issues with capacitor discharge circuit causing system resets -- help wanted

Thread Starter


Joined Jan 25, 2011
Hi Guys,

I have been working on a circuit for a white now and ive just managed to get the chance to assemble it to test and im having some weird issues.

Description of Circuit:
The circuit is basically a regurgitation of a SiliconChip Magazine design "High-Energy multi Spark For performance Cars" from their December 2014 issue, i will not link the whole article here or pics directly from their magazine in order to avoid any copyright issues here for the message board, i will however post a picture of the circuit i have drawn up (or at least the relevant sections to this topic). The circuit is basically a High voltage generator using a TL494 to drive some fets that drive a coil to produce +300V DC. This 300V then gets fed to a L6571 High Voltage Half Bridge Driver with oscillator to charge a capacitor. The L6571 works like a SPDT switch, charging the 1uF cap in on direction and discharging it in another. upon Trigger, the 1uF cap is discharged through Q4 collapsing the magnetic field in the primary winding of the coil and you get your spark(s). So, sounds easy enough right.

Description of Problem:
i have the trigger to this circuit being controlled by a microcontroller, driving a 5V digial signal to a BC817 (NPN) (T2 in schematic), i have this trigger signal running at 14Hz. the issue is when i start the program to output the 14MHz signal to T2, the system constantly gets reset, as if the microcontroller is being either browned out, or a high voltage spike is causing it to reset. This kind of spurious output continues for a while until the microcontroller gives up and just resets to the start of its program (where no 14Hz output is given to T2).

Testing and Limitations:
The 300V HV circuit works fine i have tested the voltage with a DMM, it looks like the High voltage half bridge driver works when i signal a trigger to T2 however.. this is where the problem begins.

Below you will see two logic captures (they are only at 4Mhz, i have a crappy logic analyser) :

The first is what the signal "should" look like, the channel on the bottom is the output trigger to the CDI circuit. this is the only one that really means anything here.. everything from channel 0-6 is just diagnostics for my own peace of mind. Channel 7 shows a 1ms pulse to drive T2 which then trigger the Half Bridge Driver (U7), this is WITHOUT the R36 in place.. so nothing is getting driven.. this is what we should see.

The second logic capture is what i get when R36 is installed and the half bridge driver (U7) is being triggered. This is all in between the microcontroller being reset and picking back up the Trigger signal (channel 0) and sending the "CDIout p79" signal (Channel 07) to the half bridge driver.

What the problem could be?
1. there is some high voltage transients moving onto the GND plane after the 1uF cap gets discharged causing a host of problems for the power regulation circuits of the microcontroller.
2. Once the spark plug is extinguished, the collapsing field of the ignition coil develops a reverse current flow and this sucks out all the juice from the circuit causing resets?
3. There is not enough supply current and the circuit is being brown'd out when the cap is getting charged by the HV generator (there is atleast 10A of 12V supply availabel to this circuit, so i doubt it?)

limitation --> i dont have a decent enought scope that will record the GND plane voltage when the 1uF cap is being discharged.

Idea of Possible Solution:
Add some filter caps and diodes to the circuit to remove any HV spikes.. i need your help for this.. where and what exactly.
Gain some more information about the functional state of the circuit when the problem happens to help get a better idea of whats going on (data gathering)

Other info and Thoughts:
Throughout the SiliconChip article they mention trigger signals coming from automotive devices. like Hall Effect, Points, Reluctor pickup, Piranha optical picups... never a digital signal. i suppose that these traditional methods of triggering in engine are more "rugged" and might not be so susceptible to spikes. however.. when we are dealing with a circuit involving 3.3V and 5V logic these digital signals need to be isolated from high voltage spikes for sure.


Joined Jan 18, 2008
Some of the hobbyist projects with which I have been involved used and optoisolator in the MCU trigger line. They were much simpler designs.