DC-DC converter design problem

[manve_13]

Hello,

I've designed a system that includes 3 small dc motors, pic18f and an optical drive. When i did fist prototype version i was using a lab style power source and I was using 12V, 5V and 3.3V from that source but in last version it has 12V adapter input and i was converting voltage 12V to 5V and 3.3V with 2 LM78xx voltage regulators.

The problem is voltage regulators are heating up like crazy in 30 seconds. As i've checked it drawn between .7 A to 1.5 A from 12 V source. Then i connected a huge heatsink to the both regulators but I'm sure there should be another way to solve heating problem.

Can you suggest me a good way to convert 12V to 5V and 3.3V, which kind of dc-converter should I choose?

Thank you.

 

[BradtheRad]

It sounds as though you are operating the 78xx chips at their limit of current, and/or limit of dissipating heat.

It is possible to add a transistor (called a pass transistor), to take some load off the 78xx's.



There may be versions of the 78xx which are rated for higher current/power.

Or you can try the 317 adjustable regulator. It handles 1.5 A.

You may still need heatsinking on some devices.

There is also the method of running a DC motor by sending rapid pulses through it. Did you rule this out? It does not waste so much power as the resistive drop method does.

 

[Tahmid]

A lot of power is dissipated by the linear regulators. So, you could use DC-DC buck converters/regulators (switching) instead of linear regulators. You will gain better efficiency and thus lower heat.

For the 3.3V, you can use L4971 (taking input directly from 12V).

For 5V, you can use LM2576, L4960, L4964, L4972, L4973, etc. For 3.3V, you may just use a linear LDO regulator eg LM2937.

Hope this helps.
Tahmid.

 

[E-design]

The 7805 can be configured as a switching regulator. Look at the datasheet.
Another option https://www.gravitech.us/35v1aswvore.html although a bit expensive

 

[manve_13]

A lot of power is dissipated by the linear regulators. So, you could use DC-DC buck converters/regulators (switching) instead of linear regulators. You will gain better efficiency and thus lower heat.

For the 3.3V, you can use L4971 (taking input directly from 12V).

For 5V, you can use LM2576, L4960, L4964, L4972, L4973, etc. For 3.3V, you may just use a linear LDO regulator eg LM2937.

Hope this helps.
Tahmid.

Thank you Tahmid, i will use L4971 and L4972 or L4960 to use heatsink probably.

Do you think they will heat up if I use surface mount L4971 and L4972?

Thank you.

 

[manve_13]

It sounds as though you are operating the 78xx chips at their limit of current, and/or limit of dissipating heat.

It is possible to add a transistor (called a pass transistor), to take some load off the 78xx's.





There may be versions of the 78xx which are rated for higher current/power.

Or you can try the 317 adjustable regulator. It handles 1.5 A.

You may still need heatsinking on some devices.

There is also the method of running a DC motor by sending rapid pulses through it. Did you rule this out? It does not waste so much power as the resistive drop method does.

I've tried 1.5A version of another linear regulator. the result was same, it was heating up.

Do you mean PWM for dc motors?

E-design, i've just saw that on 7805 datasheet, is it going to be work with same effiency if I use switching regulators?

 

[Tahmid]

I think it's better to use L4960 and use a small heatsink. Then, all your overheating problems should be gone.

 

[marce]

12V to 5V SMPS use 5V for 3V3 using an LDO, this is a pretty common voltage combination for a lot of equipment. Using the 5V for the 3v3 less voltage to drop and using an LDO keeps it quiteter for the logic. This has been done on the last several designs I worked on.