MC34063ABN boost converter doesn't regulate the voltage with small loads.

[Electro Dodo]

Hello

I'm trying to build a 5V to 12V boost converter using the MC34063ABN.
This is just something I build to understand boost converters better, so I have completely ignored efficiency, and used parts that I had on hand.
I previously had problems with the MC34063ABN, so this time I kept very close to the Data-sheet.
The problem that I had, was that as soon as I used a smaller load, the voltage of the output rose above the 12,6V I wanted.

As I said I used what I had lying around for the components, so that was the place where I started looking for the problem.

1. The diode I used is not a Schottky diode and has a large(1.1V) voltage drop. But from my understanding, if the Diode was the problem the output voltage would be to low, and not to high. So I don't think the diode is the problem.

2. The inductor is a selfe-wound coil with an air-core that I had lying around. The Inductance (measured with a cheap component tester) is twice what I need as minimum inductance, but I don't know how much the inductance will drop with the current. But again a lower inductance should cause a lower and not a higher output voltage. The high resistance of the coil(0.7Ohm) is bad, but if it's a problem it should result in lower and not higher output voltage.(at least to my understanding)

3.The timing capacitor I used consists of two ceramic capacitors(470pF+100pF) and together slightly lower then the 617pF I calculated using the formulas in the data-sheet.(f min 50kHz)
I don't think it causes the problem, but I'm confused by the readings from the frequency I measured at the timing capacitor and at the switch directly.
The frequency  on the timing capacitor was stable 58.8kHz, but the frequency at the switch varied from 165kHz to 130kHz depending on the load.(Higher load, lower frequency)
Which is higher than the frequency the chip is rated for.

4.The resistor divider I used was also made from resistors I had lying around, so the output voltage should be 12.6V instead of 12V. In addition, the resistors are +-5%.
The second resistor over which the 1.25V voltage drop for the reference occurs, is exactly as big as the one in the data-sheet. So the current should also be the same.

Now after all my digging, I think the problem is that the boost converter needs a higher minimum current than occurs with some of the test setups.
The boost converter works fine with a 1kOhm and with a 470Ohm load, but fails to regulate the voltage down to 12.6V with a load larger than 10kOhm, and reaches voltages of 20V and higher.(up to 28.6V with no load)

So now my questions.

1.Is my reasoning correct and the problem is the low current? Or have I completely overlooked something else and the problem is something else?

2.What's up with the frequencies I've measured? They don't make any sense to me. So I would be grateful if someone could explain them to me.

3.If the low load current is the problem. How do I find out what the required load current is, and does it make sense to design the resistor divider in a way that the voltage converter always regulates the voltage properly, or should I use a separate resistor for that.

TL;DR
A boost converter I build using the MC34063ABN doesn't regulate the voltage with a small load attached to it. Why?(probably a to low load current) and how do I prevent that?

[MrAl]

Hello

I'm trying to build a 5V to 12V boost converter using the MC34063ABN.
This is just something I build to understand boost converters better, so I have completely ignored efficiency, and used parts that I had on hand.
I previously had problems with the MC34063ABN, so this time I kept very close to the Data-sheet.
The problem that I had, was that as soon as I used a smaller load, the voltage of the output rose above the 12,6V I wanted.

As I said I used what I had lying around for the components, so that was the place where I started looking for the problem.

1. The diode I used is not a Schottky diode and has a large(1.1V) voltage drop. But from my understanding, if the Diode was the problem the output voltage would be to low, and not to high. So I don't think the diode is the problem.

2. The inductor is a selfe-wound coil with an air-core that I had lying around. The Inductance (measured with a cheap component tester) is twice what I need as minimum inductance, but I don't know how much the inductance will drop with the current. But again a lower inductance should cause a lower and not a higher output voltage. The high resistance of the coil(0.7Ohm) is bad, but if it's a problem it should result in lower and not higher output voltage.(at least to my understanding)

3.The timing capacitor I used consists of two ceramic capacitors(470pF+100pF) and together slightly lower then the 617pF I calculated using the formulas in the data-sheet.(f min 50kHz)
I don't think it causes the problem, but I'm confused by the readings from the frequency I measured at the timing capacitor and at the switch directly.
The frequency  on the timing capacitor was stable 58.8kHz, but the frequency at the switch varied from 165kHz to 130kHz depending on the load.(Higher load, lower frequency)
Which is higher than the frequency the chip is rated for.

4.The resistor divider I used was also made from resistors I had lying around, so the output voltage should be 12.6V instead of 12V. In addition, the resistors are +-5%.
The second resistor over which the 1.25V voltage drop for the reference occurs, is exactly as big as the one in the data-sheet. So the current should also be the same.

Now after all my digging, I think the problem is that the boost converter needs a higher minimum current than occurs with some of the test setups.
The boost converter works fine with a 1kOhm and with a 470Ohm load, but fails to regulate the voltage down to 12.6V with a load larger than 10kOhm, and reaches voltages of 20V and higher.(up to 28.6V with no load)

So now my questions.

1.Is my reasoning correct and the problem is the low current? Or have I completely overlooked something else and the problem is something else?

2.What's up with the frequencies I've measured? They don't make any sense to me. So I would be grateful if someone could explain them to me.

3.If the low load current is the problem. How do I find out what the required load current is, and does it make sense to design the resistor divider in a way that the voltage converter always regulates the voltage properly, or should I use a separate resistor for that.

TL;DR
A boost converter I build using the MC34063ABN doesn't regulate the voltage with a small load attached to it. Why?(probably a to low load current) and how do I prevent that?

Hello there,

This isn't one of the best regulators around to start with.  What I didn't like about it is it causes a somewhat irregular ripple output.  That's because the way it regulates the output is very different from a more typical switching regulator.  A more typical regulator will show a ripple that is about the same for a given input and output, but this one has a sort of randomness to it, so the output ripple will vary in amplitude sometimes.

As to non-regulating at low loads, that could be because of this randomness but I never looked into that.  It's usually used for simple applications that do not need super accurate output levels.

[Electro Dodo]

The output had very low ripple, mostly because I put the largest capacitor on the output I had lying around (220microF).
I checked the output on the oscilloscope and found no relevant output ripple, so I don't think it has anything to do with my problems.

Also what regulator would you recommend me using when I want to make a boost and flyback converter with an external mosfet as a switch? (preferably wit a easy to hand solder package)

[magic]

If you have a scope it should be possible to figure out what's going on.

AFAIK this chip shouldn't produce any output cycles as long as feedback pin voltage exceeds 1.25V.

[shapirus]

The boost converter works fine with a 1kOhm and with a 470Ohm load, but fails to regulate the voltage down to 12.6V with a load larger than 10kOhm, and reaches voltages of 20V and higher.(up to 28.6V with no load)

I had a similar issue. It turned out that the inductance I used was too low. Increasing it solved the problem.
What was the value that you measured?

[Electro Dodo]

The inductance I used had a value of 300µhenry and 0.7Ohm. The values were measured with a cheap component tester and aren't that reliable. The min required inductance I need calculated from the data-sheet was 79µ henry.
But I can try to use a bigger one.

I reply again after I tried it with the larger inductance.

[Electro Dodo]

If you have a scope it should be possible to figure out what's going on.

AFAIK this chip shouldn't produce any output cycles as long as feedback pin voltage exceeds 1.25V.

While testing with a load of 10kOhm I measured at Vref 2V and ton of 1µs and toff of 5µs at the switch.(Vout was 20.7V)

the voltage at Vref was with no noticeable ripple.

I don't really understand why though. Maybe the controller has a minimum duty cycle?

[MrAl]

The output had very low ripple, mostly because I put the largest capacitor on the output I had lying around (220microF).
I checked the output on the oscilloscope and found no relevant output ripple, so I don't think it has anything to do with my problems.

Also what regulator would you recommend me using when I want to make a boost and flyback converter with an external mosfet as a switch? (preferably wit a easy to hand solder package)

I would recommend looking on the TI site or other sites for a chip with the requirements you want.  There are so many other there these days, but yes the trend now is SMD, but there are some DIP's left out there.

Another thing I forgot to mention is that I remember needed a larger than usual inductor for the circuit because of the way it regulates.  The timing is unlike most other controllers.  Most of them use a set pulse width for a given set of input and output conditions, but this chip can end up varying that and actually skipping an entire pulse period.  That means that the filtering has to be better for this chip than most others all other things equal.

The attraction to this chip is it's easy to use.  There are a lot of other chips out there too though that have a set frequency and some of them have a much higher frequency too which makes the required parts smaller.

[kimballa]

You mentioned using a regular diode instead of a Schottky. I built a boost converter circuit recently based on the LMX62421 IC, and my initial lazy breadboarding involved a plain 1N4148 and did not work; switching that out for a BAT54 was a massive improvement in regulation. The chip I used switches at 1.6MHz so at least for the faster switchers out there, diode type matters.

Some regulators do require a certain minimum load - or otherwise hiccup with noise on the +ve rail if the load is too light. I've found about 10mA does get them reliably into the normal operating regime. If your "main load" is too light you can always rig a resistor and a "power on" LED to pull 10mA in parallel with your circuit load.

[MrAl]

You mentioned using a regular diode instead of a Schottky. I built a boost converter circuit recently based on the LMX62421 IC, and my initial lazy breadboarding involved a plain 1N4148 and did not work; switching that out for a BAT54 was a massive improvement in regulation. The chip I used switches at 1.6MHz so at least for the faster switchers out there, diode type matters.

Some regulators do require a certain minimum load - or otherwise hiccup with noise on the +ve rail if the load is too light. I've found about 10mA does get them reliably into the normal operating regime. If your "main load" is too light you can always rig a resistor and a "power on" LED to pull 10mA in parallel with your circuit load.

Who said 'regular' diodes work in switching converters?  You left no reference to that statement.

Regular rectifier diodes are the worst kind and really cannot be used in a switching regulator.  Rectifier diodes are made for frequencies up to about 400Hz, but more typically used at 50Hz and 60Hz.

The 1N4148 diodes is fairly fast, but the forward current is very limited, and the voltage drop is more than with a Schottky diode.  That's another reason why Schottky diodes are used in switching regulators they allow a higher efficiency for the converter.
They do have higher reverse leakage current though, but that's not usually a factor in a power converter.
There are other suitable diodes too with zero reverse recovery, but I think they all have a higher voltage drop.
From what I understand, Schottky diodes have no reverse recovery time, but unfortunately there is a parallel capacitance that acts in a similar manner although less reverse current is involved because the capacitance is very small.

[magic]

While testing with a load of 10kOhm I measured at Vref 2V and ton of 1µs and toff of 5µs at the switch.(Vout was 20.7V)

the voltage at Vref was with no noticeable ripple.

I don't really understand why though. Maybe the controller has a minimum duty cycle?

Looking at the logic diagram in the datasheet, it shouldn't. The output should turn off when the oscillator goes down and stay off until the oscillator goes up AND the comparator goes up.

Furthermore, 1μs+5μs is 6μs and 166kHz. You said that you see this frequency on the switching node, but the oscillator is working at 58kHz? This is another abnormality, the output should only turn off when the oscillator goes down, so frequency shouldn't be higher. It's not entirely clear how the Ipk limiting works, but your schematic shows that it's disabled?

There appears to be some unintended mechanism by which the chip oscillates. Is the oscillation synchronized with the Ct pin in any way at all? Maybe the chip is defective, did you try another one?

Are you running this on some solderless breadboard? When probing with the scope, are you probing (and grounding) exactly at the pins of the IC, not somewhere else?

[shapirus]

Are you running this on some solderless breadboard?

It must be noted that running it on a solderless breadboard does not guarantee failure. I have been using them with this IC to verify my boost converter circuits with consistent success (as long as the inductor is right).

[Electro Dodo]

So good news first.
I switched the inductance with a proper one, and switched the diode for a schottky diode.
The regulator now works without a problem.

The output voltage is 12.6V now even without a load attached.
The frequency at the switch is now a more explainable 44kHz, and after seeing how it's supposed to work, I realized, that the switch was constantly switching before, instead of just when necessary.

So magic seems to be right with his theory that the controller oscillates for some reason if the components (either inductance or diode) are not properly chosen.

There appears to be some unintended mechanism by which the chip oscillates. Is the oscillation synchronized with the Ct pin in any way at all? Maybe the chip is defective, did you try another one?

And because someone asked, I didn't use a solderless breadboard.
The construction of it was questionable nonetheless. (I attached some pictures of the final circuit)

So thanks everyone for the help and advice.

TL;DR.It works now, but I use a different controller next time.

[jwet]

I see that you have it working but there is one other point worth noting.  The Rsc is not just for short circuit protection, it sets how much peak current goes into the inductor.  With nothing here, the part won't limit inductor current- it will just rise to Vin/R (inductor), your old inductor being .7 ohms, worked out OK but at low currents, there is just too much energy stored in the inductance  and the part can't regulator.  The threshold across Rsc is about 275 mV so the .22 values give you a pk of 1.25 A.  At lower currents, you could increase this R and keep the peak current down.  The inductor in a boost converter needs to store energy (1/2Li^2) and the power will be the rate at which it does this- rep rate- to give you power.  If you don't watch values, you can push this balance out of bounds and the circuit won't regulate.