band pass filter for inverter output current and voltage signals

[anotherbrick]

hello dear forum,

I try to design an 20 KHz 1 KW inverter for ultrasonic welding

In order to run the inverter at resonance I pick output current and voltage signals and feed them to comparators
The comparator detects zero crossing of both signals and according to them the STM32 uC increase or decrease inverter frequency
in order synchronize both signals

however the current and voltage signals itself are not pure sinus becouse of harmonics and parasites
becouse of this the comparators make more than one zero cross and this in turn mislead the software

my question ; will it help to band pass filter ( 20 KHz center ) the current and voltage signals before the comparators
what is your advice ? I think of making an active filter with opamps
what shall I be careful about - for example does phase difference of filter make an difficulty for me

thank you

 

[KlausST]

Hi,

What is your frequency range?

Maybe a low pass filter is more suitable.
Or a controlled rectifier built with analog switches. (Or a lock in amplifier)

Klaus

 

[BradtheRad]

The only method I've found to detect resonant frequency automatically is to read voltage across an ohmic resistance in the current path. (If you read voltage across a reactive component you don't catch zero current crossisngs.) The resistor being a linear component indicates voltage and current linearly, thus it catches zero current crossings.

Simulation has an op amp reading voltage across a low-ohm resistor in the LC resonant circuit. At each zero current crossing the op amp changes state. Thus it reverses bias to the H-bridge at exactly the proper moment, automatically maintaining oscillations at the resonant frequency.

 

[FvM]

At 20 kHz, current can be most efficiently sensed with a current transformer. No voltage measurement necessary, just relation of current phase to inverter switching signal, best detected with a sychronous rectifier as suggested. Current magnitude gives additional information about output power.

 

[anotherbrick]

Hi,

What is your frequency range?

Maybe a low pass filter is more suitable.
Or a controlled rectifier built with analog switches. (Or a lock in amplifier)

Klaus

hi Klaus,

I have 3 questions

can there be frequency components with lower freq than the inverter switching freq at the current and voltage signals

second what is a controlled rectifier - do you mean thyristor ? for inverter body instead of IGBTs ?

third what do you mean built with analog switches

thank you

 

[FvM]

Depending on the inverter design, there may be DC and subharmonic current components, but you don't necessarily want to measure it.

Controlled rectifier: you multiply the signal with 0 and 90 degree phase shifted square wave and low-pass filter to get real and reactive current component, respectively current phase after processing.

https://www.edaboard.com/threads/analog-amplifier-with-gain-switchable-between-1-and-1.289173

 

[KlausST]

Hi,

can there be frequency components with lower freq than the inverter switching freq at the current and voltage signals

Of course. An inverter maybe switches with 20,000 Hz, but producing 50Hz low frequeny.

Both exist. And a suitable way is to use a LPF to attenaute the higher frequencies = 20kHz switching frequency and it´s harmonics. Then the 50Hz remains.

second what is a controlled rectifier - do you mean thyristor ? for inverter body instead of IGBTs ?

third what do you mean built with analog switches

This is one: "a controlled rectifier built with analog switches"
No Thristor no IGBT. No high current. Just small signals.

Then use an analog switch (like a DG419) maybe add an OPAMP .. and you can build a signal rectifier.

Klaus

 

[anotherbrick]

Depending on the inverter design, there may be DC and subharmonic current components, but you don't necessarily want to measure it.

Controlled rectifier: you multiply the signal with 0 and 90 degree phase shifted square wave and low-pass filter to get real and reactive current component, respectively current phase after processing.

https://www.edaboard.com/threads/analog-amplifier-with-gain-switchable-between-1-and-1.289173

hi ,

I read the messages in the web page link you gave above

however I couldnot find a circuit schematic or paper in google or in "art of electronics book" about how to get real and reactive components of current signal
can you give web page adress if possible

thank you