IR2153 350 watt RMS Audio SMPS with current limiter --based on Borysgo2 design--

Silvio

Well-known member
Dear Silvio,

I have built the SMPS, but the current limit circuit seems to be working incorrectly in my case. I have placed R10=4.7K and still it does not resist to a short circuit. Transistors get burned. Also, at about 2A load current, a few volts of peak-to-peak ripple appear on the output rail.
Are there any considerations about C10 capacitor? Maximum voltage, type?

Thank you and best regards,
Mihai

Hi Naminator200. The current limit is a current limit not a short circuit protection. This way if it is set correctly with the amplifier load it will also save the amplifier by limiting the voltage to its input. I also tried to short out the output in this smps and it did the same for me. If you want a short circuit protection then another method has to be used.

The voltage ripple is always present on high load. This also depend on the input capacitance ( the higher the better ) usually we use 1uF per watt across the 320v supply. This means if you have 2 capacitors rated at 200v X 470uF to form the half bridge. The capacitance value will be half because the capacitors are hooked in series. In this case 470 /2=235uf. The ripple voltage usually varies between 5 to 10%.
 

naminator200

New member
Dear Silvio,

Thank you for your answer.
1. I am using DC loads on the output, not resistors or bulbs. I have also noticed that it cannot start with DC loads tied at the output rails. For example, if 1.5A of load on each rail are present before powering up the smps, the transistors get burned at startup. Is this normal, based on your experience?

2. The current limiter actually pulls the Vcc (pin 1) of the IR2153 to ground. If voltage at pin 1 falls below 10V, gate drive signal stops. So in your build, with current limit correctly adjusted, the output current can reach 4A?
 

naminator200

New member
Later edit:

I have replaced C10 capacitor with new one. Adjusted current limit to almost maximum from the potentiometer.
I have powered-up the board with no load, then increased DC load current to about 4A on each rail. It seems to be working ok. The heatsink is warm, but at an acceptable temperature.
There is ripple on the output with 10Vpp amplitude and 50Hz frequency. It checks your previous post.
Please see attached pictures.
IMG_20190422_1819081.jpgIMG_20190422_1819251.jpgIMG_20190422_1818529.jpg
 
Last edited:

Silvio

Well-known member
Later edit:

I have replaced C10 capacitor with new one. Adjusted current limit to almost maximum from the potentiometer.
I have powered-up the board with no load, then increased DC load current to about 4A on each rail. It seems to be working ok. The heatsink is warm, but at an acceptable temperature.
There is ripple on the output with 10Vpp amplitude and 50Hz frequency. It checks your previous post.
Please see attached pictures.
View attachment 6921View attachment 6922View attachment 6920

Some considerations. The output 50Hz ripple is inevitable. This can only be tamed down with more input and output capacitance. On the input it will be better. Usually we take as a rule of thumb is 1uF per watt of power across the 320v rail. This means that if you have 2 capacitors of 500uf X 200v to make the half bridge then the value of these capacitors will become 250uF because they are hooked in series. Do not forget music does not take a constant load and usually the peak voltage varies only a little with the music playing. In smps due to the switching frequency being 60 or 70Khz then it will fill back the output capacitors a 1000 times faster than a normal iron transformer. Due to this we use smaller value capacitors on the output.

Check also that the triac is working and shunting the input soft start resistor. You can do this by touching the PTC thermistor after running the smps with a 2 amp load on the output for a couple of minutes and then switch off. This should feel cold to the touch. If it feels hot then you do not have enough voltage arriving to the gate. In that case try to add another half turn or another full turn to the coupling loop as needed. You can also alter the resistive divider at the gate of the triac. Check the datasheet and see the voltage needed for the triac gate to operate. You should have this voltage (0.8 to 1V) or maybe a little more on the gate while smps is running.

Regarding the overload protection cap C10 You can change this to 330pF or 470pF as needed. The concept here is to sense the ripple voltage on the half bridge and as the amplitude rises with load it will create enough voltage on the base of the transistor to shunt down the supply rail. Usually we shut down the IR2153 from Pin 3 to ground. This has to be done very quickly by using an SCR or mosfet otherwise if it is done gradually as with when using a transistor the chip changes to a lower frequency. By this happening the flux in the transformer rises and it may saturate. The limiter was made this way using a transistor and pulling down pin 1 instead for the power supply to reset itself on high current peaks like when using very high volume. Otherwise what will happen if using pin 3 and scr the supply latches on overload and shuts off completely. To restart you have to switch off for 15 seconds and switch on again. This will be a pain in the a......

Setting the overload. Here you have to load the smps gradually to the peak current needed using an ampere meter on the output rails. Load both rails together. Adjust the trim pot until you see the ampere meter starting to back down. Here you will find the point of the current fold back. Note that the led will illuminate before this point is reached. Do not leave on the current limiter working more than a few seconds as the fets may blow. With an amplifier hooked up the music peaks come on intermittently and the limiter will cope well with this.

Regards Silvio
 

Silvio

Well-known member
Dear Silvio,

Thank you for your answer.
1. I am using DC loads on the output, not resistors or bulbs. I have also noticed that it cannot start with DC loads tied at the output rails. For example, if 1.5A of load on each rail are present before powering up the smps, the transistors get burned at startup. Is this normal, based on your experience?

2. The current limiter actually pulls the Vcc (pin 1) of the IR2153 to ground. If voltage at pin 1 falls below 10V, gate drive signal stops. So in your build, with current limit correctly adjusted, the output current can reach 4A?

Answer 1 That is because the current limiter is not working. The reason for this is as soon as the bulk charges up and the chip starts to oscillate (do not forget this chip has no soft start) it has to charge up the secondary capacitors and then the load itself this will be too much. If the current limit is working then its a different story because the current limit will limit the current during charge up of the secondary capacitors and then handles the load. In fact on start up you will notice the led blinks for a very short while.

Answer 2 Yes my smps could deliver 4 amps. I will post a link for you on youtube to see the load test. Mind you this was my first smps and if you read the file there is the whole story of this build and what I went through.

https://www.youtube.com/watch?v=qWtICMHrYM4&t=1s
 

Silvio

Well-known member
Dear Silvio,

Thank you for your answer.
1. I am using DC loads on the output, not resistors or bulbs. I have also noticed that it cannot start with DC loads tied at the output rails. For example, if 1.5A of load on each rail are present before powering up the smps, the transistors get burned at startup. Is this normal, based on your experience?

2. The current limiter actually pulls the Vcc (pin 1) of the IR2153 to ground. If voltage at pin 1 falls below 10V, gate drive signal stops. So in your build, with current limit correctly adjusted, the output current can reach 4A?

I would like to point out something. Using a DC electronic load will not be suitable for the adjustment of the current fold back because no matter how much the output voltage drops the load current will stay the same. In reality for a given load resistance when the output voltage of the smps drops due to the current fold back so does the current with it. With the the electronic load it doesn't as it will maintain the set current.
 

naminator200

New member
Dear Silvio,
About the ripple: I think it is ok for now. I will test it with an audio amplifier as a load. Hopefully, if it does not affect the amplifier, I accept the discussed ripple voltage.

About the triac: Thank you for the advice. You were right, one turn of wire around the transformer was not enough in order to energize the triac. The PTC thermistor was quite hot. Since I have used BTA12, Vgt=1.3V according to the specs. Now, I have placed two turns around the transformer and the thermistor is cold.

About the overload protection: I see. So shutting down via pin 3 (IR2153 oscillator capacitor charge and discharge) was not a good idea for audio, since each overcurrent spike creates a latch condition. Over current via pin 1 may be a better solution.

About adjusting the overcurrent protection adjustment: I don't have resistors as load, and I do not want to buy or improvise from light bulbs yet :). You are right about constant current mode (CC) of DC load. Current is maintained constant no matter of the voltage.
But the DC loads can also work in constant resistor mode (CR), and disipate up to 350W each.
I have made the following steps:
1. Potentiometer to maximum current
2. Power up.
3. Loads ON, at high resistance, let's say 60ohms.
4. I have decreased the resistance until approximately 11.2ohms (note that my output voltage is 45V, one secondary turn less than your design). Now, approximately 4A of current are drained from each output.
5. Turn off smps, turn OFF loads, adjust potentiometer a little anticlockwise.
6. Power up smps, loads ON.
Repeat steps 5 and 6 until LED is barely lit.
Now, OCP starts to kick in at approximately 4A.

I do not want to try powering up the smps with the DC loads ON, even in CR mode, since it might burn the fets again.
I understand that there is no efficient short-circuit protection.
The smps runs nicely under load for a long time, with my heatsink at 60 deg. C.

I have watched your video. You have worked a lot on this project. Thank you for your work and your valuable advices!
I will soon test the smps on an audio amplifier and let you know of the results.
 

Silvio

Well-known member
Regarding the over current setting the led will light up before the over current starts dropping the output voltage. This is because the current passing through the LED has to reach a certain point when it over comes firstly the auxiliary voltage and after that it has to drain also the constant current startup circuit. This would be around 20mA or maybe a bit more. Be careful don't be fooled as the actual tripping current adjusted your way may be in excess of 5 or 6 amps.

You must calculate a load on the output no more than 450 watts. You have to consider that this load is around 20% more on the input because I am assuming an efficiency of 80% in your smps. More than this the transformer may saturate and will present a dead short to the fets and they will surely blow.

Good luck
 
Top