Converting 20 Volt to 3.3 Volt in two stages

[sean87]

Hi all!

I want to get 3.3 volts from a 20V DC battery. I already got 2x "L7805" and one "LM317" regulators on my desk.

According to 7805's datasheet it can convert 20 volts to 5 volt with no additional components. Does this mean I don´t have to use any filter caps?

Also, For LM317, same thing applies?

Can I just make a simple voltage divider with 1/4 watt resistors that I have, so I get 3.3 volts using only one LM7805?

Would be nice if you tell me how can I get  a trusty stable 3.3 volts after all! I have no experience with converting voltages and I really cant afford to blow up so many micro controllers that I am going to power up using this 20 volt supply (converted to 3.3 of course!)

Thanks!

[firewalker]

Trusty stable 3.3 volts?

1) Use a dedicated 3.3 volts regulator.
2) Configure LM317 for 3.3 volts.

Discard any ideas for voltage divider ASAP! And... Always use your filter caps!!!

[PSR B1257]

Can I just make a simple voltage divider with 1/4 watt resistors that I have, so I get 3.3 volts using only one LM7805?

No. If you do this, you can put nearly no load to the voltage divider.If you still do it, you get a voltagedrop of course.

As firewalker said, use the LM317.

But you have to be beware of the powerdissipation. Already at low currents the power dissipation is significant do toe the high voltage difference.
A free standing TO220 can handle (rule of thumb) about 1W --> 1W/(20-3.3)V=60mA

[mariush]

Both 7805 and LM317 are linear regulators - this means the difference between 20v and 3.3 or 5v is converted to heat.

So let's say you have a microcontroller and a small lcd screen... you'd use about 100mA of power... The linear regulator would have to take 20v and get it down to 5v, so it will turn into heat (20-5) * 0.1 = 1.5 watts  That's a lot of wasted power wasted.

A microcontroller is not really that sensitive to the quality of the DC voltage, as long as it runs at few Mhz and you filter the voltage a bit using some capacitors. You should be able to use a cheap and simple switching regulator to convert that 20 v down to 5v or 3.3v - switching regulators are more efficient and don't waste as much power turning it into heat, but you do need a few extra components.

A very simple chip for something like this is the 34063 (made by various companies) :  http://uk.farnell.com/jsp/search/browse.jsp?N=2025+203946&Ntk=gensearch&Ntt=34063&Ntx=mode+matchallpartial

Dave even made a video about how these regulators work and how to calculate all the values needed for these regulators here :



and there's also an online calculator for these chips which makes it easy for you : http://dics.voicecontrol.ro/tutorials/mc34063/

(later edit ) The datasheets for this chip even have examples with this up to 25v input and 5v / 0.5 A output, which is enough for most home circuits/small prototypes... see page 11:  http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00001232.pdf

There's also a video Dave made about the differences between linear regulators and switching regulators :



So you could use the switching regulator to go straight 20v to 5v or 3.3v, and the microcontroller will probably run just fine with it.

If you want to be extra sure and have as smooth voltage as possible (for example if you want to use the 5v voltage as reference for analog to digital converters instead of the built in reference voltage), you can always configure this switching regulator to generate 6-6.5v, then use a linear regulator to take this 6-6.5 down to a smooth 5v.  Most linear regulators need the input voltage to be 1-1.5 volts above their output (but some can require as low as 0.2v above the output, like MIC2941)

[ptricks]

You left out an important part of any power supply design , how much current do you need ?

[perfect_disturbance]

Yeah an LM317 is easy and effective.  But I wouldn't use one on a battery powered 20v circuit unless I was comfortable burning away 5 times as much power as I was using.

[Psi]

or, for $2.80, get a switchmode module that's 6+ times more efficient

http://www.hobbyking.com/hobbyking/store/__17159__3_3V_3A_UBEC_2_5S_Lipoly_6_23v_.html

Might want a little filtering on the output though.

[sean87]

Thanks guys.

I need this for a developement board from Texas Instruments, which has MSP430 on it. It should be pretty low power chip (maybe around 100mA or less?).

If I just use the LM317 with a good heatsink and some resistors to cofigure it to output 3.3 volt, is it going to be fine?!

[kripton2035]

or, for $2.80, get a switchmode module that's 6+ times more efficient

http://www.hobbyking.com/hobbyking/store/__17159__3_3V_3A_UBEC_2_5S_Lipoly_6_23v_.html

Might want a little filtering on the output though.

+1
just bought this one recently : http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=180830601650
you have both 5v and 3.3v
it supports more than the written 12v input.
regards,

[perfect_disturbance]

Thanks guys.

I need this for a developement board from Texas Instruments, which has MSP430 on it. It should be pretty low power chip (maybe around 100mA or less?).

If I just use the LM317 with a good heatsink and some resistors to cofigure it to output 3.3 volt, is it going to be fine?!

Ok so another beginner here someone may want to back this up.

But if you have no concern for efficiency then it may work.
I don't know what exact part you have but I'm looking at this spec sheet.
http://www.onsemi.com/pub_link/Collateral/LM317-D.PDF
And I'm looking at the TO-220 package.

max input voltage is 40V so your under that.
max output current is 1.5A so 100mV is way under that

The only real concern is the heat.  if your going from 20V to 3.3V at .1A your going to need dissipate 1.67 Watts
the max value for Tj is 150C and without a heatsink the TO-220 has a head disipation junction to ambient of 65C/W so that's 108.55 degrees above ambient

So as long as your ambient temp is roughly below 40C you shouldn't melt anything. Which shouldn't be too tough.

That said It's probably going to be one hot puppy.

[mariush]

Yeah, it will work with just a linear regulator.

But you can make a simple heatsink for it so that it won't be that hot. For example, it can be as simple as screwing a piece of  an Aluminum coke can. Grab a can, cut a rectangle out of it, dry it up, fold it two-three times and make some fringes on one end with a scissor ... then make a hole for a screw in the middle... make sure the metal tab is pressed well against the aluminum.

This simple heatsink would be enough to keep the chip hot but stable.