for reference, the ~2 degree error I worked out would produce an uncertainty of ~1 nm on the location of the SAM, if looking at it from 25nm away (taking just delta_L = D*sin(theta), where theta is the PSF of our detector, D the distance of the emitter, and delta_L the uncertainty on its position). So basically in the Hornet in TOO, you would get your target box in the HUD, point your TGP at it, and have the same could be anywhere 1nm to the left or right of where you have pointed the FLIR. And that obviously neglects the fact you can't estimate distance either, so you would have an even bigger area to search.
Instead in DCS, we can point the TGP at the box in the HUD and always instantly spot the SAM site.
edit: to avoid derailing the thread and asnwer the OP, at what time of the day were you flying? I noticed the lock range of the Walleye is a strong function of time of day rather than light conditions, even though the scenario may look the same between, say, 7 and 8 am. The -H also has a TV sensor, so maybe that's what's going on.
TLTeo's post in What is the typical lock range for an AGM-65H? was marked as the answer