Decoupling capacitor

Hi,

I currently design an small electronic circuit. The circuit is mainly a 5V regulator and an accelerometer connected to a comparator (LM393) that drive a MOSFET switch. Do I need decoupling capacitors for the accelerometer and the comparator? I thought that maybe just one 0.1 uF capacitor after the regulator would do the job... By the way, this is a automotive application, the output of the accelerometer is filtered to 1Hz and the board has about 1 square inch.

Thanks

Reply to
Ziguy
Loading thread data ...

One cap should be OK for a small thing like this, but be aware that many regulators will oscillate if they don't have enough output capacitance. See the reg datasheet.

Other posters will give you the three-bears line: big cap plus medium cap plus small cap. Ignore them.

John

Reply to
John Larkin

If there is no other sources of noise being powered from this regulator, there is little need for lots of capacitance on the 5 volt bus. Your idea of .1 uf sounds pretty good. The bigger problem will be noise from the 12 volt side that gets past the regulator. I would worry more about what you do about the + or - 50 volt spikes that will occasionally appear there. Perhaps, a diode to prevent reverse voltages from getting to the regulator follower by an RC or LC filter. This capacitor should be more like 10 or 100 uf in parallel with a .1 uf film or ceramic cpacitor.

And make sure yout 1 second low pass filter is referenced to exactly the same node as the reference voltage it is compared against.

--
John Popelish
Reply to
John Popelish

Before the regulator, I put in series a diode and a resistor, and in parallel a zener, a 100uF and a 0.33uF tantalium capacitors.

Later,

"John P> >

5V

drive

and

application,

about 1

Reply to
Ziguy

I'd advise against the tantalum. They're expensive and have a bad habit of exploding when used across supply rails. In a car, where serious overvoltage spikes are likely, tantalums will likely die. A

100 uF aluminum cap should be enough.

I recently did a strain-gauge signal conditioner with eight excitation supplies, and foolishly bypassed each regulator output with tantalums. Even there, it had enough peak current to blow the caps occasionally. We had to replace them with aluminums, and fortunately found some that fit the tantalum surface-mount footprint.

John

Reply to
John Larkin

Do you suggest me to forget the 0.33uF cap at all? Not even a 0.33uF ceramic cap! Are they not needed for high frequency noise?

Thanks again,

"John Larkin" a écrit dans le message de news: snipped-for-privacy@4ax.com...

Reply to
Ziguy

No matter what, put a .01 ceramic disk cap across every IC, as close to the pins as you can get. I have seen too many strange things happen without them that it is cheap insurance against power line induced noise.

I'll spend the extra 5 cents per chip so I don't get hassled a year from now and have to retrofit a pile of units in the field.

-- Dan Fraser

From Costa Mesa in sunny California

949-631-7535 Cell 714-420-7535

Check out my electronic schematics site at:

formatting link
If you are into cars check out
formatting link

Reply to
Dan Fraser

The small cap mentioned above is a inrush C, the switching action (charging mos transistors) in the i.c. needs for a short time a lot of charge (current, due to the impedance of the power supply lines on the board you'll see a dip in the voltage, which might result in a reset of a piece of the circuitry. The small ceramic C will solve this problem.

The problem with 78 or 79 series regulators is that they are quite noisy over a large piece of the spectrum up to 30MHz or even higher. A small 10nF ceramic capicitor close to the regulator, i.e. on the "legs" of it will solve this. It's good engineering practice to put small ceramic 10nF C close to non switching regulators both on input and output.

In a car a load dump might occur which takes the voltage on the 12 V supply upwards to 80 volts or more for a few seconds. And the spikes from the ignotion are anoying as well.

Cees.

--
# Javascript is a powerful language, except it mutates faster than a
# fruit fly in a X-ray machine    (Andrew TanenBaum)
 Click to see the full signature
Reply to
Cees Keyer

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.