I've built a couple of boost converters, one using a 555 timer and another using an attiny85 microcontroller. Both are working and boosting 4v from an 18650 cell up to 70+ volts. I've been able to regulate the voltage down to 12v. I tried running a 10 watt led and it lights up, but it isn't very bright and the voltage drops down to about 8v. Is it reasonable to expect to get 10 watts from a simple boost converter? I've tried running a simulation in ltspice using a 15ohm resistor to simulate a load. I've chosen components that match what I have except for the diode which is a sr360 schottky rectifier. It seems to hold steady in the simulation providing 16 volts. The attiny85 is producing a 27khz pulse with about 75% duty cycle. I've tried increasing the duty cycle but it doesn't seem to have much of an effect.
Is there anything I should try to increase current that this circuit can provide? Or is pulling almost 1 amp too much?

 

What you need to do is set your expectations correctly. The governing proposition for all DC-DC converters is as follows:

The output power in watts will always be less than, sometimes much less than, the input power"

Great you say, but how does that work in practice. Glad you asked -- here is an example.

The output power is +70 VDC @ 1 Ampere or 70 watts. Let us assume that you boost converter is 80% efficient. This means that the required input power is 70 watts divided by 0.80 which is 87.5 watts. This means that an input source which supplies +4VDC must provide 87.5 watts divided by 4 Volts or 21.875 amperes. I don't know many batteries that can supply that amount of current for an extended period of time without serious degradation in their output voltage.

When it comes to boost converters, more current out requires much more current in.

 

Absolutly agree RE power however:-

When simulating an led use an led or other diode, a resistor is not the same.

Batteries...
look to RC lipo packs 21A is not high, comparitivly speaking.
SLA's will be cheaper but much bigger, and again 20A will not be a problem.

Of course you will need a big battery to get any reasonable run time at nearly 90W.
What's your application?

Al

 

I don't have an application in mind yet. Although making a flashlight with that 10 watt led would be neat. At first I had a 12v solenoid valve that I wanted to open with a 3.7v battery. I've done that now I'm just seeing what else I can do with it.
I've also tried connecting two batteries in parallel. That should increase the amount of current I can get out of them. But the voltage still dropped to about 8v. I'm not sure the battery is the limitation here. If it was nearly maxed out I would expect it to start getting warm after a few minutes but it stays cool. Also I have a cheap Chinese boost converter I got on ebay that powers the 10 watt led from the 18650 cell at full brightness just fine. It does have a dedicated chip though just for that purpose. Also I've seen 18650 cells provide over 100 watts in some cases. So I'm pretty sure the battery can do it, at least for a short while.
So is there anything else I can look at to improve the efficiency and get a higher output?

 

...
So is there anything else I can look at to improve the efficiency and get a higher output?

You could switch to a dedicated controller chip. If your efficiency is at 80% and you want to try for 95%+, I think this is the way to go. You can use an external switch with a very low rds(on) and a toroidal inductor with larger diameter wire to minimize IR losses in the inductor. In a boost converter the peak inductor current can be twice the input current, so 42 Amperes or so. You need the "Crocodile Dundee" model inductor.

 

I'll be building one using a dedicated chip soon. I ordered some mc34063a chips that should arrive in the next week or so.
As for what I have now, I've got some IRF640, and IRF3205 mosfets that I've pulled from some atx power supplies that advertise low rds(on) resistance but I think they need a higher voltage to switch them. I'll get some low rds logic level mosfets as well.
I also have some larger inductors that I've pulled out of power supplies but I don't know what their value is. Do I need to run at a much lower frequency to use a larger inductor?

 

At a higher frequency, would a mosfet driver improve efficiency? Also, does the frequency itself have any effect on the efficiency of the circuit? Is there any other benefit to higher frequencies other than being able to use a smaller inductor?

 

The FET has losses when its on, see RDS ON, and losses asociated with switching when it is partially on/off
The first place to look is probably your gate drive, if it cant charge/discharge the gate quickly enough then the FET will run hot/ineffiaciant.

Higher freequency will allow a small inductor but comes with its own issues, not least increesed switching loss.
Think about an extreame case where by the time the FET was on it needed to turn off again...
OK so you probably arnt there but it illustrates the issue.

Do you have a scope? you are going to struggle to pin down the problems without one. It is essentally a prerequisite for building anything with magnetics, that is fast or switches a high current at almost any apriciable speed.

I would recomend yoiu read up on buck boost/buck design, there is plenty of stuff out there and to be honest I am no expert, although others here are much better than I am at this stuff.

Al

 

Trying to build switch mode power supplies with salvaged components is like trying to build a house with clarinet reeds and scotch tape. A serious effort would involve detailed component characterization, selection, and acquisition. Those components are then placed on a carefully designed printed circuit board. I'm all for learning by doing, but to achieve superior results you will need to change your approach a bit. I like the oscilloscope suggestion; it would be a good first step.

 

Are you saying you're converting 4V up to 70V, then down to 12V for the output? What's the point of that?

Regardless, the circuit needs feedback to maintain a constant output voltage. Without that the voltage will indeed drop under load.

 

The FET has losses when its on, see RDS ON, and losses asociated with switching when it is partially on/off The first place to look is probably your gate drive, if it cant charge/discharge the gate quickly enough then the FET will run hot/ineffiaciant.

Higher freequency will allow a small inductor but comes with its own issues, not least increesed switching loss.
Think about an extreame case where by the time the FET was on it needed to turn off again...
OK so you probably arnt there but it illustrates the issue.

Do you have a scope? you are going to struggle to pin down the problems without one. It is essentally a prerequisite for building anything with magnetics, that is fast or switches a high current at almost any apriciable speed.

I would recomend yoiu read up on buck boost/buck design, there is plenty of stuff out there and to be honest I am no expert, although others here are much better than I am at this stuff.

Al

The Tip41c does get quite hot. I've ordered some IRLU024 which has an rds(on) of .1 at 5 volts. I don't have a scope but I've been looking. I'm not sure I can fit it in my budget yet. I know it would make this much easier.
I've read quite a few pages and even watched some youtube videos about boost converters. I get the basic idea, but too many of these articles just tell you how to put it together like its a bunch of legos.
Here are a couple of the links that I did find informative:
http://www.simonbramble.co.uk/dc_dc_converter_design/boost_converter/boost_converter_design.htm
http://www.dos4ever.com/flyback/flyback.html
And on youtube the post apocalyptic inventor has a series on switch mode power supplies:

I can't pretend to understand it all yet but it's been informative.

Trying to build switch mode power supplies with salvaged components is like trying to build a house with clarinet reeds and scotch tape.

Actually its more like building a switch mode power supply out of a switch mode power supply. As long as I have the right components why worry about where they came from? Although I do need another mosfet as the ones I have are for higher voltage.

Are you saying you're converting 4V up to 70V, then down to 12V for the output? What's the point of that?
Regardless, the circuit needs feedback to maintain a constant output voltage. Without that the voltage will indeed drop under load.

What is the feedback going to do to keep the voltage up? I've implemented a feedback loop to limit voltage output to 12v but I don't see how the feedback can increase voltage.

 

The feedback signal is used to vary the pulse width, so that as the output voltage attempts to fall, the pulse width adjusts to maintain the desired voltage. Integrated converter chips will have feedback input pins for this purpose.

 

I've hard coded the attiny85 controller with various duty cycles. I can get 12v from a 20% duty cycle. If I attach a load I have to increase the duty cycle. I've increased the duty cycle to 90% and still can't reach 12v with a load. I don't think feedback is the problem as I have been manually adjusting the duty cycle. I've already tried with max duty cycle.
As discussed I think the main problem may be the switch I'm using. The tip41c gets quite hot and I'll need to replace it.

 

Why don't you just boost +4V to +12V and regulate that process. There is no reason on earth I can think of why you need 70V to run a 12 V device.

But, without seeing the schematic of what you are doing and why, we are just guessing at the cause of your problems.

 

I certainly don't need 70v. I was just saying that that is what it will put out at about 80% - 90% duty cycle. But with a load, even with a 90% duty cycle, the highest voltage I get is about 9v. I havn't added a feedback to the attiny85 yet as I have been adjusting the duty cycle manually by uploading a new program to the microcontroller. I have added a schematic in the first post, as well as the ltspice circuit. It's just a very basic boost converter. I only need about a 30% duty cycle to get 12v but that will drop down to about 5v - 6v with a load attached.

 

So what you have is an unregulated boost converter with a fixed duty cycle. When the load demands a particular value of current the output voltage conforms to that current demanded at precisely the duty cycle you have selected. The whole idea is to select a fixed output voltage, and vary the duty cycle to supply the required current to the load at that fixed voltage.

 

Yes, and if the load demands more current than can be provided at that duty cycle then the controller would compensate by increasing the duty cycle. Correct? So what I'm saying is I'm already running at 90% duty cycle. Whether its fixed or not there isn't much more room to go up from there. That's why I don't think that feedback is my current problem. I could be wrong. I don't think that if I added a feedback and then programmed the controller to vary duty cycle as needed, that it will be any different than just setting a fixed duty cycle for testing. When I know that the circuit can provide the necessary current then I will add feedback and write the rest of the program.

 

Yes, and if the load demands more current than can be provided at that duty cycle then the controller would compensate by increasing the duty cycle. Correct? So what I'm saying is I'm already running at 90% duty cycle. Whether its fixed or not there isn't much more room to go up from there. That's why I don't think that feedback is my current problem. I could be wrong. I don't think that if I added a feedback and then programmed the controller to vary duty cycle as needed, that it will be any different than just setting a fixed duty cycle for testing. When I know that the circuit can provide the necessary current then I will add feedback and write the rest of the program.

You believe what you want to believe. I'm done for now.

 

I don't believe anything. That's why I came here to ask a question. If I'm wrong about something, or don't understand, you haven't really pointed out where i'm not understanding. I'm not sure what the attitude is for.

 

You might need a larger inductor. There are equations for switching converters to determine the proper inductor size. I don't have them in front of me at the moment.