Sounds about right. I forget exactly what happens, but the lights dance around for a while. At this point, the GPS is pulling one way, the OCXO is pulling another way, and the poor little micro is just totally confused. Give it an hour or two to gain some semblance of sanity. Once the GPS figures out where it is and the OCXO properly warms up, things will start to settle down and you'll see the 'On' light on Ref 0 and the 'Stby' light on Ref 1 both on solid. If you don't have an RS422 adapter, use the hack described in the first message on the Time-Nuts mailing list. Get the unit hooked up to a PC so you can see what's happening. I recommend you use the modified Z38xx program. It includes some data collection functions so you can track the performance of the unit. Ultimately, the performance will be dependant on just how good your particular OCXO is and how carefully written the software was.
The picture below shows some of the things Z38xx can show. You open up each of these views under the View tab on the main screen.
The top left graph shows the number of satellites locked and the Holdover Uncertainty Prediction (HUP). I like the HUP. The unit is estimating how far off it would be if it lost all satellites for 24 hours. The value here is okay, but not great. My Z3801A moves between about 3.5 us and < 0.5 us. As the OCXO improves, this should drop. Less than 1 us is really good.
The top right graph shows the error between the 1 PPS and GPS in blue and the EFC voltage to the OCXO in red. The way the EFC is bouncing around explains the relatively high HUP and the jumpiness of the 1 PPS. Hopefully, as the OCXO works the kinks out after its long nap, it will settle down and start showing a smooth graph that slopes either up or down and just runs straight. This could take 30 days or more. Although I don't see it on the datasheet, many OCXOs specify 30 days of continuous operation before they meet their aging specs. Some really high precision units specify at least 90 days. The longer it runs, the better it will perform. It's best to leave it running permanently. The unit's ultimate performance will be determined by your particular OCXO and the design parameters of the software.
The bottom left chart shows the signal strength of each satellite. If your view of the sky isn't great, or your antenna or cabling isn't very good, you'll see it here.
The bottom right screen is the status report from the unit. You could run Hyperterm or putty and get this with the command :SYST:STAT? . It might be a good idea to start with a program like that just to ensure that your communications are working properly.
I won't bother posting another graph of the Allan Deviation. It has improved some, but until the OCXO settles down, the results aren't particularly meaningful. Even so, the results shown earlier are good.
Ed