LC Low pass filter design for SPWM filtering

[longcrystal]

I did simulate it with different values already.


I am not angry. I am just trying to teach one simple thing:
An LCR low pass filter is only called a "Butterworth" filter if the response is -3dB at the corner frequency. Any other LCR low pass filter is not a Butterworth filter.


Agreed. A damped filter, as you showed in post 34 is one way to deal with it. The alternative, as pointed out by Goldsmith, is to use feedback to control the response.


The current drawn by a load like that will have high levels of harmonics at frequencies well above 50Hz, so it is important for the inverter to have a low output impedance even at those frequencies. I guess that's one good reason to make the inductor smaller rather than larger in value.

hi thank for godfrey and goldsmith discussion .. i am sorry about that ... i dontknow a lot thing for the filter something... i need to study more..then will be come back and ask ~~~~~now just wondering....for the inverter design... if i predict the delay on and delay off ..to generate the pulse as i show in below

**broken link removed**

Uploaded with ImageShack.us




the off pulse i setting 0.2ms and On pulse i setting various of period 0.5ms ,1,0ms ,1.5ms , 2.4ms

then make the spwm pulse full cycle in 50HZ range........but i have asked my lecturer...my lecturer ask me back...what is ur spwm pulse frequency i said 50Hz...he said ...i am wrong...i can sense that i am wrong ..understanding...but for my own concept...

i am trying to predict ( 0.2ms X No.off pulse) + ( 0.5ms + xxxx+xxxx) =20ms
so F=1/t = 1/20ms = 50Hz <----for spwm range as the picture show.......

but the main problem is coming for my concept maybe totally is wrong.....because i refer from other website....they used triangle wave and sine wave ....combine together ....and measure the intersection for sine wave with the triangle wave.....so for my own understanding triangle wave = carrier frequency ......
sine wave = modulation frequency and the spwm is depend on the intersection for this both frequency only can draw it out.........for my case....i predict spwm 1st....but i didnt know ....how many carrier frequency for my spwm...i just can answer that i use off pulse 0.2ms and 0.5 1.0 1.5 2.4 ms do the 50 Hz pulse....but i want to understand that .....who can explain for me ...how to i find carrier frequency for the spwm.....because this is the purpose for i further data value for design filter...without this ..maybe very hard to continue ..to design LC

So i read this website for small understanding a bit ..but i am still confused...http://www.imakenews.com/eletra/mod...09&show=F,T,T,T,F,Article,F,F,F,F,T,T,F,F,T,T

so in my case i predict spwm 1st....but for the real theory in inverter..must produce triangle and sine wave 1st..only can do next step for the spwm pulse...but in my case i jump the step....because i dontknow how to generate the triangle wave in H-bridge application...and i saw some where for the concept of inverter .they used bubba oscillator to generate triangle wave..but i hope i dontwant to use this method to build out my inverter..because it look like not necessary to use this way ....So.....hope anyone can help with this ...explain simple and reasonable concept to me .......and for my case ...after spwm is LC filter i had read this low pass filter concept for cut the high frequency of my carrier into low frequency http://www.corollarytheorems.com/Design/filter.htm

and last time post the comment..godfrey is reply for me cut off frequency is higher than the 50Hz...i didnt get this meaning.. Cut off frequency should be much higher than 50Hz. If the PWM frequency is 5KHz, then the cut off frequency could be about 500Hz..i dontknow why cut in 500Hz...as i expected the output is 50Hz...why i need cut at 500Hz..but i am know u are right...maybe i didnt understand about that..hope u can explain to me simple way........

thank for help and regard

 

[goldsmith]

can answer that i use off pulse 0.2ms and 0.5 1.0 1.5 2.4 ms do the 50 Hz pulse....but i want to understand that .....

Hi Longcrystal
Do you know what duty cycle means ? your frequency is constant , just d.c has variations thus it is frequency of your SWPM ! you shouldn't say 50 hz is your SPWM frequency !

because i dontknow how to generate the triangle wave in H-bridge application

Easily ! there are several ways available ! the first one is generate an square wave with two 40106 schmitt trigger not gates . and then take an integral from that ( e.g with lf356 )
or many other ways . such as a current source and then a capacitor in loop ! it depends on you .

they used bubba oscillator to generate triangle

Yes bubba oscillator has a pretty pure sine wave , it is the reason of using that .
And an advise , if you increase your carrier frequency , your way will be easier .
Best Wishes
Goldsmith

 

[longcrystal]

Hi Longcrystal
Do you know what duty cycle means ? your frequency is constant , just d.c has variations thus it is frequency of your SWPM ! you shouldn't say 50 hz is your SPWM frequency !

Easily ! there are several ways available ! the first one is generate an square wave with two 40106 schmitt trigger not gates . and then take an integral from that ( e.g with lf356 )
or many other ways . such as a current source and then a capacitor in loop ! it depends on you .

Yes bubba oscillator has a pretty pure sine wave , it is the reason of using that .
And an advise , if you increase your carrier frequency , your way will be easier .
Best Wishes
Goldsmith

Yes i know duty cycle just simply % Duty Cycle =( pulse Width/ T period ) x100

or % Duty cycle = ( ON / ON + OFF ) X100

**broken link removed**

Uploaded with ImageShack.us



and this image show in constant pulse period ...mean constant duty cycle

**broken link removed**

Uploaded with ImageShack.us






and in my case the spwm .......is depend on ur design ...so the pulse width and period T always is different in certain frequency range......so mean that i got Various of Duty Cycle...it is not constantly duty cycle...may i misunderstanding ..or correct ???????????????????????????

so that the reason u said that....cannot represented my carrier Frequency of Spwm pulse is 50Hz.....50Hz just output range of the pulse...proportional with sine wave or spwm....


so in my case must do ??????? the triangle wave first to match with Sine wave 230VAC for the pure sine wave in the AC supply..... and use intersection of triangle wave with sine wave....to catch the intersection point to setting the spwm pulse when i need to on and when i need to off ????


but why i saw some ppl design they wont use any op-amp bubba oscillator idea and u approach for me .. use the two 40106 schmitt trigger not gates.............


For my own design ...I got two way ..either use MOSFET H-bridge with IR2112 mos driver or Transistor H-bridge with OPtocoupler..............but issit i can avoid the triangle wave method.....to make a good sine wave for my inverter ???? sorry let u too much trouble and headache..............

 

[goldsmith]

Hi again
I don't know why i can't open your files . can you attach them here using original uploader of this site , please ?

Yes i know duty cycle just simply % Duty Cycle =( pulse Width/ T period ) x100

or % Duty cycle = ( ON / ON + OFF ) X100

That's correct .

so the pulse width and period T always is different in certain frequency range

T is constant not different . perhaps i couldn't get your exact meaning ?

not constantly duty cycle...may i misunderstan

Of course D.C isn't constant for this aim . it is definition of SPWM or PWM .

the triangle wave first to match with Sine wave 230VAC

What ? i hope your mean wasn't to compare 230 v AC sine wave with a triangle wave .

but why i saw some ppl design they wont use any op-amp bubba oscillator idea and u approach for me .. use the two 40106 schmitt trigger not gates.............

Of course each designer has an optional way to design . that is one of the available ways ! you can choose many other ways ! for example abusing from a PWM driver for SMPS , to create SPWM , but i don't suggest this way for a beginner because it needs more experience .

For my own design ...I got two way ..either use MOSFET H-bridge with IR2112 mos driver or Transistor H-bridge with OPtocoupler..............but issit i can avoid the triangle wave method.....to make a good sine wave for my inverter ???? sorry let u too much trouble and headache..............

Of course you can prevent from using a triangle wave . an MCU can be used as i have notice at past . ( however i don't use MCU in my designs and if i have to use , i will ask my friends , because i'm a complete analog designer with low experience in digital . and i have a group for design process , each people with specific skill )
But you can use MCU too . by the way why headache ? these discussions are very sweet ! ha ha ! ;-)
Good luck
Goldsmith

 

[FvM]

Please compare the waveform in post #61 with the ("two-level") SPWM waveforms in your link. It doesn't look right, you have varying PWM frequency, but an almost constant duty cycle which doesn't correpond to the shown sine wave. Both are supposed to have same average value.

Most tutorial waveforms are using a low PWM frequency for clarity of the picture, e.g. 1 kHz in the linked paper. It can work for a real inverter but enforces a very low filter cut-off frequency respectively a high filter effort to remove switching frequency residuals. As previously discussed in this thread, a low filter cut-off frequency will cause large output voltage changes with varying load. So you'll prefer a higher switching frequency like 5 kHz - 20 kHz. Switching speed of available power semiconductors and respective switching losses are setting an upper bound however.

 

[longcrystal]

Please compare the waveform in post #61 with the ("two-level") SPWM waveforms in your link. It doesn't look right, you have varying PWM frequency, but an almost constant duty cycle which doesn't correpond to the shown sine wave. Both are supposed to have same average value.

Most tutorial waveforms are using a low PWM frequency for clarity of the picture, e.g. 1 kHz in the linked paper. It can work for a real inverter but enforces a very low filter cut-off frequency respectively a high filter effort to remove switching frequency residuals. As previously discussed in this thread, a low filter cut-off frequency will cause large output voltage changes with varying load. So you'll prefer a higher switching frequency like 5 kHz - 20 kHz. Switching speed of available power semiconductors and respective switching losses are setting an upper bound however.

**broken link removed**

Uploaded with ImageShack.us

Please view this picture i attached .....maybe i wrong just correct to me ..

 

[FvM]

For various reasons, usual sine pwm uses constant pwm frequency respectively period time.

 

[ROCKET SCIENTIST]

Hi,
I've been following this thread for a long time. Discussions (specially between goldsmith and godfreyl) was most interesting.Thanks to them. However, I've learned many thing from this topic, but still there's a lot to go for. Well, the reason for I'm here, is I'm gonna make a class D audio amplifier which will drive a subwoofer. 200Hz bandwidth is enough for me. Last week I've designed an audio amp (consisting class A + AB push pull). It's working nice but low frequency response isn't good enough. So, I'm looking forward to build a class D woofer amp. I know about sampling the signal and driving mosfet with it, but I guess for the output section, this has similarities with inverters. That's why I need help. I don't know much about butterworth filters but I think this would be best. I'm gonna use 30KHz as triangular carrier.

Therefore, Experts please suggest me how to do all the needed calculations and passive component choosing. Maybe a 100dB attenuation of the carrier would give HI-FI sound. Waiting for reply. Thanks in advance.

rockeT/Scientist

 

[goldsmith]

Hi,
I've been following this thread for a long time. Discussions (specially between goldsmith and godfreyl) was most interesting.Thanks to them. However, I've learned many thing from this topic, but still there's a lot to go for. Well, the reason for I'm here, is I'm gonna make a class D audio amplifier which will drive a subwoofer. 200Hz bandwidth is enough for me. Last week I've designed an audio amp (consisting class A + AB push pull). It's working nice but low frequency response isn't good enough. So, I'm looking forward to build a class D woofer amp. I know about sampling the signal and driving mosfet with it, but I guess for the output section, this has similarities with inverters. That's why I need help. I don't know much about butterworth filters but I think this would be best. I'm gonna use 30KHz as triangular carrier.

Therefore, Experts please suggest me how to do all the needed calculations and passive component choosing. Maybe a 100dB attenuation of the carrier would give HI-FI sound. Waiting for reply. Thanks in advance.

rockeT/Scientist

Hi rockeT/Scientist
Well , you are going to make an audio amp . a rule is calling us , that Ftr should be at least 10 time larger than Fsine . hence 200*10=2KHZ ( at least ) but 30 KHZ is very good choice !
you'll need a half bridge arrangement with two mosfet and two mosfet driver . and then a dead time creation section to prevent overlap problem . and then a 2nd order filter . ( LC ) with cut off frequency around 200 HZ . it will be easily achieved .
Now what is your problem ? i think you can easily design such a amplifier .
Best Wishes
Goldsmith

 

[ROCKET SCIENTIST]

Originally posted by:

Hi rockeT/Scientist
Well , you are going to make an audio amp . a rule is calling us , that Ftr should be at least 10 time larger than Fsine . hence 200*10=2KHZ ( at least ) but 30 KHZ is very good choice !
you'll need a half bridge arrangement with two mosfet and two mosfet driver . and then a dead time creation section to prevent overlap problem . and then a 2nd order filter . ( LC ) with cut off frequency around 200 HZ . it will be easily achieved .
Now what is your problem ? i think you can easily design such a amplifier .
Best Wishes
Goldsmith


Dear goldsmith,
Thanx for your reply. I know the rule you've mentioned. However I choose such high carrier frequency to eliminate any kind of audible hum and to reduce my inductor size. well,driving mosfet wouldn't be a great problem for me I guess, but the filter designing is. I don't feel comfortable with those highly complicated maths. I need to learn some simpler approaches. I know you're an expert, that's why I'm knocking you. I need to start from beginning. With the info of f(cutoff) and dB attenuation at a particular stopband frequency, I need to know how to chose filter order and component values. I googled but still couldn't understand much. I think you can give me a short tutorial on that.

Regards,
rockeT/Scientist

 

[goldsmith]

Well , how much is impedance of your speaker ? ( Subwoofer ?)

 

[ROCKET SCIENTIST]

I'll be having a 4 ohm speaker.

 

[goldsmith]

I'll be having a 4 ohm speaker.

Well here the rules comes !
1-you should select one of the values randomly . then going through the formula : 200 = 1/2*pi sqrt ( Lx*Cx ) then you'll find another one .
2- next step is finding the out put impedance of a filter with such values . hence zout = 0.5sqrt(L/C) then you'll see the result isn't 4 ohms ! ( that's possible ! )
3- normalization ! m= Rx/R nominal RX is out put impedance of your filter , and R nominal is what you need ! ( 4 ohms )
4- optimizing values ! Cnew= Cx *m and Lnew= Lx/m

Next step is enjoying from out put sound !!
Good luck
Goldsmith