Can't apply over 50 volts to h-bridge

Thread Starter

myil

Joined May 2, 2020
101
Hi Everyone,

I have been testing my inverter design recently. I tried to apply 150V to h-bridge as a bus voltage but it started to pull a lot of current and the 150V dropped down to 50 volts. The circuit doesn't pull much current below 50 volts. Why could this be happening? I am using ir2110 driver IC and didn't apply any dead time. Could this be the problem? If so, how can I solve that? It works well with the simulation as well.

Regards
 

Attachments

Last edited:

Thread Starter

myil

Joined May 2, 2020
101
The inductor at output filter of my inverter gets warm and makes noises.The LC filter also pulls a lot of current. If I remove the LC filter and replace a 20k resistor, it pulls low current. I have a problem with design. The switching frequency is 50khz ,the inductance is 300uh and the capacitance is 20uf. Can anybody help me with a better design ? Thanks
 

Papabravo

Joined Feb 24, 2006
14,701
The inductor at output filter of my inverter gets warm and makes noises.The LC filter also pulls a lot of current. If I remove the LC filter and replace a 20k resistor, it pulls low current. I have a problem with design. The switching frequency is 50khz ,the inductance is 300uh and the capacitance is 20uf. Can anybody help me with a better design ? Thanks
Can you define better in other than qualitative terms?
 

ronsimpson

Joined Oct 7, 2019
1,024
Your filter starts to role off at 180hz? Resonant point = 2khz? I think the 20uF on the power line seem large.
Please post more information.
Have you looked to see if the 300uH coil is at saturation.
 

Thread Starter

myil

Joined May 2, 2020
101
Your filter starts to role off at 180hz? Resonant point = 2khz? I think the 20uF on the power line seem large.
Please post more information.
Have you looked to see if the 300uH coil is at saturation.
How can I figure out if my coil at saturation? I only observed noise and temperature so far
 

Ian0

Joined Aug 7, 2020
768
keep the filter resonant frequency well away from the switching frequency and the output frequency. So choose the filter frequency as the geometric mean of the switching and output frequencies.
The filter should be LC second order, any higher order will create too much phase shift for the feedback to deal with.
If L is too large then there will be too much voltage drop across it at the operating frequency, and if it is too small there will be too much circulating current in the filter.

Is there any aspect of the design of your inverter that you haven’t contacted this forum about?
 

Thread Starter

myil

Joined May 2, 2020
101
keep the filter resonant frequency well away from the switching frequency and the output frequency. So choose the filter frequency as the geometric mean of the switching and output frequencies.
The filter should be LC second order, any higher order will create too much phase shift for the feedback to deal with.
If L is too large then there will be too much voltage drop across it at the operating frequency, and if it is too small there will be too much circulating current in the filter.

Is there any aspect of the design of your inverter that you haven’t contacted this forum about?
It works fine under low voltage. Pure sinewave I get. if I increase the input voltage, the inductor makes noise and gets warm. I added another core now. So I have two cores for inductor. I also changed the output capacitor to 1uf. But sinewave's shape changed somehow. Does it mean core saturates?
 

Attachments

Last edited:

Ian0

Joined Aug 7, 2020
768
My guess is that you would need a lot more L and a lot less C, but I suggest you run it on SPICE. Have the input be a 50:50 mark:space squarewave at your voltage and PWM frequency and observe the current in the inductor, and try various values of load.
Maybe an inductor in both live and neutral would be a good idea, or two windings on the same inductor.
 

Thread Starter

myil

Joined May 2, 2020
101
My guess is that you would need a lot more L and a lot less C, but I suggest you run it on SPICE. Have the input be a 50:50 mark:space squarewave at your voltage and PWM frequency and observe the current in the inductor, and try various values of load.
Maybe an inductor in both live and neutral would be a good idea, or two windings on the same inductor.
I think you are right about more L. I have changed my powder core to ferrite core which has around 5mh inductance. it improved it :) So I applied 150VDC, still had the sinewave. But I need to apply 340VDC at the end. The sinewave isn't good enough after 220VDC.
 

Attachments

Marc Sugrue

Joined Jan 19, 2018
167
I think you are right about more L. I have changed my powder core to ferrite core which has around 5mh inductance. it improved it :) So I applied 150VDC, still had the sinewave. But I need to apply 340VDC at the end. The sinewave isn't good enough after 220VDC.
See the filter in Post Number #29 https://forum.allaboutcircuits.com/threads/proteus-filter-design.169353/page-2 as a start point. Your exisitng output capacitor will pull more current the higher you take you voltage which is a contributory factor.

The Sine wave shape is probably as good as its likely to get without redesigning the circuitry driving your H bridge. This is because the peak of the waveform is clipping and distorting the voltage output.

Also as the zero crossing glitch is dificult to fix - IC manufacturers of power factor controllers do an awful amount of work to try and ride through this region and minimise this glitch around the zero crossing.
 

Thread Starter

myil

Joined May 2, 2020
101
See the filter in Post Number #29 https://forum.allaboutcircuits.com/threads/proteus-filter-design.169353/page-2 as a start point. Your exisitng output capacitor will pull more current the higher you take you voltage which is a contributory factor.

The Sine wave shape is probably as good as its likely to get without redesigning the circuitry driving your H bridge. This is because the peak of the waveform is clipping and distorting the voltage output.

Also as the zero crossing glitch is dificult to fix - IC manufacturers of power factor controllers do an awful amount of work to try and ride through this region and minimise this glitch around the zero crossing.
Hi Marc,

Today, I figured out that the sharp edges of my ferrite core makes short with copper wire. That is increasing the current and heating the inductor. I isolated it with a tape now. So, the problem has disappeared. But, i also had to remove snubber network which pulls a lot of current as well. I am having this problem only 50khz switching side. Do you have any recommendation to solve this problem?
 

Marc Sugrue

Joined Jan 19, 2018
167
Hi Marc,

Today, I figured out that the sharp edges of my ferrite core makes short with copper wire. That is increasing the current and heating the inductor. I isolated it with a tape now. So, the problem has disappeared. But, i also had to remove snubber network which pulls a lot of current as well. I am having this problem only 50khz switching side. Do you have any recommendation to solve this problem?
I would recommend removing the filter and snubbing for the time being and power a non inductive Resistive load. Try and get that to 300+V first. Once your there determine how many dB of attenuation you require to get the desired ripple current.

As pointed out by others on the thread if you have 20uF in the C8 position then its far too much. Assuming you are generating 50Hz then the impedance of the the Capacitor will be 160R @ 50Hz which will act as a 1.8A Load @ 300V (550+W) on your supply without a load even being present.

Go back to go forward
 

Marc Sugrue

Joined Jan 19, 2018
167
yes, it's
Is your maximum output load still 300R? I did a crude simulation of your filter and it says your targeting 50dB attenuation of 50kHz is that correct? I think to achieve this level of attenuation with an LC you maybe need to consider a multi stage filter as the components may be difficult to source for a single stage.
 

Attachments

Last edited:
Top