variable output design by using top-switch-gx

This should be doable...but will be obviously more complicated than a fixed output voltage. I don't have the required time to create the design for you but can point you in the right direction.


Using PI Expert, do you initial design for 40V, 1A. This will let the software and you choose an appropriate transformer that is sized for the maximum power you intend to deliver.


Once you have this, your final output voltage (using a TL431 feedback circuit) will depend on the ratio of the resisters used on the reference pin of the TL431. The output voltage will largely be dependent on the ratio of the two resisters connected between Vout+, Vout- and the reference pin on your TL431. There are different ways you could implement this but I will leave that up to you.


If our reference designs and datasheets, you'll see that with a TL431 and optocoupler feedback type circuit, there is a resister above the optocoupler (typically tied to the bulk cap before your post filter). This resister sets the initial bias point for the optocoupler and TL431. You need at least a mA or so flowing through the opto and TL431 for the feedback circuitry to function correctly. You'll need to size these components so that everything will stay biased at all the different output voltage you intend to use.


To save yourself some time and effort, create additional designs in PI Expert with some different output voltages (using the transformer picked in your Pout(max) scenario). This should give you an idea of the range of values for the feedback components you have to work with.


Please read (and re-read) the following as it is INCREDIBLY important:
An often overlooked aspect of power supply design is verifying the proper operation of your control loop. Many people will tell you that simply doing a step response and adjusting components for a critically damped response is adequate. This is false.


Get a gain/phase analyzer (not negotiable) and evaluate your control loop at several different line voltages, load currents, output voltages AND operating temperatures. Capacitor values can change wildly at cold/hot operating temperatures, directly affecting the tolerance of your feedback circuit. Make sure you have proper gain and phase margin at ALL possible operating conditions of your power supply. I would also recommend testing multiple units to bring in the variable of production tolerances.

-The Traveler

dear sir,
i designed a supply by using opto coupler& tl431 as feedback circuit.i used a pot in the feedback circuit for variable voltage.
input voltage =110v

my ic is top246YN

np =24

ns =14

i am getting variable voltage from 40v to 18v.but after that the output voltage does not varied based on pot changing.i think at this stage top-switch is operating in min duty condition.plz suggest me to get 12v output.

Controller duty cycle is inversely proportional the the control pin current. As you lower your output voltage, your optocoupler is probably not getting biased correctly which means you probably have a resister value (above the opto) that is too large to bias the feedback circuitry at 12V.


You also might want to check the pot you're using to make sure that it will provide the resistance range you need for this application. Some have min/max resistance ranges due to their physical construction.


Again, this is going to be a MUCH more complicated project than I think you realize. There is a reason we don't have any public reference designs with variable output voltages. If you had a smaller voltage window or could create discrete designs optimized for the different output voltages you need, the process would be much easier. There are just so many issues that come up, it becomes terribly complicated to design.


For example:
Feedback circuitry biasing: You have to bias the feedback circuitry for operation at 12V but at 40V, your feedback gain will be so high, you can easily run into loop stability problems.
Transformer Selection: You have to choose your transformer based on your max Pout operating condition (40V, 1A) but then at 12V, your design will become so continuous you might run into minimum duty cycle issues.


If you don't have a means of verifying loop stability with a gain/phase analyzer, I would abandon this project all together. The amount of design verification required for this design will probably take over a week just to complete the gain/phase plots for all the different line voltage, load voltage, load current, operating temperatures and transient conditions analysis alone. Attempting to perform these measurements with a scope and a step load will NOT be adequate.


Add into the situation that you're guessing about whether or not the IC is running at minimum duty cycle instead of verifying it directly tells me that you might not have the equipment necessary to complete this design.

-The Traveler