Ultracapacitor powered arduino - feasable?

Hi,

Arduinos are typically switched off simply by disconnecting the power. However, I've worked on several projects where this isn't ideal, the system really needs to go through a shut-down routine where SD files get closed and connected devices are told the device is going off line.

One of my recent projects could be improved greatly if it would send one last SMS when it shuts down, the problem is that I've really got no control of when a user pulls the plug! - no matter what instructions are given, there's really nothing stopping someone switching a unit off at anytime. So, what I'm thinking, is adding one or more ultracapacitors to the project, the idea being that when external power is pulled, the ultracapacitor will have enough charge to power the arduino for a few extra seconds so it can send a final SMS.

Does this sound plausible so far?

Lets say my project normally operates at 12 volts, and the arduino requires 5volts minimum. Ultracapacitors normally operate at 2.7 volts (see datasheet http://www.maxwell.com/products/ultracapacitors/docs/maxwell_technologies_product_comparison_matrix.pdf ) So i have 5 ultracapacitors in series that charge to 12 volts, and as soon as the external power gets disconnected then these capacitors will discharge into the arduino and power it for few seconds.

Here's the problem as I see it, an SMS can take up to 10 seconds to send, and requires quite a bit of power (2amps). Battery capacity is measured in mA/hour, but capacitor capacity is measure in farads (not something I understand), so how many and how big a capacitor would I need to provide 2amps for at least 10 seconds?

On a related note, anyone know where I can purchase (retail) ultra capacitors? The only distributors I can find are wholesalers.

All help appreciated

Thanks.

Two problems there.

  • Ultracapacitors are capacitors, so as they deliver current, their voltage drops in direct proportion to the current. You could use two ultracapacitors for 5V, but at 1 Farad (total - two 1F caps in series), you drop 0.25V for every second you draw 250mA. You really need to use a switchmode regulator to maintain a steady voltage to the electronics from the continuously reducing voltage.

  • If put in series, ultracapacitors do not discharge evenly due to difference in capacitance, nor do they charge evenly, so have to be controlled by special charging circuits like lithium ion batteries to control each individual capacitor. The more capacitors you put in series, the more of a problem this is. It is much more practical to use higher voltage capacitors, and parallel them if necessary.

OhMyCod: Hi,

Arduinos are typically switched off simply by disconnecting the power. However, I've worked on several projects where this isn't ideal, the system really needs to go through a shut-down routine where SD files get closed and connected devices are told the device is going off line.

One of my recent projects could be improved greatly if it would send one last SMS when it shuts down, the problem is that I've really got no control of when a user pulls the plug! - no matter what instructions are given, there's really nothing stopping someone switching a unit off at anytime. So, what I'm thinking, is adding one or more ultracapacitors to the project, the idea being that when external power is pulled, the ultracapacitor will have enough charge to power the arduino for a few extra seconds so it can send a final SMS.

Does this sound plausible so far?

Lets say my project normally operates at 12 volts, and the arduino requires 5volts minimum. Ultracapacitors normally operate at 2.7 volts (see datasheet http://www.maxwell.com/products/ultracapacitors/docs/maxwell_technologies_product_comparison_matrix.pdf ) So i have 5 ultracapacitors in series that charge to 12 volts, and as soon as the external power gets disconnected then these capacitors will discharge into the arduino and power it for few seconds.

Here's the problem as I see it, an SMS can take up to 10 seconds to send, and requires quite a bit of power (2amps). Battery capacity is measured in mA/hour, but capacitor capacity is measure in farads (not something I understand), so how many and how big a capacitor would I need to provide 2amps for at least 10 seconds?

On a related note, anyone know where I can purchase (retail) ultra capacitors? The only distributors I can find are wholesalers.

All help appreciated

Thanks.

How would you propose to switch power sources?... ie, if the power goes out and the super caps (2 in series to give 5v) where connected to the vin or whichever it was that bypasses the regulator onboard the Arduino... when the power is out, the caps automatically take over and reduce in voltage, yes it keeps it going if someone unplugs it but you need a way to know if the power is "off"

but you need a way to know if the power is "off"

I think I can monitor the external power supply via a diode on an analogue input pin.

You really need to use a switchmode regulator to maintain a steady voltage to the electronics from the continuously reducing voltage.

Yes, that was my intention.

It is much more practical to use higher voltage capacitors

Ideally, yes, but all ultracapacitors seem to work at about 2.5 volts (I guess it's something to do with the physics of the dialectric). Conventional capacitors do operate at higher voltages, but have a fraction of the capacity.

Preferably, use a single 2.7V ultracapacitor and a DC-DC converter to generate 5V or 3.3V for the Arduino from it. You can charge it from the 5V or 3.3V supply using some sort of current limiter (e.g. a resistor) and a voltage limiter. Start the DC-DC converter when you sense the incoming power going off. If necessary you can use a conventional capacitor or small 5.5V ultracapacitor to power the Arduino for the short time it takes the converter to start up.

PS - work out the energy you need as voltage * current * time. For 5V and 2A for 10 seconds, this comes to 100 joules. The energy stored in the ultracapacitor is 0.5 * C * V^2, so each 1F of capacitance charged to 2.5V stores 3.125 joules. In practice, your DC-DC converter will only work down to e.g. 1.25V, so you can only extract around 75% of this energy; and you need to allow for the losses in the converter. But it sounds to me that a 100F 2.7V capacitor may be sufficient.

Why dont you just use a rechargable battery?
Have the powersupply power on a relay, and when disconnected it switches the relay off and then over to the battery. You can find lithium ion batteries all over the place that would supply stable current for quite a while after the power was cut.

Several people have described projects where they use a capacitor to power the system long enough to close files and so on when the power fails, so that part is certainly feasible. Getting the SMS sent seems like the hard part to me - it will take a significant time to complete and require a lot of power while it's happening. So, that particular part of your requirement makes the problem a lot harder and requires a much bigger capacity power source such as a rechargeable battery. If you know how to use an ultracapacitor as a power source and have one with sufficient energy capacity and are capable of charging and discharging it correctly then you could use that instead - but a conventional rechargeable battery does seem simpler.

Co-incidentally I was reading a bit about ultra-capacitors as "stay-alive" devices for DCC model trains recently. It seems to me (in both applications) that a rechargeable battery is better suited to the application.

I presume ultra capacitors are both fast and efficient at charging and discharging - neither of which seems relevant to keeping an Arduino alive occasionally for a few minutes.

...R

One of the reasons for choosing an ultra capacitor was that I had thought it would last longer (more charge/discharge cycles) than a rechargeable battery. However, it seems from this discussion that it might be the only practical option. Bummer - I've been looking for an excuse to play with some new technology for ages!

OhMyCod: One of the reasons for choosing an ultra capacitor was that I had thought it would last longer (more charge/discharge cycles) than a rechargeable battery.

You are right, a battery has a limited number of charge/discharge cycles. A lithium battery also requires relatively complex circuitry to control the charging and protect against over-discharge. The ultracapacitor requires a DC-DC converter to get the best out of it, but the charging circuitry (for a single capacitor) can be very simple.

OhMyCod: However, it seems from this discussion that it might be the only practical option. Bummer - I've been looking for an excuse to play with some new technology for ages!

Not at all! I've already pointed out that a 2.7V 100F capacitor should be sufficient (if your figure of 2A for 10 seconds is accurate), and you can pick these up on eBay or from the usual component distributors.

What you need is a part like this http://www.linear.com/product/LT3580

It will take the capacitor output and boost it to a steady level until the cap droops too much.

You will need to use Low ESR caps that will support higher current flow. High ESR caps will look like they have a big resistor in series and their output voltage may look like its drooping way too early.

http://parametric.linear.com/html/Internal_Power_Switch_Boost?s=&o=&cols%5B%5D=2167&cols%5B%5D=1032&cols%5B%5D=1033&cols%5B%5D=1035&cols%5B%5D=1034&cols%5B%5D=1040&cols%5B%5D=1036&cols%5B%5D=1367&c%5B1032%5D%5B%5D=2.5&x=17&y=6

Also take a look at Design Notes DN450, N485, DN487, DN498, all dealing with using supercaps.

The concept you're referring to is called "dying gasp" BTW, which might help you search for other solutions to this problem.

My first thought would be to try a bunch of traditional electrolytic caps in parallel. They're cheap, ubiquitous, and you can charge them to a relatively high voltage -- 12v for example, but if you have a good switching regulator that will handle higher input voltage, then you can raise that even higher and pull less current.

I don't know that sending an SMS really pulls 2A for 10s, so your real-life energy consumption is probably much lower -- especially given the duty cycle of GSM, etc.

Whilst hopefully the "way of the future", ultracapacitors clearly have a long way to go yet in development.

There are projects to use them as storage for electric vehicles but serious power storage means getting them to operate at many *kilo*volts (and not as series arrays either).

The self-discharge rates in that datasheet suggest that they are pretty useless for power storage beyond a few minutes, perhaps an hour or so.

Lithium-ion batteries are going to be the most practical for the present. In light use, and as a "backup", they last many years.

UltraCaps are best used for substaining low-power CMOS state during battery changes - not active electronics during shutdown.

Use a latching relay (or "e"version of same) and have the power switch On state Latch the relay, the Off state is issued by the uC or other control logic after the pre-shutdown requirements are met. Ex: think how your cellphone shuts Dow.

Ray

mrburnette, you are thinking of the older technology carbon layer capacitors. Those do have a rather high ESR, causing long charge times and only allowing low power use. The newer 2.5 to 2.7V capacitors are -very- low ESR, allowing very fast charge times, high discharge currents, and typically come in much higher capacitances than the older carbon layer caps.

I hate this whole ultra-super-duper etc. just like I hate it for LEDs. The first of the LEDs that were much brighter were "only" something on the order of 300mCd, but they called them super, ultra, etc. -bright, even when they later got to be 10 and 100 times brighter. So all those names mean nothing.

you can pick these up on eBay

You're right, i've just checked - didn't realise how expensive they were!

think how your cellphone shuts

Yeah, but the difference is a cell phone has an internal battery, so you can'd just physically disconnect it. My project has an external battery and I have no way of preventing someone just pulling the plug!

The concept you're referring to is called "dying gasp"

thanks for this, it's amazing how much easier research becomes once you know the correct term. I once wasted several hours researching how to make a fountain like the Bellagio fountain in Los Vegas, it suddenly got a lot easier once I discovered they were called 'linear flow fountains'

I've come up with a short cut for my project, rather than attempting to send an SMS on shutdown, (taking several seconds and lots of power) I'll write the shutdown data to a temporary file, and then send the contents of the file via SMS next time the unit is switched on. This should massively reduce the power/time requirements.

OhMyCod: Yeah, but the difference is a cell phone has an internal battery, so you can'd just physically disconnect it.

Haven't you just described the solution for your own project?

...R

Quote from: OhMyCod on Today at 08:37:08 am Yeah, but the difference is a cell phone has an internal battery, so you can'd just physically disconnect it.

Haven't you just described the solution for your own project?

...R

My project is supposed to be a physically secure (i.e. tamper proof) device. Putting a battery inside it requires opening/closing it periodically, requiring the tamper-proof seals to be broken.

I hate this whole ultra-super-duper etc. just like I hate it for LEDs. The first of the LEDs that were much brighter were "only" something on the order of 300mCd, but they called them super, ultra, etc. -bright, even when they later got to be 10 and 100 times brighter. So all those names mean nothing.

I think the prefix is justified for this technology. Traditional capacitors have values values of millionths or billionths of farads, ultra capacitors have values of multiple farads.

How often is periodically? Once a year?

If that really is a problem why not make the device with two compartments - one for the secure stuff and one for the battery. Like in my portable radio, but with the back glued on rather than screwed.

If what you are really looking for is a secure system of telling you the power has been disconnected (such as a tag for a prisoner) - that’s a different matter. In that case a couple of NiMh batteries within the secure compartment should work reliably for a few years. One of them powers my shaver. I think I’ve replaced it twice in several years - and it is used as the main power source, not as a standby.

…R