Some short not on an idea for the 600m band.
The typical SDR, as we all know, uses 4x the center frequency so that the 90 phase-shifts can easily be created by flip-flops.
In 2012, the Netherlands will most likely open the range 501-505kHz for ham radio operators. I figure a simple RX (maybe also TX) solution could be a 2.000MHz canned oscillator. This will get us spot on 500kHz center frequency, just as you may want. Comfortable 1 to 5kHz audio, which any sound card can handle easily, with a sample rate of only 24kHz. A further experiment should show if side-band suppression is required at all. I figure, a decent pre-selector should be enough already.
Should however, following a decision at the WRC-12, the range open to hams change to the range proposed by CEPT (472 to 480kHz), a 2.000MHz SDR-LO would be somewhat too high. In this case, the oscillator could easily be swapped with a 1.8432MHz one. Resulting in a center frequency of 460.8kHz. Audio up to 20kHz would still be somewhat a challenge for cheap sound hardware, never the less, a sample rate of 48kHz would cover it all.
No to the TX-part of it. One could either use a sound card generated signal, as provided by some software solution. One could also thing of generating an I/Q modulation signal at 4x the audio signal, divide and phase shift similarly to the LO chain. Unfortunately, we would now have a rich audio square wave. I figure some severe filtering will be required here, in order to end up with a sine wave.
I would not consider an AF phase-shifting network. I believe the frequency range is to great as provide accurate phase-shifting.
However, as in the RX part, it may be conceivable to filter the side-band at the RF range. A series of tank and trap circuits could possibly be enough. Mind you, the aerial matching itself is very selective too.