesign.com
Reply by ●October 26, 20162016-10-26
On Tue, 25 Oct 2016 23:38:42 -0700, gyansorova wrote:> My understanding is that within the audible frequencies - we don't like > the annoying sound for starters and for higher frequencies the ripple is > less (for current in the motor). Disadvantage is that faster switching > in the MOSFETs make then hotter. > > One more thing - does operating at a low a frequency PWM frequency in a > control loop introduce phase-lag?While you're agonizing over tradeoffs, don't forget iron losses in the motor. I have _not_ gotten a good handle on what's to be expected from them, alas. So far I'm stuck on 10kHz, and it seems to be working well. I've had one in my office for the last three years or so that generates a 10kHz PWM and samples the current loop at 5kHz. I don't hear any whine from it, and the folks in my family with much more sensitive ears than I have don't complain either. I arrived at 10kHz by measuring the motor inductance, then more or less arbitrarily declaring that the peak-peak current sawtooth would be 50% of the maximum motor current (at least, that's what I remember declaring to myself...). It consumes about 3% of full power when it's just sitting there at zero torque (meaning zero average current, but the FETs are whizzing away at 50% duty cycle). At the moment the major tradeoff is between EMI and cooking the FETs -- we have the FET gate resistance dialed up to round off the voltage pulses, and have found a nice compromise between a hot board and an unhappy EMI testing house. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com I'm looking for work -- see my website!
Reply by ●October 26, 20162016-10-26
On Thursday, October 27, 2016 at 6:42:12 AM UTC+13, Tim Wescott wrote:> On Tue, 25 Oct 2016 23:38:42 -0700, gyansorova wrote: > > > My understanding is that within the audible frequencies - we don't like > > the annoying sound for starters and for higher frequencies the ripple is > > less (for current in the motor). Disadvantage is that faster switching > > in the MOSFETs make then hotter. > > > > One more thing - does operating at a low a frequency PWM frequency in a > > control loop introduce phase-lag? > > While you're agonizing over tradeoffs, don't forget iron losses in the > motor. I have _not_ gotten a good handle on what's to be expected from > them, alas. > > So far I'm stuck on 10kHz, and it seems to be working well. I've had one > in my office for the last three years or so that generates a 10kHz PWM > and samples the current loop at 5kHz. I don't hear any whine from it, > and the folks in my family with much more sensitive ears than I have > don't complain either. > > I arrived at 10kHz by measuring the motor inductance, then more or less > arbitrarily declaring that the peak-peak current sawtooth would be 50% of > the maximum motor current (at least, that's what I remember declaring to > myself...). It consumes about 3% of full power when it's just sitting > there at zero torque (meaning zero average current, but the FETs are > whizzing away at 50% duty cycle). > > At the moment the major tradeoff is between EMI and cooking the FETs -- > we have the FET gate resistance dialed up to round off the voltage > pulses, and have found a nice compromise between a hot board and an > unhappy EMI testing house. > > -- > > Tim Wescott > Wescott Design Services > http://www.wescottdesign.com > > I'm looking for work -- see my website!I also settled on 10kHz.
Reply by ●October 27, 20162016-10-27
On Wed, 26 Oct 2016 17:22:55 -0700, gyansorova wrote:> On Thursday, October 27, 2016 at 6:42:12 AM UTC+13, Tim Wescott wrote: >> On Tue, 25 Oct 2016 23:38:42 -0700, gyansorova wrote: >> >> > My understanding is that within the audible frequencies - we don't >> > like the annoying sound for starters and for higher frequencies the >> > ripple is less (for current in the motor). Disadvantage is that >> > faster switching in the MOSFETs make then hotter. >> > >> > One more thing - does operating at a low a frequency PWM frequency >> > in a control loop introduce phase-lag? >> >> While you're agonizing over tradeoffs, don't forget iron losses in the >> motor. I have _not_ gotten a good handle on what's to be expected from >> them, alas. >> >> So far I'm stuck on 10kHz, and it seems to be working well. I've had >> one in my office for the last three years or so that generates a 10kHz >> PWM and samples the current loop at 5kHz. I don't hear any whine from >> it, and the folks in my family with much more sensitive ears than I >> have don't complain either. >> >> I arrived at 10kHz by measuring the motor inductance, then more or less >> arbitrarily declaring that the peak-peak current sawtooth would be 50% >> of the maximum motor current (at least, that's what I remember >> declaring to myself...). It consumes about 3% of full power when it's >> just sitting there at zero torque (meaning zero average current, but >> the FETs are whizzing away at 50% duty cycle). >> >> At the moment the major tradeoff is between EMI and cooking the FETs -- >> we have the FET gate resistance dialed up to round off the voltage >> pulses, and have found a nice compromise between a hot board and an >> unhappy EMI testing house. >> >> -- >> >> Tim Wescott Wescott Design Services http://www.wescottdesign.com >> >> I'm looking for work -- see my website! > > I also settled on 10kHz.It's a nice round number. -- www.wescottdesign.com
