What power source setup should I use?

Projects Discussion and Showcase Robotics
1

I have read multiple threads spanning 10+ years in this forum to find an answer, and it seems the most efficient way to find one is being super transparent with my intent :grin:. So here it is:

I am looking to build a multiped robot that can be trained to walk using ML/simulation. Potentially also computer vision capabilities. Hence it would need a raspberry pi, multiple servos, and an arduino. Why both raspberry pi and arduino? Well, in this video, this man makes a very compelling argument for them being better together i.e. a "best of both worlds" approach.

I know it's a huge project with a ton of both hardware and software hurdles and lost sleep but today I'm here to ask about power sources.

So now I need to figure out how to power multiple servos, arduino, a raspberry pi, (including their peripherals). The question is, should I use three batteries (servos, arduino, rpi)? two batteries? one battery? I've read a ton about different types and arrived at LiPo being the best power:weight ratio so I was hoping I could use a single 11.1v 5200mah lipo battery and then step down the voltage to simplify and power everything, unless that's a terrible approach?

I know I will get questions on servo specs, etc. but this is chicken and egg. I am flexible on servo strength, robot size (have 3d printer), and budget, because if I get told I need a ton of big batteries then obviously everything needs to be bigger/stronger. Hopefully not though, I envision the robot being smaller than a toddler. In short, I can get stronger/weaker servos/robot depending on this answer.

I watched this video on powering Arduino and rpi with LiPo but he goes straight into PCB and also does them individually, not together + servos. My first version will be breadboard, then if the project doesn't crash and burn like my other aspirations, move to PCB, etc.

What is the best approach in terms of battery(ies) and other components (voltage converters, regulators, fuses, or more)?

2

Yes that is the way I would do it. Use switching regulators rather than linear ones for maximum efficiency and minimum waste heat.

3

Start with an umbilical, and spec the supply from that

4

I've recently built a machine with numerous sensors, actuators, motors, servos, etc, so will share my experience. Life is much easier if you power the whole thing from one battery. Then you only need one charger (or one type of spare battery in your kit bag). The LiPo you mentioned sounds great.

Get an idea of what voltages you will require: for example it might be 11.1V for motors, 6V for servos, 5V for microcontrollers, 3.3V for sensors, and so on. This is just by way of example - make your own list. Then see if you need two of any of those; for example, you might choose to have two 5V rails, one for each microcontroller in order to keep their power rails clean.

Next, estimate a ballpark figure for how much current each supply must deliver. For example, your 11.1V motor and 6V servo supplies may well run to several amps, whereas your 5V rail might be no more than 3A, and less than 1A for the 3.3V sensors. Remember, these are just made up examples - you should estimate these for your own robot.

Then, take a look at the vast number of DC-DC converters available on sites like Banggood. Buy a set that meet those various requirements (consider variable voltage ones, if you like), plus several spares. Variable voltage converters reduce the number of different spares you must keep.

Mount the battery, charger (if on-board) and converters into a removable power module if you can. Leave space in the box for a couple of extra converters and output connectors in case you want to add new features to your machine in the future. Add an LED for the power input and each power output so you can see at a glance that the entire power module is working correctly.

The outputs from the power module should be considered as a set of power rails, or busbars, at the specified voltages and current capabilities (not unlike a PC power supply) which are extended out into the various parts of the machine. Then you can wire each module in your machine to it's appropriate power rail. Think of the busbars/rails as distributing the power around the machine. It is useful to regard the power supply and distribution rails as a distinct module in a design sense.

The other approach is to have just one bus going around the machine (i.e. the 11.1V from the battery), and then install small DC-DC converters local the the point of need. This is a distributed model, whereby the power supply is - in effect - distributed around and amongst the rest of the robot.

I recommend the first method because it is easier to manage during the design and build of your machine. Both are used, and both will work, but a distributed power system is more likely to lead to mistakes because you tend to build it ad hoc.

I'm a fan of DC-DC converters, because most of them don't need any fuses, and will look after themselves in terms of over-voltage, over-current, over-temperature, etc.

1 Like
5

Thank you all for the prompt replies - it is reassuring to know having a single battery source is the right direction. It seems like planning will take some time, I definitely cannot make it too heavy/big or I'll have to start thinking about brushless dc motors instead of simply high torque servos. I will make make a list.

Special thanks to @SteveThackery for such a detailed reply, covering multiple aspects of this question and then some. The methodology with the 2 approaches was especially useful and should be the accepted solution I think.

I agree with your recommendation on the first approach, also on using DC-DC converters. Final related noob question: For the prototyping stage, would it be as simple as connecting the battery --> main breadboard power rail --> different converters --> for each converter, their own breadboard power rails with LED? (i.e. this is the final set of power rails that you described here)

The outputs from the power module should be considered as a set of power rails, or busbars, at the specified voltages and current capabilities (not unlike a PC power supply) which are extended out into the various parts of the machine.

If so, I am happy that I have a way forward, although it sounds I'll have to come up with a way to reduce bulkiness from multiple breadboards

6

Are you talking about solderless bread boards? If so they can only handle up to about 1A, Anything bigger should be soldered up with strip board. Even strip board has a limited current capability and everything over about two amps should be reinforced by soldering tinned copper wire on the back of the strips carrying large currents.

1 Like
7

I don't think you'll need to do all that much breadboard prototyping. Your machine will be complex mechanically, and the software will be complex, too. But I think most of the electrical work will be sensors and actuators. The actuators will need to be driven by switching transistors, and there is some minor development work there (there are endless circuits showing how an Arduino can drive a motor or servo - just do a bit of searching), but that circuit will probably be repeated throughout your machine, so won't require extensive breadboarding.

Breadboards are really intended for developing electronic circuits, where you need to experiment with resistors, capacitors, ICs, etc. I may be wrong, but I don't think there'll be an enormous need for that in your project.

1 Like
8

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.