If it is fully functional, an Agilent 1683AD for $200 seems like a fairly reasonable price. That's not much above what an older 1673G with similar acquisition specs might sell at.
I'm not sure why, but the 168xA / 168xAD series seem to have rather high asking prices on eBay compared to older 167xG series or newer 168xxA series self-contained portable logic analyzers.
Sometimes run this search on eBay and the asking prices seem a bit crazy:
"agilent (1680ad, 1681ad, 1682ad, 1683ad, 1690ad, 1691ad, 1692ad, 1693ad)"
The AD models are more desirable than the A models due to the 4x sample memory depth. The 168xA / 168xAD series and the 169xA / 169xAD series use identical logic acquisition modules. The acquisition modules interface to the controlling PC through a 1394 port. In the case of the 169xA / 169xAD series the controlling PC is an external PC with a 1394 card. In the case of the 168xA / 168xAD series the controlling PC is integrated into the analyzer with a standard PIII motherboard and PCI 1394 card internally cabled to the acquisition module. One thing about these series is that for some reason the 1394 ports seems like they can die. Maybe the 1394 phy chip fails. I have seen several 169xA / 169xAD series listings on eBay where the photos show a 1394 cable being routed to one of the internal 1394 ports instead of the front panel 1394 port. If the 1394 port fails, then the logic analyzer application obviously can't connect to the logic analyzer at all.
If the logic analyzer application connects to the analyzer and all of the self test pass, that is a pretty could sign that it is fully functional. The only other thing that can go wrong that might not be detected by the self tests is the failure of any channels in the front end logic level comparators, for example from exceeding the maximum voltage input levels, or static discharges. You need to feed some sort of input signal (for example a TTL level oscillator clock signal) into each input channel and check for activity instead of stuck low or high levels.