Some months ago i purchased a defect Yaesu FT-817 radio. It turned out that PA and driver were defect. I was able to buy the PA transistors (RD07MVS1) and driver transistors (RD01MUS1) at DL2JWL (thanks!). I got some really good help in the 817-onair forum.
After the TX was operational i was interested in the output spectrum.
I did a singletone measurement (mode CW, 500mW) and a twotone measurement (mode DIG, 500mW, twotone signal from soundcard). Find the pictures below. The spectrum was measured with the DG8SAQ VNWA in spectrum analyser mode at 28MHz.
Please note that the RBW was 750Hz and you may want to recalculate noise levels.
One word about the twotone spectrum: The VNWA uses rectangular filters. Therefor the discrete tones apear to be rectangular spectrum. The gap inside each of the tones comes from the AC coupling of the measurement system which causes a notch at zero beat frequency.
The measurement results appear to correspond with spectral measurements by some OM in the US done some years before and measurements done by DF9IC. In fact the TX signal is not usable for a real contest because the sideband noise is around -80dBc within a bandwith of 3kHz.

FT-817 twotone spectrum 28MHz (750Hz RBW!)

FT-817 singletone spectrum 28MHz (750Hz RBW !)