Peak detector with discharge

[Harvs]

Gday all,

I'm designing a new ignition module for the old Ducati 2V engines, primarily for interest’s sake.

At this stage the plan to use a uC to control everything (PIC32 at the moment), with a pretty standard inductive/flyback setup using an ignition coil driver IGBT.

However, rather than just use the maximum internal current limiting of the coil drivers (which isn’t suitable under a number of circumstances), I've been putting together some circuits to add:
- Variable peak current limiting to the coil primary set by a DAC output from the uC,
- Feedback to the uC of when it has hit that current limit (so the uC can actively manage dwell time),
- An ADC input of the peak primary coil current, so the uC knows what current was reached if it is not hitting the target current.

Anyhow, for the third point I’ve been experimenting with a peak current detector circuit that came out of the TI datasheet.  Works well for the application with careful selection of diodes and cap for low discharge.  However, what I want to do is add a uC controlled discharge to the cap.  So the idea will be, after the uC turns off the IGBT, wait a hundred or so uS’s for the main peak spark to occur (as it generates a pile of EMI), then sample the peak current, then discharge the detector for the next coil charge cycle.

Obvious thought was to add a small FET across the Cap.  However I’m struggling to find anything with a small enough off state drain-source leakage current. Particularly when you look at the higher operating temp leakage current.  Needs to be in the order of several nA for the cap values not to end up too large for the opamp to drive, and still maintain reasonable accuracy.

So I’m hoping for some better ideas?  Maybe a completely different peak detect & hold circuit?  Or am I looking at the wrong type of devices to discharge the cap?

[Harvs]

Sorted... I think.  Increased the holding cap to 12nf, added a 300R resistor to the output of the first opamp stage for stability and put a 2N7002 mosfet across the cap.  At this stage, all appears well. 

I guess the real test will be once I warm it all up in an oven and see how that effects the leakage rates.