Input Bridge rectifier Heating up in a Flyback converter

Dear All,

I designed a flyback converter with folllowing specification: input voltage of 230V and output 20V, 3A.
For testing the circuit I am giving input through a variac. However, even at 20V Ac input the diode rectifier (using 1n4007) is getting damaged. There is also heating in the current sense resistor. But at this low voltage the circuit should not even have started right. The output load that I have connected is 50 ohms. The controller used is UC 3842. I have simulated it in PSIM and the system is working fine. I then removed, input CM choke, NTC and Capacitor however, there is still no effect. What could be the issue and how to resolve it. Thanks

Don't start up an offline SMPS at brownout voltages. The SMPS controller chip may not have enough head voltage to start and may be permanently stuck in startup mode.

Also, please draw the transformer correctly, as an actual transformer symbol, showing the windings and its polarity relationship. On a flyback, this polarity relationship is of utmost importance.

sabu31 said:

Dear All,

I designed a flyback converter with folllowing specification: input voltage of 230V and output 20V, 3A.
For testing the circuit I am giving input through a variac. However, even at 20V Ac input the diode rectifier (using 1n4007) is getting damaged. There is also heating in the current sense resistor. But at this low voltage the circuit should not even have started right. The output load that I have connected is 50 ohms. The controller used is UC 3842. I have simulated it in PSIM and the system is working fine. I then removed, input CM choke, NTC and Capacitor however, there is still no effect. What could be the issue and how to resolve it. Thanks

Click to expand...

Please check your Transformer Polarity.
You can implement soft-start technique for the IC.
If possible disable the IC operation till the input is below certain threshold

The startup hysteresis is very low on UC3842...you may need a much bigger C6......and c10 reduce to 100nf.
UC3842 takes loads of bias current, so will drain down c6 before your bias coil takes over...try a UCC38C42 instead...its pin-for-pin, and takes much less bias current.
Change D4 to a super fast diode , or at least put say 10R in series with it.
Your schem doesnt fully show extent of vout net.
Can you divulge your turns ratios...?..is your aux diode getting overvoltaged...its schottky and not snubbed...beware killing it with OV spikes.

Have you used fast sec diode? (<75ns)

Why not try getting it working with a isolated lab psu feeding vcc first.

50ohm load resistor is much less. IN4007 diode is 1 amp . you may have to use at least 2-3 amp rectifier diode.
2nd check the output filter capacitor, put a load resistor of 330 ohm/1 watt, check polarity of xformer
& also check power FET. if still the problem not solved then remove the ic from circuit then check the diode.

60W converter - the power factor on the input will be about 0.55 thus the input current, rms = 500mA at full power

for 230Vac input, at 115 Vac in I = 1A for same power

at 40Vac in ( assuming the converter can still work at full power here ) Iin = 2.9 amps rms ....

you see the issue here? also the peaks will be about 4-5 times the rms, so 500mA rms invites 2.5A pk ...

also check your build carefully - and the polarity of the transformer winding ( phasing - dot notation )

Easy peasy said:

60W converter - the power factor on the input will be about 0.55 thus the input current, rms = 500mA at full power

for 230Vac input, at 115 Vac in I = 1A for same power

at 40Vac in ( assuming the converter can still work at full power here ) Iin = 2.9 amps rms ....

you see the issue here? also the peaks will be about 4-5 times the rms, so 500mA rms invites 2.5A pk ...

--- Updated Jul 16, 2020 ---

also check your build carefully - and the polarity of the transformer winding ( phasing - dot notation )

Click to expand...

Thanks for the reply. This means that i should quickly go to 230W incase of 60W load. However, my doubt is When input is 20Vac rms, the device should not turn on , then how is the diode getting damaged. What I am still not able to understand is that the even if I remove the IC and power on the circuit, the diode is still getting hot. Ideally when IC is not there, there should not be any path for the current flow.

codemaster11 said:

50ohm load resistor is much less. IN4007 diode is 1 amp . you may have to use at least 2-3 amp rectifier diode.
2nd check the output filter capacitor, put a load resistor of 330 ohm/1 watt, check polarity of xformer
& also check power FET. if still the problem not solved then remove the ic from circuit then check the diode.

Click to expand...

The diode is stil getting damaged when the IC is removed from the circuit and when I give 20 Volts AC.

treez said:

The startup hysteresis is very low on UC3842...you may need a much bigger C6......and c10 reduce to 100nf.
UC3842 takes loads of bias current, so will drain down c6 before your bias coil takes over...try a UCC38C42 instead...its pin-for-pin, and takes much less bias current.
Change D4 to a super fast diode , or at least put say 10R in series with it.
Your schem doesnt fully show extent of vout net.
Can you divulge your turns ratios...?..is your aux diode getting overvoltaged...its schottky and not snubbed...beware killing it with OV spikes.

Have you used fast sec diode? (<75ns)

Why not try getting it working with a isolated lab psu feeding vcc first.

Click to expand...

I have replaced 1n4007 with UF4007 for both aux diode and snubber diode. The turns ratio is 52(input): 8(output):5 (aux). At present I have only UC 3842 with me, so will have to resolve the issue with it. I will try with supplying vcc of ic separately

treez said:

The startup hysteresis is very low on UC3842...you may need a much bigger C6......and c10 reduce to 100nf.
UC3842 takes loads of bias current, so will drain down c6 before your bias coil takes over...try a UCC38C42 instead...its pin-for-pin, and takes much less bias current.
Change D4 to a super fast diode , or at least put say 10R in series with it.
Your schem doesnt fully show extent of vout net.
Can you divulge your turns ratios...?..is your aux diode getting overvoltaged...its schottky and not snubbed...beware killing it with OV spikes.

Have you used fast sec diode? (<75ns)

Why not try getting it working with a isolated lab psu feeding vcc first.

Click to expand...

C6 i have increased to 100uF. However, there is not much effect.
Why should C10 be reduced to 100 nF.
Similar circuit I tried simulating in PSIM and its working.

it would appear you have a failed diode or other fault ...

Hi,

Did you connect a scope ... causing a short circuit to earth ground?
Or anything else connected to mains or erarth ground?

Klaus

replace all the diode of main AC bridge rectifier with IN5406 diode which is 3amp/600v.
then check. possibly the power MosFET is shorted.

OK yes, youve removed the UC3842 and your circuit is drawing much current. Have you got a thermal cam?...you can see which component is damaged short.
Otherwise you just have to remove components one by one till the short goes.

here is recomendation for thermal cam

Let's say that for a 230 Vac input voltage, you have 185Vac = 185*1.414 ~ 262Vdc as low line. But now you are running it at input voltage of 20 Vac = 20*1.414 ~ 28.3Vdc. This is enough voltage for the IC to start up considering UVLO but you shouldn't be running it at that input voltage level. Normally, the flyback converter draws Iin = Pout/(Vin*(1-D)*efficiency) amps average current through the transformer, sense resistor and MOSFET.

Also duty cycle increases as Vin decreases and the duty cycle of the UC3842 can go as high as 100%, which of course doesn't make sense for a flyback converter.

When you calculate for the average MOSFET current during the D interval (coupled with the fact that D is much higher at Vin = 28.3V than at 262V, of course with constant turns ratio), you'll see that this current is now very much higher at 28.3V than it is at 262V. You can also estimate how much the diode bridge average current has scaled up.

You'd be designing your transformer for a much lower peak current than you have at Vin = 28.3V. So your transformer is very likely going to saturate, and that means even higher peak and average currents through the MOSFET and diode bridge, respectively. Corresponding RMS currents scale up too.

So, I did this assuming CCM, although these scalings is not very much different with DCM.

Your bridge rectifier is getting damaged and your sense resistor is getting hot because because of this high currents. I'd be surprised if your MOSFET doesn't also get damaged. Also, I was looking at your sense resistor and wondering why it is 1ohm.

Run the converter at low line. Around 185Vac. Put back the NTC in the circuit. Use maybe 180uF for your bulk capacitor to achieve about 3uF per watt of input power. Take a closer look at your current sense resistor. You can report back.

BTW what's your turns ratio?

Dear All,

The following are the changes that I have done.
a. Replaced the bridge rectifier diodes with UF5408
b. Removed the auxillary supply diode, thereby disconnecting Vcc from transformer output. I have directly given Vcc from a PSU at 15W.
c. The Isense resistor is 0.33 ohm ,2 W
d. The gate resistor was getting heated when input is 80V ac.

Now some noise is observed from variac and the gate resistor burnt out it was rated twice (10 ohm 0.25W, 20Ohm 1 W). The load i connected is 50Ohms. This would mean output power is 8W. I am attaching the updated schematic.

In my understanding, the transformer should not get saturated at this power output. I have used SPW17N80C3 (800V 17A mosfet). The turns ratio is 52:8.
What else could be done to resolve the issue. Any waveforms that I should record that would help in resolving the issue.

What size gap are you using on that ETD 29 core?
(you have 52 sec turns).
I will calculate the Isat for you with Isat = Bsat.A.N/L

Is your FET ok?

Is it low z from gate to drain and source to drain.

I mean, it sounds like a dead fet. Do you use ESD protection?

change main bulk capacitor 33uf/400vdc, replace D4 with uf4007, change C12 to 3.3nf it will change the fosc to 54khz
, gate drive resistor with 15ohm/1watt also you place 1uf/400v non electrolyte capacitor
replace it with 2.2nf/400vac. place its other connection on main ground instead of main positive.
check power mosfet + feedback section components.

seems like the mosfet is dead from ages ago - likely there is no gap in the Tx - you need to calc Bmax and gap - along with the turns ratio - did you do this?

Hi all,

The power Mosfet was damaged. I have replaced it. However, I could not test the system with Vcc of the UC3842 supplied by external PSU as the DC bus voltage after the rectifier was somehow getting charged upto 77V DC even when input is 10V ac. So I removed PSU supply and connected auxillary winding. Now the system is not getting damaged (no heating of diodes/resistors) thankfully , however there is no output either.

I have captured the Vcc waveform (output of axuilary diode) and also the Vds. The MOSFET is not switching even though DC link voltage is 325V. The capacitor across Vcc is 100uF. The value of R10 was changed to 39k (2 Watts) but it didnt affect much. R10 was getting heated after 1 minute or so but didnt burnout.

The Bmax used for calculation of the Transformer is 0.3T, Material is N97. The air gap is 0.25mm. Turns ratio (52:8:5(aux)). Magnetizing inductance is 880uH. At present I am using only 50 ohm load at output. So may be saturation effect will not come .

I am attaching the schematic and waveforms (Vss, Vds). Is the issue now due to transformer.

put a zener on pin 7 - likely you have now killed the chip - the electro on pin 7 will be a bit suspect now too ...

make sure the phasing (dot notation ) of the transformer is absolutely correct.

Yes, as Easy Peasy implies...when a mains flyback fet fails...the drain and gate can get shorted together with low Z, so the high voltage on the drain can rip through the controller circuitry, via the gate of the fet.

by the way, can you post gerbers.
Attached is an smps pcb layout doc too.

Here is ETD25 datasheet..
https://www.tdk-electronics.tdk.com/inf/80/db/fer/efd_25_13_9.pdf

…You say you have a gap of 0.25mm…..that is not a standard gap size….how did you gap these?
With 52 turns and a gap size of 0.22mm, and N87 ferrite, your i(sat) would be 1.04A.
What is your primary sense resistor…it should be more than 1/1.04 = 0.96R
Are you sure your CS pin cap is correct value…is it 100p?.....you have it as “0.1n”, and maybe mistook and put a 100n cap there instead?

Easy peasy said:

put a zener on pin 7 - likely you have now killed the chip - the electro on pin 7 will be a bit suspect now too ...

--- Updated Jul 17, 2020 ---

make sure the phasing (dot notation ) of the transformer is absolutely correct.

Click to expand...

The voltage across Vcc is postive as seen in the Figure, therefore the polarity should be correct. However, the IC may be damaged when the gate resistor got burnt during earlier testing. Will have to check with updated transformer with turns ratio of 60:9:8(aux) and also perhaps replace the IC.

treez said:

Yes, as Easy Peasy implies...when a mains flyback fet fails...the drain and gate can get shorted together with low Z, so the high voltage on the drain can rip through the controller circuitry, via the gate of the fet.

--- Updated Jul 17, 2020 ---

by the way, can you post gerbers.
Attached is an smps pcb layout doc too.

--- Updated Jul 17, 2020 ---

Here is ETD25 datasheet..
https://www.tdk-electronics.tdk.com/inf/80/db/fer/efd_25_13_9.pdf

…You say you have a gap of 0.25mm…..that is not a standard gap size….how did you gap these?
With 52 turns and a gap size of 0.22mm, and N87 ferrite, your i(sat) would be 1.04A.
What is your primary sense resistor…it should be more than 1/1.04 = 0.96R
Are you sure your CS pin cap is correct value…is it 100p?.....you have it as “0.1n”, and maybe mistook and put a 100n cap there instead?

Click to expand...

The core I have used is ETD 29.

As I mentioned earlier , though the converter was designed for 20 V, 3A. I have not yet connected that rated load. At present the load is 50 Ohms as somehow I want to get the system to get running and hence saturation effect should not come now at low load. Sense Resistor is 0.33 Ohms.