Stabilize arduino power supply

[T_guttata]

Hi

When I'm using an arduino nano on 12V with various other devices, it happens, that the voltage drops due to high inrush currents (motors for example).

I measured the voltage drop with my DSO, it goes down from 12V to 3.5V within 10ms. As a result, the arduino restarts.

Now, I would like to use a capacitor to sabilize the input. With a 1000 uF cap charged to 12V, the arduino nano (without anything attached) will run for 600 ms until it turns off.

Will it work? Is there anything to improve further?

[Adrian_Arg.]

I in arduino the motors, and other components feed them separately with respect to the arduino, I do not take the voltage from the arduino, if I put the common negative to the peripherals and the arduino board but they do not work, since the current that the arduino can withstand it is little and can be burned.


yo en arduino los motores, y otros componentes los alimento por separado con respecto al arduino, no tomo el voltage desde arduino, si pongo el negativo comun a los periferico y la placa arduino sino no funcionan, ya que la corriente que puede soportar el arduino es poca y se puede quemar.

[Kim Christensen]

I would use a 1n400x (x = any number 1 to 7) instead for D5..

[David Hess]

Your solution will work but I agree with Kim Christensen; use a larger diode to handle the surge current charging the capacitors.

[gcewing]

If the other devices include inductive loads such as motors and solenoids, I'd feel safer running them from a completely separate supply, so that the Arduino can't be damaged by inductive spikes.

[T_guttata]

I would use a 1n400x (x = any number 1 to 7) instead for D5..

I'm not familiar with all the nomenclature for electronic devices. There are thousands of diodes. What are the important requirements? Diode type? Reverse voltage? Average current? Peak current? Will there be any disadvantage using diodes with higher voltage rating (several 100V)?

If the other devices include inductive loads such as motors and solenoids, I'd feel safer running them from a completely separate supply, so that the Arduino can't be damaged by inductive spikes.

Well, often, that's not an option. There must be a protection solution available.

[tooki]

I would use a 1n400x (x = any number 1 to 7) instead for D5..

I'm not familiar with all the nomenclature for electronic devices. There are thousands of diodes. What are the important requirements? Diode type? Reverse voltage? Average current? Peak current? Will there be any disadvantage using diodes with higher voltage rating (several 100V)?

They didn’t tell you to choose from thousands, they told you to pick one of seven: 1N4001 through 1N4007.

[T_guttata]

But they are not SMD components? 1N4148W is in SOD-123 package!

[Doctorandus_P]

Such capacitors will draw very high peak current during switch on, and this may damage connectors and switches.
You can add an extra series resistor to limit this the input surge current.
For example, a 10 Ohm resistor will limit the peak current to 1.2A, which is probably a safe surge current for the 1n4148.
If your arduino draws a constant current of 50mA, then this will also cause a steady voltage drop of 500mV though, so this resistor is a compromise.

The power suplies of the arduino's have always been overly symplistic, which is quite sad unfortunately.
If you replace the lineair voltage regulator with a small SMPS module, then you can more then double the time that your capacitors can power the arduino board.

The ATMEGA328 works with voltages down to 2V (Some versions 1.8V I think) and it draws less current at lower voltages. So if you reduce the output voltage of your SMPS module, then you can again increase the time that you can buffer from your capacitors.

And of course other factors that influence the current consumption of your arduino pcb.
The simplest is to make as much use of sleep modes as possible. With this you can reduce the current consumption to (much) less then a mA. If your arduino has an always-on-led then removing that will also contribute to the buffer time, and all other connected circuitry will also influence it.

[T_guttata]

Thanks for your explanation.
Energy consumption of the arduino is currently not my main concern. So, I will try with a 10 Ohm 0603 series resistor. This will end up in 
a voltage drop of approx. 0.7+0.5=1.2V? If the input is 12V, this should not be an issue.

I would like to stay with the SOD-123 package, because it’s small and I have enough diodes in stock. Currently, it‘s a pain to choose electronic components, because even standard components are often out of stock.

[rpiloverbd]

What about using a motor driver IC/module?

[Steffalompen]

Yeah, I've also wondered about this established 'truth' that one must have a separate supply. How far back does it go, will I have to go to another substation?

I'd say it probably will work, try it.
Also I have high hopes you can bypass the inrush resistor when the caps have charged with a MOSFET in polarity protection mode, as that conducts when voltage is reached.
And is there no way of softstarting that ruddy motor?

[Kim Christensen]

Energy consumption of the arduino is currently not my main concern. So, I will try with a 10 Ohm 0603 series resistor. This will end up in
a voltage drop of approx. 0.7+0.5=1.2V? If the input is 12V, this should not be an issue.

It might be a bit more if you look at the 1N4148W's datasheet. It could drop up to 1V at 50ma in some cases for a total of 1.5V... Still not an issue I would think. But if the max current consumption of the arduino is more than 50ma its going to be even more....

I would like to stay with the SOD-123 package, because it’s small and I have enough diodes in stock. Currently, it‘s a pain to choose electronic components, because even standard components are often out of stock.

Definitely a reason to use the parts you already have, so that makes Doctorandus_P's resistor solution a good one.

[tooki]

But they are not SMD components? 1N4148W is in SOD-123 package!

Then the equivalent is the S1 series, e.g. S1J ≈ 1N4004.

[David Hess]

I would use a 1n400x (x = any number 1 to 7) instead for D5..

I'm not familiar with all the nomenclature for electronic devices. There are thousands of diodes. What are the important requirements? Diode type? Reverse voltage? Average current? Peak current? Will there be any disadvantage using diodes with higher voltage rating (several 100V)?

JEDEC (Joint Electron Device Engineering Council) defined a series of part numbers.  1N devices have 2 leads, 2N devices have 3 leads, 3N devices have 4 leads, 4N devices have 5 leads, etc.

So 1N devices are mostly diodes, like the 1N4148 shown in your schematic.  The 1N4001, 1N4002, 1N4003 series of diodes are the most common power rectifers and have a peak current more suitable for handling the inrush current into a large capacitor.  Higher voltage diodes usually have a higher forward voltage drop and slower speed, but that is irrelevant in your application.  I usually rely on the 400 volt 1N4004 in place of all of the lower voltage versions just to simplify parts inventory.

2N devices are mostly transistors.  3N devices are mostly 4-lead transistors like dual gate MOSFETs and FETs with a separate body connection.  4N devices (5 leads) are mostly transistor optocouplers.

[T_guttata]

Thanks, your post is very appreciated, did not know that. That's exactly the type of post which gives added value! 

I have a mechanical background and it's not that easy to find the right components. There are a quite infinite number of abbreviations and different nomentclature systems and usually it's not explained. Even for the footprints of SMD components there are many variations, sometimes very similar, sometimes there seem to be even several notations for the same. I usually like to know the meaning of all abbreviations which I use, because it's easier to memorize. But in terms of electronic components, it's almost impossible to know all^^

I usually order components from digikey, but I like to look for components in smaller stores first. In the smaller stores there are usually only the more common types of components, which makes it a lot easier for beginners.

So 1N4004 means: 1N --> 2 leads, the last digit means higher = higher voltage. But what's about the digits in between? (..400.)?

[tooki]

They don’t mean anything in particular, and the last digit referring to voltage is specific to the 1N400x series, it is not something defined by the nomenclature.

Ultimately, with components in general, some types have structured nomenclature, but most are at most structured within the particular manufacturer. There’s simply no way around using datasheets to verify what something is.

[Steffalompen]

Here's something I doodled up while I should be wrapping christmas gifts, to bypass that pesky inrush resistor.
If MOSFETs didn't have body diodes this would be much sleeker. (I suppose you could use two back to back)
I tried it in a breadboard, and after charging caps the inrush resistor had no voltage over it anymore, with a 20mA load.

[David Hess]

So 1N4004 means: 1N --> 2 leads, the last digit means higher = higher voltage. But what's about the digits in between? (..400.)?

The digits except for the first two are issued in the order that the part was registered with JEDEC, so older parts have a smaller number, and not all registered parts were put into production so there are gaps.

And of course JEDEC part numbers are only issued for parts that are registered with JEDEC.  There are plenty of parts which have nothing to do with JEDEC and use the manufacturer's, or sometimes the customer's, numbering scheme.  For instance Motorola registered plenty of JEDEC parts, but also has their own MUR series of fast switching rectifiers, their MJ series of transistors, etc.  International Rectifier used IR.

There are other part number schemes as well.  The EU has BC for transistors.  Japan has 2SA, 2SB, 2SC, 2SD, etc.