Kawai K3 -1986    Weight=? Lbs./? Lbs. Number mfg.= ? MRP=$1095

Circuit Overview:
     Digital ROM based waveforms and all digital signal process until DAC to analog filters and chorus effect. User waveform with 128 harmonics. Can't really hear all of them but..oh well. :-) 10Khz maximum frequency generated I hear.

Plutoniq9/Ryan notes: I've done some more investigating with the ROM's. I guess that patend had nothing to do with it after all. (Yup...I didn't think so :-) ) Seems ROM#10 (like ROM#9), has interleaved waveforms within 1024-bytes per waveform (512-byte, 256-byte, 128-byte etc.). In order to find a relationship between the two ROM's, I overlayed screenshots of each pair showing waveshape & FFT spectrum analysis. I covered this for six of the K3's waveforms, (green = ROM#9, red = ROM#10);

Wave 31 (sine) :: http://audiosyn.com/files/K3/Sine%2031.jpg
Wave 30 (Saw):: http://audiosyn.com/files/K3/sawtooth%2030.jpg
Wave 29 (Square) :: http://audiosyn.com/files/K3/square%2029.jpg
Wave 28 :: http://audiosyn.com/files/K3/wave%2028.jpg
Wave 27 :: http://audiosyn.com/files/K3/wave%2027.jpg
Wave 25 :: http://audiosyn.com/files/K3/wave%2025.jpg

What's the relationship between them? To me, the red wave's look like (evaluating FFT) that they are hipass filtered versions of the green, but they also seem to be out of phase or shifted? The K3/m uses a hardware based HPF, so it's performing that function of the synth.....but they are definetley realted.....i just can't put my finger on how (completley) & why.

Harry notes: Here is an interesting sdiy discussion where the ROM contents are unveiled!
Don't know why thy'd do this but it looks like the red wave is the derivative of the green one. They might use this to do adaptive filtering that allows for fast response on a large change and better anti-alias on a slow moving wave. Note: the sine wave has almost noting on the red...

And Ryan concludes: harry win's the mystery challenge! Indeed, the second waveform ROM on the Kawai K3/m is proportional to the derivative of the input. I whipped up a quick differentiator in Synthedit and ran a single-cycle waveform from ROM#9 through it, result looks identical to the mstery waveform on ROM#10. I imagine this is for linear interpolation, to reduce aliasing on the oscillators.

This is good! I'll hopefully code up something soon that will take a 512-byte 8-bit waveform and covert it into the Kawai K3's waveform system...... my plan is to make a ROM that contains a broad spectrum of the the best analog & digial waveforms of different classic synths. Should be a nice match for the SSM2044 filters......

Thanks for eveyones help on this,ryan

Service Tips:
     This is my first synth. I'm not aware of any real problems but can imagine that like the Chroma Polaris of it's day some of the membrane switch assemblies are starting to die.