Cascade two voltage regs

For any good reason ? Or just because National IIRC said to do so ? I can tell you that tantalums are certainly not needed.

And yes, capacitance adds. The importance for regulators is *where* the cap is for the most part though.

Graham

"Largo"

** The data sheets ASSUME that a large value electro (or a battery) is connected in parallel with and close to the input of linear regulator ICs. ** LOL - chocolate dipped is better.

** The LM317 needs a low impedance at its input at frequencies between 100 kHz and 10 MHz.

A 100 uF, 25 volt electro should do that fine.

Best avoid tantalum bead caps completely.

....... Phil

So it's okay to use unpolerized multilayer ceramics at the output?

Tantalums on power rails are bad news; they tend to blow up. Aluminums are much safer.

I'd use 'lytics at the input and output of both regs, which would be three total, with the middle one shared. 22uF 16v or something.

John

Multilayer ceramics should work fine so long as you have decently large values to hand - equipment manufacturers tend to be too stingy to use these!

use these! """

What does this mean? Please suggest values. Thank you. Largo

>

At least 0.47uF better yet 2.2uF - I personally would still use alu-electro about 22 or 47uF as well just to make sure!

So what I gain from this relative discussion is the more capacitance the better but put something in. Small is okay, only because "manufacturers tend to be too stingy". Larger is better--"just to make sure".

i.e. I'd be well off to use aluminum electrolytic [on the output], up to 47uF. What about the input? Use .33uF electrolytic as the datasheet says? or 22 or 47uF? Or does this apply on both input and output?

Thank you for your help. Largo

Of course, you realize that the 7809 will have to pass both the 9V load current and the 5 volt load current. Make sure that the heat sink is adequate for the total power dissipated.

--
Dave M
MasonDG44 at comcast dot net  (Just substitute the appropriate characters in the 
address)

Some days you\'re the dog, some days the hydrant.

It is also a good idea to make sure the capacitance on the upstream side of the regulator is higher than those downstream, because if the input voltage of the regulator drops below the output, damage can occur. Another way to prevent this is by adding a diode from the output to the input.

Most regulators are stable with a 0.1 to 1 uF low ESR capacitor close to the output, and up to several hundred uF in parallel downstream, but it is better to have the large capacitance upstream for energy storage, and smaller capacitors distributed near components which need a low impedance supply source for switching transients.

Paul

Paul

There are certain rules to observe, the 3-TRs require small-ish capacitors from both IP & OP to GND which must have very low values of ESR, this is to ensure stability as the internal error sensing amplifier that controls the voltage regulation has a high gain. If you are running from the secondary of a transformer you have to have a large enough reservoir capacitor (in addition to the decoupling caps) to smooth any ripple out of the rectified voltage, but excessively large capacitors on the OP can actually cause instability!

[snip]

It is also a good idea to make sure the capacitance on the upstream side of the regulator is higher than those downstream, because if the input voltage of the regulator drops below the output, damage can occur. Another way to prevent this is by adding a diode from the output to the input.

Most regulators are stable with a 0.1 to 1 uF low ESR capacitor close to the output, and up to several hundred uF in parallel downstream, but it is better to have the large capacitance upstream for energy storage, and smaller capacitors distributed near components which need a low impedance supply source for switching transients.

(sorry if this is a repeat. My previous post seems to be lost in cyberspace)

Paul

Leave the "reservoir" cap - the BMF filter at the output of the unreg. supply. Then, as close as possible to the input lead of the 7809, put a 0.33 uF. As close as possible to its output, put a 0.1 uF. As close as possible to the input of the 5V reg, put a .33 uF - they will look like they're in parallel, but you've got that circuit trace between them, you see. Then, at its output, as close as possible to the chip, put another 0.1 uF. Or whatever the data sheet recommends, as far as value, for each.

Yes, multilayer ceramics will be fine - maybe better than aluminum electros, because I've heard that they do good at high frequencies, which is what we're trying to bypass here - you don't want your regulators oscillating. :-)

Good Luck! Rich