MC34063/33063 power supply problem

[ShahbazAli]

I have used 34063 is my power supply design for an automotive gadget design. Circuit diagram is attached. Constant power requirements are around 300 mA with temporary surges of 1500mA at 4.2V. But I have been so far unable to get a clean supply. There is too much noise and voltage fall over a cliff when current goes above 200mA. I have tried both 100uH and 330uH but to no avail. I put a 1000uF on the output side and noise got better, but voltage drop is still there. Any help will be appreciated.

Thanks

 

[willi65]

Hi


the MC34063 not easy to handle...how low u need ur peak of 1.5A? ur current normaly 300mA thsi are 500% more...mean u need 5times more Outputpower..

About the Indictor u cna caluculate the inductor L = (Uin-Uout) * Uout / 0,3 * 100*10^3 * 0,3A * Uin

when u calulate it for ur two currents u see the problem...The value of inductor are different.An the MC34063 an current Mode controller..if u make the Value hige ur Output Ripple low ( nomaly calulate with 30% ) but the current mode Jitter..if u make low u have high ripple but ur response time very good...mean peak..

if possible to change ur design to L5973?

 

[FvM]

Generally you shouldn't expect extraordinairy performance, when you choose one of the cheapest available switcher chips. 1.5A is clearly out of reach, 500 mA is specified for similar datasheet reference designs. You can't utilize higher peak currents than 1 A and keep the 1.5 A maximum rating by design. A certain amount of current ripple (e.g. 30%) is suggested for an economic inductor dimensioning, so you can't go above e.g. 800 mA output current in a serious design.

Output ripple is another point, a MC34063 doesn't achieve a clean PWM operation, also switching losses require to use a moderate frequency. so ripple and noise are higher than with a today's high frequency MOSFET switcher.

 

[haker_fox]

Thanks

I can recommend you to read http://www.intusoft.com/onsemipdfs/AN920-D.pdf.
There are much of a theory and calculations.

 

[ShahbazAli]

Thanks for the helpful comments.

It now seems that 34063 is not a good choice for this circuit. There are wild variations in the current required. As it runs off car battery, so the current usage needs to be at minimum, especially when it is not charging. So in sleep mode, current usage will drop to around 75mA. Regular usage is around 300mA and at the time of GSM transmit, small bursts of 2A are expected.

At the moment I am using LM317T as backup in my design, which is stable so far. But I worry about the wastage and heat. LM317 wastes about 62% in converting from 12V to 5V i.e. 42% efficiency. Can someone suggest what efficiency can be realistically expected from normal PWM regulators?

What can be the economical replacements for 34063 in such scenario? I have LM2575/LM2576 in mind. Any suggestions?

 

[FvM]

I have used different LM267x chips in various designs, their application is really simple. The 90 % efficiency for a 5V regulator said in the datasheet is realistic, I think. For automotive application however, you have to provide additional overvoltage protection.