I am building a variable output power supply using the LM2576 switching regulator. Please see the image which shows the circuit diagram here:
The circuit uses a negative voltage generation sub-circuit to pull the ground pin of the LM2576 below 0 Volts so that the output of the LM2576 can vary from 0 Volts to some positive value. In the absence of this negative bias on the ground pin of the LM2576, the output would vary from its minimum value (1.23V) to some positive value.
The negative voltage generator consists of: C2, D2, D3, C3, C4, IC2, P3 and R3.
Voltage at C3 without connecting IC2 (which is a LM337) is observed at Vpeak (1.414 X 25V = 35V). However, when I connect the IC2 input, this voltage drops to 1.23V! (Interestingly, 1.23V is the Vref of the LM337.)
I am unable to understand why. The manufacturer specifies LM337*LZ* whereas, I am using a LM337 (from ST). The datasheets of these two ICs don't show any difference. Could the LM337 be very different than the LM337LZ?
I tried replacing the LM337 with a LM7905 and it works quite fine (-5V output is observed). Another thing that I tried is to use the center tap of transformer that generates the input AC voltage as ground. In this case, the LM337 works fine!
Any suggestions on how to debug this further? Any suggestions for an alternate method of generating a negative bias in this circuit?