On SineWave inverters.
To areza:
"1 . i believe the class d with iron core , can be scaled to 350v full bridge to eliminate the iron core, i m on the rite path ?"
Yea, sure. Why not?
"2. what do you mean by “add a discharger to the "rectified" wave “ ?"
I am adding some schematics of what i have posted previously.
On your question about the discharger, look at the image lf_bridge.GIF. See C1+L1? You must have them both. The first stage here is a "simple" SMPS, except of course, that instead of a stable DC reference like the TL431 in "normal" regulators it gets a "rectified sine", which, of course this circuit will attempt to reproduce, multiplied by 1:N of the trafo. The second bridge merely inverse rectifies this waveform at 50Hz to get a full sinewave at the output.
Remember that the DC/DC-HF side of the converter is UNILATERAL, that is it cannot DRAW current FROM C1, only to charge it. It is up to the LOAD to discharge C1 to 0V at the start and the end of each half sine.
Now, if your LOAD is resistor, you have no troubles. But if your load is reactive/nonlinear, you need to have a reliable bleed to keep the capacitor C1 in the correct voltage.
The bleeder MOSFET is opened by the comparator each time C1 charges too high, keeping nice sinewave at the load.
"3. which topology would allow open loop or less complex loop control/feedback ?"
Open loop SINEWAVE inversion? To my opinion.... impossible! A nice feedback can be by a simple low power 50Hz trafo.
"4. wanna know your opinion on “ one of the MIC topology “ named as “DIRECT DC-AC INVERTION “ outlined as , pwm modulated sine wave feed into a isolated DC/AC converter , the content is retrieved through a set of low pass filter"
What do you refer MIC? For me its Microwave-Integrated-Circuit, which means mostly a chip&wire hybrid technology, not related to power electronics at all. If you refer to the circuit with C1, i may add that it is very nice as long as you wont wenture with it more than, say 500Wrms. The bleeder starts to be heavy after this.
To areza:
"1 . i believe the class d with iron core , can be scaled to 350v full bridge to eliminate the iron core, i m on the rite path ?"
Yea, sure. Why not?
"2. what do you mean by “add a discharger to the "rectified" wave “ ?"
I am adding some schematics of what i have posted previously.
On your question about the discharger, look at the image lf_bridge.GIF. See C1+L1? You must have them both. The first stage here is a "simple" SMPS, except of course, that instead of a stable DC reference like the TL431 in "normal" regulators it gets a "rectified sine", which, of course this circuit will attempt to reproduce, multiplied by 1:N of the trafo. The second bridge merely inverse rectifies this waveform at 50Hz to get a full sinewave at the output.
Remember that the DC/DC-HF side of the converter is UNILATERAL, that is it cannot DRAW current FROM C1, only to charge it. It is up to the LOAD to discharge C1 to 0V at the start and the end of each half sine.
Now, if your LOAD is resistor, you have no troubles. But if your load is reactive/nonlinear, you need to have a reliable bleed to keep the capacitor C1 in the correct voltage.
The bleeder MOSFET is opened by the comparator each time C1 charges too high, keeping nice sinewave at the load.
"3. which topology would allow open loop or less complex loop control/feedback ?"
Open loop SINEWAVE inversion? To my opinion.... impossible! A nice feedback can be by a simple low power 50Hz trafo.
"4. wanna know your opinion on “ one of the MIC topology “ named as “DIRECT DC-AC INVERTION “ outlined as , pwm modulated sine wave feed into a isolated DC/AC converter , the content is retrieved through a set of low pass filter"
What do you refer MIC? For me its Microwave-Integrated-Circuit, which means mostly a chip&wire hybrid technology, not related to power electronics at all. If you refer to the circuit with C1, i may add that it is very nice as long as you wont wenture with it more than, say 500Wrms. The bleeder starts to be heavy after this.
To Golam
Thank you for your help.
Back in 1996-7 i tried to make a 24VDC->230VAC inverter. I built it by using a PWM peripheral of 4MHz PIC16C72 modulated by 100 steps half sine and synchronized to the 50Hz mains by the software. The actual PWM was SG3524 with an AC coupled Full-Bridge of, 60V MOSFETs driven at 50KHz by 2*IR2110. The uC was used to supply a rectified half sines, RC filtered, and fed to an error-amp of this SG3525. It also commanded a 50Hz IRF840 bridge to invert the half sinewaves. I used 3C85 ETD44 or ETD49 (...don't remember) transformer with silicone-laminations inductor at the output.I partially got it working, but cancelled it all, after a while, by lack of my boss interest.
High efficiency power electronics was always one my favorite interests, because it is so complex and such a cruel and unforgiving challenge. But now, as the energy crisis is upon us, humans, we, the power electronics experts can do a lot for everyone. Leave alone Al-Jazeera false horror stories garbage. So now i am interested in inverters also from their "green" aspects.
And you actually made a working unit. So i am obviously curious about it!
I am a little confused about the uC issue. You say, that you made a 90 degrees portion of a sine by 256 points. Thats 0.35 degree resolutuion! But your actual PWM was 4 bits, thats 16 combinations. For a 310Vpp sinewave you have 19.375V steps. I am kinda confused about the very high phase resolution against so crude an amplitude resolution.
Is that what made your THD=7%?
Also 4 bit PWM times 50KHz is 16*50e3=800KHz not 8MHz as the chips clock...
Also WDT is merely a primitive prescaled timer that needs to reset every X mS. Failure to do so, makes the WDT to roll-over, resetting the uC by this. It cannot check if a FLSH table has been written... or i am missing something here?
Why do you need to rewrite the sine into flash? I am kinda thought, that self rewritable FLASH is very good for distant programming upgrade by internet or so...
The stated endurance of program FLASH is 100K times. Doing so 50 times per second gives 2000 seconds lifespan for the inverter. I am most probably missing here something.
Regarding my own stuff, i am interested in 3-phase space vector modulated Inverters an Rectifiers. Especially the algorithmic aspects that can be used by some uC, or, even better for me, an FPGA.
Also i still prefer the analog PWM chips like UCC2800 or UCC2808 series. They have an infinite resolution in terms of output voltage, and are very reliable.
I kinda cannot understand the very usage of voltage-mode-control like SG3525 in push-pull, like so many statements here...
And also i am very concerned about using a uC with PID algorithms, when, computing more than one output, it creates a huge REAL-TIME multiprocessing burden, and the bugs are just sitting there, between the lines of my very code, waiting for the opportunity to spoil my day.
In that, respectfully, me and you, as engineers, are different indeed in our views!
For me, uC management of SMPS on the other hand is a most welcome blessing. Using stuff like 16F884 with 13 10-bit A/D is very easy and saves a lot.
All the best.
Alex.
Thank you for your help.
Back in 1996-7 i tried to make a 24VDC->230VAC inverter. I built it by using a PWM peripheral of 4MHz PIC16C72 modulated by 100 steps half sine and synchronized to the 50Hz mains by the software. The actual PWM was SG3524 with an AC coupled Full-Bridge of, 60V MOSFETs driven at 50KHz by 2*IR2110. The uC was used to supply a rectified half sines, RC filtered, and fed to an error-amp of this SG3525. It also commanded a 50Hz IRF840 bridge to invert the half sinewaves. I used 3C85 ETD44 or ETD49 (...don't remember) transformer with silicone-laminations inductor at the output.I partially got it working, but cancelled it all, after a while, by lack of my boss interest.
High efficiency power electronics was always one my favorite interests, because it is so complex and such a cruel and unforgiving challenge. But now, as the energy crisis is upon us, humans, we, the power electronics experts can do a lot for everyone. Leave alone Al-Jazeera false horror stories garbage. So now i am interested in inverters also from their "green" aspects.
And you actually made a working unit. So i am obviously curious about it!
I am a little confused about the uC issue. You say, that you made a 90 degrees portion of a sine by 256 points. Thats 0.35 degree resolutuion! But your actual PWM was 4 bits, thats 16 combinations. For a 310Vpp sinewave you have 19.375V steps. I am kinda confused about the very high phase resolution against so crude an amplitude resolution.
Is that what made your THD=7%?
Also 4 bit PWM times 50KHz is 16*50e3=800KHz not 8MHz as the chips clock...
Also WDT is merely a primitive prescaled timer that needs to reset every X mS. Failure to do so, makes the WDT to roll-over, resetting the uC by this. It cannot check if a FLSH table has been written... or i am missing something here?
Why do you need to rewrite the sine into flash? I am kinda thought, that self rewritable FLASH is very good for distant programming upgrade by internet or so...
The stated endurance of program FLASH is 100K times. Doing so 50 times per second gives 2000 seconds lifespan for the inverter. I am most probably missing here something.
Regarding my own stuff, i am interested in 3-phase space vector modulated Inverters an Rectifiers. Especially the algorithmic aspects that can be used by some uC, or, even better for me, an FPGA.
Also i still prefer the analog PWM chips like UCC2800 or UCC2808 series. They have an infinite resolution in terms of output voltage, and are very reliable.
I kinda cannot understand the very usage of voltage-mode-control like SG3525 in push-pull, like so many statements here...
And also i am very concerned about using a uC with PID algorithms, when, computing more than one output, it creates a huge REAL-TIME multiprocessing burden, and the bugs are just sitting there, between the lines of my very code, waiting for the opportunity to spoil my day.
In that, respectfully, me and you, as engineers, are different indeed in our views!
For me, uC management of SMPS on the other hand is a most welcome blessing. Using stuff like 16F884 with 13 10-bit A/D is very easy and saves a lot.
All the best.
Alex.
The lower the number of turns, the higher the frequency has to be.
The higher the number of turns, the lower the frequency can be.
Otherwise there can be problems of core saturation, pulse imbalance and improper output.
Actually switched mode system was patronized by NASA for their requirement of small power supply for their spacecraft. That's why high frequency low number of turns high power SMPS has evolved.
Reply to Alexsch
Probably I could not make you understand regarding my sine wave modalities. I am attaching a diagram to make you understand.
Pic 16F884 is a beautiful piece of Microcontroller. It is of 40 pins. Due to programming difficulties like page changing etc. and specially writing data table like sine table in programme memory I have shifted to 18F series. As the program counter low(pcl) is of 8 bit it is very difficult to write more than 256 bytes of data in programme memory in 16F Series.
For small program, I use pic 12F675/683 of 8 pins, 16F884/888 for 14 pins, 18F1320 for 18 pins and 18F4550 of 40 pins. Earlier for 40 pins, I used pic 16F 877A.
There is problem in making sine wave inverter with 16F series. PCL is 8 bit, moreover max. frequency is 20 MHz, so your ADC result conversion is slower and lot of other difficulties which you will face while using. All these difficulties are not there in 18F series. So, using more than 14 pins, I switch to 18F series.
Sine table is not of 4 Bits. It is of 8 Bits. To offset 8 Bit PCL of 16F series, 8 bit-that is 256 combination in total is used different way 4 times in 4 quarters of a full cycle. Shown in the sketch attached in detail.
As I have used 16F Series Pic, I resorted to 8 Bit resolution for my sine wave table. Thats why efficiency is < than 90% and harmonic distortion is almost 7%. I just used my programs in 18F 1320 without modification.Now I am modifying my sine table to 10 Bits as I can use the whole program memory in 18F for storing Data. So, I will get 1024 combination and total sine table will be 1024x4. I have tested the 1st prototype with new program and found that efficiency can be as high as 94%. I require more 1 month to finalize the new one.
I have not used internal 8 MHz oscillator - rather I used 40 MHz pLL oscillator for getting my ADC result in 100nsx12 TAD=1.2 Micro sec.
Since I used 40 MHz oscillator, my instruction cycle is 100 nano sec. I used 50KHz freq for my inverter, so each pulse for inverter is taking 10 micro second for half cycle. After each PWM- ed pulse from ECCp, before providing next pulse, checking /rechecking/error calculation/modifying error count, Adc reading adjustment etc. required only 5 micro sec and doing every thing ready, just after each 10 micro sec, PWM signals are given by ECCP module to the gate of Mosfets through drivers if you use those at all and you are getting half of a cycle and with this operation ultimately you are archiving what you are intending for.
WDT in 18F series is different than 16F series and you can use it in 101 way if you use. Here, after each predefined time schedule, WDT goes to a Macro intended for it and according to that Macro some parameters will be ascertained and this checking requires only 2.6 micro second and it will not clash with normal calculation with in 10 micro sec time. WDT is not doing anything, every thing is done by Programmer but WDT is used as a tool for achieving his goal.
I tried to make you understand. I am still having lot of shortcomings regarding Power electronics and Microcontroller. I am trying utmost to overcome and suggestions from all members are welcome.
I think the sketch will help you to overcome some of your doubts.
Probably I could not make you understand regarding my sine wave modalities. I am attaching a diagram to make you understand.
Pic 16F884 is a beautiful piece of Microcontroller. It is of 40 pins. Due to programming difficulties like page changing etc. and specially writing data table like sine table in programme memory I have shifted to 18F series. As the program counter low(pcl) is of 8 bit it is very difficult to write more than 256 bytes of data in programme memory in 16F Series.
For small program, I use pic 12F675/683 of 8 pins, 16F884/888 for 14 pins, 18F1320 for 18 pins and 18F4550 of 40 pins. Earlier for 40 pins, I used pic 16F 877A.
There is problem in making sine wave inverter with 16F series. PCL is 8 bit, moreover max. frequency is 20 MHz, so your ADC result conversion is slower and lot of other difficulties which you will face while using. All these difficulties are not there in 18F series. So, using more than 14 pins, I switch to 18F series.
Sine table is not of 4 Bits. It is of 8 Bits. To offset 8 Bit PCL of 16F series, 8 bit-that is 256 combination in total is used different way 4 times in 4 quarters of a full cycle. Shown in the sketch attached in detail.
As I have used 16F Series Pic, I resorted to 8 Bit resolution for my sine wave table. Thats why efficiency is < than 90% and harmonic distortion is almost 7%. I just used my programs in 18F 1320 without modification.Now I am modifying my sine table to 10 Bits as I can use the whole program memory in 18F for storing Data. So, I will get 1024 combination and total sine table will be 1024x4. I have tested the 1st prototype with new program and found that efficiency can be as high as 94%. I require more 1 month to finalize the new one.
I have not used internal 8 MHz oscillator - rather I used 40 MHz pLL oscillator for getting my ADC result in 100nsx12 TAD=1.2 Micro sec.
Since I used 40 MHz oscillator, my instruction cycle is 100 nano sec. I used 50KHz freq for my inverter, so each pulse for inverter is taking 10 micro second for half cycle. After each PWM- ed pulse from ECCp, before providing next pulse, checking /rechecking/error calculation/modifying error count, Adc reading adjustment etc. required only 5 micro sec and doing every thing ready, just after each 10 micro sec, PWM signals are given by ECCP module to the gate of Mosfets through drivers if you use those at all and you are getting half of a cycle and with this operation ultimately you are archiving what you are intending for.
WDT in 18F series is different than 16F series and you can use it in 101 way if you use. Here, after each predefined time schedule, WDT goes to a Macro intended for it and according to that Macro some parameters will be ascertained and this checking requires only 2.6 micro second and it will not clash with normal calculation with in 10 micro sec time. WDT is not doing anything, every thing is done by Programmer but WDT is used as a tool for achieving his goal.
I tried to make you understand. I am still having lot of shortcomings regarding Power electronics and Microcontroller. I am trying utmost to overcome and suggestions from all members are welcome.
I think the sketch will help you to overcome some of your doubts.
To Golam
4 Quadrants, not 4 bits. Yes i understand now. Your 89% efficiency, in my opinion is excellent, as you actually doung a DOUBLE-CONVERSION inverter, so the efficiency figures get multiplied. Regarding 7% distortion at 8 bit PWM resolution:
8 bits per quadrant give you 1.21V steps. I think it is very fine, and (without due expirience, i admit..) guess that it can be 1-3% at the most. Respectifully, i would suggest to look deeper into the switching algorithm or the control algorithm (did you used PI? PID?) for the explanation about those 7%.
Specifically, the 4-quadrant algorithm looks unidirectional to me. In that case it can easily correct for lack of voltage on the output final filter cap, but cannot withdraw excess charge, as i posted yesterday on a different type of inverter.
PIC18F indeed have a contiguous program memory, and in PIC16F one needs to observe his look-up tables carefully (...been there, seen that). I wanted to use PIC18F44J10, but got reluctant to learn 70+ assy instructions. Maybe in future, who knows.
All the best.
Alex.
4 Quadrants, not 4 bits. Yes i understand now. Your 89% efficiency, in my opinion is excellent, as you actually doung a DOUBLE-CONVERSION inverter, so the efficiency figures get multiplied. Regarding 7% distortion at 8 bit PWM resolution:
8 bits per quadrant give you 1.21V steps. I think it is very fine, and (without due expirience, i admit..) guess that it can be 1-3% at the most. Respectifully, i would suggest to look deeper into the switching algorithm or the control algorithm (did you used PI? PID?) for the explanation about those 7%.
Specifically, the 4-quadrant algorithm looks unidirectional to me. In that case it can easily correct for lack of voltage on the output final filter cap, but cannot withdraw excess charge, as i posted yesterday on a different type of inverter.
PIC18F indeed have a contiguous program memory, and in PIC16F one needs to observe his look-up tables carefully (...been there, seen that). I wanted to use PIC18F44J10, but got reluctant to learn 70+ assy instructions. Maybe in future, who knows.
All the best.
Alex.
Alexsch
From 35 instructions in 16F, 77 instructions have been incorporated in 18F Series for the Compilers of High Level Languages like C . Most of the instructions are easy and simple to understand. Do not afraid of 77 Instructions. Electronics is a subject, which requires lot of hard work, patience and dedication. No short cut in Electronics. Instead of sticking to old and obsolete technology, adopt new and modern technology. You know better than me regarding this, as you are from a technologically advanced country.
From 35 instructions in 16F, 77 instructions have been incorporated in 18F Series for the Compilers of High Level Languages like C . Most of the instructions are easy and simple to understand. Do not afraid of 77 Instructions. Electronics is a subject, which requires lot of hard work, patience and dedication. No short cut in Electronics. Instead of sticking to old and obsolete technology, adopt new and modern technology. You know better than me regarding this, as you are from a technologically advanced country.
Confused with the WDT
Hi everybody,
I stumled upon this thread and really did not read it in entirety. But I am seriously confused with the use of WDT in the PIC18F1320. Please help.
Golam could you please clarify how you setup the WDT for a 2.6 microsecond period and how to invoke a macro at timeout.
Regards
Sougata Das,
Andig Technologies,
Kolkata,
India.
Hi everybody,
I stumled upon this thread and really did not read it in entirety. But I am seriously confused with the use of WDT in the PIC18F1320. Please help.
Golam could you please clarify how you setup the WDT for a 2.6 microsecond period and how to invoke a macro at timeout.
Regards
Sougata Das,
Andig Technologies,
Kolkata,
India.
Dear Golam,
Thanks for your information regarding sine wave inverter. I also use Pic18F1320 and trying to make a sine wave inverter with that Pic. I have understood your modalities but not clear regarding use of WDT. Did you want to mean that after a predetermined period, say- 20 mili sec, the WDT moves to a Macro and perform some checking task within 2.6 micro sec? Or what Saugata tried to mean that is correct?
Whatever your answer is, I am not clear regarding the use of WDT in Pic 18F1320 regarding making sine wave Inverter. If you get time in U.S.A., please try to make us clear. It will be of great help.
With Thanks and wishing your enjoyable journey to U.S.A.
Thanks for your information regarding sine wave inverter. I also use Pic18F1320 and trying to make a sine wave inverter with that Pic. I have understood your modalities but not clear regarding use of WDT. Did you want to mean that after a predetermined period, say- 20 mili sec, the WDT moves to a Macro and perform some checking task within 2.6 micro sec? Or what Saugata tried to mean that is correct?
Whatever your answer is, I am not clear regarding the use of WDT in Pic 18F1320 regarding making sine wave Inverter. If you get time in U.S.A., please try to make us clear. It will be of great help.
With Thanks and wishing your enjoyable journey to U.S.A.
For Areza,
If you are new in Pic Microcontroller, You should start with PIC 16F84A. This is available in Stadium market. You can try in Pacific Traders in Stadium(swimming pool) or New Mitsui in Moulana Bhasani outer stadium market. This will cost Tk. 110 to 140 each. You will get 8 pin Pic 12F 675 costs Tk. 55 to 65, 28 pin pic 16F873A costs TK.150 and 40 pin 16F877A costs Tk.250 to 300.
PIc 18F series are advanced level and I am in doubt about availability of those. I am collecting those from China.
If you are new in Pic Microcontroller, You should start with PIC 16F84A. This is available in Stadium market. You can try in Pacific Traders in Stadium(swimming pool) or New Mitsui in Moulana Bhasani outer stadium market. This will cost Tk. 110 to 140 each. You will get 8 pin Pic 12F 675 costs Tk. 55 to 65, 28 pin pic 16F873A costs TK.150 and 40 pin 16F877A costs Tk.250 to 300.
PIc 18F series are advanced level and I am in doubt about availability of those. I am collecting those from China.
Dear Areza,
I normally collect my required parts from abroad as I collect those in bulk for my production house.
But everything is available in Dhaka if you are willing to spend money and put some effort to source those.
However, from which shop you collected your Ferrite Core for 1KW smps inverter? I require the address as I require those.
From local market, I could collect only ETD 39/ECR 40 cores, not other types as your one and each piece costs Tk.55.
I normally collect my required parts from abroad as I collect those in bulk for my production house.
But everything is available in Dhaka if you are willing to spend money and put some effort to source those.
However, from which shop you collected your Ferrite Core for 1KW smps inverter? I require the address as I require those.
From local market, I could collect only ETD 39/ECR 40 cores, not other types as your one and each piece costs Tk.55.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.