>doesn't it look like you get the ripple from the input bias? you can check
>this and use the photodiode as a current source into the inv. input (diode
>between gnd and inv.inp., plus a feedback resistor from inv.inp. to the
>output and noninv.inp. to gnd) without any bias from the supply. (usually
>you use the bc-diode of a transistor and leave e open in this configuration
>if you have a phototransistor.)
>
>ge

You got it.  (I thought it was a phototransistor though)  In any case,
loosing the negative side will fix it. A phototransistor is an optically
driven current source, any noise on it's supply will be greatly attenuated.