hi there…
i know this was discussed some days ago…but i dont know if there ever was a solution…the sawtooth oscillator has a bug: when playing low octaves, the saw is changing the pulswidth, so that there are 2 spikes in the waveform and the saw is nomore a clean saw…
so had some fun with my oscilloscope last night…i noticed, when changing to SAW and turning the pulswidth either -64 or +64 (doesnt matter) the spikes disappear even in the lower octaves… before knowing this, i used the TRI wave with fully turned pulsewidth to mimic a sawtooth—and this results in a inverted sawtooth compared to the “real” saw…
just for info 
cheers, nik