OK, obviously I'm misunderstanding something.
I have a RedPitaya, which is an oscope+generator. Its output impedance is 50\$\Omega\$.
If I send a signal from its output to its input (which is 1M\$\Omega\$) and use a BNC T with a 50\$\Omega\$ terminator (to match impedance), I expect a signal drop of about half, since this is a voltage divider: 50\$\Omega\$ out, in series with 50\$\Omega\$ in, with the 1M\$\Omega\$ in parallel not affecting the signal much.
However, I do not see this on the trace. With the terminator, a 1Vpp (nominal) signal is almost exactly 1Vpp (1.003Vpp). Without it, it is about 1.05Vpp.
This makes no sense to me. The only thing I could think of is perhaps the RP "knows" when the terminator is connected and is doubling the voltage trace? Or, perhaps I don't understand how terminators work.