Psım Full Bridge Feedback & Output Voltage Problem

[fethiyeli]

I've simulated my full bridge circuit via Psim but i cannot get a stable 370VDC at output. I calculated my feedback network according to Switching power supply design and Power supply cookbook. I've also used Basso's book. When i set the input voltage to lower value, output also goes down, and if i set the input to higher value, output also goes up, so feedback does not work. What's wrong with my circuit ? Could u pls help and give some advice ? I also do not know how to calculate the feedback network accuretly without having any network analyser. I've also tried to get bode plot but it will get something like 8 hours to get whole bode plot.

 

[FvM]

Before thinking about feedback compensation details, you should verify that the feedback loop is working DC-wise. I don't think it can with a 200 ohm optocouplerr load.

 

[fethiyeli]

Before thinking about feedback compensation details, you should verify that the feedback loop is working DC-wise. I don't think it can with a 200 ohm optocouplerr load.

Actually i dont remember which book i've used for calculating 200R but i will check it again. Did you mean that i need to check the feedback loop seperately from whole circuit by just giving DC ?

 

[FvM]

Did you mean that i need to check the feedback loop seperately from whole circuit by just giving DC ?

I meaned checking the DC levels in operation, but yes, you can check part of the functionality safely without operating the power stage. This is particularly interesting if a non-working feedback can damage part of the circuit.

 

[BradtheRad]

Looking at the 653k resistor above your opto led. This dropping resistor limits current to 560 uA.
Are you certain the led lights up at all, to activate the phototransistor?

It is understandable if you wanted to use a high ohm value. It does not dissipate so many watts as heat.

However the led could need a milliamp or two, depending on its efficiency. This would require a lower value for the dropping resistor (about 150k). But then you'll need a higher watt rating.

By the way, what volt level did you set the TL431 to provide at its upper terminal? 2.52 V appears at the node between the 365k and 2.5k resistors.

 

[fethiyeli]

Looking at the 653k resistor above your opto led. This dropping resistor limits current to 560 uA.
Are you certain the led lights up at all, to activate the phototransistor?

It is understandable if you wanted to use a high ohm value. It does not dissipate so many watts as heat.

However the led could need a milliamp or two, depending on its efficiency. This would require a lower value for the dropping resistor (about 150k). But then you'll need a higher watt rating.

By the way, what volt level did you set the TL431 to provide at its upper terminal? 2.52 V appears at the node between the 365k and 2.5k resistors.

There is nothing about current for opto led in Psim. I set the TL431 as 2.5V.

Actually i do not know how to calculate feedback network accurately. I am really confused that which book is good for calculating. Since I've used marty brown's power supply cookbook, presmann's switching power supply design and basso's book. Pressmann's book does not tell anything about TL431 feedback with optocoupler. Marty Brown's book is pretty good but i am really confused to calculate some equations about feedback network. Basso's book is great but it needs some parameters such as needed gain and phase margin. So I've used Marty Brown's book for calculating but according this book, one resistor in compensation network in TL431 appears as 1054Mohm which is not realistic. Could you tell me which book you have been using for calculating feedback network and compensation ? I really need that. And I've also tried to simulate whole circuit via OrCAD but OrCAD always gives an error during the simulation. Here is the schematic and error.

INFO(ORPROBE-3183): Simulation running...

** Profile: "SCHEMATIC1-Full-Bridge" [ D:\Electrical & Electronics\Power Electronics\Simulation & Schematics\Orcad\full-bridge-pspi

Reading and checking circuit

ERROR(ORPSIM-16276): Can't find library

Circuit has errors ... run aborted

See output file for details

INFO(ORPROBE-3188): Simulation aborted



I am using UC3825 but it also says as follows:

WARNING(ORNET-1018): Connection to unmodeled pin U8 pin 'RT'

WARNING(ORNET-1018): Connection to unmodeled pin U8 pin 'CT'

 

[FvM]

There is nothing about current for opto led in Psim. I set the TL431 as 2.5V.

PSIM can hardly model an optocoupler with assuming a current transfer ratio, either fixed or current dependent. Standard optocouplers have mostly CTR values up to unity (100 %).

The other problems are about reading and understanding error messages.

"Unmodelled pins" means that it's a functional model that sets the oscillation frequency in a different way, e.g. by a model parameter.

 

[BradtheRad]

There is nothing about current for opto led in Psim. I set the TL431 as 2.5V.

Actually i do not know how to calculate feedback network accurately.

The TL431 acts as a zener, the same way you would make an expanded scale meter. The aim is to accentuate changes in led brightness as the volt level goes up or down.

Try a TL431/zener level of 340V. Select the upper resistor value so that enough current goes through the led, to activate the photosensor.

The values shown below result in led current going from 0 to 8.6 mA, over a range from 340 to 400V.

 

[fethiyeli]

The TL431 acts as a zener, the same way you would make an expanded scale meter. The aim is to accentuate changes in led brightness as the volt level goes up or down.

Try a TL431/zener level of 340V. Select the upper resistor value so that enough current goes through the led, to activate the photosensor.

The values shown below result in led current going from 0 to 8.6 mA, over a range from 340 to 400V.

I tried your solution but it didnt work. Even i put simple voltage divider feedback, output voltage never sets itself to 370V. I used 1k and 147k for voltage divider. When the input voltage is 30VDC, output becomes 357VDC, when the input voltage is 43.2VDC (three batteries in series), output becomes 376VDC. So, duty cycle cannot be set. It's always fixed i think. I cannot change it accurately. By the way, i didnt use anything for pin. So, error amplifier is now being used as open loop. Here are the grahipcs;

 

[FvM]

I must confess that I overlooked the point. TL431 has a maximum anode voltage of 36V. For higher power supply output voltages, an additional voltage limiting circuit or a HV-transistor cascode must be used.

See e.g. https://www.edaboard.com/threads/143544/#post620204

 

[BradtheRad]

Even i put simple voltage divider feedback, output voltage never sets itself to 370V. I used 1k and 147k for voltage divider.

The led will not light sufficiently with these values. And if you reduce the resistors so it does light, then its brightness stays about the same over a range of 340 to 400 V.

Looking at R3-4-5-6. These receive feedback from your optocoupler. These add up to 40k. The phototransistor has to respond to the led, so the current changes sufficiently and alters the volt levels between R3-4-5-6. You will need to do some more experimenting with the optocoupler. And you need to make sure the led lights sufficiently to activate the phototransistor.

By the way, your second image shows the volt level rising, then it stops sharply and levels off. So it would look as though it is being regulated in some fashion. The first image does not act this way.

 

[fethiyeli]

The led will not light sufficiently with these values. And if you reduce the resistors so it does light, then its brightness stays about the same over a range of 340 to 400 V.

Looking at R3-4-5-6. These receive feedback from your optocoupler. These add up to 40k. The phototransistor has to respond to the led, so the current changes sufficiently and alters the volt levels between R3-4-5-6. You will need to do some more experimenting with the optocoupler. And you need to make sure the led lights sufficiently to activate the phototransistor.

By the way, your second image shows the volt level rising, then it stops sharply and levels off. So it would look as though it is being regulated in some fashion. The first image does not act this way.

I've calculated again as; resistor for opto's bjt is 200R and opto's led is 71.5k. Martyn Brown's book calculates it reverse. Resistor for opto's bjt is 71.5k and another one is 200R. So both of them are not working in Psim. I mean even i connect simple voltage divider and do not use TL431 and opto in order to get 370VDC, duty cycle remains same. The problem is duty cycle of UC3825 cannot be changed.

 

[BradtheRad]

I've calculated again as; resistor for opto's bjt is 200R and opto's led is 71.5k. Martyn Brown's book calculates it reverse. Resistor for opto's bjt is 71.5k and another one is 200R. So both of them are not working in Psim. I mean even i connect simple voltage divider and do not use TL431 and opto in order to get 370VDC, duty cycle remains same. The problem is duty cycle of UC3825 cannot be changed.

Calculations are fine and dandy...
However the component in the formula apparently is different from the one you have in the simulatior.

You must do more experimenting, and find out how the optocoupler performs..

The phototransistor needs to adopt a certain range of resistances, in order to produce sensible volt levels at R3-4-5-6.

The led needs to adopt a certain range of brightness, in order to make the phototransistor operate correctly.

Remove the optocoupler, and where the phototransistor is, substitute different resistances.
Get volt readings at pins 1, 2, 16 of U8.
See if the duty cycle changes.

 

[FvM]

I've calculated again as; resistor for opto's bjt is 200R and opto's led is 71.5k.

How did you calculate this? What's the assumed opto coupler current transfer ratio?

Did you notice that TL431 would be immediately fried when applying the high voltage output voltage to it? I'm not sure how the Psim model handels voltages beyond maxim ratings, but the circuit can't work in real hardware.

 

[fethiyeli]

Calculations are fine and dandy...
However the component in the formula apparently is different from the one you have in the simulatior.

You must do more experimenting, and find out how the optocoupler performs..

The phototransistor needs to adopt a certain range of resistances, in order to produce sensible volt levels at R3-4-5-6.

The led needs to adopt a certain range of brightness, in order to make the phototransistor operate correctly.

Remove the optocoupler, and where the phototransistor is, substitute different resistances.
Get volt readings at pins 1, 2, 16 of U8.
See if the duty cycle changes.

As i said, even when i dont use the optocoupler, duty cycle almost does not change except little bit. Thats why, i can not set the output to 370V. I have already seen the duty cycle did not change accurately at the output of the UC3825.

How did you calculate this? What's the assumed opto coupler current transfer ratio?

Did you notice that TL431 would be immediately fried when applying the high voltage output voltage to it? I'm not sure how the Psim model handels voltages beyond maxim ratings, but the circuit can't work in real hardware.

I calculated the whole calculations according to Marty Brown's Power Supply Cookbook. Ctrr was assumed as 130% max in the calculations. I dont know whether it's true or not. Actually, I am soooo confused how to calculate the feedback compensation network correctly. Also i need to simulate it but i am not able to simulate. Ofc this circuit will not work in real hardware. I've used OrCAD but it always give me very annoying errors and its nearly impossible to fix them. That's why, i have been using Psim for simulation. Here is the optocoupler in Psim;