|
|
|
![]() |
|
|||||||
|
||||||||
![]() |
|
|
Thread Tools |
Rating:
|
Display Modes |
|
|
|
#1
|
||||
|
||||
|
Add a couple of capacitors in parallel to the raspberry pi's power input. If your problem was the rippled power, this should solve your problem.
|
|
#2
|
|||
|
|||
|
Re: Battery powered raspberry pi
http://www.mouser.com/ProductDetail/...2fVA l1Xs8%3d
KA278R05CTU - Used this part to substitute my power supply in senior design. Low dropout, .5v @ 2amps. Basically, keep the input voltage to 5.5v or more and your pi should work. There's also an adjustable version if you're so inclined. Good luck. |
|
#3
|
||||
|
||||
|
Re: Battery powered raspberry pi
I think it's probably important to frame this for the reader as to the advantages and disadvantages:
1. Use a portable computing device that is designed with an integral battery: Advantages (limiting this only to the power issue): A1. No effort to design additional power circuits required. B1. Complete isolation from the battery used to power the robot motors. C1. Battery life that will exceed 30 minutes. D1. Easily replaced year on year (performance advantages outside scope). Disadvantages: A2. Limited opportunities to reduce the weight of the unit. B2. Something else to remember to charge. C2. Limited shelf life. 2. Create a battery powered COTS version of various boards like the Raspberry Pi, Arduino, BeagleBone, Panda... Advantages: A1. No effort to design additional power circuits required. B1. Complete isolation from the battery used to power the robot motors. C1. Slips past FIRST approval process if it is legal to do at all (unknown criteria). D1. Allows weight tuning to optiminal levels. E1. New market territory as not many things like this exist. Disadvantages: A2. We don't know how far you have to go to make something like this a COTS portable computing device. You might have to include parts at the time of sale that can be removed later to reduce the weight. B2. You get whatever you get power wise. How long is long enough to run on battery? If your target market exceeds FIRST there's a real design issue in battery run time. C2. Cost - you are bundling the cost of the computing device into the cost of the battery package. You must absorb both costs at the time of manufacturing and stocking. D2. New market means uncharted territory and risk. 3. Create a super capacitor based circuit to provide isolated power to whatever you connect to it: Advantages: A1. No effort to design additional power circuits required. B1. Potentially complete isolation from the battery used to power the robot motors. C1. You can provide power to whatever works within those limits increasing the target market. D1. If a new version of a computing device ships you don't care as long as it needs the same power. E1. Super capacitors do not contain as many nasty or volitile chemicals. F1. Bleeding edge market opportunity. Disadvantages: A2. Don't know if FIRST is willing to approve something like this. B2. Approval is a process that commits the maker to certain requirements. C2. The cost of the super capacitors mean this will likely cost at least $15 but more likely closer to $25. D2. Super capacitors generally will provide less run time then a battery for the same volume. E2. Cannot be expanded with a mere pass transistor - besides it stores power not merely regulates the power. F2. Bleeding edge markets might leave the seller bleeding. 4. Use a step-up then step-down power supply: Advantages: A1. Limited effort to design the power circuits. B1. Cheap - There are piles of Chinese boards for this on E-bay and through North American / EU supply houses. C1. The power supply maker doesn't lock you into a computing board as part of the package. D1. When the robot is off this will not hold power very long. E1. Can retain operation at proper output voltage down around a 3V robot battery. F1. Someone could mitigate many of the disadvantages below just be bundling these items into a COTS item. G1. If FIRST requires approval for this at least it has wide potential market. H1. At least 85% efficient and easily past 90% efficient. I1. It is possible to exploit existing power supply circuits in the PDB to achieve the first stage regulation. J1. Good isolation but when the input from the battery is charging the output filter of the first stage noise can pass. K1. The odds of 2 separate regulators synchronizing with the noise on the robot battery are very low. Disadvantages: A2. Very few COTS step-up then step-down modules prefabricated (boost - buck) B2. Generally single stage (boost - buck) converters have lower power limits than 2 seperate systems chained. C2. The circuits themselves are much more complex than a 7805. D2. End user packaging is probably required. E2. If someone bundles this into a COTS item FIRST may require approval for it. F2. Regulation can not be expanded with external pass transistor without inheriting the disadvantages from next choice.. 5. Use a simple 7805, 7809, 7812 or adjustable reference regulator: Advantages: A1. Simple. B1. Generally a small circuit. C1. Cheap - most of the car power supplies that output 5V are 7805. D1. A low dropout version will work to fractions of a volt at the input over the output voltage. E1. Great for limited currents. F1. Low weight when the current limits are low but high currents require heat sinking. G1. If you use mica insulators you might be able to heat sink to an aluminum or copper robot component. H1. Available in various packages with various current limits: surface mount, TO92, TO220, TO3 I1. It is possible to exploit existing power supply circuits in the PDB to provide initial regulation to reduce heat. J1. It is possible (but counter-productive) to make a circuit to bypass the regulator if the input voltage drops lower than the regulator input voltage requirements. Disadvantages: A2. Gets rid of excess energy as heat. B2. Increased current demands increases the heat generated. C2. At increased currents needs a heat sink so weight increases unless you can manage to use the robot components as a heat sink. D2. Does not deal with highly inductive or capacitive loads well. E2. Can sustain physical damage under high loads. F2. Can not produce a higher voltage than available at the output. G2. Versions that are not low drop out require at least 1V more at the input than the regulated output. Comments or suggestions let me know. Last edited by techhelpbb : 08-08-2013 at 15:53. |
|
#4
|
|||
|
|||
|
Re: Battery powered raspberry pi
Quote:
Quote:
The regulators that are not low-dropout (78xx, LM371, etc.) require that the input voltage be at least 2-3 Volts higher than the selected output voltage. When the input voltage drops too low, the output voltage also drops, with the voltage across the regulator being in the 2-3 Volt range, depending on the regulator type and the output current. With low-dropout types, the input voltage can be as little as 0.5 Volts above the selected output voltage. It would still be best to install a capacitor bank and an input diode on the input of the regulator to provide "ride through" time. The anode of the diode would be connected to the battery. The cathode would be connected to the capacitor and regulator input. The diode becomes reverse biased when the battery voltage drops and "disconnects" the regulator input from the battery leaving it to be powered from the capacitor. Without the diode, the other loads on the battery will discharge the capacitor and you are no better off. It would be best to use a schottky diode such as a STPS1545FP from ST since it has lower forward voltage drop than the normal rectifier diode. With the maximum input current of 0.7 Amps for the Raspberry Pi, a battery voltage of 11 Volts, a forward diode drop of about 0.3 Volts, a minimum input voltage of 6.0 Volts to the regulator and a ride through time of 1 second, one would need a 150,000 microFarad capacitor (C = current x ride through time/ voltage change). Nichicon makes a suitable part that is 51 mm in diameter and 120 mm long and is rated at 6.3 Volts. This capacitor is rated for 15.3 Amps so one would not get very much voltage drop when discharging it. One should use an oscilloscope to monitor the battery voltage, the regulator input voltage and the regulator output voltage during high load to verify that the capacitor chosen is large enough. http://www.digikey.com/product-searc...ds=LNR1E154MSE There is still the problem of getting this past inspection. Allowing for a lower battery voltage, lower regulator input voltage and shorter ride through time will reduce the capacitor value required and the stored energy in the capacitor. A smaller part would be: http://www.digikey.com/product-detai...579-ND/1882056 You will still have to deal with the ride through time issue with this scheme. You would want to choose a step up converter with a very minimum input voltage (at the desired load current) yet still has a high enough maximum input voltage that will accept the maximum battery voltage without damage. The output voltage should be as high as possible (check with next year's rules). You may want to choose a converter that has a maximum output current rating that is several times higher than the maximum load current. The rated minimum input voltage is usually with the maximum output current and will be lower with lower output currents. To get the most out of your system, it would be best to actually measure the maximum current draw of your Raspberry Pi while running the software you want it to run since it may be quite different than what the maximum current in the specification. You would want to choose the step down converter with a maximum input voltage that is equal to or just greater than the output voltage of the step up converter. You would also want the minimum input voltage to be as low as possible at the desired output current. Again choose a converter with a maximum output current that is several times higher than your maximum load current. You may need to connect a capacitor on the output of the step up converter to get the ride through time you want. The step down converter may have an input capacitor that will help with this. Use the same calculation shown above. You will not need the series diode since the step up converter will isolate you from the battery already. It would be best to use an oscilloscope to monitor the battery voltage, the output voltage of the step up converter and the output of the step down converter during high load conditions to ensure that the system is working correctly and the ride through cap is sufficiently large. When choosing the step up and step down converters, you do not need to choose isolated types since they are usually more expensive than the non-isolated types. The input reference terminal (0 Volt) of the isolated type is not connected to the output reference terminal but is connected in the non-isolated type. |
|
#5
|
||||
|
||||
|
Re: Battery powered raspberry pi
Quote:
The Nichicon super caps that arrived yesterday are rated for 200mOhms ESR. I could have gotten them with 100mOhms at higher cost. My intention was to charge them higher than the 5V and then accept the loss that will be present anyway because of the on-board Raspberry Pi linear regulators. My previous example feeding from a 7805 was merely an example. The rest is a great contribution to the topic. Thank you. Last edited by techhelpbb : 11-08-2013 at 13:15. |
|
#6
|
||||
|
||||
|
Re: Battery powered raspberry pi
I heard back from FIRST and they are willing to review a product using super capacitors. Of course you'd need to meet the same requirements with this project to get it on the approved hardware list as any other. Further they won't confirm or deny the rules in the future.
So there's the matter of making 10 of them. Disclosing the testing. Figuring out the costs. Determining distribution so everyone can get one. Potentially being asked to give FIRST enough for everyone free. Explaining who your company is and what resources it has. Of course the other take was see about making something like a Raspberry Pi laptop. Since there's no official approval for that I will now see if I can figure out how to approach that question. In the meantime this does not solve the issue but it is neat (the plans are COTS what you build would not be): http://blog.parts-people.com/2012/12...ble-rpi-to-go/ Further communication from FIRST representatives suggests that one could create a Raspberry Pi portable device that integrates a battery and a Raspberry Pi and it would likely fall under COTS rules of previous years. Again rules change so I am not sure how much risk someone is willing to take with this financially. I do have an idea of how to start. Last edited by techhelpbb : 14-08-2013 at 15:57. |
|
#7
|
||||||
|
||||||
|
Re: Battery powered raspberry pi
Our controls team is looking at integrating a raspberry pi with a vision system for a fall project. One question they have is on boot up time - the specs are quoting about 45 seconds. Has anyone had any issues with this, does it determine how it needs to be powered (breaker or direct), do you power up the robot in queue to give it time, and any other advice on integrating it to maintain connection through the match?
|
|
#8
|
||||
|
||||
|
Re: Battery powered raspberry pi
Quote:
|
|
#9
|
|||
|
|||
|
Re: Battery powered raspberry pi
FWIW, the Beaglebone Black ($45) boots up in under 10 seconds using the native Angstrom Linux.
|
|
#10
|
||||
|
||||
|
My Pi server (devstuff.no-ip.biz) takes 1.5 minutes to boot to a point where I can SSH to it. However, it takes up to ten minutes to get to max efficiency, where it is done launching every application! Is there a way how I can benchmark the boot?
|
|
#11
|
||||
|
||||
|
Re: Battery powered raspberry pi
To solve the boot time issue - perhaps put a circuit inline with the Pi's power that will use a dedicated battery until the robots mains come online? Two diodes and a capacitor will do or one can get fancy with a circuit based on something like the MAX6236.
HTH |
|
#12
|
||||
|
||||
|
Re: Battery powered raspberry pi
Quote:
If someone wants to make such an item they need to be prepared to supply anyone that wants it to make it legal as COTS. To get it approved outside of COTS they need to make several and go through FIRST engineering which will eat up this entire season between field testing and internal testing. Capacitors are a grey area. For a real storage capacitor you should treat it like above. People have managed to slide high value capacitors into competition but run time would be uncertain. Last edited by techhelpbb : 28-11-2013 at 10:48. |
|
#13
|
||||
|
||||
|
Re: Battery powered raspberry pi
Quote:
You missed the point of SWITCHING power sources. I'm not talking about storing energy anywhere illegal during competition. The battery (or whatever power source) used to boot the Pi is not on the robot "during the competition". You remove it after turning on the robot mains. If necessary to comply with the rules a second approved battery can be the source during boot time, similar to how many of us power the driver station or Ethernet switches (for debugging) between matches. |
|
#14
|
||||
|
||||
|
Re: Battery powered raspberry pi
Quote:
Last edited by techhelpbb : 28-11-2013 at 11:53. |
|
#15
|
||||
|
||||
|
Quote:
Off topic: Make sure you strip the Pi of all the software that you do not need because everything means greater power consumption and lower performance. You want to squeeze the juice out of the Pi during the few competition minutes. Put good load on the Pi, the Pi will do the work slowly but efficiently. For example: My Website run on a Pi. I have all sorts of other crapware on the pi too, but it still works with a decent result. Also, why are you using a Raspberry Pi? It is one of the lowest power dev boards available, just better than the cRIO by a few times. You might try looking at a development board like the oDroid. The specs make it seem like a good one to use. Along with removing the crapware to squeeze the juice out of the Pi, you might find it useful booting off USB with an HDD or even an SSD. I reduced my boot time twice by doing this. Also, SD cards die quickly when running an OS on them because that is against their nature. Booting from USB will allow you to use a more OS "Friendly" disk drive. Also, an SSD will be better becauause of the G-Forces on the drive in a moving robot. Also, how will you keep the SD card in place. TO RDP to it, use SSH. It eliminates the requirement for an entire server like xRDP. That helps with development. |
![]() |
| Thread Tools | |
| Display Modes | Rate This Thread |
|
|