Need help with sawtooth vertical scan on homemade video camera

[Aussie Bloke]

G'day all.

I need help from any of you technicians who are cluey with circuitry design for scanning and deflection.

Basically I have a few months ago completed my vidicon camera project http://www.troysvintagevideo.741.com/mycamera.html and the camera is working but there's problems, most notably with the vertical scanning. I am however starting to understand what's happening on the vertical scan. As I understand TR19's base is fed a 1v pulse every say 20ms and during that pulse period TR19 goes to saturation and shorts the current at the collector to ground and then during the pulse off period the 4.7uF capacitor charges up to about 14.5V, and then when the next pulse occurs the process is repeated which is suppose to create a sawtooth waveform. The problem is however the capacitor charges up exponentially giving a non-linear ramp which results in a picture that's vertically squashed at the top but stretches as the vertical scan progresses down the screen. I have experimented with increasing the capacitance which resulted in a better slope that looks more like a sawtooth but the peak voltage drops as capacitance is increased which is the nature of capacitors as they filter out AC ripple, this results in a better picture with more vertical width but more darker due to drop in amplitude.

I need to find a way to generate a sawtooth waveform from that 20ms pulse coming from the MC14528 chip with a linear ramp, so I was wondering if anyone here knows of what circuitry I could implement between TR19 and TR21 that would give me a linear charge up ramp to a peak voltage of about 14.5v in a period of 20ms? Any help would be MUCH appreciated.