help needed on current control SMPS

[jegandren]

hi all,

attached is a current controller circuitry that i have build using UC 3846, but with no success..

the problem here is i guess with the current sensing part as for an open loop circuit without feedback i would get +/- 40V with no noisy coming from the switchimg transformer, with a nice pulse.

but when the feedback is given im getting only +/- 17V with a noisy sound coming from the switching transformer and the mosfet gets heated up quite fast.

Is it the rite way to sense the current. please advise.

[mflorin]

just few comments:
- the way you sense the output current is compromising the isolation barrier.
- the voltage error amplifier output pin (COMP) is floating!!! . You have no compensation network for EA. The system is oscillating.
- No current limit protection is implemented (pin 1 is floating)
- the optocoupler and series zener diodes are connected in a strange way
- you have no duty cycle control. Most probably the two outputs are generating the maximum duty cycle value, internal T type flip-flop doing its job and toggling the outputs.

Search for UC3846 app notes and read them carefully...

[zilog]

Except for the obvious, that you have connected primary mains and secondary side together in a most lethal way - you miss the most advantageous part of current mode which is to detect and avoid flux imbalance in the transformer.

If you instead sense the current (in an isolated fashion), you will get the advantage that the transformer doesn't walk into saturation.

[guitar_joe]

Are you aware this IC does not have slope compensation built in?

Have you ensured adequate dead time for the push pull outputs?
I think this is programmable with CT.

[switchmodepower]

You may would do some research on design notes/app notes on UC3846. And if you truely want to do a peak current mode half-bridge then it's a little more involved.

[jegandren]

Originally posted by mflorin
just few comments:
- the way you sense the output current is compromising the isolation barrier.
- the voltage error amplifier output pin (COMP) is floating!!! . You have no compensation network for EA. The system is oscillating.
- No current limit protection is implemented (pin 1 is floating)
- the optocoupler and series zener diodes are connected in a strange way
- you have no duty cycle control. Most probably the two outputs are generating the maximum duty cycle value, internal T type flip-flop doing its job and toggling the outputs.

Search for UC3846 app notes and read them carefully...

hi,

This is my 1st design for current control...

as you said that the way i sense the output current is compromising the isolation barrier, that means for an isolated SMPS is it we should only sense the current from the primary part and not from the secondary part... because the common ground for the IC is from the primary ground..

The optocoupler and zener diode, is actually a diagram mistake.

[mflorin]

sensing the current in the primary side is the most common way, however, sensing the output inductor current is also possible as far as you pass the information to the pwm controller in a isolated manner.

[sawreyrw]

The UC3846 is a peak current mode chip as it responses to the peak current rather than the average current. What this means is that the chip is expecting to see a ramp of current during the on time. Therefore, it will probably not work well if you sense the output current.

Take a look at some of the applications notes to understand how to use the UC3846.

[megajocke]

And also keep in mind that half bridge with peak current mode control doesn't work without special balancing circuits.

That's the probably the "more involved" part switchmodepower is referring to.

I recommend two switch forward instead if you want to use peak current mode control.

[switchmodepower]

Originally posted by megajocke
That's the probably the "more involved" part switchmodepower is referring to.

yep