Pete: I think that you can do it by creating a global variable to store a1. Then, in each a-rate pass you can add the value from the global, multiplied by some value if you want to scale it, back into the frequency input of the carrier, and then save the output again in the global variable for the next pass. I did something similar recently and it seemed to work. Is that what you had in mind? I just heard about sound forge recently. Someone suggested a compositional method where you create snippetts of music using csound and then use sound forge to actually put the piece together. Does that sound workable to you? Any comments about sound forge? Thanks. >>> pete moss 01/09 11:51 PM >>> i was fooling around with sound forge and fm synthesis and created a sound i like. i have been able to implement this sound in csound with the following orc+sco. however, i would also like to add variable feedback to the last oscil. how can i send the output of a1 back into a1? is there an opcode that would handle this for me? i have never used feedback. thanks pete sr = 44100 kr = 44100 ksmps = 1 nchnls = 1 instr 1 iamp = p4 a3 oscil iamp*.1, 2200, 1 k2 linseg 0, .128*p3, .92, .19*p3, .3, .682*p3, 0 a2 oscil iamp*.0708*k2, a3+110, 2 k1 line 1, p3, 0 a1 oscil iamp*.5957*k1, a2+220, 1 out a1 endin f1 0 8192 10 1 f2 0 8192 7 0 4096 1 0 -1 4096 0 i1 0 2 32000 e !