Kai Deng GTI300W Grid tie inverter reverse engineering

[vasonline]

Changes of today:
(had some time & some sun on the solar panels)
+ Improved mppt tracker. - The vasonline way (tm)

Die kiste lauft güt
(just joking )

Still to to do:
- WDT
- Would like to reduce the cpu load of the CCPRL formula.

[vasonline]

+ Changed the mppt a bit again.

Not much changes for the rest. Will have to wait a bit.
No bright sun over here currently
But it's no more cycling in low light

[vasonline]

+ Changed the mppt to a fixed value around 16,8V

Will try to find if we can get a current reading from RA2 & RA3

They share the same signal. But it's pulsing, with the same rithm of the PWM...
Not quite sure how does one read/interpret that.
Will try to make a movie from it.

[vasonline]

New code update:

+ Improved mppt tracker...Still testing.
+ Improved grid detection: compares time of high period with previous (high) period.
+ More "#define <value>" to make modifications easier.

Wish list:
- Find out why WDT does not work.
- Find out what's the purpose of RA2 and RA3's input. It show a (sort of) delayed version of the PWM pin.

[vasonline]

Here is the input of RA2 (RA3 is largely identical) Voltage is 1V or less, excluding the spikes.
I'm not sure which useful function I can do with it.

What would the original manufacturer to do with it?

[bernhard8]

Are this pins connected to the circuit outside? Maybe you see just spikes and influences from outside? Just an idea...

[vasonline]

Hi Bernhard,

Thank you for your reply.

I'll have to look to where they are connected to. But one of the inputs is definitly coming from around the DC side.
I did not mention it: my movie is what you see when it's increasing the current from 0A to 3A. (the max my psu could deliver)

It could be indeed noise, and not anything usefull. I "saw" the same voltage with my multimeter.
Curious that they went the trouble of connecting these pin's of the pic to the circuit.
If the PWM is off, then there is nothing interesting to see on them.

if you discard the spikes:
- It's a low signal: well below 1V at max. Even not 0,5V if I recall correctly.
- Since the PIC's microcontroller's ADC voltage reference is 5V / 8 bit there is limited resolution in detecting a precise value.
One could detect if the signal is present or not...but I could be wrong...it's not increased much - in voltage - at 3 amps.

[vasonline]

+ Mppt update/improvement: evaluates after tacking back a step - vs - or - vs +

[vasonline]

+ Watchdog (WDT) now works
+ Slight modification of temparature control and mpp tracker.

Reason was this:

HI-TECH C Compiler for PIC10/12/16 MCUs Version 9.81 Release Notes
Delays and the watchdog timer Note that if the _delay feature is used to insert a simple delay sequence or
loop, the CLRWDT instruction may be used in this sequence. If the watchdog timer is not being used, this
is of no concern. If the watchdog timer is enabled, then be aware that delays may reset its timer.
Deze delay_us doet een CLRWDT() ! Getest & werkende bevonden.

[vasonline]

Latest updates
Development was halted for quite some time.

+ Best mpp tracker to date.

- Fan is disabled for now. ---> important!!
It interfered with the mpp tracker.....
I might or might not fix that. The original uses a "software" pwm like signal (I guess).


My best mppt code:
(never tought it got so long! I made a flowchart first!)