Puzzling Problem of Using a LDO Regulator for the Power Supply of a Comparator

Recently I am using a NEGATIVE output low-dropout (LDO) linear regulator, TI's TPS72301, for the NEGATIVE power supply of a comparator (designed as: -9V input, -8V output). By the way, the positive power supply terminal of the comparator was connected to the output of a positive LDO regulator keeping okay all the time.

Initially, TI's TLC3702 was chosen as the comparator (dual per package). When turned on, as observed by an oscilloscope, the voltage at both the output and the FB terminal of TPS72301 creeped from the designed value, -8V, towards ground. The output of TPS72301 stops creeping finally at -1.5V ca, while the FB voltage at -0.6V ca.

I'm sure my usage of TPS72301 is okay, since another TPS72301 with the same circuitry (but used for power supply of some op amps) on my PCB works well. Thus the comparator, TLC3702, was suspected. As mentioned in its datasheet, TLC3702 was designed to operate from a SINGLE supply. To implement dual supply operation, another type of comparator, LM393, was substituted for TLC3702. However, its output needs additional push-pull. Furthermore, TPS72301 still did not work properly and this time its output turned to be -1.0V (I also replaced TPS72301 with a new piece before the 2nd test). All external components including the power supply security has been checked and no fault was found. Now I have to doubt whether TPS72301 can be used with general comparators.

I required TI's technical support by e-mail and the reply was that TPS72301 had better not be used here and the positive supply must be ensured during use of such a comparator. That called up an important detail in my design: a deliberate turn-on time gap between the positive supply and the negative (hundreds of microseconds). Is this the cause of my problem? Or TPS72301 itself fooled me? As for me, the latter is more likely.

Here list my next steps to study the problem:

1. Remove the comparator and see whether TPS72301 works properly;
2. If yes, replace LM393 with LT1018CN8 (a device of unique design, see pg. 8 of its datasheet, 'NEGATIVE Voltage Regulator') and see whether it is okay;
3. If it fails, remove TPS72301 and connect the negative supply of the comparator directly to -9V input.

I would appreciate any instruction or information about the cause of the problem or the possible solution.