Current mode Flyback converter error amplifier unstable

[chaitanyab]

Hi,

I am having a problem in my error amplifier for current mode control scheme. It is providing 6 PWM pulses and then there is a gap and then the sequence repeats.

I am attaching my schematic, It is a flyback with DCM and current mode control with separate Aux winding.

Regards,

Chaitanyab

 

[treez]

you do not show whole circuit so it is difficult to help you.
But your connection to comp pin is wrong, it should be as an integrator, you are just using the internal error amp as an amplifier....
Also, your vin is a pulsed voltage?
Also, you use 1n4007 in RCD clamp, this not good, you must use ultra fast diode there.
Do you know uc3844 is 50% max duty

What is your vin, vout, pout? Cannot help you without these

- - - Updated - - -

Also, what simulator are you using?...some of them cannot handle smps.

- - - Updated - - -

Also, the uc384x series of controllers really need a schottky diode to be placed from pin 6 (gate drive pin) to ground, cathode to pin 6, this is to stop that pin from going too negative

 

[BradtheRad]

6 PWM pulses and then there is a gap and then the sequence repeats.

This sounds like 'hiccup mode'. Output is sporadic. Possibly caused by slight delay of the feedback signal, and/or too much gain. The feedback signal overshoots. I could be wrong.

 

[treez]

sounds like also maybe the converter needs duty cycle above 50% but uc3844 cant give that

 

[Warpspeed]

This can also happen if the output capacitor is very large, and the loading very light.

The dc output voltage overshoots, and the system shuts off for a brief time until the output falls back within feedback range.

 

[treez]

true too, of course, as we all know, the main point is your error amp is connected up like an inverting amplifier.....big problem....that's the first place to start

 

[Easy peasy]

C11 100pF is too small, try 100nF in series with 100 ohm instead, your feedback loop is oscillating, but unless you run the sim for longer you won't see it...

 

[chaitanyab]

Hi Tony,

Yes I have large output capacitors, for 55 mA I have 20uF. Do you think that is problem? or is it oscillating? what is the optimum capacitor value?

Regards,

Chaitanya

- - - Updated - - -

Hi Treez,

The datasheet was showing it in that way so I have added, I thought it is a integrator, which compensation do you recommend https://www.onsemi.com/pub_link/Collateral/UC3844-D.PDFdo you think it is a mistake in the datasheet? Page 12

Regards,

Chaitanya

- - - Updated - - -

Hi Treez,

Going beyond 50% does it means I need to design for CCM and not DCM?

Regards,

Chaitanya

- - - Updated - - -

.
It is a 11Watt Power supply with 190-500V DC input, Inductor 1.57mH, Dmax (by designed) 37%.I am using simmetrix.

 

[treez]

That method of feedback compensation is single pole compensation, and discussed on page 211 of Browns book “power supply cookbook”, 2nd Ed.
It has a low gain at DC which isn’t desirable from load regulation pojnt of view, I agree with Easy Peasy that the feedback cap is too low in value.
I am not sure why anyone uses that type of feedback compensation method, when they could just spend another 0.1c on an 0603 capacitor and have pole-zero compensation.

ok so vin=190-500vdc, pout=11w, what is vout?....& how many secondaries are there?, what is vout and pout of each secondary?

CCM can be with d>50% or indeed less than it, we need to see your vout to decide.

if power is low DCM can be preferred to ccm because you don't need to bother with slope compensation with dcm.

- - - Updated - - -

of course it could be a pcb layout issue, your gate drive pulses are getting pulled out of c4, which can mean gound bounce if c4 not properly located, etc

 

[pradhan.rachit]

Not related to the E/A issue, but doesn't the diode in the RCD snubber need to be an ultrafast diode instead of a 1N4007?

 

[Easy peasy]

Yes a fast diode would be much better...