Hi, everyone. I'm building a 110VAC-12VDC 120W Flyback converter using UC3842, and have some problems with the start-up circuit and output PWM signal.
In the beginning, the design spec at first:
Input: 110VAC
Output:12VDC 10A
Switching frequency: 107kHz
Transformer core: PQ35/35 (unknown manufacture, but reference the TDK datasheet to design)
Ae:196mm2
The turns number of primary, secondary, and auxiliary is 33:4:5
Switch: P20NM60 NMOS 600V/20A/0.25Ohm
Diode: STPS30150 Power Diode 150V/30A/0.69V
I reference the datasheet schematic and other friend's circuit, draw the schematic like this:
Schematic pdf
edit 20/12/16
P.s Modified as #14
New issue #17 about the PWM duty cycle problem

And the PCB layout looks like this:


And the whole board looks like this:



I'm facing two problems:
1. The start-up circuit
I'm using a 50kOhm 1/2W resistor as a start-up resistor, and an 18V Zener diode to limit the Vcc voltage of IC.
There is two 100uF capacitor parallel to maintain the Vcc voltage.
Thought I think it can handle the power consumption of the UC3842, but actually, the IC repeat startup and shutdown.
The voltage of the IC is like below, CH1 is the oscillator of UC3842, which is the Rt1 and Ct1 pin in the schematic, and CH2 is the Vcc pin of UC3842.
As you can see, the IC repeat start-up and shutdown means the IC can not start switching the input voltage, so auxiliary winding is not working.
The datasheet of UC3842 says that the start-up current of IC is just 1mA, so can anyone help me check the circuit and find out where the problem is?



2. No PWM output
To bypass the start-up circuit problem, I try to use another Lab power supply to provide the UC3842 power.
Like the picture below, the IC seems to work fine because the reference voltage of IC is established,
so the oscillator works like the picture below.



But the output of the PWM pin is still not working.
So I check out the layout and components, there is nothing wrong.
Then I try to build the circuit on a breadboard, which contains Rt1, Ct1, and Vcc.
The UC3842 is also no PWM output, too.

Conclusion:
Is the start-up circuit of my schematic any wrong or any clue to find out the problem?
Why the PWM is not working even I import the external DC power supply with no current limit?

Very thanks for your reply
Have a nice day.

 

Hi, everyone. I'm building a 110VAC-12VDC 120W Flyback converter using UC3842, and have some problems with the start-up circuit and output PWM signal.
In the beginning, the design spec at first:
Input: 110VAC
Output:12VDC 10A
Switching frequency: 107kHz
Transformer core: PQ35/35 (unknown manufacture, but reference the TDK datasheet to design)
Ae:196mm2
The turns number of primary, secondary, and auxiliary is 33:4:5
Switch: P20NM60 NMOS 600V/20A/0.25Ohm
Diode: STPS30150 Power Diode 150V/30A/0.69V
I reference the datasheet schematic and other friend's circuit, draw the schematic like this:
Schematic pdf

And the PCB layout looks like this:
View attachment 224603

And the whole board looks like this:

View attachment 224607

I'm facing two problems:
1. The start-up circuit
I'm using a 50kOhm 1/2W resistor as a start-up resistor, and an 18V Zener diode to limit the Vcc voltage of IC.
There is two 100uF capacitor parallel to maintain the Vcc voltage.
Thought I think it can handle the power consumption of the UC3842, but actually, the IC repeat startup and shutdown.
The voltage of the IC is like below, CH1 is the oscillator of UC3842, which is the Rt1 and Ct1 pin in the schematic, and CH2 is the Vcc pin of UC3842.
As you can see, the IC repeat start-up and shutdown means the IC can not start switching the input voltage, so auxiliary winding is not working.
The datasheet of UC3842 says that the start-up current of IC is just 1mA, so can anyone help me check the circuit and find out where the problem is?

View attachment 224605

2. No PWM output
To bypass the start-up circuit problem, I try to use another Lab power supply to provide the UC3842 power.
Like the picture below, the IC seems to work fine because the reference voltage of IC is established,
so the oscillator works like the picture below.

View attachment 224606

But the output of the PWM pin is still not working.
So I check out the layout and components, there is nothing wrong.
Then I try to build the circuit on a breadboard, which contains Rt1, Ct1, and Vcc.
The UC3842 is also no PWM output, too.

Conclusion:
Is the start-up circuit of my schematic any wrong or any clue to find out the problem?
Why the PWM is not working even I import the external DC power supply with no current limit?

Very thanks for your reply
Have a nice day.

Hi,
Its not a great startup circuit. You'll need to support the 20mA needed for the UC3842 via the Rstart until you get to a steady state on the Vaux. A better solution would be to consider using a transistor to regulate the VCC and switch off the primary side feed once the Vaux is established.

 

I would build the circuit in the datasheet first and then modify it.



If there is no pwm then it won't self start, so power it up from a bench PSU, and fix the pwm first, then try it on mains.

Ps that is a terrible drawn circuit diagram.

 

The start up circuit is what I have used many time. The VCC supply should slope up to 16V, the IC starts working, slopes down to 10V and stops working, repeat. What should happen is winding pin 7-9 should hold up VCC.
1. Maybe the phase of aux winding is backwards.
2. Maybe duty cycle is very low.
3. Maybe bad part.
Almost any thing wrong will cause what you are seeing.

The power line is dangerous. First time power up: I connect a 17V supply to VCC and no power line! See if the IC looks good. PWM=very high duty cycle, Vref=5V, Comp=5V.
Then I connect a power supply to C2 and slowly bring up the voltage. Even 10V will work. (no load or small load) Q1 should fly back. Check the output voltages. They should be small and check the phase on the transformer. Because I have 0-400V power supplies I can safely bring up the supply. If all you can do is supply 25V, you can see much of the power parts start to work.

 

I think this is your first time PCB for power line. There are many problems.
Check what is the spacing for 220vac power line to line. Check where I made arrows.
Also; C4 will get charged up to 200V on top of the charge on C3. Maybe 500V at D2+. Copper spacing on D2 looks small for 500V.

You must keep wide spacing between Primary and Secondary.
U2 should not be "normal" but "wide". Never have traces under the isolator. You need spacing of the width of the isolator between P and S!


Here is a example. Blue line shows P to S isolation. Red arrows show where there is too small room so they added a slot in the board to increase voltage rating.

 

Another example of Primary Secondary spacing. Bottom side of board there is a open area to the right. Top side the open area is under the isolator and transformer.

 

Is C4 rather too large? It will have to charge up to the flyback voltage before the output and therefore the auxiliary winding can start.
Try using a 200V transil or a BZT03C200 zener instead of C4 and R3

 

Is C4 rather too large? I

nF not uF It is OK. It will charge up by 2A in a big hurry.

 

Hi,
Its not a great startup circuit. You'll need to support the 20mA needed for the UC3842 via the Rstart until you get to a steady state on the Vaux. A better solution would be to consider using a transistor to regulate the VCC and switch off the primary side feed once the Vaux is established.

I'll replace the resistor with a smaller one, thanks for your reply!
But I try to use a bench power supply and the IC didn't work, too

 

If there is no pwm then it won't self start, so power it up from a bench PSU, and fix the pwm first, then try it on mains.

Yes, I had tested the circuit with a bench power supply, but the IC still can not work, though the reference voltage and oscillator is running (as the picture shows)
Thanks for your reply!

 

The start up circuit is what I have used many time. The VCC supply should slope up to 16V, the IC starts working, slopes down to 10V and stops working, repeat. What should happen is winding pin 7-9 should hold up VCC.
1. Maybe the phase of aux winding is backwards.
2. Maybe duty cycle is very low.
3. Maybe bad part.
Almost any thing wrong will cause what you are seeing.
The power line is dangerous. First time power up: I connect a 17V supply to VCC and no power line! See if the IC looks good. PWM=very high duty cycle, Vref=5V, Comp=5V.
Then I connect a power supply to C2 and slowly bring up the voltage. Even 10V will work. (no load or small load) Q1 should fly back. Check the output voltages. They should be small and check the phase on the transformer. Because I have 0-400V power supplies I can safely bring up the supply. If all you can do is supply 25V, you can see much of the power parts start to work.

Your test step is a very good example for me to follow! I'll try to do it.
As you mention, there are too many things that make the PWM not working,
and I checked the winding polarity is correct, but not sure is there any part damaged(they should be brand new components from the storehouse)
I'll check all of the components including the MOSFET and DIODE.
Thanks for your reply!!

 

I think this is your first time PCB for power line. There are many problems.
Check what is the spacing for 220vac power line to line. Check where I made arrows.
Also; C4 will get charged up to 200V on top of the charge on C3. Maybe 500V at D2+. Copper spacing on D2 looks small for 500V.

You must keep wide spacing between Primary and Secondary.
U2 should not be "normal" but "wide". Never have traces under the isolator. You need spacing of the width of the isolator between P and S!

Here is a example. Blue line shows P to S isolation. Red arrows show where there is too small room so they added a slot in the board to increase voltage rating.

Thank you to remind me of the clearance of high voltage and the isolator!!
Yes, this is my first time PCB for power line, and after your suggestion, now I check my layout seems stupid
I'll redesign the whole circuit and check again.

 

nF not uF It is OK. It will charge up by 2A in a big hurry.

You're right - 470nF is bigger than one normally sees, but not that big. The greater stored energy is in the output smoothing so that will be the slowest one to reach operating voltage.
But I think I have spotted the problem: the polarity of the feedback is reversed. Pin 2 (FB) should go higher as the output goes higher. In this circuit is goes lower. With no output, FB will be at V+ so the output will be switched off.

The UC3842 is so old that it was invented before opto-isolator feedback became popular, so it doesn't seem to be designed with that it mind. As its error amplifier is an OTA it's probably best to connect the opto between COMP and V-, and tie FB to ground. That avoids making another pole inside the feedback loop and the whole thing going unstable.
see https://www.ti.com/seclit/ml/slup340/slup340.pdf

 

the polarity of the feedback is reversed.

Test: No ac power, Power UC3842, add 11V to the output voltage. Slowly increase Vout to above 12V. You should see the feed back parts respond at 12V.
When 12V is low, there will be less than 1mA flow in R5. When 12V is high there will be about 10mA in R5.
VR1 is unusual. I do not use VCC but Vref (5V) with 10k resistor to pull Vfb up. So in my circuit U2-4 is 5V to 0.4V.
This part is a simple gain of 5 amplifier. It is trying to regulate at 2.5V on Vfb.

You are right. Backwards!

 

The UC3842 is so old that it was invented before opto-isolator feedback became popular

I was using optos before the UC3842 was invented.
I got prototypes of the UC384x and had supplies working before we could get real parts. We used 1 million parts a month for years. When we needed more parts than could be made we helped Samsung make a clone. The old UC3842 is much better than the older TL494.

 

Very appreciate all of your replies!

The problem is as @Ian0 said,

But I think I have spotted the problem: the polarity of the feedback is reversed. Pin 2 (FB) should go higher as the output goes higher. In this circuit is goes lower. With no output, FB will be at V+ so the output will be switched off.

the feedback pin is reversed, and I modified the PCB as @ronsimpson draw,

View attachment 224697

The PWM signal and output is working fine.

Though, I still have some problems with the PWM duty cycle and the snubber design,
I'll arrange the questions and take a screenshot of the waveform.

Very thanks for your help!
Have a nice day

 

After my testing, when I connect the electronic load and take 1A current output, the duty cycle is around 22%, like the picture below.
CH1 - Vcc, CH2 - PWM, CH3 - Vds, CH4 - Secondary of the transformer.

In my design, this converter should be 10A capability, but when I increase the load current to 3A,
not only the snubber resistor temperature increase to 150 C but also the PWM expansion to 45%.

If I continued to increase the load current, the PWM signal starts to jitter, as the video link shows below:

(sorry AAC can not upload the video file, so I upload it to Youtube, It's not for commercial used)

I'm wondering if the Current Sense pin of UC3842 cause this problem, so I check the pin3 of the IC, the waveform is like below:
CH1 - UC3842 Pin3 voltage, CH5 - Vds, CH6 - PWM
The waveforms shot under output current is 3A.
The datasheet said the PWM will reset to zero when Pin3 attachs to the 1V threshold, but the voltage only increases to 145mV as the picture shows.
Also, the PWM duty is opened to 54%(the value is unstable around 45%-50%)

My new question is:
What may causes the duty cycle is unexpected?
Transformer core saturation? or the ratio of the feedback loop is not well designed? or something I missed?

Thanks for your reading!
Have a goooooood day~

 

The duty cycle varies with the output power.
Without slope compensation flyback converters go unstable above 50%.
You are getting very close to continuous conduction mode, which is to be avoided as it will result in reverse recovery losses in the secondary rectifier.
The overshoot on Vds seems rather large, which would explain the snubber losses

 

Please show how you made the transformer? It looks like the primary and secondary are not coupled well.

 

First thoughts in red. Long on time short time to dump energy into secondary (too short) Long long dead time. Now I think this is wrong.
Second thought is in blue. Current ramps up during on time. Energy dumps into clamp for a short time. Next power to secondary. Then short dead time.

If my blue thoughts are true then there is too much leakage inductance in the transformer. The "clamp voltage time" is when the leakage inductance is dumped in the a resistor. Also the dead time is short. So the duty cycle can not increase much. Bottom right blue trace I drew is the current in the transformer. Primary ramp up, leakage inductance dump, current ramp down into secondary, dead time.