3KW BLDC Controller/IR2101/ATMEGA328p Failure

Thread Starter

Driver2021

Joined Jan 21, 2021
11
Hi everyone,

As a university graduation project, I am designing a 3 kW motor driver. For this, I wanted to make a very simple design and make improvements on it. For this reason, I made the design that I shared the schema of, using datasheets and open sources on the internet, and produced the pcb.

The tests I did were very good at the beginning (all of course with no load and small bldc motors). Then I started testing with big motors and under load.

That's when things started to go wrong. After the free-whelling situations and in some other conditions, the 1n4148 diodes and IR2101 ICs started to explode. I've had burnt Atmegas a few times as well.

I wanted to refer to your experiences as to why this happened.

Regards
 

Attachments

kaindub

Joined Oct 28, 2019
70
In4148 diodes are signal diodes. Go back to the manufacturers data sheet and use what they recommend. Also 3kw is not a trivial load, given your Vin is less than 100 v.
 

Sensacell

Joined Jun 19, 2012
2,838
Post photos of your setup. At the power levels you seek, layout and wiring is critical.

Gate drive currents can reach ampere levels - 1N4148 is going to die for sure.
 

Thread Starter

Driver2021

Joined Jan 21, 2021
11
I have seen applications on the internet that use both 1n4148 and 1n5819, so I used 4148. So should I try 5819?

I have 3 different boards, it works fine at the beginning, but they are gone after several tryings around 60 V.

The gonners are always atmega chips and diodes
 

Sensacell

Joined Jun 19, 2012
2,838
Read the datasheet and application notes for the IR2101.

Don't trust random circuits that you see on the internet, so much of this is garbage.
Trust the people that make the parts.
 

kaindub

Joined Oct 28, 2019
70
I worked for a motor drive manufacturer for a number of years. Testing and finding the weak links is part of the process of development. We had a whole room full of failed designs. A good set of diagnostic tools is imperative. That’s usually beyond the budget of the backyarder. In that case you have to go overboard with the specification of devices to ensure a super safe margin.
 

RPLaJeunesse

Joined Jul 29, 2018
165
Your USB and control scheme shares the same ground as the power stage. That's a recipe for disaster. Any load inductive kick can perturb ground enough to cause unplanned ground currents and cause magic smoke to appear. Also makes safe test equipment connections difficult.
 

Thread Starter

Driver2021

Joined Jan 21, 2021
11
Your USB and control scheme shares the same ground as the power stage. That's a recipe for disaster. Any load inductive kick can perturb ground enough to cause unplanned ground currents and cause magic smoke to appear. Also makes safe test equipment connections difficult.
Actually, power ground and signal ground are seperated in pcb, they are only connecting in one point but usb ground shares the same plane with usb. Does it cause a problem ?
 

RPLaJeunesse

Joined Jul 29, 2018
165
Signal (in this case control) and power grounds on the same PCB plane are generally not a problem, provided they are kept apart. But for high power applications the power stage should be galvanically isolated from anything that 1) comes out of the box and 2) should be touch-safe.
 
Top