Holy cr@p!
I've just noticed you try to make a poly-phase rectification of all 3 phases and then try to make a PFC with a single boost regulator
How do you imagine to correct the power factor of all 3 phases together if they are diode-ORed or commonly multi rectified (don't know the exact name for this technique) ??
I've just noticed you try to make a poly-phase rectification of all 3 phases and then try to make a PFC with a single boost regulator
How do you imagine to correct the power factor of all 3 phases together if they are diode-ORed or commonly multi rectified (don't know the exact name for this technique) ??
With this type of output filter you will experience high peaks in primary current, probably not reflected into secondary DC current sense. So the current sense is useless here.
Correct it into proper LC filter.
We still cannot see the control algorithm, as it is implemeted in DSP.
And maybe even more vital: why don't you sense the primary current?
Sensing secondary current the way you did it is asking for problems IMO, I do not really see how a proper current ramp could be created this way...
Correct it into proper LC filter.
We still cannot see the control algorithm, as it is implemeted in DSP.
And maybe even more vital: why don't you sense the primary current?
Sensing secondary current the way you did it is asking for problems IMO, I do not really see how a proper current ramp could be created this way...
)
The snubber is useless that way, it must be tuned to the resonant frequency of the supply capacitors the IGBTs and the transformer. One in each full bridge leg is better.
The 4.7u 400V capacitors on the supply rail are completely useless if their terminal spacing is longer than in the electrolytics (typically 22-27mm vs. 10mm). They can't provide any HF bypass with such a long spacing (inductance).
IR2112 can only drive those IGBT with buffers.
Turn off diodes are missing from one of the IR2112.
I would never test at high power something with so much ringing, it's like calling for disaster (parasitic turn on, self disturbance, etc...)
Gate and collector waveforms must be measured directly at IGBT leads with a trimmed 10:1 probe. The ground ring of the probe near the tip has to be connected to measurement ground close to tip connection. Using only the wire with the alligator clip as ground results in erratic waveforms.
Capturing several periods and looking at the waveforms that way is useless for switching analysis. The rising and falling edges have to be captured in detail, at 10ns to 100ns/div depending on switching speed.
Couldn't the microcontroller be hanging?
Such a converter can't be reliable without a primary side current limiting signal going from the full bridge to the microcontroller.
The output filter does not allow for regulation (may be intentional, though).
The 4.7u 400V capacitors on the supply rail are completely useless if their terminal spacing is longer than in the electrolytics (typically 22-27mm vs. 10mm). They can't provide any HF bypass with such a long spacing (inductance).
IR2112 can only drive those IGBT with buffers.
Turn off diodes are missing from one of the IR2112.
I would never test at high power something with so much ringing, it's like calling for disaster (parasitic turn on, self disturbance, etc...)
Gate and collector waveforms must be measured directly at IGBT leads with a trimmed 10:1 probe. The ground ring of the probe near the tip has to be connected to measurement ground close to tip connection. Using only the wire with the alligator clip as ground results in erratic waveforms.
Capturing several periods and looking at the waveforms that way is useless for switching analysis. The rising and falling edges have to be captured in detail, at 10ns to 100ns/div depending on switching speed.
Couldn't the microcontroller be hanging?
Such a converter can't be reliable without a primary side current limiting signal going from the full bridge to the microcontroller.
The output filter does not allow for regulation (may be intentional, though).
Hi Eva.
I using now drive IR2113 for 2A. The power suplly conected in a variac and work ok with 80 vac , but in 230 vac, work with low load, then I'll increase the load and the IGBTs bummm.
I wrap new Trafo, before I get 1500 watts in 230 vac, then moved to increase the Trafo power, not now reaches 700 and is defective.
Can it be?
I using now drive IR2113 for 2A. The power suplly conected in a variac and work ok with 80 vac , but in 230 vac, work with low load, then I'll increase the load and the IGBTs bummm.
I wrap new Trafo, before I get 1500 watts in 230 vac, then moved to increase the Trafo power, not now reaches 700 and is defective.
Can it be?
Hi Orcino,
Why are you sensing the current on the secondary side?
It is far better to add a current transformer in series with the
transformer primary to sense the current. In that way you can reconstruct the current ramp and use current mode control.
Output filter: why have you placed the inductance after the capacitors? In that way you will never be able to regulate the output. On top of that the converter sees basically a dead short at power up (output capacitor discharged) and you don't have a
way to switch it off because you are sensing the current after
the output caps. Move the inductor between the diodes and the caps.
Snubber: the values sounds very strange. Is the converter working in hard switching or in phase shift ZVS? If you use phase shift adding a small inductance in series with the transformer primary may help to achieve resonance with the IGBTs Cce.
Transformers: 3xE55 cores seems quite exagerated for 1.4kW. How many volts and amps you want at the output?
Primary capacitor bank: Why so many capacitance? 4*820uF/400V is really a lot for 1.4kW. Do you need a very long holdup time or what else? The 4.7uF are more or less useless, if you want replace them with 100nF X2 Class film capacitors.
PFC:
1) Don' put the fuse there!! if the rectifier brigde fails you can have a dead short between two phases without any fuse.
Use one fuse for each phase.
2) Place a small capacitor (<10uF) before L1. Don't put too much capacitance otherwise the PFC action will be useless.
3) There are two conditions for that kind of PFC to be able to work.
- It must always be in DCM (or even better in borderline mode
so that you can make it soft switching).
- The PFC output voltage shall be regulated at a voltage higher
that the input peak voltage.
4) What is the input voltage of the AC input
Assuming this it means that you maximum input line voltage must be lower that Vin<400/sqrt(2)=282Vac (phase to phase) or if you prefer 282Vac/sqrt(3)=164Vac (phase to neutral).
If you are using you PFC on europeans mains (400Vac phase/phase) you must regulate you PFC at at least
400Vac*sqrt(2)=565V.
Instead of doing this and dealing with high DC bus voltage it is better to use a buck PFC and step the voltage down.
5) PFC inductor: 80uH/45A makes monster inductor, are you sure of the values ?
Normally to "PFC" a 3phase system without neutral point you rectify the 3 phases as Orcino did and then you connect the rectifier output to a buck regulator working in DCM. This will not give you an unity PF but something like 0.95. The idea behind this is to replicate the behavior of a 3phase brigde connected to a resistive load. Since there are no regulamentations on power factor itself but only related to line harmonics a power factor corrector like this can fullfill the harmonics specification even if the input current is far from sinusoidal.
You can try it with Spice: put 3 voltage sources out of 120°, connect them to a 3phase brigde and then connect a resistor on the brigde output. Then chech the phase current and do an FFT, you will see that the harmonic content of that current is much lower than direct rectification with a capacitor connected after the brigde.
Why are you sensing the current on the secondary side?
It is far better to add a current transformer in series with the
transformer primary to sense the current. In that way you can reconstruct the current ramp and use current mode control.
Output filter: why have you placed the inductance after the capacitors? In that way you will never be able to regulate the output. On top of that the converter sees basically a dead short at power up (output capacitor discharged) and you don't have a
way to switch it off because you are sensing the current after
the output caps. Move the inductor between the diodes and the caps.
Snubber: the values sounds very strange. Is the converter working in hard switching or in phase shift ZVS? If you use phase shift adding a small inductance in series with the transformer primary may help to achieve resonance with the IGBTs Cce.
Transformers: 3xE55 cores seems quite exagerated for 1.4kW. How many volts and amps you want at the output?
Primary capacitor bank: Why so many capacitance? 4*820uF/400V is really a lot for 1.4kW. Do you need a very long holdup time or what else? The 4.7uF are more or less useless, if you want replace them with 100nF X2 Class film capacitors.
PFC:
1) Don' put the fuse there!! if the rectifier brigde fails you can have a dead short between two phases without any fuse.
Use one fuse for each phase.
2) Place a small capacitor (<10uF) before L1. Don't put too much capacitance otherwise the PFC action will be useless.
3) There are two conditions for that kind of PFC to be able to work.
- It must always be in DCM (or even better in borderline mode
so that you can make it soft switching).
- The PFC output voltage shall be regulated at a voltage higher
that the input peak voltage.
4) What is the input voltage of the AC input
Assuming this it means that you maximum input line voltage must be lower that Vin<400/sqrt(2)=282Vac (phase to phase) or if you prefer 282Vac/sqrt(3)=164Vac (phase to neutral).
If you are using you PFC on europeans mains (400Vac phase/phase) you must regulate you PFC at at least
400Vac*sqrt(2)=565V.
Instead of doing this and dealing with high DC bus voltage it is better to use a buck PFC and step the voltage down.
5) PFC inductor: 80uH/45A makes monster inductor, are you sure of the values ?
Normally to "PFC" a 3phase system without neutral point you rectify the 3 phases as Orcino did and then you connect the rectifier output to a buck regulator working in DCM. This will not give you an unity PF but something like 0.95. The idea behind this is to replicate the behavior of a 3phase brigde connected to a resistive load. Since there are no regulamentations on power factor itself but only related to line harmonics a power factor corrector like this can fullfill the harmonics specification even if the input current is far from sinusoidal.
You can try it with Spice: put 3 voltage sources out of 120°, connect them to a 3phase brigde and then connect a resistor on the brigde output. Then chech the phase current and do an FFT, you will see that the harmonic content of that current is much lower than direct rectification with a capacitor connected after the brigde.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.