help with boost converter

cravenhaven

Senior Member
I am attempting to use the pwm output of a 20M2 to construct a simple voltage booster but have been unsuccessful in getting it working properly and was hoping that some of you might be able to point me towards the solution.

Basically my problem is that I cant get it to boost to my desired requirement of 5V @ 1 amp from an input of between 3 and 4.2V. The best I can get into a 4 ohm power resistor is about .9A, which is getting close but the efficiency is woeful at around 30-50%.

The inductor I started using is a PCB mount with a value of 47uH, capable of 3A and a resistance of .097 ohms (MULTICOMP MCBFS7330-470MU), but I also tried a 2.2uH PCB type and standard toroid style with values in the low mH range without any outstanding change in results.

My program simply uses pwmout to set a period of around 100uS and it then monitors the variable resistor to set the duty cycle. I tried other frequencies up to 100KHz but the lower frequency seems to work better. I also tried simulating the cct on LTspice and according to its result I should be achieving what I want :).

I have the cct built on a breadboard ATM and have monitored the waveforms on a little DSO quad oscilloscope.

The best result I get is with the 47uH inductor and a duty cycle of 30%, gives me an output of about 0.8A and efficiency of 50%. I have attached a representation of the scope output with these settings.

I should add that I tried have the 20m2 directly connected to the FET but it makes no difference and as far as I can tell, the scope traces seem to indicate that it is driving the FET well enough.
 

Attachments

fernando_g

Senior Member
You are not driving the Mosfet correctly, for starters, a Vgs of 3.5 volt is way too low to enhance the channel, even of a "logic level" Mosfet.
You compound the problem with a way too large gate resistor. Also, the Picaxe doesn't have the peak instantaneous current capabilities to rapidly charge a Mosfet either. Instantaneous currents may be a couple of amps.

I just built a 950 watt boost converter successfully with a 14M2 at 94+% efficiency. To drive a 1 HP 90 VDC motor from a solar array.

The trick was to use the proper gate driver IC (I used a FAN3100C, but there are many) supplied from 12 volts. The resistor gate is only 8.2 ohm, and I'm running at only 20Khz. Higher frequencies would demand lower resistors and an even more powerful driver IC.

There may be other contributors to the efficiency loss. You don't mention for instance the main rectifier diode. It has to be, at the very least, a schotky device. With the low voltages you are operating from, a synchronous rectifier would be even better.
 
Last edited:

Jeremy Harris

Senior Member
As above, you need LOTs of peak gate drive current to get fast switching, typically up to around half to one amp, to charge the gate capacitance quickly to get decent efficiency. The turn on and off time with just the tiny current a Picaxe pin can source and sink (the 100R resistor isn't doing anything in practice, neither is the 10k resistor, the limit is the ~20mA from the Picaxe pin) will be very sluggish, and is probably where most of the losses are.

Also, the FET you're using has a high Rdson, which isn't helping. Generally Rdson rises with increasing Vds capability - that's a 100V Vds FET, when you only need a much lower Vds capability. My switching controllers typically use FETs with an Rdson of around 3 mohms or so; that FET has an Rdson of 77 mohms.
 

cravenhaven

Senior Member
Thanks guys for the pointers. I had seen many discussions on FET drivers in the past (seemed to drive Dippy nuts at times :)), but when I looked at the scope traces as I tried to indicate in my sketch, it seemed that the switching was sufficiently rapid to allow reasonable performance. I'll attach photos of the actual traces here if anyone cares to comment further. The yellow trace is the gate input with the picaxe connected directly to FET bypassing the 100 ohm resistor, and the blue trace is the Drain of the FET.

Fernando, the cct diag showed the diode I'm using which IS a schottky and is rated at 1 amp.

My application is to drive a 5volt 1 amp device from a Li-Ion cell and I'm using the picaxe for other functions so thought I would try to run a boost converter from it. Guess I'll have to try harder :).
 

Attachments

cravenhaven

Senior Member
Just had a bit of a look for a low voltage FET driver and have been unable to find anything with a minimum supply voltage of less than 4.5 volts.
 

Jeremy Harris

Senior Member
I've made discrete bipolar FET gate drivers before now and they work well. Just a complimentary pair of reasonably fast switching transistors, driven by the Picaxe output. That way you can get the needed half amp or so into the FET gate to charge the gate capacitance (you may need a low value, say 10 ohm, gate resistor to prevent ringing).

As above, you can buy cheap and very efficient switch mode boost regulators for doing just this job. They are cheap as chips as they are used in all those millions of USB power packs. The efficiency varies a bit with supplier, so it's a bit of pot luck what you'll get, but if you can find a good source of them they are very useful things to have around. I've been using a (rather expensive) Sparkfun unit (this one: https://www.sparkfun.com/products/10968 ) for several years now, powering a Picaxe at 5V from a single NiMH cell charged from a small solar panel. It's only 200mA though.

There are units like this: http://www.ebay.co.uk/itm/3V-to-5V-1A-Charger-for-MP3-MP4-Phone-DC-DC-Converter-Step-Up-Boost-Module-CF-/272247684448?hash=item3f6339fd60:g:s~0AAOSwmc1XPTps with a 5V 1A output for under £1, for example.
 

fernando_g

Senior Member
Cravenhaven;
upon a second look, indeed it is a schotky diode. Thanks for pointing it out.

To your waveforms;
On the yellow one can you see it rises quite flat, then flattens and then jumps a little bit more? That is the Miller plateau. As you can see on the blue trace, that is where the Mosfet really starts conduction.
And you can see from the blue trace that said conduction is sloooooooow.

You are correct that gate drivers are not found below 4.5 volts. But you can bootstrap the output voltage to the driver to obtain the 5 volts.
all you require is a bootstrap diode for the initial startup. On the Sparkfun schematic, that is what D1 does.
 

Goeytex

Senior Member
1) The MBR1100 is only rated for 1 amp continuous current and 2.0 amp peak repetitive current. A more appropriate Schottky diode should be considered. Maybe something like an MBR735 would be more appropriate.

2) 47uH inductors can vary in resistance from about .05 ohm to up to 5 ohms depending upon the core material, wire diameter and number of turns. It is imperative to use the correct inductor. Do you know the resistance of your inductor? Do you know the peak current rating? (At what current will the inductor go into saturation?)

3) A 47uH inductor seems rather high to me but can be made to work if it has low resistance and is rated at no less than 5 amps peak current.

4) What does the current waveform through the inductor look like? Is the boost regulator operating in discontinuous mode (Current falls to zero) or continuous mode (current never falls to zero)?

5) I am an advocate of using mosfet drivers whenever possible. Advise that is often ignored here. However, at the relatively slow switching speed of 10KHz the IRL540 with a G to D threshold of ~1.5V should switch "OK" with the relatively the low gate drive current provided by the 20M2. Efficieny will suffer by about %5 or so compared to using a driver. This assumes that that IRL540 is a genuine Vishay or Infineon product and not some cheap Chinese or Asian sourced counterfeit POC. In this case all bets are off.

6) Does the inductor get hot? This suggests a high resistance or underrated inductor, or incorrect switching speed/duty, or some other problem.

A PSPICE simulation shows that the following combination "should" work OK with an efficiency of around 78%

1. IRL540 Driven by Picaxe through 100 ohm resistor
2. Coiltronics DR127-470R Inductor (47uH, 5.2 amps, .072 Ohms)
3. MBR735 Diode
3. Switchihg speed 5 KHz
4. Duty Cycle 35%

-----------------------------------------

I highly recommend that the inductor be reduced to 10uH and the frequency changed to 25KHz. The duty should be somewhere around 30% for 5V into a 5 ohm load. (1 amp)

A Coilcraft PCV-0-103-03 inductor models very well at 25KHz. Any quality branded inductor with similar specs should also work.

** A switching regulator circuit may not operate well on a "breadboard" due to high contact resistance and stray capacitance. Consider that switching currents can exceed 4 amps with this circuit. If you "must" develop on a breadboard keep all leads as short as possible. Connect the inductor directly to supply voltage instead of through a breadboard "rail".
 

cravenhaven

Senior Member
Thanks Fernando for helping me understand what I'm looking at.
And thanks Goey. in answer to your questions/statements:
- I constantly checked the temperature of the components and was surprised that the only thing getting warm was the load resistor.
- I would also like to know what the current through the inductor looks like, but I'm not sure I know how to do that. My first guess is to put a resistor in series with it and monitor the voltage profile across it, but I suspect that would alter the response of the cct.
- the inductor i was using is a multicomp MCBFS7330-470MU (http://www.farnell.com/datasheets/1679989.pdf) that is rated at 3 amps with resistance of .097ohms. They dont list the saturation current explicitly but show a current handling up to 4.5A @ 30uH so I guess it is still operating in an unsaturated mode at that current.
- I bought all the key components from element14 (Farnell) so assume they are all genuine.
- I tried a 2.2uH inductor (Bourns SRN6045-2R2Y, 6A sat) but couldnt get it to work properly at all, but upon simulating that combo I might try it again at a much higher frequency.
 
Top