Other Parts Discussed in Thread: LM76002
I've designed a symmetrical power supply based on the above mentioned TI eval Board.
Unfortunately, our samples show a rather poor -15V performance.
The +15V is stable and load independent tightly settled at 15V (bode analysis also shows sufficient gain and phase margin).
The -15V settles at about -9V in our design.
The load at +15V is about 130mA
The load at -15V net is about 90 mA (measured with an external supply @ -15V)
Our prototype design
I've made some measurements to show the performance. In the following the device is powered with +24V.
+15V is loaded with a variable DC load.
-15V is loaded with a resistance.
We do have two differences compared with the eval board.
First, we only feed back the +15V. As the eval board DCDC controller refers to -15V (what we do not prefer to do) the eval board feeds back the differential voltage of +30V. I've modified my schematic to also feed back the differential voltage.
Differential voltage feedback
This gives 75% feedback of the +30V and 25% feedback of the +15V. This modification improves the situation a bit but the outcome is still not satisfying.
Referring the device to -15V as you do in your eval board comes with another benefit. The input voltage is virtually increased to Vin + abs(-15V). This lowered the transition of DCM -> DCM. As DCM is mandatory for a good operation I guess a increase of the input voltage again improves the performance. And yes, powering the device with +48V shows better -15V performance
Increased input voltage
^this measurement has been made with a 16V setpoint without differential voltage feedback!
* Is there a input voltage requirement i am currently not aware of?
* Is there a requirement to put the controller on -Vout for this kind of application?
* Is there anything to be modified on my design based on your experience?
* Is the design the right solution for us(see asymmetrical load situation in the beginning)