Try chorus or flange with the depth and feedback on zero and mix on full wet instead of a delay. Chorus and flange are both essentially very short, modulated delay so if you turn the modulation allt eh way down you end up with a very short delay that’s well suited for K-S stuff. No need to raise the tempo or anything, although the OT’s parameters are too coarse to do anything melodic consistently with this (since most notes fall between the delay times available).
If you use two tracks and a cue in studio mode you can build an actual, proper K-S algorithm. The first track is basically just your impulse has a flex or static machine playing something like a noise burst and maybe a filter for some more control, and is sent to cue only (main volume 0, cue 127).
The second track is either a flex machine simultaneously recording and playing from cue, or a thru machine with its input set to 3/4 and the K-S feedback loop physically patched from the cue outs to the inputs. On this track you need a flanger or chorus with modulation depth and feedback on 0 and mix of 127/full wet in slot 1, a filter in slot 2. The cue level controls the amount of feedback and the main level controls the actual output level of the entire K-S algorithm (what you hear).
If you look at a basic K-S algorithm, track 1 is everything upstream of the + and track 2 is everything downstream. The only difference is that the delay and filter are before the feedback loop rather than in the feedback loop but the result is essentially the same.
If you want to feed some of the unprocessed noise burst to the output (for extra attack?) you can bring up the main volume of track 1, so you actually have a bit more control than you would in a strict K-S topology. Instead of only being able to change the level of the feedback loop you also have completely independent blend between the feedback loop and the raw impulse, so you can dial in the actual K-S synthesis with the cue level on the second track (feedback in the K-S algorithm) and then blend that with the dry impulse using the main volume controls for both tracks.
The trigs on the firs track ping the K-S loop, and you use trigless locks on the second track to control the pitch via the delay time in the chorus or flanger. In some ways that’s cumbersome, and in others it opens up a lot of possibilities for experimentation.
If you use a physical feedback loop from the cue outs to a pair of inputs, you can set up two independent K-S algorithms on two pairs of tracks, and then use the balance control for cross feedback to get some pretty ridiculous stereo textures.
I’ve gotten better results using a physical feedback loop because the offset between the record and play trigs you need to make the flex machine method work makes the range of delay times you have available just a bit longer and IME the physical loopback covers a more useful range, but that will vary with tempo since the space between the trigs is tied to tempo; the thru machine method will behave the same regardless of tempo.
I wouldn’t say it’s necessarily more useful than pinging a comb filter, but it definitely sounds different and setting it up is a good crash course in feedback in the Octatrack and in basic K-S synthesis since you have to build the algorithm from scratch.
EDIT: I haven’t actually done this in 2 or 3 years so it’s possible I got some detail wrong in my description somewhere, but I’m pretty sure that’s how it works. I just worked backward from the diagram I posted when I first figured it out.