"Ghost modulation" in an 8-stage Ropez

[Mark Hammer]

Digging through the bin-of-iniquity today, I tried resurrecting an 8-stage Ross phaser (Tonepad Ropez w/4extra stages) I had made that needed some troubleshooting.  After locating a solder bridge and broken trace, I got it working.  I found I needed to replace the suggested 1uf caps in the Regen path with 100nf to minimize the howling at higher feedback settings.  I also find that using a 9k1 resistor on the output of the LFO gets me a wider sweep, that I like.  But what I was not prepared for was the "ghost modulation" I get when I switch in the additional 4 stages.  Everything else about adding the 4 stages is nice, but there is a little bump after the peak of the upward sweep, like there is a hesitation that is finally overcome to complete the sweep.  At medium and faster speeds, it takes away from the smoothness of the sweep.

I thought that maybe it was a matter of trying to feed too many stages via the same LFO output resistor, so I used equal-value resistors from the LFO to feed each set of 4 stages separately.  I tried 10k, 9k1, and even 8k2, but to no avail.  I still get the little hiccup at the end of the sweep.  Has anyone else experienced this?  Do you have any idea why it might occur?

Does it have anything to do with differences in the OTAs I used? (LM13600 for the first 4 stages, and a JRC and LM for the additional 4)  Puzzling.