Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

More fun with a simple boost converter (3V--> 9V)

Status
Not open for further replies.

carbonzit

Active Member
Some of you may have followed my adventures here with trying to create a simple yet effective boost power supply giving 9 volts from 2 AA cells. (The last installment was this thread.) As things stood last, I was working with two competing designs: my own, consisting of a multivibrator turned into a boost converter, with the 2nd transistor acting as the switch; and a design of Mr Al's, an interesting 2-transistor switcher whose operation I still don't fully understand. (I just built Mr Al's version last night; so far, it's winning against my design, at least in terms of output.)

One of my goals was to see how much I could do with as little circuitry as possible; part of my motivation is to see if I can give the "Minty Boost" a run for its money without using a dedicated IC like everyone else seems to do). I want to see what can be done with just a few transistors.

However, it struck me last night that maybe I was going a little too far in that direction, with my multivibrator/switch combo. I may be simply asking too little to do too much. So I'm trying a different tactic. I still like the idea of basing this on a simple multivibrator, but it occurred to me, why not just use that to drive a MOSFET switch?

Here's the block diagram of the thing (your basic boost converter topology):

**broken link removed**

The circuit I'm currently simulating is posted below (Spice). After initially failing to work at all, it's giving pretty decent performance, at least in simulation. I can get nearly 6 volts at 600mA out of it (for comparison, the Minty Boost claims 5V @ 500mA), on a 2.75V power supply (partially-depleted AA cells).

Unlike many other circuits where component values aren't terribly important, everything here seems to depend on very small adjustments. In particular, the following seem critical:

  • The oscillator frequency, obviously (set by the two capacitors and two base resistors), as well as the duty cycle. I've tried to lengthen the on time for the driver (signal "mv1") as much as possible by tweaking the capacitor values.
  • The coil inductance (tied to frequency).
  • Even changing the MOSFET gate resistor seems to have a big effect on the output.

I have some specific questions about this. Rather than go off on a dozen tangents, I'm really hoping that someone can focus on just these questions and help me understand things better. Once again, keep in mind that this is the farthest thing from any kind of commercial product idea. I'm doing this because I'm interested in small, efficient power supplies, but equally important, I'm trying to teach myself some basic electronic principles. If anyone else learns along the way, so much the better!

Here are my goals and criteria for this project:

  • Small & compact; I currently build small projects in Altoids or similar tins, and would like this to fit in one (again like the Minty Boost).
  • Run off cheap, readily-available power (AA cells, alkaline or rechargeable).
  • Two target voltages:
    5 volts, with enough current to run typical USB devices (my goal, as stated above, is to rival the Minty Boost's 500mA);
    9 volts, with an arbitrary goal of supplying ~250mA, which should be plenty for most 9-volt battery-operated devices/circuits I'm interested in.
  • Well regulated, well-filtered supply that could be used with audio devices without hum, noise or other bad things.
  • Suitable for most loads; reasonably non-reactive ones, at least. Only need to handle small to moderate inrush currents; no high starting current devices like DC motors.

So my questions:

1. While it performs reasonably well as-is, there are some things which bother me about it. The first on is the driving waveform. Look at this snippet:

**broken link removed**

The top trace is the multivibrator output, while the bottom is the MOSFET gate voltage. The gate is getting an extra serving of AC here in the form of ringing on the rising edge; what's up with that? After all, this is measured on either side of a resistor. I can only guess that the ringing has something to do with the inductor current somehow spilling over in the the gate.

In any case, I don't like the waveform much. Mushy; that rising edge should ideally be like a cliff face. Is it possible to get any better performance out of a lowly 2-transistor multivibrator? Something to shape that into a cleaner square wave? and get rid of that ringing, too. That can't be doing my efficiency any good.

2. I've never worked with MOSFETs below, and just grabbed one from LTspices's parts box that looked like it might work OK. Can someone suggest some devices that would be appropriate for this, based on my power requirements? Remember, I'm not after overkill here. Ideally this would be something small, in a TO-92 package, or similar, not a big FET requiring a heat sink. I'm using a N-channel; is that the right way to go here?

3. This oscillator runs a bit on the slow side (~11kHz currently). I'd like to get that up somewhere closer to the 30-40-60kHz neighborhood if possible, but I have problems with the simulation just not starting if I decrease the caps too much. There seems to be a fairly delicate balance running a multivibrator on such a low voltage. Any pointers on why that's so and how to work around it would be appreciated.

4. Obviously this circuit is lacking regulation (I'm trying to get the boost working first). In the interest of simplicity, I'd like to use my classic poor-man's regulator circuit:

**broken link removed**

In my other design, I simply connected the collector here to the base of one of the multivibrator transistors. Works surprisingly well. Any other ideas appreciated. (Simple ones.)
 

Attachments

  • Boost conv FET 1.asc
    3 KB · Views: 214
The gate-source of a Mosfet has a fairly high capacitance because a power Mosfet is made with thousands of Mosfet cells in parallel. This capacitance resonates at a high frequency with the inductance of wire feeding the gate so a low value resistor is used in series with the gate at the gate to swamp the resonance.

Your multivibrator has its collector resistor pulling the gate high but the capacitance of the gate causes a slow ramp instead of a sharp cliff (if a high current source is used to quickly charge the capacitance).
 
The gate-source of a Mosfet has a fairly high capacitance because a power Mosfet is made with thousands of Mosfet cells in parallel. This capacitance resonates at a high frequency with the inductance of wire feeding the gate so a low value resistor is used in series with the gate at the gate to swamp the resonance.

Your multivibrator has its collector resistor pulling the gate high but the capacitance of the gate causes a slow ramp instead of a sharp cliff (if a high current source is used to quickly charge the capacitance).

OK, so that being the case, how would one go about fixing this (assuming it can be fixed)?

When you say "power MOSFET" I get a little nervous, envisioning some humungous device needing to be bolted down. Remember one of my questions was what type of FET to use: what is your definition of a "power" MOSFET (vs. a non-power one, I guess)?

Can anyone suggest an actual real-world FET that would work here?
 
An IRFZ44 "power" Mosfet has a very low on-resistance of 0.028 ohms max so it can conduct up to 50A with such a low loss that it barely gets warm. It is not big (TO-220 case).
A little 2N7000 Mosfet has a fairly high on-resistance of 5 ohms max so it can conduct a max of only 200mA.

The differences of their gate-source capacitance are huge. The power Mosfet needs a high current driver to quickly charge its capacitance.
 
Try making the inductor larger. Say 30-50Uh

Nope.

I actually started with a much larger inductor (100µH), since that was what had worked in a previous design. But the output went waaaay up when I started reducing the inductance. (The Minty Boost, by the way, uses a 10µH coil.)

So what can you tell me about FETs?
 
Last edited:
So what can you tell me about FETs?
Simply look at the datasheet of a suitable Mosfet. It shows its typical and maximum on-resistance at a certain Vgs and temperature. It shows its GS capacitance. Then you will be able to design a circuit to quickly turn this Mosfet on and off.
 
There are several things going against you right now:
1- You only have a 2.5 volts supply to turn on the FET so this one is kind of in no mans land. A logic level FET would maybe help.
2- The rise time of the signal driving the fet is very slow - you would like it to be fast so it is either on or off. This is really the problem that will be hard to fix.
3. If I remember correctly the circuit draws almost 6 amps from the battery, so as it turns on the battery voltage falls which also slows down the turn on.
4- Once the inductor current falls to zero there is ringing. This is the source of the noise that is causing it to turn on and off due to the capacitance. Capacitance is usually the reciprocal of current capability of the FET so pick one with low gate capacitance that will meet your current needs.
After all that, the only thing it really hurts is that the FET disapates more power. The ringing may eventually get all the way back into your multivibrator, messing it up.
You could add a small FET inverter between the multivibrator and the FET gate and spiff things up a bit.
 
Fairchild has a portable solutions brochure for different types of low voltage FET's. I have used an IR pFET part (irf7304) that has a really low gate charge and a guaranteed 2.5V RdsON. It will drive a 500ma load with a micro, or cmos logic, gate drive no problem. Not sure what the nFET compliment of that part would be, but worth looking into.
 
There are several things going against you right now:

Hey, tell me about it! It's not like I picked a project that's a piece of cake.

Just to address one of your points:

3. If I remember correctly the circuit draws almost 6 amps from the battery, so as it turns on the battery voltage falls which also slows down the turn on.

Nope. According to the simulation, at least, battery current is below 3.5A. Which isn't bad considering that the converter is running near the extreme. I'm trying to simulate the worst-case condition, a fully-loaded supply. Obviously, it won't be able to supply this much power for long, but it does show that the circuit is pretty robust.

Also, check my new thread out which shows the improved gate drive waveforms after a couple more tweaks to the circuit. As I said, all the gains here seem to be made at the margins, by changing component values slightly.
 
Last edited:
I suggest you can the mosfet idea! The multivibrator relies on the ramped charge voltage of the cap crossing the specific 0.6v Vbe threshold of the transistors giving fast clean switching. With a FET there is no clearly defined Vgs "switching voltage" and you will get really soft switching and terrible performance.

The 2 transistor "power" multivibrator is the best solution which I why I suggested it in your original "Colins boost converter" thread, and helped you refine it in your last thread (as lots of people did) and took the time to provide you with the single transistor regulator to add to it (that has worked well in other simple converters).

All you really need to do is increrase the inductor size and change some timing parts to tune that circuit, I even suggested a tuning methodology in your last thread.

Have you actually built one yet and done some real parts testing or are you still punching numbers into a simulator program? The building and measuring is the fun part, the simulator won't cut it there. :)
 
The 2 transistor "power" multivibrator is the best solution which I why I suggested it in your original "Colins boost converter" thread, and helped you refine it in your last thread (as lots of people did) and took the time to provide you with the single transistor regulator to add to it (that has worked well in other simple converters).

Yes, I well remember this, and appreciate the input from you and others there.

All you really need to do is increrase the inductor size and change some timing parts to tune that circuit, I even suggested a tuning methodology in your last thread.

Have you actually built one yet and done some real parts testing or are you still punching numbers into a simulator program? The building and measuring is the fun part, the simulator won't cut it there. :)

Yes, I did try building it, using the best latest result of the simulation. All I can say is that the results were extremely disappointing. When I could get any output at all from it, it was very weak.

I did try increasing the inductance: I bought myself a "kit" of coils from Digi-Key, including 250 and 500µH, but far from improving things, they actually made it run worse.

I may come back to this idea later; in fact, I fully intend to. In the meantime, I'm getting very encouraging results from this MOSFET multivibrator (should I call it "MultiMOS"?), at least in simulation.

Besides, having never worked with a single FET in my life, this is the perfect excuse to learn how them critters work ...
 
CZ,

Think about this one:
A Cmos 555 driving the Nfet. Only 50 micro amps wasted in the Cmos 555.
 
Nope. No chips allowed. Only 3-legged devices here.

Hell, if I was to do that, I might as well just give up and use one a'them Maxim chips or whatever.
 
Last edited:
How many separate Mosfet transistors will it take to make a cheap Cmos 555? Eight? Ten?
A blocking oscillator is made from 1 or 2 transistors in a cheap solar garden light. The second transitor simply turns it on and off so its oscillator is only one transistor, not a multivibrator that uses twice as many transistors.
 
Yes, I well remember this, and appreciate the input from you and others there.

My apologies Carbonzit, I was wearing my "grumpy old man" hat, being grumpy that you keep starting a new thread every week on the same project so contributions made by myself and others have just been lost on the stack instead of being all available in one good "open source 3v-9v converter" thread that could have benefitted lots of people as this is a useful hobby application.

...
Yes, I did try building it, using the best latest result of the simulation. All I can say is that the results were extremely disappointing. When I could get any output at all from it, it was very weak.

I did try increasing the inductance: I bought myself a "kit" of coils from Digi-Key, including 250 and 500µH, but far from improving things, they actually made it run worse.
...

That surprises me, since the 3 tran circuit (2 tran power mutlivibrator + reg) I presented in post #23 of your other thread; https://www.electro-tech-online.com/threads/attempt-at-a-simple-boost-converter-3v-5v.119773/?
is one that has worked well for me over the years. The multivibrator switches clean and square and can be adjusted for period and duty reasonably independent of the inductor (similar performance to SMPS ICs) and likewise the regulator holds the multi in a stable OFF mode for the entire time the output is higher than Vref, again mimicing the performance of SMPS ICs. With the right parts values it's performance is very good.

If I get some time later I'll throw some parts values in that circuit and take some measurements.

Besides, having never worked with a single FET in my life, this is the perfect excuse to learn how them critters work ...

Ok I respect that! :) The power multivibrator is not a good choice for a FET due to the slow ramping up of the Vgs voltage. The best fix might be to use 2 NPN for the multi to give good square waveform, then use that waveform to drive the FET. Total solution; 3 NPN, 1 FET.
 
My apologies Carbonzit, I was wearing my "grumpy old man" hat, being grumpy that you keep starting a new thread every week on the same project so contributions made by myself and others have just been lost on the stack instead of being all available in one good "open source 3v-9v converter" thread that could have benefitted lots of people as this is a useful hobby application.

Well, I guess that is a downside to starting new threads. I'll try to minimize it in the future.

That surprises me, since the 3 tran circuit (2 tran power mutlivibrator + reg) I presented in post #23 of your other thread; https://www.electro-tech-online.com/threads/attempt-at-a-simple-boost-converter-3v-5v.119773/? is one that has worked well for me over the years.

Did you even look at the LTspice simulation I posted in the very first post in this thread? It's pretty much exactly the circuit you suggested (link to your post here), except that it uses the MOSFET (I actually built the circuit as you suggested, with the 2nd multivibrator transistor acting as the boost switch). Here's the schematic for those who don't have LTspice:

**broken link removed**

The multivibrator switches clean and square and can be adjusted for period and duty reasonably independent of the inductor (similar performance to SMPS ICs) and likewise the regulator holds the multi in a stable OFF mode for the entire time the output is higher than Vref, again mimicing the performance of SMPS ICs. With the right parts values it's performance is very good

I went back and played with the "power multivibrator" simulation, tweaked it a bit, and then breadboarded a version. (LTspice file attached below.) Here's the schematic:

**broken link removed**

Here are the results I got from this:

With a 100Ω load:

Vo: 9.1V
Io: 91 mA

It won't give enough voltage as shown, with a 50Ω load, which is disappointing. Can't even make it at 75Ω. (The LTspice simulation works great at even heavier loads; unfortunately, the real-world model doesn't work quite as well.)

The power multivibrator is not a good choice for a FET due to the slow ramping up of the Vgs voltage. The best fix might be to use 2 NPN for the multi to give good square waveform, then use that waveform to drive the FET. Total solution; 3 NPN, 1 FET.

Butbutbut ... that's pretty much exactly what I showed above, no? (Minus the last transistor for regulation, of course).

I'm still not convinced that a MOSFET driver is not the way to go, but that's at least partly because I just want to finally work with one of those devices. I still could really use information on selecting a specific MOSFET.
 

Attachments

  • Boost conv reg 1a.asc
    3.9 KB · Views: 157
The gate of your Mosfet has a high capacitance. It is slowly charged by the 1k collector resistor R4 plus the 220 ohm gate series resistor R5. The slow charging causes the Mosfet to slowly ramp instead of quickly switch.
 
hi cz,
This is a bootstrap option, needs a little more work.

Change the 8V2Z back to your model
 

Attachments

  • AAesp01.gif
    AAesp01.gif
    40.1 KB · Views: 415
  • Boost conv reg 1x.asc
    3.8 KB · Views: 163
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top