Anyone experienced with MC34063 voltage regulator design?

Hi all…I ve been struggling with the following issue for a few months now. So i decided to ask for help from other more experienced users of the forum.

I am using the 34063 switch mode voltage regulator in step down configuration to power a stand alone arduino board. The board needs both 5V (for arduino and other TTL logic) as well as 12v for other parts of the circuit (relays).

diagram2.png

The 34063 steps down the voltage from 24V-35V input to 12V and then I am using a linear regulator (not shown) to get the 5V needed. The total power requirements at 12V do not exceed 500mA but can range from as little as 50mA to an average of 200mA.

The circuit shown below operates as expected when fed from a DC source with anything from 16V upto 35V. I have even tried it with various load currents, including short circuit to check performance.

You will notice that the circuit also has a full wave rectifier so that it can also be supplied with AC.
The unexplained now happens when I try to supply the same circuit with 24V AC (34V pk measured after rectification) : After a few seconds the regulator shuts off and the only way to reset it is to turn off the power. This happens even with minimal current drawn of 50mA.

Now, I know that there are much more advanced SM regulators out there I could use, but this one is very cheap and I have already designed a PCB for it and therefore I wouldn’t like to change it!

If anyone has any idea why this is happening and how I could fix it, It would be great help indeed!

I have not used that specific device but have worked with switched mode supplies.
The first thing is that the layout and ground plane are absolutely vital for stable performance over the full range of input voltages. I am assuming that you have a PCB layout that closely resembles the ones in the data sheet.

The other thing that is vital is the performance of the inductor, specifically the magnetic saturation of the core it is wound on. A higher voltage generates higher magnetic spikes and that could cause core saturation if the core is not up to this sort of treatment.

The other thing that is vital is the performance of the inductor, specifically the magnetic saturation of the core it is wound on. A higher voltage generates higher magnetic spikes and that could cause core saturation if the core is not up to this sort of treatment.

Thanks for the input Mike.

Yes, I am aware of the points you made. I did try to follow the datasheet guidelines as much as I could regarding PCB design and layout. Regarding the inductor used, to be honest I dont have the full spec other than the inductance, max current and the fact that its a ferrite wound inductor.

Putting a scope across capacitor C31 when the circuit is supplied with AC, (with no load) I can see that the frequency of the switcher is not stable, after some time (which can range from 5Sec to a few minutes) the oscillation gradually fades out (amplitude diminishes) and the regulator switches off.

On the contrary, I had the circuit working for about a week non stop supplied with something like 30V DC and under load with no problems.

What puzzles me is exactly why the difference in behavior when the supply changes from DC to AC.

Needless to say that I have tried to increase the input capacitor (to smooth out further), add another 0.1uF ceramic cap in parallel ,replaced the rectifier etc with no significant change.

PS: Love your quote "Solder is electric glue"

What puzzles me is exactly why the difference in behavior when the supply changes from DC to AC.

I think it is rather the increase in voltage ( 34V pk measured after rectification ) on the input rather than the fact that it is AC that would be causing the problem.

Putting a scope across capacitor C31 ........

Sounds like a classic case of what happened with the specialist switch mode supply engineers I have had to manage in the past. That is normally a layout / decoupling issue although on one occasion it was a saturation problem.

It would be great to check the signals with a scope.

The resistance values for the R21-R24 divider seem really high (susceptible to noise). I would try R21=9.1K and R24=1.2K for about the same ratio (8.608). This would be more compatible with the datasheet's Figure 11. Also, the optional LC filter (1.0µH/100µF) may help.

Whats supplying the 24V AC.
How much AC ripple is on pin 6?

Hi,
What are the voltage ratings on your electrolytics, omission on circuit diagram.
Did you try adding extra filter caps to rectifier output?

How did you arrive at the values for the inductor and the 330pF timing capacitor?

Tom.. :slight_smile:

The resistance values for the R21-R24 divider seem really high (susceptible to noise). I would try R21=9.1K and R24=1.2K for about the same ratio (8.608).

OK ... I will try out a new resistor combination and post the result.

What are the voltage ratings on your electrolytics, omission on circuit diagram.
Did you try adding extra filter caps to rectifier output?

How did you arrive at the values for the inductor and the 330pF timing capacitor?

The electrolytics are rated 50V.

Yes I did try to increase the value of the filter cap with no difference on the output.

The formula given derives the minimum inductor value. Anything above that is ok.

The capacitor determines the switching frequency. I have tried capacitors in the range 270-330pf.

Whats supplying the 24V AC.
How much AC ripple is on pin 6?

24 AC is supplied by a 230/24V mains step down transformer.

I will measure AC ripple later and post the result. However please note that I have increased C32 upto 2200uF but the regulator still shuts down as before.

Upadate: I have replaced the inductor with another of the same inductance and ampacity..but looks bulkier. The system now looks much more stable regarding the frequency. The regulator still shuts down when the system is supplied with AC but now stays on much more time.
Another thing i noticed is that the first time the unit it turned on, it stays on for a few minutes. Then, once the regulator shuts down once, even if switched off, it only lasts a few seconds back on again. It must be kept off for some time before it can fully recover .
I am now inclined to conclude that tis is an issue with inductor core saturation as Mike suggested in his first post.

Still not sure why it hapens only with ac supply and how i can fix it. These inductors were of power supply grade.

Ps: Another issue i noticed is that the sawtooth oscillation waveform is slightly modulated from what appears to be harmonics from the MCU frequency. They are both on the same pcb but as far away from each other as possible.

Still not sure why it hapens only with ac supply and how i can fix it.

Because the AC supply when rectified give you more voltage than the DC supply. Therefore the impulse is more intense.

Is the Inductor getting hot when the regulator shuts down?

Because the AC supply when rectified give you more voltage than the DC supply. Therefore the impulse is more intense.

I thought of that too Mike, but I was then able to try exactly the same DC voltage as the measured AC rectified Vpk and the problem did not replicate!

Fair enough, new information. Looks like it might have something to do with the power supply impedance then.
In my experience these sort of things are often a "black art".

Is the Inductor getting hot when the regulator shuts down?

Not at all. There is no detectable raise in the conductor's external temperature. The current drawn is far below its limits.

Looks like it might have something to do with the power supply impedance then

You mean the impedance the regulator is "seeing" towards the rectifier side?

In my experience these sort of things are often a "black art".

I know what you mean! That's why I hate these analog high frequency things. It reminds me the University years when we were designing RF circuits and you only had to barely touched the table, for the frequency to go crazy!

No the impedance of the supply feeding your circuit. The difference between the DC feed and your rectified AC feed.

The MC34063 seems an older generation (low frequency) chip.
There are more modern versions now with 10x higher switch frequencies (smaller inductors).
12volt/500mA, 42volt max input. Total size only 0.5" x 0.4".

Leo..

The MC34063 is a voltage mode control chip. Voltage mode control suffers from a time delay before the supply can respond to load changes.

More advanced SMPS add current mode control to the voltage mode control, so they have two control loops. There are several ways to do this, one is by adding the current sense ramp to the oscillator ramp, (a sort of slope compensation).

I find an MC34063 runs well as a boost converter in current limit, but that has limited application (one is to charge up a capacitor to drive a latching solenoid with).