Best 5volt power supply for atmega328...???

When you have the whole circuit connected with the Arduino with its sensors and additional circuits, what is the total current you think / measure ? A TO-92 in my opinion is not powerfull enough to provide the necessary currents of your ardiuno project.

A TO-220 : LM7805 with a heatsink will be fine, as long the current consumtion is less than 1 A ( about 750 mA in total max )

How much do you want to pay? You can get arbitrarily good power for arbitrarily large sums of money :slight_smile:

I'd recommend a low-ripple regulator, and a significantly larger capacitor, and perhaps a HF choke. Say, a 2 uH choke, followed by a 470 uF capacitor.

jwatte:
How much do you want to pay? You can get arbitrarily good power for arbitrarily large sums of money :slight_smile:

I'd recommend a low-ripple regulator, and a significantly larger capacitor, and perhaps a HF choke. Say, a 2 uH choke, followed by a 470 uF capacitor.

Paying is not the problem...
I can make one regulator with LM338 or LM317, and I have tested it,there is no ripple, atleast there no noise on the analog pin of the atmega like when I used the 7805..

But the problem is putting the regulator circuit in such a small place in my application..
There was somehow a little compact place for the LM317 but the problem is it get hotter that the 7805.. pulling 15volts to 5volts..

So I need something which wont get much hot, and will also fit in the place...

If your raw supply is 15V why not use a 7808 followed by a 7805? Share the heat, double the supply ripple rejection at least. (Unless the ripple is due to the load?)

MarkT:
If your raw supply is 15V why not use a 7808 followed by a 7805? Share the heat, double the supply ripple rejection at least. (Unless the ripple is due to the load?)

Hummm...There again the circuit is getting a little bigger... :frowning:

Well if I have to go your way, I will use a 7808 and then use a LM317 to drop that 8volt to 5volt... Thats because I found the LM317 is not generating much ripple, atleast it is not disturbing the analog pins of the atmega328... :slight_smile:

Another problem is that there is not much room for both of them, so I will have to pull them out of the main cabinet which is another problem.. :frowning: and will have to run another wire into the cabinet with the raw supply into the cabinet as the Atmega328 is measuring raw voltage in its code... :stuck_out_tongue:

Well, "small space" and "low heat" pretty much work against each other. You'll need capacitors and inductors of a switching regulator to get the heat down.
Some options:
Four bucks, slightly bigger than a TO-220, 500 mA: http://search.digikey.com/us/en/products/R-78E5.0-0.5/945-1648-5-ND/2834904
Forty bucks, five amps, a lot bigger: http://search.digikey.com/us/en/products/PT6653D/PT6653D-ND/307301

Well I finally decided of pulling out the voltage regulator outside.

So as now it is outside I do not have any issue using the 7805 with inductor and bigger caps...And heat will not be of much issue not as I will be able to use a heatsink and the package will be in constant air flow position as it will be located on a motorcycle, so not much tension of heat.

So now if you have to use an inductor and caps what are the values you will use...??

With a 7805, you just need an output capacitor. Typically, minimum 10 uF, but more is generally better (up to a limit of a few thousand uF).

The reason I talked about inductors and capacitors is that you may want to change to a switching power regulator, which is a lot more efficient -- it doesn't "burn off" the excess voltage, like a linear does, but instead switches power on and off very quickly, and then filter it out with an inductor and a capacitor to get the desired output voltage. This can in theory be 99.9% efficient, but in practice the lower-cost versions are about 80% efficient. This means that you'll only lose 20% in heat, as opposed to the 60% or more heat loss you'll get from a linear regulator giving out 5V from an input of 12V.

Check out that $4 component I linked to, though -- it's a miniature switching regulator in a package that fits in the same footprint as a TO-220 7805, but it generates less heat.

With a 7805, you just need an output capacitor. Typically, minimum 10 uF, but more is generally better (up to a limit of a few thousand uF).

With the 7805 I tried caps till 1000uf. But still then it was generating ripples which was causing the analog pins of the AVR to generate noise. Thats the reason I asked how to cancell that amount of ripple with the inductors..

Check out that $4 component I linked to, though -- it's a miniature switching regulator in a package that fits in the same footprint as a TO-220 7805, but it generates less heat.

Yes..! I have gone through this unit... I it good, but the problem is that it has the maximum current output till 500ma, where as I did a rough calculations and found that I may be drawing about 750ma..

How about using a LM317 or LM338...

I am only afraid of the R1 here, As soon as it gets disconnected the output voltage gets equal to the input voltage... :frowning:
This is how one of my Atmega got fried up...:stuck_out_tongue:

Huh... The power supply is driving me crazy... :frowning:
More over the circuit will be connected to a motorcycle, So I thing there may me a little noise in the input voltage of the regulator.. :frowning:

How about this in here..??
http://www.onlinetps.com/shop/index.php?main_page=product_info&cPath=12_50&products_id=103

The data sheet is here..

Several things here do not make sense to me:

  1. Many, many 7805s get used with ATmega328s without trouble.
  2. I'm not clear whether you are actually measuring the alleged ripple, or are making an assumption based on something seen on the analog inputs, please clarify that.
  3. The ripple rejection specs for 7805s and LM317s are fairly similar.
  4. I don't know why an LM317 configured to output 5V would run significantly hotter than a 7805, all other things being equal.

Have you tried a different 7805? The fact that a 78L05 works makes me wonder if you have a bad part.
Please post your power supply circuit, and describe the source of the input voltage.
How much current does the load pull? If it's just a 328 and a couple LEDs, it can hardly be more than 50mA, yes?

The powersupply circuit which I am using is preety simple..

I am using a 2200uf capacitor on the input of the 7805 then on it output I am using a 10uf cap..
When I solder the 7805 on the PCB there is problem, but when I am using the same 7805 outside the PCB and supplying only 5v from the 7805 to the circuit with a long wire, then things are just fine...

If it's just a 328 and a couple LEDs, it can hardly be more than 50mA, yes?

You are correct it just that much of load I have. So you are sure that it will only dray 50ma...??

I am using a 2200uf capacitor on the input of the 7805 then on it output I am using a 10uf cap..

If you want to filter the signal that comes out of the 7805, you want the big capacitor on the output side.
You can try adding a choke coil (inductor) in series, after the output capacitor. However, if the ripple is in the 60 Hz range, you will need a REALLY BIG CHOKE to reject any significant amount of ripple.

So, the circuit should be something like:

Motorcycle -> 100 uF capacitor -> 7805 -> 100 uF capacitor -> BIG INDUCTOR -> BIG CAPACITOR -> output

Another option is to use a 4.5 volt Zener diode in series with a 2 ohm, 1W resistor, tied in parallel after the output capacitor. This will work as a second regulator, burning off some more voltage to reduce ripple some more.

However, are you SURE it's the power? It may just as well be the sensor you're hooking up to the analog input. Or the analog input itself. Are you buffering it? Does the sensor provide enough current? Is the sensor interference free?

However, are you SURE it's the power? It may just as well be the sensor you're hooking up to the analog input. Or the analog input itself. Are you buffering it? Does the sensor provide enough current? Is the sensor interference free?

Yes ..!! I am sure of it because when I am powering my own build atmega board with the 5 volt power supply of an original arduino duemilanove it works fie(no noise on the analog pins), or when I am using a LM315 to power my custom board it again works fine..
It only with the 7805 when there is noise on the analog pins... :frowning:

And you're sure the ONLY thing that's different is what the power regulator is? Same power source, same sensor, same engine on, etc?
I'd try another 7805 then. Perhaps there's something wrong with the one you have. Or just use the LM315 switcher and go on with life :slight_smile:

Joy:
I am using a 2200uf capacitor on the input of the 7805 then on it output I am using a 10uf cap..
When I solder the 7805 on the PCB there is problem, but when I am using the same 7805 outside the PCB and supplying only 5v from the 7805 to the circuit with a long wire, then things are just fine...

There is a clue in there somewhere I think. Your description of the circuit sounds OK, although I would add a 100nF ceramic capacitor in parallel with the 2200uF on the input, and another in parallel with the 10uF on the output. Attached is a schematic of a typical 7805-based supply. The electrolytic on the input can be arbitrarily large. This one shows two 100nF capacitors on the output, these are physically located near different ICs on the board. It's very important to have a 100nF capacitor from Vcc to ground located as close as possible to the ATmega328, and another from AREF to ground. In a noisy environment like this one, I'd put one 100nF near the 7805 along with the 10uF, another close to the ATmega328, and a third on the AREF pin, just for starters.

If it's just a 328 and a couple LEDs, it can hardly be more than 50mA, yes?

You are correct it just that much of load I have. So you are sure that it will only dray 50ma...??

I was asking more than making a statement. Hard to be sure about anything without a proper schematic.

schematic.png

jwatte:
So, the circuit should be something like:

Motorcycle -> 100 uF capacitor -> 7805 -> 100 uF capacitor -> BIG INDUCTOR -> BIG CAPACITOR -> output

That's a new one on me, my understanding is that the regulator shouldn't require all that hardware on the output side. If the input voltage was bad enough to need an LC pi filter, I'd put it on the input side. Give the regulator relatively decent input and it will do its thing just fine.

Thank you for the second last post of yours...
I will follow that...

That's a new one on me, my understanding is that the regulator shouldn't require all that hardware on the output side. If the input voltage was bad enough to need an LC pi filter, I'd put it on the input side. Give the regulator relatively decent input and it will do its thing just fine.

But I will not agree with you in this...

I tried a lipo battery on the input of the 7805 which had the cleanest input with no noise but then still the 7805 generated a noisy output...
So I believe that the 7805 itself is generating some rippls which is generating noise on the analog pins of the atmega...

Joy:
Thank you for the second last post of yours...
I will follow that...

That's a new one on me, my understanding is that the regulator shouldn't require all that hardware on the output side. If the input voltage was bad enough to need an LC pi filter, I'd put it on the input side. Give the regulator relatively decent input and it will do its thing just fine.

But I will not agree with you in this...

I tried a lipo battery on the input of the 7805 which had the cleanest input with no noise but then still the 7805 generated a noisy output...
So I believe that the 7805 itself is generating some rippls which is generating noise on the analog pins of the atmega...

Hmmm, I'd really have to be convinced of that, probably by seeing it demonstrated with scoped output, and from more than one particular 7805 (don't think you ever said whether you'd tried more than one). Lacking that, I'm perfectly happy to agree to disagree on this one.

So I believe that the 7805 itself is generating some rippls which is generating noise on the analog pins of the atmega...

You are really talking about oscillation and is usually dealt with by the small value capacitors that are specified across the input and/or output pins.

Don

If the 7805 has ripple on the output, there are a few possibilities:

  1. You are not using the recommended capacitors, or the wires between them and the 7805 are too long, so the 7805 is oscillating.

  2. You are drawing a lot of current from the 7805 and you don't have an adequate heatsink, so it is getting too hot and the thermal current limiting is kicking in. [This happened to me once with a 7912. Fitting a better heatsink fixed it.]

  3. Your 7805 is faulty.

How much current are you drawing from the 5v line? Bear in mind that some things can be run from the unregulated input to the regulator, e.g. LEDs switched by transistors, LCD backlights, relays etc. So you may be able to reduce the current needed from the 5v supply to a value low enough to use a TO-92 regulator.