SMPS using UC3842 - And related problems

[vantusaonho]

Hi all!
I am designing a smsp using UC3842, but first I test this IC.
The IC work fine with load as Led, but I put smps windings transformer, the IC turn off output, I sense pin has 0 volt. Please explain to me this phenomenom.
Thanks

The schematic


Uploaded with ImageShack.us

[FTL]

I'm afraid that I have to start off with the standard answer: You have not given anywhere near enough information for somebody to help you.

Please provide a schematic of how you have it hooked up. You could just be missing some small thing, but without all the details it is impossible to help.

[enzine]

In the flyback converter (your circuit), the energy is stored in the transformer inductance during the "on" phase and released to the load during the "off" phase.
The diode D3 in the off phase is in conduction and "block" this energy.
You must use a voltage clamp snubber instead od D3 (see application note of UC3842 or similar).

Ciao!

[vantusaonho]

In the flyback converter (your circuit), the energy is stored in the transformer inductance during the "on" phase and released to the load during the "off" phase.
The diode D3 in the off phase is in conduction and "block" this energy.
You must use a voltage clamp snubber instead od D3 (see application note of UC3842 or similar).

Ciao!

Exactly! The circuit work when I removed D3 diode. But Output voltage is 25VDC meanwhile output voltage not rectified is 8VAC. I think it should be 8V*sprt(2) =11.3 VDC



Uploaded with ImageShack.us

[enzine]

... I think it should be 8V*sprt(2) =11.3 VDC

NO!
This formula is valid only for sinusoidal waveforms.
You must also take into account the duty cycle.
I suggest you study a little bit theory of switching converters

Ciao!

[vantusaonho]

Hi everybody!
I made a smps successfull. My smps has 33VDCt input and 12VDC output 1A. I checked the output about 12.66V. This is the schematic.


Uploaded with ImageShack.us
But there is a problem that is the 56K 3W resistor of snubber circuit is too hot and the mostfet too. I measured voltage at pin 3 (Current sense), it is about 0.143V. That proofed that is the current through primary windings about 0.143A. Please explain to me why the resistor and mosfet heating.
Thanks in advance

[Mr. Spock]

Check leakage inductance of transformer.

[Gallymimu]

Check with a scope, you might have a lot of ringing.

[vantusaonho]

Check with a scope, you might have a lot of ringing.

Yes, please tell me how to remove ringing

[Gallymimu]

Yes, please tell me how to remove ringing

Tell us what is ringing. Where are you measuring.

Be sure to check the gate and the drain of your MOSFET.  I am not a snubber expert but have designed a few RC snubbers before.  Not as sharp at the RCD style one you are using.

Typically you can put a series RC across the mosfet such that it is tuned to maximally dampen the ringing frequency.  That might be redundant with the RCD though, not sure without thinking about it harder.

I've used these equations with success for inductive switching placing the snubber across the load or across the mosfet depending on the circuit:
http://www.ti.com/ww/en/analog/power_management/snubber_circuit_design.html.

something seems odd though, why would a 56K resistor be getting hot unless your ring voltage was REALLY REALLY high?  If you were applying 100VRMS to that resistor it would only be 200mW.

What is your peak ring voltage???

[vantusaonho]

Voltage at Capacitor snubber is 317V, I don't understand why it is high like that. I rewound flyback transformer with less turns than, primary winding only wound 50% length bobbin. The rest of space, I wound secondary winding and bias winding. The result that the resistor not overheat more, but when smps work with load, mosfet overheat.

[xenix]

Energy stored in the leakage inductance of the transformer charges C1 when MOSFETis turned off. Snubber resistor R1 discharges C1 until next switching. If you can decrease leakage inductance of the transformer, you can decrease capacitance of C1 and the heat on R1 will be decreased.

If you have a scope, decrease C1 value and check Vds. As you decrease C1, Overshoot voltage on Vds will increase, stop at the point where MOSFET is safe.

If you can share transformer design details, we can help to improve it.

What is the switching frequency? if it is high, you should decrease 100Ohm gate drive resistor to decrease switching losses on the MOSFET.

[vantusaonho]

Energy stored in the leakage inductance of the transformer charges C1 when MOSFETis turned off. Snubber resistor R1 discharges C1 until next switching. If you can decrease leakage inductance of the transformer, you can decrease capacitance of C1 and the heat on R1 will be decreased.

If you have a scope, decrease C1 value and check Vds. As you decrease C1, Overshoot voltage on Vds will increase, stop at the point where MOSFET is safe.

If you can share transformer design details, we can help to improve it.

What is the switching frequency? if it is high, you should decrease 100Ohm gate drive resistor to decrease switching losses on the MOSFET.

Your explain is easy to understand. The flyback transformer include: EE25 core, primary winding 19 turns, wire 0.4mm - secondary winding 9 turns wire 0.6mm - Bias winding 10 turns, wire 0.3mm. All windings are wounded follow clockwise. switching frequency is 109KHz. Please help me

[xenix]

Heat on the MOSFET is related with the high gate drive resistance, try decreasing it to a lower value like 33 ohms. The gate driver inside UC3843 can deliver ±1A maximum, that means lowest gate drive resistor value is 12 ohm @12V IC supply voltage.

Primary and secondary windings should be as close as possible. In the below figure, prefer the winding arrangement like the transformer at left of the figure. The transformers on the right of the figure has separated primary and secondary windings. In this configuration, leakage inductance is high, so avoid this design.

[vantusaonho]

I find out that diode D1 is overheat, too. I wonder FR207 have break voltage 1000V, current forward 2A meanwhile the current in secondary winding only 1.5A (peak), so why it overheat?
I think that air-gap large (is about 0.75mm, meanwhile is required 0.16mm) is the cause to high leakage inductance, so Vsn is high (317V). I fixed air-gap and result that R1 not overheat. The resistor for gate drive, I replace with 10 Ohm, mosfet seem less overheat but it still is hot.

[zab]

Mosfet heat is result of normaly two things,
1 on resistance of mosfet
2. insufficient drive volt.
The mosfet used here has RDS(ON) = 3.8Ω@VGS = 10V. on resistance is high you can select some other transistor.I have used mosfet having on resistane in m ohms.You are using  high volt 600v mosfet for 33v. bring it down to 100v to get  good on resistance mosfet drive volt are coming from ic. so ensure volt for ic more than 10v all the time.

[vantusaonho]

so ensure volt for ic more than 10v all the time.

When smps have a load, drive volt is about 7.2V (45%*16V -Vcc), normal without load drive volt is about 3V. How to ensure it more than 10V? Please

[Gallymimu]

At 100KHz switching losses are going to dominate over Rds losses.  Be careful reducing the gate drive resistor.  It will likely increase the ringing as your switch edges will be sharper.  This may actually increase your switching losses rather than reducing them.

You should be able to measure your leakage inductance.  Do you have an LCR meter?  To a crude estimation you can measure the primary leakage inductance by shorting the secondary and measuring the inductance of the primary.  This PDF gives a pretty good explanation:
http://0071c19.netsolhost.com/support/articles/55/Measuring%20Leakage%20Inductance%20(104-105).pdf

with 300V ringing you have set up a really nice resonant tank!

For leakage inductance reduction.  No air gap if you have an air gap in the core.  primary and secondary overlaying will help, winding in hand will also help.  Winding in hand means having your primary and secondary wound together at the same time as a pair.  For a 2:1 transformer wound in hand it is a bit of a pain but you can wind 3 wires together and use 1 of them to make the primary and 2 of them to make the secondary.  This helps a lot with coupling but your mileage may vary depending on other design constraints.  I hope that makes sense.  I don't think I explained it well.

Core material can also affect leakage but most ferrites should be pretty good.

[vantusaonho]

I replaced resistor for gate mosfet down to 10 Ohm, Mosfet is less hot than (with a led as load, no overheat), but It still hot when I put load as fan 12V 0.7A. The diode for rectifier at secondary side is overheat too. I dont know why it behave like that although the load is only a led with 10mA in current. I don't have oscilloscope even LCR meter because I am a newbie in smps and they are so expensive.
I designed a new circuit with 110-220VAC, output 12V 1A. I hope it work fine

[Gallymimu]

I replaced resistor for gate mosfet down to 10 Ohm, Mosfet is less hot than (with a led as load, no overheat), but It still hot when I put load as fan 12V 0.7A. The diode for rectifier at secondary side is overheat too. I dont know why it behave like that although the load is only a led with 10mA in current. I don't have oscilloscope even LCR meter because I am a newbie in smps and they are so expensive.
I designed a new circuit with 110-220VAC, output 12V 1A. I hope it work fine

Check and see if the ringing voltage has gone up since reducing the gate drive resistor.  It is important to monitor that.  If your ring voltage is 300V and your input voltage is 30V I'd be nervous about using the same/similar design at 100V-200V input until I really understood why had had such high overshoot voltages.

You are playing a tough game working on this without at least a scope.  Like playing basketball with one eye covered.  Good luck.  We'll help as we can.

[zab]

When smps have a load, drive volt is about 7.2V (45%*16V -Vcc), normal without load drive volt is about 3V. How to ensure it more than 10V? Please

Have you check these volt with muti meter on out of ic,If yes then you are mistaken by the values. what is important is the amplitude of the drive pulse.this can be checked only with oscilloscope.
with multimeter you can check the vcc of 3842 if the value is more than 10v it means gate drive pulse will also be of more than 10v . that is ok. if your vcc come down than it need correction.

[vantusaonho]

I only have Volt Ohm meter, no multimeter, no oscilloscope. Poor me!

[zab]

No problem the method will remain same for your meter.So go ahead. Please make clear the point of checking 300v and method as well. check and  share the result.

[Faros]

you can disconnect the UC power pin (pin 7) and connect it to external power supply of 12V, make sure to share the power supply GND with your circuit GND. check your MOSFET for heating. if it is lower in temp this means that the MOSFET gate voltage was under driven, if not then alter the frequency by changing C5 to 10n and R7 to 1.5K (should give 100 kHz), recheck and feed us back.

[vantusaonho]

I will check this. Please tell me how much do I should coefficient and duty cycle with input 110-220 output 12V? And what equation say this? Thanks