Summing node/S+H generator question

[halfmast at clubmet.metrobbs.com]

As my first experiment with sample and hold, I sampled some white noise
and applied the output to an attenuated CV input of my VCO.  A problem
arose as I lessened the attenuation.  A peak detector connected to the
summing amplifier of the VCO would register 2.5v (which is really only
half of the s+h's theoretical swing--it gives +-5v unattenuated), the
VCO's frequency would go up past an audible level, and then the peak
detector reads 12.5v. The thing is, the oscillator can handle such a
swing when my LFO is controlling it; so why can't I get the full
swing from the S+H?  I should probably also add that it would never
reach the high frequencies without becoming unstable, and that the S+H
never has problems when its not connected to the VCO.  Any ideas?
Thanks in advance for any input.

Evan Keller
halfmast at clubmet.metrobbs.com