Building a temperature tolerant SMPS

[Jules_Theone]

I've solved the diode heating problem, but having new problems... The modified spec is as follows:

40v - 75v input
10v @ 200mA output, stepped down to 3v3 through a L200CV regulator
5v @ 10mA (with isolated gnd) output, stepped down to 3v3 through a LM1117 regulator.

I'm using a LM5020MM-2 controller, Wurth Elektronik power over ethernet tranformer part number 749119133, FCD4N60 MOSFET for the power switch. The circuit is attached.

The circuit powers up ok, with stable 10v regulated, 5.2v outputs with desired resistive loading. This works fine at ambient temperature (22deg C) but the output voltage gradually decreases with increasing temperature (maximum 150deg C). I have tested the L200 regulator with 10v input and a 22R load, ramped up the temp and measured the output voltage. this decreased with temp, i believe it's a feedback resistor issue. The output voltages from the SMPS also decreases with temp by approx the same amount so when I connect it all together the problem is compounded. Although the LM1117 3v3 reg doesn't change at all (I didn't build it, so naturally it works perfectly!) But i cannot use a LM1117 for the 200mA because the input voltage is too high and the current it needs to pass causes it to drop out at 135deg C.

Are there any ways to stabilise or compensate for the output change?

Any ideas?

Thanks.

 

[Jules_Theone]

I've since given up with the L200 regulator on the output and are focusing on the SMPS now. I am still having the trouble with the voltage decreasing with temperature, some ideas i have include increasing the inductance in the transformer so there is more flux stored as i believe that the capacity of inductors decrease with temperature and saturation. I have connected two tranformers together with each winding in series with it's counterpart (this increases inductance?). But the output is worse when loaded. I've changed the ocsillator frequency up and down but it's really a stab in the dark.

Any other ideas?

 

[dknguyen]

These voltage errors that occur as temperature gets farther away from ambient...are they still within the specs of the IC over temperature? Because if they are...not much you can do about that.

You have never specified how much the voltage is off by, just that it is.

 

[Nigel Goodwin]

It's usual to have a precision voltage reference on the feedback, so it has a much more stable reference - this may well cure your problems?.

I presume you know the design isn't isolated?, usually there's no directly connection between primary and secondary.

 

[Jules_Theone]

Firstly the voltages are 5.2v and 18v at ambient and decrease to 4.1v and 12.5v at 150deg. The 18 -12.5v is ok as the motor controller IC can handle 10 to 20v, but the 5v needs to stay above 4.5v as it powers a 3v3 regulator with a 1.2v drop out. (Although thinking about it, why not generate 3v3 to start with and fiddle the other windings to get 12v ish and 3v3 isolated...?)

I dont understand how do use a precision reference, where would it connect and how would it function? would it be used to compare with the output voltage and keep it stable?
I've found that the feedback voltage stays at 1.25v the whole time. Does that mean that the controller IC is struggling to control the outputs?

Lastly, the primary and secondary are not isolated from eachother so the IGBTs can power a motor from the 60v and be controlled from the lower voltage secondary.

 

[Nigel Goodwin]

Yes, the output voltage is compared to a precision reference, and the error sent back to control the chip.

Check out the TL431, the most common reference used - here's a datasheet showing one use of it.