How this DCDC converter works? AZ431-A PRECISION SHUNT REGULATORS

[Zucca]

I just successfully lowered the output of a 12V LED PSU.
Unfortunately I can't fully understand how it works, I will really appreciate if somebody can push me in the right direction.

Here the datasheet of the output PRECISION SHUNT REGULATORS AZ431-A.

This is implemented in the PSU:



My question is simple and stupid I suppose, that's why I post it in the beginner section.
If Vout is 12VDC how can the optocoupler generate PWM impulses? I am assuming the optocoupler must generate some sort of PWM impulse but I see only DC constant voltages at the input light diode (right part of PS521).

Many thanks in advance.

[ledtester]

It doesn't generate PWM impulses -- it just disables the PWM when the voltage is too high.

[wasedadoc]

It adjusts the ratio of on to off time of the drive to the switching transistor/MOSFET on the primary side of the transformer.

[magic]

431 and optocouplers are linear parts that generate feedback current/voltage/whatever proportional to Vout deviation from nominal.

The PWM controller watches the feedback signal and adjusts its duty cycle. This is the actual "brain" of the switching PSU.

Look at datasheets of PWM chips.

[Zucca]

Of course the PWM is adjusting itself depending on the output voltage at the end of the day.
My question is how?

optocouplers are linear parts

That's what I do not understand, how can an optocoupler be linear?
This is the relationship between the two currents:


Maybe it is working in a small signal condition were the curve can be tolerated as a line....

It doesn't generate PWM impulses -- it just disables the PWM when the voltage is too high.

I accept that, make somehow sense.

Thank you so much for helping me, sorry I am slow sometime....

[magic]

Well, linear enough. There is some minimum current which causes the PWM to start reducing the power it's pumping into the transformer, say 0.5mA, and then more feedback current limits it further until it's completely turned off at 1mA or whatever, the numbers are made up. It doesn't take much difference in REF voltage for TL431 to go from sinking 0.5mA to 1mA or even 10mA, so regulation is fairly accurate no matter what specific feedback current and PWM duty cycle is required to meet load demands.

CTR of the coupler must be accounted for in the design, but it doesn't change the scheme meaningfully. It needs to meet some minimum/maximum limits over a small range of output currents and be monotonic, i.e. more LED current sends more feedback current to the PWM.

[MrAl]

Of course the PWM is adjusting itself depending on the output voltage at the end of the day.
My question is how?

optocouplers are linear parts

That's what I do not understand, how can an optocoupler be linear?
This is the relationship between the two currents:


Maybe it is working in a small signal condition were the curve can be tolerated as a line....

It doesn't generate PWM impulses -- it just disables the PWM when the voltage is too high.

I accept that, make somehow sense.

Thank you so much for helping me, sorry I am slow sometime....

Once you set the output voltage the opto only needs to tell the controller when the output needs to be lowered and it works at nearly the same current all the time.  Same thing with a thermistor temperature controller, once you set the temperature the thermistor sees almost constant temperature so it doesnt need to be linear either and that is because the system keeps the temperature constant.  As long as there is only one set point the deviation will be small.  In cases where there is some hysteresis that doesnt matter either because each threshold is a constant also, so once you set the upper and lower temperatures the thermistor always sees those temperatures.

What you can't do is assume that you can set something without seeing the output.  You cant assume that either the opto or the thermistor will go to some set value unless you calibrate the system over the entire range of operation, which you dont need here.  When you set the voltage or the temperature and you see the output go to that value, the system is calibrated for that one point and that one point only, but that's all you need.

Oh another thing is that if the output goes up the sensor must react by going up, and if the output goes down the sensor must react by going down, and the set point of the sensor must be repeatable or nearly so.  If the sensor has to react by going down when the output goes up, then it has to react by going up when the output goes down, if that is by design.  The sensor can not react by going down for both when the output goes up or and when it goes down, and it can not react by going up for both output changes either.  This means the sensor can not operate at the peak of it's curve if it has that it has to be either before the peak or after it.

[Zucca]

Got it many thanks!

Basically is a good old negative feedback loop, and VFB on the PWM controller is an ANALOG input not a PWM one...
Now everything is crystal clear, also yes you do not need linearity....