Boost converter design. Switch getting hot.

[esgeroth]

I've made a few boost converters mostly to get 12v out of a 3.7v 18650 cells to power small motors or led strips, but now I'm trying to make something a bit bigger. I'm using a 12v lead acid battery and boosting to 34v to power a 100watt led. I've attached a picture of the circuit I've built. The mosfet is actually a IRF3205 but it doesn't exist in LTSpice so I selected one that has similar characteristics. The converter can boost up to about the needed voltage with the led attached. The switching signal comes from an atmega328 microcontroller. There is a driver circuit that will open the mosfet with the 12v from the battery. The mosfet gets very hot after a few seconds and will start smoking if I don't disconnect. It is attached to an aluminum heat sink to try and keep it cool. The IRF3205 datasheet says it should be able to handle up to 110 Amps. I've measured the input current at only about 7 Amps. Is there any other reason the IRF3205 would not be suitable for this circuit? The 12v should be enough for it to turn on completely.

[slugrustle]

It looks like your mosfet is on when the MCU I/O pin is either low, floating, or pulling up weakly. Be mindful of that during MCU startup, sleep, and reset.

Slow switching might be an issue. In particular, the IRF3205 datasheet figures for switching time assume a 10V source and 4.5ohm gate resistor, for a max current of ~2.2A. Q2 is probably not giving you more than 200mA, due to the way 2n3904 gain drops sharply at higher collector currents. Ditto for Q1. There might also be additional delay in the mosfet off->on transition if Q1 enters saturation.

You probably don't have to meet EMC requirements, so upping your switching speed will only help.

What does your control scheme look like? Is it continuous conduction mode or discontinuous conduction mode?

[David Hess]

If nothing else is wrong, then L1 which you did not describe is saturating.

[ahbushnell]

I would use a gate driver chip. 

Do look at the gate voltage.  Gate resistor?  The gate might be oscillating. 

How carefully did you lay out your circuit?  I hope it's not on a bread board.  That can cause problems.

Andy

[esgeroth]

Unfortunately my scope situation is inadequate. I've been using my sound card with soundcard scope software. I can see wave forms at low frequencies but it wont give me very good readings on actual voltage or current levels. This is high on my shopping list but I can't do it yet.
Slugrustle you're right, I'll look into pulling Q1 high instead of low to fix this.
I have some TIP41 transistors that I could swap in instead of the 2N3904.
I'm not sure about the control scheme, without a scope I can't tune it fine enough to determine.
The circuit started on a bread board but I moved it to a proto board once it was working.
The inductor could be saturating, wouldn't that cause the inductor itself to start getting hot as well? I pulled it off of a ups battery backup board so I don't know much about it other than its inductance. I have some others that I purchased that are rated at 6 Amps. I'll try those instead. I already had some gate drivers on the way so I'll be trying those when they arrive.

[T3sl4co1l]

Some reference may be useful:
https://www.eevblog.com/forum/projects/building-a-simple-switching-circuit/msg1252706/#msg1252706

Your circuit has an unforgivable sin: no current control.  This alone makes Arduino* unsuitable for purpose, because a digital control needs to sample that input very frequently (at least once every few cycles).

*Assuming the base AVR model. Upscale versions might be able to, though probably not from within the Arduino libraries.

Much better to do it in hardware, and it really doesn't take a whole lot, as you can see.  The circuit pictured above is actually exactly what you need, without any changes to the controls; you'd add a stronger gate driver IC to drive a large transistor, and a snubber to the transistor to protect it from itself, and all the other power stuff scales up, obviously.

Tim

[rx8pilot]

Why not use any of the many boost controllers with integrated gate drivers, current controls, loop compensation, etc, etc.

Easy and reliable.

Short and misplld from my mobile......

[james_s]

You're really going to need a decent scope if you want to design switching regulators. Actually "decent" can be used loosely here, even an old 15MHz analog scope that you can get for a song with a bit of luck will do the job.