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.

My BJT/FET multivibrator boost converter beats Minty Boost!

Status
Not open for further replies.

carbonzit

Active Member
Finally, I has did it.

For those who've been following, this is the latest installment of my attempt to make a simple yet effective boost converter (3 volts--> 5 or 9 volts) using only discretes (only 3-legged parts allowed), kind of a poor man's version of the Minty Boost (a popular 5-volt boost converter built in a mint tin to power USB devices, for those who aren't familiar).

(By the way, I realize I have several other threads on this subject, but thought this was remarkable enough to warrant a new one. But if folks think I'm starting too many new threads with this, let me know; I don't want to be a thread polluter here.)

Previously I came up with a very workable 3-transistor boost converter design (regulated) that produced either 5 or 9 volts from 2 AA cells. I've built several of these and have been using them to power small projects that would normally use a 9-volt battery. (I say "I came up with", but must give credit where credit is due: I would never have made as much progress as I did without the help of others, particularly Mr. Al and Roman Black, one of whom supplied me the "power multivibrator" circuit.)

"My" design was one that I especially liked, since it was a mash-up of two classic circuits, a 2-transistor astable multivibrator and an inductor-based boost converter, using the second transistor both as part of the oscillator and the boost switch.

Problem is, that little BJT (a BC337) takes an awful beating in this application, and runs very hot, close to overheating. I tried replacing it with a small MOSFET, played around with a couple simulations in LTspice, but couldn't get anything to work.

Well, tonight I sat down and took it up again in the simulator. After just a little tweaking, the thing started working incredibly well. The figures in the box at right tell the story:

**broken link removed**

So I'm happy to announce that at this point, at least in simulation, this thing beats the Minty Boost, and does it without any fancy-schmancy chips! Of course, this is just a friendly competition: I have no illusions about trying to usurp the people who produce that device. I just like the idea of being able to do this with readily-available parts a hobbyist might have lying around. The most "exotic" part, really, is the MOSFET, which I believe is available easily and cheaply from Digi-Key.

So far I haven't added my simple regulator (zener diode, another transistor tied to the gate), but don't anticipate any problems doing so. And I haven't yet checked this for efficiency; I'm very curious to see just how much of a drain this puts on those poor AA cells. But at least according to LTspice, it works like a charm. It's a little finicky: I could only get it to start at all by keeping the inductor within a very narrow range (about 8-10µH). This may just be an artifact of the simulator, and the real circuit might not be quite so demanding. Hopefully I'll be able to breadboard this soon.

Just did a quick efficiency check: not too good, only 52%. Some tweaking is needed here ...

So, any comments, advice or (constructive) criticism would be welcome. LTspice file attached below for you to play with.

Later: I added my simple regulator, which works flawlessly (gives 9.24V @ 92mA). LTspice file attached. At this low output, efficiency goes up to about 70%.

Hmm; efficiency goes up further, to 76%, when I increase the load by reducing RL to 50Ω.
 

Attachments

  • Boost conv FET 3.asc
    3 KB · Views: 485
  • Boost conv FET 3 reg.asc
    3.9 KB · Views: 390
Last edited:

Without it there are large negative-going pulses to Q1's base which I thought might be harmful to it. It does reduce efficiency somewhat. The LTspice trace of Q1's base voltage is attached below; the maximum negative pulse is more than -6V, with the average seeming to be somewhere around -3.2V. (The 2nd picture shows a closeup of the pulses, including ringing on the trailing edge.) Should I worry about this or not?
 

Attachments

  • negative-going pulses.gif
    negative-going pulses.gif
    17.3 KB · Views: 439
  • negative-going pulses 2.gif
    negative-going pulses 2.gif
    15.4 KB · Views: 444
Last edited:
The battery voltage is too low to cause problems, it's a standard technique when you use higher voltages to prevent reverse Vbe breakdown, but it's not required at this voltage.

By the way, if you wanted to add regulation, the worlds first domestic use of a switchmode PSU (the Thorn 3000 colour TV) used a monostable - it had two timing capacitors on one half, connected by a diode - the regulation worked by switching the diode ON or OFF and so altering the pulse width by paralleling the two capacitors.
 
The battery voltage is too low to cause problems, it's a standard technique when you use higher voltages to prevent reverse Vbe breakdown, but it's not required at this voltage.

I have no doubt that this is true. I went ahead and removed the diode and re-ran the simulation. Turns out that it actually increases the efficiency significantly (from 61% to 73%, at an output of 180mA), probably by removing the ringing and cleaning up the waveform. So I guess it should stay.

Your description of regulation by changing the multivibrator frequency sounds intriguing; I'm going to have to try this out. I assume this can be done simply by biasing the diode to turn it on and off?
 
Clever, I can see why you don't want something complicated like an IC. Here is a 1.2V to 3V LED driver. It uses a complicated IC.
 

Attachments

  • ZXSC380.pdf
    383.6 KB · Views: 484
Last edited:
Clever, I can see why you don't want something complicated like an IC. Here is a 1.2V to 3V LED driver. It uses a complicated IC.

Guess you didn't read this part:

carbonzit said:
I just like the idea of being able to do this with readily-available parts a hobbyist might have lying around.

I respect any desire to learn the lost art of circuit design.
 
I have no doubt that this is true. I went ahead and removed the diode and re-ran the simulation. Turns out that it actually increases the efficiency significantly (from 61% to 73%, at an output of 180mA), probably by removing the ringing and cleaning up the waveform. So I guess it should stay.

Your description of regulation by changing the multivibrator frequency sounds intriguing; I'm going to have to try this out. I assume this can be done simply by biasing the diode to turn it on and off?

I'll see if I can find the circuit at work and scan it, I couldn't find it on google - but it LONG predates the Internet and home computers so that's no surprise.

Bear in mind, I haven't seen the circuit for probably 30 years!, but I can still remember it reasonably.
 
I think if you connected R2 to the output capacitor, with a parallel bleeding resistor to get things started, you might actually get PWM regulation. There is some math that would be involved, but I'm too lazy to do it.
 
Last edited:
Guess you didn't read this part:
carbonzit said:
carbonzit]I just like the idea of being able to do this with readily-available parts a hobbyist might have lying around.
I respect any desire to learn the lost art of circuit design.[/QUOTE]

Thanks for pointing that out, Brown.

To Ron, while that little chip is interesting, I think it's fair to say that we may have different approaches to electronics. If I read between your lines correctly, you're telling me "Get real--why bother with all those parts when you can just use this chip that does it all". (Sorry if I put any wrong words in your mouth.)

Don't know what your background is, but I'm not an engineer, don't work in the electronics industry (though that would be nice!), and am not designing anything for the marketplace. I feel not a bit of shame admitting I'm a hobbyist and an experimenter. So I get a good deal of satisfaction designing (or at least trying to design!) circuits such as this, and hopefully learning something in the process.

Besides, while that chip is pretty good, I'm trying to do a little more here than just drive a LED. Although it has a respectable output (up to 200mA if I remember correctly), I'm not sure it would serve as well as my own designs for my intended purpose (5 and 9 volts supplies).

And of course, as I've said several times, the design that I'm trying to "beat" (the Minty Boost) is probably much more efficient than mine will ever be. But the idea of at least rivaling it with a circuit made of lowly discretes is intriguing to me.
 
The battery voltage is too low to cause problems, it's a standard technique when you use higher voltages to prevent reverse Vbe breakdown, but it's not required at this voltage.
...

When regulating, the FET on-duty can be quite small by percentage and C1 can charge to the full boost voltage (the highest voltage that appears at the inductor) then when FET turns on C1 causes the Q1 base voltage to go below ground by that voltage.

That will mean around -9v Vbe each cycle which may be too high for safe operation, which is why I included the Vbe diode in the original circuit.
 
OK, I've dug the manual out of the attic at work :D

Here's a couple of scans showing the monostable and feedback system.

The two capacitors are C614 and C615, with W615 switching them together under the control of VT608.

I built a jig for repairing the panels from these sets, I can't imagine how many PSU's I repaired over the years.

I hope this gives you some extra ideas? - late 60's, early 70's technology, the worlds first domestic SMPSU.
 

Attachments

  • 3000PSU.jpg
    3000PSU.jpg
    112.4 KB · Views: 554
  • 3000_1PSU.jpg
    3000_1PSU.jpg
    71.7 KB · Views: 525
Hey, Nigel--thanks for digging out that old service manual! Very interesting. Nicely documented.

So this seems to be where the action takes place:

**broken link removed**

C614, C615 and W615. (Never before seen diodes marked "W".) Not exactly sure how this all works, though.

When the regulator transistor turns on, it forward-biases the diode, correct? Which--what?--then adds the smaller capacitor (C614, 2000pF) in parallel with the 0.022uF? I'm never sure about polarities and such in circuits like these.
 
Hey, Nigel--thanks for digging out that old service manual! Very interesting. Nicely documented.

So this seems to be where the action takes place:

**broken link removed**

C614, C615 and W615. (Never before seen diodes marked "W".) Not exactly sure how this all works, though.

When the regulator transistor turns on, it forward-biases the diode, correct? Which--what?--then adds the smaller capacitor (C614, 2000pF) in parallel with the 0.022uF? I'm never sure about polarities and such in circuits like these.

Have a read of the text I scanned, which explains how it works - it's an interesting old idea, and could make a simple multivibrator design easily regulated. The original TV's worked really well.
 
Carbonzit,
I'm halfway there at building something similar, came across this place a bit too late else would have just copied yours ;)

I'm using only transistors as the active device and the main goal was to have something build from parts lying around and at the same time efficent enough to not suck the life out of the battery the moment it was connected.

I'll post it soon for peer review, hopefully we can borrow from each other's designs.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top