Car power to constant current 3.6A 50v boost converter

[CommanderLake]

I gave up on the AL3353 for the above reason and went back to the MP4008 and finally settled on the PCB design and ordered it and the components, now we wait 10-15 days and hope it works after I assemble it.

 

[CommanderLake]

I made a circuit on a breadboard using the inductor and MOSFET for the prototype board with the 20Ω gate drive resistance of the MP4008 from the driver and a Picaxe to generate the PWM signal so I could see what kind of efficiency to expect.

I dont have a reliable way to measure the low current level reliably so with the LED firmly attached to a BIG heatsink to keep the temperature stable I first measured the voltage at the LED at 41.6V with 473mA at the bench supply going into the boost converter after subtracting the 20mA overhead from other stuff.
That's 5.676 watts going into the boost converter with a 12V input.

Then I connected the LED directly to the bench supply and set it to the 41.6V I measured at the LED with the boost converter and verified it with the oscilloscope, the current at the bench supply was 125mA which makes 5.2 watts at the LED.

So dividing 5.2 by 5.676 gives us 0.916, 91.6% efficiency!

The Fairchild FDP150N10A is a fantastic MOSFET for this application, the switching losses seem to affect the efficiency more than the on resistance (up to a point) with the 20Ω drive resistance of the MP4008, of course this is without the shunts but its still pretty good.

 

[CommanderLake]

Sorry to keep posting by my self but just for the record, upon further analysis using a 70mΩ 1% shunt the efficiency at 42.15V 0.1279A at the LED and 12v 0.478A into the boost converter mockup seems to be about 94%.

 

[Martaine2005]

Don’t apologise. It’s still an interesting read!.
I know I have personally enjoyed and learned from this thread.
Keep up your trial and error, that’s what this hobby is all about.

Martin

 

[CommanderLake]

I got the PCB and soldered most of the components then realised I didn't order the output caps... D'oh
So now I have all but the MOSFET and diode, 12V 1A regulator for the regulated fan output and the 2 big ceramic output capacitors soldered...

 

[CommanderLake]

I somehow committed to using a rare capacitor of which there only 4 in the world available to order and they are in japan and will take another week to arrive ordered from RS Components.
KTD101B107M99A0B00 try finding that.

 

[CommanderLake]

So I finally got the capacitors and finished it but theres a problem, it starts boosting but after about 1.5-2ms at somewhere around 20v the MP4008 gate output goes haywire then it shuts down.
After conversing with Gary Church of MPS I have narrowed it down to the ringing at the MOSFET gate when it goes low causing the MP4008 to become unstable so I'm going to cut the gate trace and bridge it with the ferrite bead.

 

[CommanderLake]

IT WORKS!!!
I replaced the 100Ω CS damping resistor with a ferrite bead in addition to the gate ferrite bead and then it was just a matter of supplying it with sufficient power as I'm using current limited bench supplies and the MP4008 latches off if the supply drops below 12v when its starting up or in operation.
At 13.6v 9A input its about 81% efficient and surprisingly the inductor gets quite hot even though it has a 2.4mΩ DC resistance and plenty of saturation current headroom.
There is also a bit of coil whine when adjusting the current limit but thats to be expected I suppose.

Actually my power wiring has a bit of resistance, the efficiency is more like 85%

 

[CommanderLake]

So theres a bit of a stability problem, it doesn't affect the light output but when I turn up the LED current to a certain point the gate drive signal goes crazy.
I think I've established by watching the inductor voltage and changing the LED current that if the MOSFET turns on again while the inductor is still at high voltage and discharging into the output capacitors it causes a large spike as the excess voltage discharges through the MOSFET and its interfering with the MP4008 on the current sense pin which monitors current going through the MOSFET.
I put a 180Ω at 100MHz ferrite bead on the gate and current sense pins but the spike is still making it to the CS pin and causing it to become unstable.
I need to block this spike going to the CS pin somehow but my options are limited on the PCB, I tried a 10nF film from the CS pin to ground but it does nothing.

Green is the MOSFET Drain voltage and red is the spike at the CS pin, the spike is about 9v and gets wider as I increase the LED current:

 

[CommanderLake]

I fixed the stability issue with a 100Ω/10nF RC to CS and now its about 90% efficient!