Hi everyone,

In my design, there are optocoupler and TL431 to control my voltage feedback circuit. I only wondering that how can I check my feedback work or not before supply circuit with 24V in input part??

In IC voltage feedback input is Pin2 and Pin8 has reference voltage which is 5V. In Pin2 tend to stay in 2.5V. Optocoupler is K1010 and
Vce(sat)max=0.2V,
VF(for led side) is 1.2V.
(CTR：Min. 50% at IF=5mA VCE=5V
Also I am not sure that R3 and R2 have chosen accurately.

 

Make R3 120 ohms, this will give you 11mA led current , you can omit R1, you can test it on a bench psu at 4 to 6V, to see if the opto led fires at 5V, you can put a led+120r resistor in parallel with it.

 

Here's the LTspice simulation of the circuit with a 4N25 optocoupler which should be similar to the K1010.



Here's the simulation with Dd's suggested changes.



In either case the circuit appears to do what you need.

 

Here's the LTspice simulation of the circuit with a 4N25 optocoupler which should be similar to the K1010.

View attachment 157579

Here's the simulation with Dd's suggested changes.

View attachment 157580

In either case the circuit appears to do what you need.

There is difference which in Pin8 give constant 5V after IC start.

 

There is difference which in Pin8 give constant 5V after IC start.

Unless you're using an isolated output, you don't need the TL431, just use the internal reference on pin14, and set the voltage using pins 1,2 with fixed resistors and and a preset from your output.

This circuit uses pins 15,16 as current limit.

 

There is difference which in Pin8 give constant 5V after IC start.

Of course when the feedback loop is closed, the output will stay at 5V with the output of the opto at 2.5V.
My simulation just shows the open-loop transfer function of the feedback loop.

Below is a zoom of the transfer function showing the V+ supply voltage is 5V when the opto output is 2.5V, the error amp reference voltage.

 

It is quite easy to get "DC" performance. What is far more critical and often difficult is to get correct dynamic performance for loop stability - placement of the required "poles" and "zeros" in the feedback network to shape the closed-loop frequency and phase response of the converter. When a TL431 is used, it is common to find some of the required compensation at the TL431 and some at the error amplifier. The objective is to have at least 45 degrees of phase margin when loop gain passes through zero dB and a "-1" slope - all while getting high gain at DC.

I suggest looking on TI's website and ON's website for useful documents (you'll find them in the power control section - TL431 datasheets and ap notes are generally far too old and lack the detail to be useful). I saw a good discussion in a doc recently, but I didn't keep a copy or bookmark it. I think it was as ON. There is a section from a Power Supply Design Seminar book at TI somewhere with a title something like "Closing the Feedback Loop" with a good overview of several techniques and general info on the various types of error amp compensation required.

With flyback converters it makes a difference if the converter is operated in "discontinuous current mode" or "continuous current mode." The latter has a characteristic called a "right half plane zero" that can't be compensated for in any useful way, simply worked around at the cost of dynamic performance.

The K1010E is unsuitable for this application. The others would all work, but there will be differences in the circuit according to which one is chosen.

 

The K1010E is unsuitable for this application.

And why is that?

 

And why is that?

The gain spread is too large which makes it even harder to optimize the feedback. The part is available binned with a 2:1 max:min current transfer ratio in a particular bin, which, while not ideal is not too bad,, but the E version is unbinned with a 12:1 CTR ratio. Most of the opto manufacturers have at least a few types that are similarly binned and intended specifically for use in switcher feedback. Isolation methods with very low gain spread are a lot more complex and expensive and usually not warranted.

 

Make R3 120 ohms, this will give you 11mA led current , you can omit R1, you can test it on a bench psu at 4 to 6V, to see if the opto led fires at 5V, you can put a led+120r resistor in parallel with it.

Pretty much what I'd have said, that resistor looks too much.

Everything else looks pretty much appnote example - the divider needs to put 2.5V on the control pin when the rail is correct. that input is about 100k.