Voltage-mode control using UC3842

[kathmandu]

Hello,

I wonder if I could use the UC3842 PWM controller in voltage-mode control in a two-switch flyback topology. This is the block diagram:



If I tie the VFB pin to ground I'll get a constant threshold of 1V at the CURRENT SENSE COMPARATOR noninverting input.

This way, by applying the output feedback voltage to CURRENT SENSE pin, I could (theoretically) control the PWM duty cycle.

The output feedback voltage should be designed using a voltage divider, to get an output of around 1V for the required output voltage.

Is it going to work this way? I have some concerns regarding loop compensation, as the output voltage (feedback) is read during transistor ON time, when the flyback diode is not conducting thus the output capacitor is discharged by the load.

- - - Updated - - -

Another variant it's to tie the CURRENT SENSE input to a fixed voltage (0.9V or something lower than the clamped 1V threshold) then to use the VFB input as ususal.

That way, when VFB is greater than 2.5V (ERROR AMP fixed threshold), the PWM pulse is turned off.

Any (other) ideas?

 

[treez]

It cannot be done as you have to use the CS pin to feed in the voltage mode ramp (eg derived from the OSC pin)...and that means you then have no current sense cutoff protection........There are some ti.com active clamp controllers which you can fit up to a 2 switch flyback design and they are voltage mode. LMxxxx or something like that.

 

[Easy peasy]

run the gate drive thru an RC filter to (with back diode to discharge the C) this will give you a ramp you can run into the current sense pin, then you can design the rest as a volt mode loop, for 7oKHz say use R = 1k5 (with a back diode across it) , C = 4n7 (with 1k5 in parallel to limit the max volts) and try, if the volts on the ramp come up too high, increase C, when the V ramp reaches 1V it will terminate the switching cycle, so gives a max PWM limit which may be helpful...

ta da.... volt mode operation of the 3842...!

 

[treez]

i see your point, but surely you would want some sort of FET current limit?

 

[kathmandu]

Buck/boost converters didn't have current limit neither.. anyway, that's not a problem as I have a fuse for abnormal operation.

I'll try to trick the currente sense input with a voltage ramp. I presume I won't get a very good voltage regulation but that's not a big problem.

 

[treez]

Buck/boost converters didn't have current limit neither

Sorry but all converters usually have a current limit. There are some exceptions but theres usually some exotic over-riding control algorithm.

 

[kathmandu]

My mistake, I was trying to say that for buck/boost converters it's unusual to have a current-mode control (with its intrinsic current limiting feature).

 

[Easy peasy]

Not all power circuits need a current limit, if for example there is a fuse, or a current limited supply, or a power limited supply, or a poly fuse....

 

[kathmandu]

.. or for solar battery chargers, where the solar panels act as current sources.

 

[treez]

Beware of solar panel smps’s not using current limiting….
https://www.bbc.co.uk/news/uk-england-sussex-32382795

…solar panels often feed into huge electrolytic capacitor banks, and this is low impedance and the inductor current can really staircase high if you don’t have current limit.
As discussed, if you don’t have a current limit in the smps itself, then its usually in the system somewhere..(there are of course, exceptions)

 

[Easy peasy]

Inductor current can only "stair case" in a shorted buck converter, the solar panels could have caused a fire simply due to the V & I available at the terminals in full sunlight, a current limit wouldn't prevent that, but a fuse on the o/p of a buck converter would limit max current in the event of a shorted load...

 

[treez]

as you know, inductor current can also staircase at startup. But as you rightly suggest, current limit is very useful for the case of shorted output. Obviously you are aware of situations where current limiting isnt needed, and you have the experience to handle that, but i believe and i believe you do too, that the OP is better served by having some kind of current sensing in his SMPS.
Maybe OP could have an additional comparator and sense resistor nxt to the uc3842 to get the current sense in there. Maybe the comparator can stop the uc3842 pulsing when overcurrent happens...eg, put 2v (via various cctry) or so on the sense pin in such a case.

 

[Easy peasy]

Or, put in a poly fuse... cheap and reliable (PTC)