The TEK 475 has 10-Megohm inputs, so you don't have to solve the problem there. There are special 50-ohm termination adapters that one uses with that 'scope to make it process 50-ohm signals.

It's possible that the FPGA isn't putting out what you think it is, though. Could you perhaps have overlooked something? Just because the "OK, I'm done programming now" light cycles, doesn't mean it's putting out at the pins. Perhaps if you simply reduce the frequency at the outputs to a range in which you can actually "see" them on your 475, it will build confidence in what you've got.

When I mentioned that 20 ns interval, I wasn't referring to frequency, but, rather, to a phase shift between two of the outputs, therefore allowing you to observe how your new digital instrument handles a 2 MHz signal that's shifted 20 ns from another at the same frequency. If you have four channels, it should be able to display them. How it deals with the 20, 40, and 60 ns phase shift might be illuminating. Further, you can try 40, 50, and 60 ns phase shifts as well. At some point you should see some meaningful things, possibly things even the designers didn't consider. This won't be a problem for you, in that it will simply tell you when, in the course of a project, it's time to get out the 475 and use it.

I really like FPGA and CPLD eval boards for the type of work I do, since I often have to make them do specific things for each project, but I use the older ones because (a) the old Spartan-II boards had internal tristate resources, (b) they were 5-volt tolerant, (c) they used the older DLL's which I learned to use satisfactorily a decade ago, (c) I acquired a number of CPLD boards with lots of space a decade ago, and also a number of the associated CPLD's, so I can reuse some of them as-built, for things like test pattern generation, special-purpose counters, modulator/demodulators for custom modulation schemes, encryption/decryption, etc. I have in some cases combined these with certain MCU's for really specialized applications which seem to come up again and again.

A couple of decades ago, I believe I found an op-amp from National Semiconductor that was capable of driving a 50-ohm load, but I don't recall whether it was able to support a true swing of +/- 15 volts, which I rather imagine I wanted. I'll have a look ... if I can find any of the doc's on that arbitrary waveform generator we were building at the time, to see what it was. It was a fairly wideband buffer amp, though I don't know that it reached beyond 30 MHz or so at full amplitude. The one that you might consider today is the LM6181, though it won't perform to full voltage swing beyond +/-5 volts into 50 ohms because its current is limited to 100 mA. I've not used that one myself, but a colleague has mentioned it as an adequate driver for your circumstances, provided you don't need the higher voltage swing.

RE