5V to 3V3 capacitor question

I have a project in which I need both 5V and 3V3 rails for different chips.

For the 5V supply I have before the 5V regulator a 47uF capacitor, then after the 5V regulator another 47uF capacitor as well as a 0.1uF capacitor.

The 5V supply described above feeds the 3V3 power supply, which has before the 3V3 regulator a 0.1uF capacitor and after the 3V3 regulator a 0.1uF capacitor.

Each chip has a 0.1uF decoupling capacitor between VCC and GND.

Will this be sufficient or would I need more/different value capacitors?

Wouldn't know for sure without reading the regulator specs - I take it you have?

  • but it sounds fine.

regards

Allan

The 5V supply described above feeds the 3V3 power supply, which has before the 3V3 regulator a 0.1uF capacitor and after the 3V3 regulator a 0.1uF capacitor.

Post the part number of the 3.3V regulator.

And the 5v regulator.

Also, the type of each capacitor being used (ceramic, tant, or aluminum).

Or you could, you know, just read the datasheet for the regulators - they tell you what you need. The input and output capacitor requirements vary greatly between individual parts.

The part number is LP2950 3.3v in a TO-92 package.
Page 1 of the datasheet shows a “simplified schematic” with a 1uF on the input and a 2.2uF on the output.
I’m not sure what is meant by a “simplified schematic”…

LP2950

The 5V supply described above feeds the 3V3 power supply

I couldn't find the "headroom" voltage on the datasheet so I'm not sure if 5V is a high enough input voltage for a 3.3V output. If the "headroom" is 1.7V or less then it should be ok.

Headroom? There's no headroom in datasheets!

Drop out. Its called Drop out. and this chip is called a LOW DROP OUT regulator.

The chip claims ~30-50mV drop out under ideal conditions and 500mv under worst conditions...

So to get 3.3V out you only need to supply 1/2 a volt more for "typical" current draw... but of course its not a perfect world, so I would round up and say 4.3v is worst case (or best case if you are a glass half full person) and should be a design goal. It is easily met by the 5V regulator.

Yeah, that's the word I was looking for. Thanks

You should be fine - maybe add a 10uF cap after the 3.3V regulator (mind that cap is rated around 10V or better). You don't need two 0.1uF capacitors between the regulators - one should do well, but keep the 47uF.

Be sure not to exceed the current ratings of the regulators with your loads (keep in mind the current for the 3.3V load also flows through the 5V regulator and adds to that load).

rosscopicotrain:
The part number is LP2950 3.3v in a TO-92 package.
Page 1 of the datasheet shows a “simplified schematic” with a 1uF on the input and a 2.2uF on the output.
I’m not sure what is meant by a “simplified schematic”…

Simplified schematic is a block diagram of the chip internals which doesn’t leak too
much intellectual property, I think they meant “typical application” for that diagram.
Shows even TI engineers can be a bit crap sometimes, or perhaps when the datasheet
was converted from NatSemi to TI format it got a bit mangled. (That’s got to be a
mind-numbing job thinking about it)

I always followed Bob Pease's stuff in the old NatSemi days and still have a very well thumbed old copy of the linear databook...

ah well

regards

Allan.

Alan,

To me, Bob Pease was the Patron Saint of Analog Electronics. He will not soon be forgotten.

To this day, I still say that National Semi had the best databooks. I see him as more a National Semi guy than a Ti guy, but he worked at both companies.

Sorry for off topic...

Now go read your datasheets...

Another sad loss was Jim Williams at LT......

sorry - well off post.

Allan

There is no easy way to determine the exact value of capacitor to use. The manufactures often list a minimum size to keep the regulator stable. They have no idea of the frequency or amplitude of noise their regulator might see in your application. Although, if one follows the manufactures recommendation, it is unlikely to see a regulator oscillate, but I've see it happen. There are a lot of variables that one doesn't have access to but one can figure that if there are large the current changes on the output, one might want to use a larger capacitor on the output. The more noise on the input, one might want to add to the input filter. The choice of capacitor type should be considered as well. Electrolytics tend to have poor high frequency response so one often sees a 47 uf electrolytic paralleled with a .01 ceramic. Dwight

it is unlikely to see a regulator oscillate, but I've see it happen.

Theoretically, the regulator should shutdown when overloaded and remain in the shutdown mode until the excessive load is removed but if there is a delay between the time the load has been shutdown and the time it is "seen" on the output, the regulator can shutdown, see the load is gone and restore power just before the load reengages and begin a repetitive cycle , essentially ossillating. This can occur if the load has some initialization period before enabling the output.

The can oscillate with thermal shutdown as well as current. I was talking about changing loads within the rated values. Dwight

They can oscillate because they are high gain ampliers. If you set things up to satisfy Nyquist's criterion, it oscillates. The capacitors affect the poles and zeroes of the feedback loop network. When a regulator oscillates you typically take out all the chips it is supposed to be powering - seen it, been there, replaced all the components with a hot-air rework station. (Though that was caused by accidental parallelling of two regulator outputs, but its the same basic issue, regulator sees the wrong impedance at the output).