Hi,

I am producing PnP ECUs for mk1/mk2/mk2.5 Miatas and encountered following issue I cannot resolve

Im using following alternator control originally designed by Jason Cuadra. Usually it works very good. But some customers (one out of 20?) have overcharging issue above around 5000rpm. Protection circuit activates, and voltage drops. But its not a solution.

Thinking for pretty long time what is cause and remedy and found nothing.
Experimented with increasing C111 value from 1nF to 2nF, C112 from 22nF to 27nF, added 220nF+47uF electrolytic blocking +12V

I attached logs sent by one of customers. This is mlg file with .txt extension added to be allowed for upload, just cut it to watch

do you have any ideas what is going on and what to do?

I know that installing mk1 NA alternator will work, but searching for something else

 

finally I have answer for myself ;-)

I analyzed circuit with Falstad.com online simulator and found following:
- circuit has very small hysteresis. Not 50mV like Jason Cuadra wrote but rather 10mV. Not ideal. Measurements confirm this.
- TL431 regulator is prone to oscillation that usually gives higher duty cycle than desired (=higher voltage). According to Texas Instruments application note slva482a.pdf it should NOT but simulation says it DOES

graph show: input voltage output voltage voltage on TL431




so I made following modifications:
- added zobel circuit to output of TL431. Found 22ohm and 100nF optimal
- added input filter (10ohm 47uF). Series resistor 10ohm also adds hysteresis to circuit.

simulation show circuit is stable now, regardless of conditions. Even strong noise on input does not drive it crazy



also real world tests by customer prove it works much better now