Open Solar MPPT Charge controller

[Pineapple Dan]

A few years ago I made an attempt to build one of these -  https://hackaday.io/project/4613-arduino-mppt-solar-charge-controller. I think it was even before the updated version that has 3 IR2104 half bridge drivers but it did not go well. Just like a lot of people in the comment section I had problems with MOSFETs blowing

V3.0 schematic: https://github.com/deba168/MPPT_Master/blob/master/mppt%20controller_schema.pdf
V3.1 with extra IR2104: http://cdn.instructables.com/ORIG/F76/5DJ1/IEOP45ZK/F765DJ1IEOP45ZK.pdf

I recently built the half bridge part of his circuit with IR2104 for a fan controller and it works fine but I would be slow to try and build another charge controller with IR2104s as they seem to behave unpredicatably when the power source is removed which is bound to happen if it's driven by solar panels

Could I get rid of the IR2104 altogether and use a P Channel mosfet with low Rds(on) for the high side? I'm using a PIC16F1509 which has a complementary wave generator that can generate the waveforms for both sides including dead band so I'd only need the mosfet driver to generate a high enough voltage to switch on Q2
If I get a +12v supply with two opposing diodes from the battery and panel, and a regulator and use this to switch on a P channel and N Channel mosfet using a pair of BC547's controlled by the PIC would it have a better chance of working?

What about Q1? As far as I remember he added this to stop current flowing back towards the panel but is it still necessary if we get rid of the IR2104?

[fourtytwo42]

Looks astonishingly overcomplicated to me yet no attempt at overcurrent protection anywhere!
Could you start by stating your panel and battery charecteristics ?

[Pineapple Dan]

The panel is a normal 30W one that is to operate around 18 or so volts, open circuit 20-21v, though I'm hoping for a bit of headroom for expansion.

 I'll be charging 20AH 12v LiFePo4 batteries off it. This: https://files.i4wifi.cz/inc/_doc/attach/StoItem/2099/GWL-LP12V20AHB-Product%20Spec.pdf , but the version with built-in balancing circuit which I can't find the sheet for at the minute

[fourtytwo42]

If you have not invested to much money on components I would try and find a better charge controller design to use. There are many things wrong with this circuit on many levels and basically it needs re-designing. There is nothing wrong with IR2104's if used correctly but they are not here, for example the 1N4148 boost diodes are completely inadiquite for the purpose but thats just one problem of to many to count!

[jbb]

... Just like a lot of people in the comment section I had problems with MOSFETs blowing ...

What about Q1? As far as I remember he added this to stop current flowing back towards the panel but is it still necessary if we get rid of the IR2104?

• Q2 and Q3 form a synchronous buck regulator, which is good. The SR5100 across Q3 may or may not be worth the trouble (depends on frequency, dead time, layout etc.).
• Q1 seems to be some kind of reverse current block to prevent the battery feeding back into the panel.  It's in a poor place. It should be done at the battery terminal (where it will only see DC current) and not in the middle of the buck converter.
• No gate resistors on the MOSFETs! This could lead to spikes, ringing and shoot-though.
• The 43V Zener across Q3, and the RC snubber across the buck inductor suggest problems with switching waveforms, e.g. big spikes and ringing.  This indicates a) bad topology - putting Q1 where it is adds a load of unnecessary leakage inductance, b) possibly bad PCB layout and c) poor decoupling - I only see C1 and C2 electrolytics, when there should be some ceramic or film capacitors as well for the high frequency stuff (e.g. clamp voltage spikes).

So, the problem isn't so much the IR2104 per se. It's everything else...

Looks astonishingly overcomplicated to me yet no attempt at overcurrent protection anywhere!

Yes, it's got problems. In all honesty, I think there are enough problems that starting over makes sense.  Additional comments:

• There is a fair bit of stuff which is always live from the battery (especially 12V regulated rail, voltage dividers).  I suggest that the battery should be loaded as little as possible to reduce power draw.
• Generic buck converter modules may waste lots of power during standby.
• There should probably be a fuse on the battery terminal (note, battery voltage divider should tap off before the fuse and use multiple resistors so that single shorted resistor doesn't cause destruction).
• ACS712 current sensors are Not for New Designs. The suggested replacement, ACS723, has quite high operating current (like 10mA each) which will lead to unnecessary battery drain.
• No current sensing inside buck converter will force the use of voltage mode control, which is harder to do than current mode control