lightweight, inexpensive speed controllers

[yash101]

Are you programming this yourself? If so, it will be complicated because I am implying that you have no flying experience. You might try a KK board controlling the copter, being controlled by the Arduino. The KK board will stabilize the quad and do all the important functions. The Arduino can be the controller. You may wind that easier. How much experience do you have in Arduino programming? I would use something like the Propeller chip because you can code the multicore environment to do the flight controlling in a core, and the rest of the stuff in another! However, this all is based on what MCU you are most experienced in using!

Also, using a linear voltage regulator at those currents would cause the magic smoke of death to puff out of the regulator. Powering my RasPi from an LDO gets the chip so hot, I have it connected to the aluminum chassey of the robot for a ginormous heat sink!

[mman1506]

Originally Posted by Andrew Schreiber

One of the guys at my local space has a series of nasty scars up his right side from when his quad turned on and the blade caught him. Turns out CF blade vs skin… the CF wins by a lot.

Don't forget these are machines with a simply crazy amount of power.

When working with your batteries be careful not to short them. Be careful flying near people, even a small quad dropping at someone can hurt.

Don't even need CF. The APC props on my quad sliced through my finger like butter.

[efoote868]

Originally Posted by yash101

Also, using a linear voltage regulator at those currents would cause the magic smoke of death to puff out of the regulator. Powering my RasPi from an LDO gets the chip so hot, I have it connected to the aluminum chassey of the robot for a ginormous heat sink!

Depending on the current draw of the controller, as well as the supply voltage of the battery, a low dropout regulator may be just fine. Basically what you need to remember is the difference in the battery voltage and supply voltage is just waste at the linear regulator.

If you have a 12V battery with a 5V linear regulator, drawing 700mA from it will cause it to act as a 4.9W space heater (gets hot very quickly).
If you have a 7.4V battery with a 5V linear regulator, drawing 50mA will only waste .12W, which is much more manageable from a power dissipation perspective.

[yash101]

Yeah. Thee Arduino should draw about .2-.5A. The Gyro and other sensors may use up more, so you may use close to 1A/Hr!

[ILAMtitan]

Originally Posted by yash101

Yeah. Thee Arduino should draw about .2-.5A. The Gyro and other sensors may use up more, so you may use close to 1A/Hr!

An Arduino running at full tilt will only consume about 20mA, and only 12mA of that is the Atemga itself. Gyros will pull maybe 4mA (x3 though). Toss on an LCD like the KK2 model uses (which I HIGHLY recommend), and your power budget for the control board is still in the sub 50mA range.

On a quad, or any aircraft, the bulk of the power is consumed in the motors, and the rest is such a small percentage you can basically ignore it. With a 20A ESC, that control system is taking less than one percent of the total power. A small UBEC will work just fine.

[Ian Curtis]

Originally Posted by efoote868

If I were to approach this problem, I might start by looking at what's available to the average consumer / near hobbyist level. I believe brushed DC motors are used on the popular Syma S107G model and knockoffs, and searching Ebay I can get a pair of them from China for $6 (including shipping).

One could also try to cannibalize more of that helicopter to get appropriate (and proven) gear ratios, as well as mounting hardware.

Controlling a brushless DC motor is not trivial, which is to say that as a hobbyist you wouldn't build one from scratch as you might an H-bridge or other brushed DC motor controller.

It is about 100 times easier to do it properly and just buy the correct hobbyist gear. The pager motors used to drive the $30 mall helicopters are nothing compared to the brushless hobbyist motors that are *very* easy to use. This stuff is meant for casual users, so it is way easier than trying to hack the cheapo helicopters. (not that that isn't fun too)

General note for people purchasing R/C toys: Buy the real thing! A $100 beginner heli from Horizon Hobby is more durable and orders of magnitude more controllable (and thus more fun) than the cheap-o ones at Brookstone.

/Soapbox

[Al Skierkiewicz]

Ryan,
This is really a much more simple project than what you have been thinking. Your motors are only going to run in one direction and then only need to be controlled for speed. In brushed motors, it is a simple matter to control the drive current to control speed. A simple amplifier is all that is required. Either a transistor or MOSFET can be used. It is also possible to control the speed by controlling the on time of the transistor which is also easy to accomplish with a micro-controller doing all the control. The transistor then becomes a simple switch turned on and off by the controller. MOSFETs are usually the device of choice since they have a much lower "on" resistance and therefore are more efficient. No caps are needed in this case. Usually a simple resistor and the transistor are all that are needed. The three terminal regulators (7805) are available in small packages (TO92) to supply the controller. And they are very cheap.

[yash101]

Use Brushless motors. With the controller, they MAY cost a little more, but will be around the same price range. At those high RPMs, brushed motors are going to rip their brushes off! It will be easier to control brushless motors, as compared to brushed motors because brushed motors don't react to signals linearly, so the small irregularities in the motor will cause them to react differently. With brushless motors, you get absolute control because these motors are commutated by you, so you can possibly set the exact RPM you want! In other cases, you could use an encoder on the brushed motor. However, that would still be quite hard to program correctly, especially on the Arduino. The amount of data coming in from the encoders could be so much that it could overwhelm the DSP!

[philso]

Originally Posted by efoote868

Well, if you're still in the planning stage, you need to first figure out what you want as the end product, and then you can begin to fill in the details of how to get the end product.

FRC is an "Engineering Competition". I know your quadcopter is not for FRC but the same though process apply.

Originally Posted by tilky

thanks for the pointer about the diode. finally, any suggestions about specific mosfets to buy (don't forget, will be operating at 5v)? and do i need to use resistors at any point, whether from the PWM signal etc.

oh, and is a "snubber diode" a specific type of diode or will any type do?

The answers to these questions are not rocket science but to properly address them (and the interactions between these choices) would require more than a few posts on a forum like this. If you were in the Houston area, I could help you out. Otherwise, there may be a hacker/maker space near you with someone who has hands-on experience doing this. If you are feeling adventurous, you might consider reading application notes from MOSFET manufacturers such as Fairchild, Infineon, Vishay and others. If you really want to get into it, there are books such as the one by Mohan that are classics in the power electronics field (http://www.barnesandnoble.com/w/powe...an/1101194755). Unfortunately, quite a bit of the advice you have been given in this thread will likely cause the release of a lot of "magic smoke" and lead you to waste time and money.

Originally Posted by nuttle

Bottom line is there's a ton of stuff already done for this particular application and you probably want to at least start with what is readily available and in widespread use. The underlying technology is similar, but you most likely do not want to try to build your own controller -- lots of optimization has already been done, just as for motors for this application.

Originally Posted by efoote868

If I were to approach this problem, I might start by looking at what's available to the average consumer / near hobbyist level.

Controlling a brushless DC motor is not trivial, which is to say that as a hobbyist you wouldn't build one from scratch as you might an H-bridge or other brushed DC motor controller.

Even though I design large motor controllers for industrial applications in my day job, I would still buy a ready made one for a project like yours since it would be very difficult for me to beat the price. You have more than enough challenges to overcome in other areas of your project.

Originally Posted by Andrew Schreiber

I'm using http://www.hobbyking.com/hobbyking/s...ontroller.html for mine.

I think these are the ones that several of my co-workers use for their electric airplanes and helicopter.

[yash101]

Originally Posted by philso

FRC is an "Engineering Competition". I know your quadcopter is not for FRC but the same though process apply.




The answers to these questions are not rocket science but to properly address them (and the interactions between these choices) would require more than a few posts on a forum like this. If you were in the Houston area, I could help you out. Otherwise, there may be a hacker/maker space near you with someone who has hands-on experience doing this. If you are feeling adventurous, you might consider reading application notes from MOSFET manufacturers such as Fairchild, Infineon, Vishay and others. If you really want to get into it, there are books such as the one by Mohan that are classics in the power electronics field (http://www.barnesandnoble.com/w/powe...an/1101194755). Unfortunately, quite a bit of the advice you have been given in this thread will likely cause the release of a lot of "magic smoke" and lead you to waste time and money.







Even though I design large motor controllers for industrial applications in my day job, I would still buy a ready made one for a project like yours since it would be very difficult for me to beat the price. You have more than enough challenges to overcome in other areas of your project.




I think these are the ones that several of my co-workers use for their electric airplanes and helicopter.

I will more-or-less have to agree with you in many ways and disagree in a few ways. The fact that you mentioned, that you need to go by the same engineering process as in FRC is nothing but correct! Ryan, you may want to skim through this. However, where you said that a lot of stuff we told Ryan wouldn't cause magic smoke to appear, if done right. And yes, this ISN'T rocket science (though that is just an exaggeration because rocket science isn't as hard as developing a nuclear reactor) . Also, what is this company you work in? I am designing a microcar, and I am having a difficult time in finding a good speed controller that is a lot more advanced than those crappy DIY ones available! Everything will be controlled by MCUs, like the QC we are talking about!

Ryan, after you build this with Arduino, I'd like to see you post a video of it, and for a challenge, you should place an RPI doing vision tracking to automate this QC!

Goog luck!

[philso]

Originally Posted by yash101

I will more-or-less have to agree with you in many ways and disagree in a few ways. The fact that you mentioned, that you need to go by the same engineering process as in FRC is nothing but correct! Ryan, you may want to skim through this. However, where you said that a lot of stuff we told Ryan wouldn't cause magic smoke to appear, if done right. And yes, this ISN'T rocket science (though that is just an exaggeration because rocket science isn't as hard as developing a nuclear reactor) . Also, what is this company you work in? I am designing a microcar, and I am having a difficult time in finding a good speed controller that is a lot more advanced than those crappy DIY ones available! Everything will be controlled by MCUs, like the QC we are talking about!

Ryan, after you build this with Arduino, I'd like to see you post a video of it, and for a challenge, you should place an RPI doing vision tracking to automate this QC!

Goog luck!

Yash

You are welcome to disagree.

I am interested in what process you used to arrive at the component selection recomendations you have made regarding the MOSFET, capacitor, diode and regulator. Please also post a schematic showing how you feel these components should be connected to clarify your descriptions in your previous posts in this thread.

There are a number of manufacturers of motor controllers for the EV conversion market. In your thread about your microcar, I had already directed you to your local branch of the HEAA where it is much more likely that someone with the appropriate expertise to help you.

I am not aware of Toshiba manufacturing any 3-phase motor drives suitable for the automotive market. If they did, there would be an exclusive supply agreement with an OEM such as Ford or Toyota and one would only be able to purchase those drives from a car dealership. If you are really curious, the following is a link to one of the products that I have been doing R&D work on. These are aimed at heavy-industrial and oil-field applications.

www.toshiba.com/ind/product_display.jsp?id1=7&id2=861

[thinker&planner]

Please pardon if I accidentally restate something that has been previously said, I have foolishly not taken the time to read ALL of the replies yet.

I have been working on a similar project, but instead of a quadrotor, I'm building an ROV (underwater). Basically, MOSFETS are the way to go for a small, compact yet extremely powerful speed controller. Here is a video that helped me a lot in understanding MOSFETS that I hope will help you. I strongly recommend checking out the other videos in the series; they are very valuable and well thought out electronics tutorials.
http://www.youtube.com/watch?v=CFt8h...ACE&index= 26

If you use MOSFETS, you need a snubber diode and really should have some capacitors for decoupling and signal noise reduction. You should use a tiny ceramic cap for very fast reactions paired with a larger electrolytic cap to handle the large loads.

I can send eagle files/circuit diagrams if that would help, just pm me.

[Al Skierkiewicz]

The video is a nice discussion but the diode in the power MOSFETs is a fall out of the manufacturing process. The FETs used in our speed controllers all have them. The real benefit is the very low ON resistance source to drain compared to a normal transistor. One of the effects of the diodes is that the speed controllers light up when you push the robot with the power off. The current generated by the motors turning, ends up passing through the diodes and out to the power supply.

[yash101]

I want to put this in a 4S quad . I love magic smoke

That's what I call the king of batteries! I searched HobbyKing and this is their biggest battery

Btw, I bookmarked that channel cause I liked that video!


I think that you will like this, Ryan:
http://www.hobbyking.com/hobbyking/store/__26944__IDEAFLY_IFLY_4_Quadcopter_with_Motor_ESC_ Flight_Controller_PNF_.html

I forgot how I got the link. Maybe I found it on this forum?!