MC34063 Buck Converter - Efficiency and Transistor Heating Problem

[sbdada]

Hello, I made this buck converter with MC34063 Ic to charge my phone from a solar battery.


And here is the schematic diagram


This circuit can deliver about 1.5A of current at 5.2V, and it charges my phone like it's original wall charger. But the problem is with the efficiency and the pass transistor. Please, have a look at this table



I connected my multi meter at the input connector, and An USB Charge Doctor at the output to measure the output voltage and current. Please see the last two row for the usb doctor's own power consumption.
The efficiency is too low. I used a good quality and well designed pcb, while in some article on the internet i found that peoples made their circuit with this IC on a vero board, and got the efficiency up to 80%.

Another problem is with that Transistor. After a few minutes of operation, it became too hot. I can't even touch it. I tried with IRFZ44N and P55NF06 MOSFETs and the result (heating) is same.

My guess is the transistor is dissipating power for unproper biasing, and so the efficiency drops. Not sure though What is wrong in my diagram? How can I solve this problem? Please help me.
Thanks in advance

[bktemp]

You have cascaded 4 transistors: MC34063 has a darlington transistor and you have added another one. Therefore you are wasting at least 4x 0.7V. And R3 is too high. 1k should reduce the switching losses.
There are much better ICs than 34063, but if you want stick to it, I would use a PNP transistor or a P-MOSFET. This should increase the efficiency under load, but slightly increase the idle losses.

[sbdada]

Hello bktemp, thank you for your reply. In my local market only 3 IC available. This one, LM2596 and LM2576, but the last two IC is pretty costly and not so available everywhere. That's why I chose this IC. I tried with N-MOSFET but that didn't solve the problem. So I used this TIP122 as it is cheaper than MOSFET. Didn't tried with normal NPN transistor or PNP or P-MOSFET. I'll try with NPN transistor tomorrow and will inform about the result here. Have to modify the PCB for PNP or P-MOS. Will do it soon. Thanks again

[SaabFAN]

The TIP122 has 200pF input capacitance. Coupled with its high amplification and the fact that the MC34063 has a current-control loop, I'm guessing it is working like the pass transistor of a linear regulator.
Have you checked with a scope if the base-emitter voltage of the TIP122 decreases to 0 when the Output of the IC is turned off?

[Monadnock]

Try increasing the timing capacitor by 50-100% and see if the efficiency improves. You could plot efficiency vs. frequency over a limited range (don't want the ripple current in the output choke to get too high) and see if there is a point where the efficiency peaks. If the efficiency doesn't vary much with lowering the frequency then switching loss is not much of a factor and you need to look elsewhere.

[Ammar]

Is the TIP122 as a high-side switch necessary? It looks weird to me having an NPN transistor as an emitter follower in this context. Can you try a low side topology? Unless the IC is designed to drive a high-side switch.

[batteksystem]

If I don't remember wrongly, the internal transistor is rated to 1.5A. So I don't understand why we need the external transistor there. I only add external switch if the current rating exceed 1.5A. In my application the requirement is to replace a 3A simple switcher we are using (high costs at that time). But the efficiency is really not impressive.

A very old photo for my prototype:

[Miyuki]

TIP 122 is too slow

220p timing capacitor gives about 100kHz
So you have 10us cycle with about 5us on time
TIP122 have rise time about 0.6us and fall time bigger than 2us
And steady state saturation voltage 2V gives you almost 20% loss even without transitions
Overall efficiency about 50% is expectable, transistor will be more time at transition state than in steady state

[T3sl4co1l]

Is it any wonder why I maintain that 34063 always costs more in the end?...

If you literally can't get any controller/regulator chips for cheap, then why bother using any at all?  This has way better performance,
http://seventransistorlabs.com/Images/Discrete_Tube_Supply.png
obviously it's not directly applicable, but simply removing the high voltage winding, and adjusting feedback for 5.0V output, would do what you need.  (I don't have a discrete buck circuit handy...)

Tim

[Miyuki]

Is it any wonder why I maintain that 34063 always costs more in the end?...

If you literally can't get any controller/regulator chips for cheap, then why bother using any at all?  This has way better performance,
http://seventransistorlabs.com/Images/Discrete_Tube_Supply.png
obviously it's not directly applicable, but simply removing the high voltage winding, and adjusting feedback for 5.0V output, would do what you need.  (I don't have a discrete buck circuit handy...)

Tim

Why not use KA7500 and TL494 ? They are quite universal and plentiful

[T3sl4co1l]

Is it any wonder why I maintain that 34063 always costs more in the end?...

If you literally can't get any controller/regulator chips for cheap, then why bother using any at all?  This has way better performance,
http://seventransistorlabs.com/Images/Discrete_Tube_Supply.png
obviously it's not directly applicable, but simply removing the high voltage winding, and adjusting feedback for 5.0V output, would do what you need.  (I don't have a discrete buck circuit handy...)

Tim

Why not use KA7500 and TL494 ? They are quite universal and plentiful

True, but arguably worse as the appnotes only give a low frequency, bipolar, voltage mode buck converter.  34063 at least has... current... feedback, of sorts?

You can put in a MOSFET without too much difficulty (even better with the TL598 that includes totem pole drivers rather than "uncommitted transistor" outputs), but implementing proper average current mode control is still a pain.

Tim

[Kleinstein]

The MC34063 is not that well suitable to drive a FET output.  It may be acceptable if the internal 1,5 A switch is OK, but difficult for higher power.  The efficiency values are that bad, one could have used a 7805 as well.

[niekvs]

You can improve the efficiency by using a MOSFET (instead of the TIP) and improve the way you drive it using a PNP and diode, similar to the red parts in this schematic: https://threeneurons.files.wordpress.com/2011/08/mc34063_mk15cm1.gif

This ensures that the switch off of the MOSFET happens much more quickly by draining its gate capacitance. Also lower that resistor to 1K or so.

Let us know if it helped!

By the way, can you also post a picture of the bottom side of your PCB? (showing the traces..)

[sbdada]

Hello, Sorry for late. Did some modification and test on this circuit. Here it is


First I decreased the R3 from 4.7K to 1k. then tested with 3 transistor D880, P55NF06 and TIP122, only these three I have in my box right now.
D880 didn't worked. It heated up too much that can fry my board. Then moved to other two transistor. I don't know why, but the TIP122 worked better than P55. Changing the inductor increased the efficiency and also the heat decreased. I don't have any temperature sensor or meter, but I can now hold the heatsink, while I couldn't even touch it before or with the MOSFET. Adding a PNP transistor and diode in the Gate pin didn't worked so well.

[johansen]

try this circuit.
http://www.romanblack.com/smps/smps.htm