Electronics > Electronics

Power Conversion Questiions

(1/4) > >>

seal killer:

(Hi. I am a newbie.)

I am an electronics novice. However, I can pretty much read a schematic and I built a lot of stuff when I was younger, including a few robots. (Think 30 years ago.)

It is time for me to get back into the robot building hobby. There is so much more available now than before! It is amazing.

My current design will be large (50 to 250 pounds), slow (walker with 8 legs) and I would like to off-load all the processing to a PC or a network of PCs via wireless link. (I am looking at Robot Electronics RF04/CMO2 combination with the GPIO14[ for I/O.)

Here is my question. It looks like I will definitely need 5v for logic, 6v for motors, and MAYBE 12 volts for . . . something. Can I just use a couple 12 volt batteries in parallel and hook a 5 volt DC to DC switching converter to them AND a 6 volt DC to DC switching converter? I cannot figure out a reason to run the leg motors--which will be very low rpm and include encoders--from a converter. Therefore, my plan is to power them directly from the batteries.

I would really appreciate any thoughts you may have for me.


Sounds expensive  ;)

Lets see, 8 legs, 3 motors each, thats 24 motors . . . Each motor will be like $50 . . . that equals . . . a lot.

As for your battery question, it really depends on how much current your components will draw. You must use a 5V voltage regulator for logic, and you can use any battery you would like for it. I recommend a small 6V NiMH battery.

For the motors, you do not need to regulate voltage and you should run it directly from the batteries (the exception is with servos). Depends on the voltage rating of your motors, but I recommend getting a 7.2V NiCad or Lithium battery.

As for wireless links, search this forum for 'wireless', 'blue tooth,' 'easy radio,' and 'laptop.' There are several posts you may be interested in.

And a side rant:
Try really hard to make your robot weigh as little as possible. Weight is very important with legged bots . . . Use carbon fiber, plastics, etc - avoid all use of metal if possible.

seal killer:

I am here to learn and I have already learned a bit! (Power the motors directly, as I suspected.)

My motor source is currently Banebot. I bought one of their RS545 motors (MP-36256-545, 12v, 61 rpm, $41.30). It is far too much motor for even my heavy machine . . . which will not weigh anywhere near 250 pounds. The truth is, I do not know how much it will weigh at this time. Nor, at this highly experimental, learning stage do I feel a need to know how much it will weigh.

I have two more of Banebots' motors coming. Both of these are on the other end of the scale; two FF-050 series (MS-16118-050, 6v, 126 rpm and MS-16574-050, 6v, 26 rpm). My structural parts will be mostly aluminum. I will turn all the joints on a lathe. (A side note: My machine's leg joints will all be driven by screw drives. My amatuerish guess is this will allow me to use very small, gear-drive motors since there will be additional "gearing" via the screw drives.)

I will search this forum for the wireless links you suggested. I will also post something soliciting comments on the RF04/CMO2 combination, on which you did not comment. Coupled with the GPIO14, which provides 14 input/output pins (and other "must have" stuff) and is daisy-chainable over the I2C bus up to 8 units, it seems ideal for my purposes. (Other than what I REALLY wanted: A wireless version of the Elexol Ether IO 24! I wrote them and suggested they offer such a device. They wrote back and said they had been thinking about it. Since then, they have quit answering my e-mails.)

I know I can learn a lot here as this forum does not seem to be oriented to any one microcontroller series or language, or other "main" thing.

In reply to your (gentle) "rant": It seems to me that the more legs the less one must worry about weight. Obviously, the more legs the less power required of leg motors. There is an (indirect) inverse relationship between number of legs (or wheels!) and the amount of weight the structure will support and the power consumed by individual drive motors. Of course, the larger the machine, the more total power will be required to move it a given distance. My machine, which is all in my mind at this time, will be powered by 12v, lead-acid garden tractor batteries. That way, I get lots of amps, easy recharging, and zero learning curve due to familiarity of use.

I am a high-level programmer by trade (OK. I USED to be. Since I have been in the business for well over a quarter century, I have joined the dark side and am in management) and am fascinated by the idea of terabytes worth of bit-state joint position tables available over a wireless link. I am also a somewhat successful GA programmer. ("Somewhat successful" means that I have written GA programs that work and actually solve "problems.") I would like to couple the bit-state joint position tables with some GA programs and see what happens . . . other than my movable components wearing out very quickly!)

But, here I am to learn, not teach. Hopefully, you and others will do the latter for me.


The reason why I dont recommend a lead acid battery is they generally have low current output.
Looking at the MS-16118-050 motor, it says .3A no load current and 1.8A stall current. Lets say each of your motors
only draws 1 amp, thats 24 motors x 1 A = 24A. A lot! You either need to put a lot of batteries in parallel, or use lithium polymer batteries. Check the current output rating on the battery you use . . . I am a big fan of a minimum learning curve too, but just know that lead acid weighs a lot more and outputs a lot less when you design your robot. I used 2 motorcycle batteries on my very first robot (which failed miserably), and I refuse to use lead acid again  :P

Yea you are right about more legs means less force carried by each leg. Also to note, you only need to design one leg, and can just make 8 of them using mass manufacturing techniques like CNC. The effort of making eight legs is about the same amount of effort in making one . . . And how many motors you plan to use per leg? I assumed 3, but perhaps 2 instead?

You may want to consider using servos as they already have built in feedback position controllers. Otherwise, if you get a regular DC motor, you then need to incorporate a leg joint angle sensor with PID control. Hitec is now making some really nice powerful 7V servo motors. Check the partlist http://www.societyofrobots.com/robot_parts_list.shtml

Sorry . . . I didnt comment about the wireless parts because I am not that experienced in them myself and dont want to say something stupid. Dunk, if you read this, this is your cue!

Since you are a seasoned programmer, I was curious of the algorithms and sensors you plan to use. I encourage you to keep us updated with your project!

seal killer:

Again, I learn! I live to learn.

I DID NOT KNOW that lead acid batteries provide relatively low amperage. I thought just the opposite. Thank you for saving me a configuration/design mistake at the very outset. (Initial mistakes are the most expensive, not necessarily in terms of cost for me, but in terms of TIME.)

I am not embarrassed to say that I really know nothing about PID controllers. (If you divest yourself of all the ego possible to get rid of in the first place, you are in an ideal position to admit your mistakes and ignorance and therefore, learn.) From what I can tell a PID controller is a device which supplies a firmware-based negative feedback capability to your machine. A PID controller also may utilize an error history to refine its feedback. Is that anywhere close?

I am an inventor with awarded USPTO patent(s). There will be five motors per leg due to something I am planning to patent that will enable far more NATURAL movement of the leg. The chance of a successful patent is fairly low. The chance of a successful patent if I describe it publically one year or more of the patent FILING date is ZERO! I hope to get some time to build a full scale leg model (the model will be simple and look almost nothing like the actual leg) that will allow me to test motors (and sensors). I will post pictures and videos of that model here. However, I will take them such that the potentially patentable part is not obvious or actually covered up. If and when I do so, I will re-explain to everyone and make full apologies at that time.

I have looked at Hitec in the past, but will look at them again thanks to your guidance.

I am a very seasoned programmer, but I think I am living on my past laurels! I REALLY need to start cutting some code again since I have discovered that it is NOT like riding a bicycle! I would like to write everything in VB6.0 or some version of BASIC because those are the last languages I used. However, if ABSOLUTELY necessary, I can write anything in pure binary. (Ha! I guarantee you that it will never be absolutely necessary, though! <g> )

My algorithms will be fairly simple since I will depend on terabytes of learned postions stored in tables as bit on/bit off entries. When my network of desktop PCs discovers an error in machine position caused by driving the machine via a table, they will simply update the appropriate table(s). My plan is to "teach" (this is not as tough as it might sound) the machine how to perform all discrete, fundamental movements, store the sensor positions in a bit table and then access that table either in its entirety or partially to recreate the movement/position called for by low level motivation logic and/or real-time sensors. (It will be a house machine, as well; at least initially.)

My sensors will initially be very simple: Bit bumpers are easy for me because I employed such things almost 30 years ago. The only modern stuff I have used is an SRF08. Since my household environment and my machine design and concept only requires 6 inches, or even less, of "notice", I need to become familiar with IR on IC2.

What I REALLY need is to figure out a way to employ a LOT of bit sensors. I am fairly certain I can do just over 100, but I would really be happy to be able to do much, MUCH more, both input and output. All ideas are welcome!

Remember, I am a novice and need to be treated that way or I will not learn . . . fast enough.



[0] Message Index

[#] Next page

Go to full version