Standard Motor Driver Schematic for FRC

[Alan Anderson]

Originally Posted by ProfessorAlekM

You would be surprised would students can do. It's really not hard to make motor controllers, I've made them myself before and I'm only in 9th grade.

Ah, to be a teenager and know everything again.

It's not hard to make a motor controller.

It's hard to design one that interfaces with standard hobby servo signals, provides sustained performance at 40 amps and occasional transients approaching 200 amps, is proven to perform as advertised so the robot can be certified as following the rules, is rugged enough to withstand the abuse it will take on an FRC robot, doesn't pose shock or fire hazards to its users, etc. It's hard to build one to those specifications without having resources, skills, and experience.

...our team has spent hundreds of the dollars on motor controllers.

If you're expecting to be able to save "hundreds of the dollars" by building your own motor speed controllers, I suspect you're failing to consider a lot of details.

[Oblarg]

Originally Posted by Alan Anderson

It's hard to design one that interfaces with standard hobby servo signals, provides sustained performance at 40 amps and occasional transients approaching 200 amps, is proven to perform as advertised so the robot can be certified as following the rules, is rugged enough to withstand the abuse it will take on an FRC robot, doesn't pose shock or fire hazards to its users, etc. It's hard to build one to those specifications without having resources, skills, and experience,

Should anyone doubt this, I'd suggest they direct their attention to the (now thankfully defunct) Jaguar motor controllers, which were made by a professional company who (ostensibly) knew what they were doing and still suffered from all sorts of reliability issues.

[ProfessorAlekM]

I was looking up the encoders on AndyMark. The encoders they sell are only $20 each when you buy them in bulk (50), and they sell them for less if buy more (it says I have to ask for a quote if more then 50), but they sell them for $40 each.

And I'm sure shipping +$5-10 in profit is not $20.

They also sold every team a base kit that can't even clear any of the obstacles, so that you would then have to buy another base kit or upgrades (Making them more money).

On top of that they knew what this years competition was, and they couldn't even keep Rhino Treads in stock during build. Some supply and demand could be going on.

I know that I'm getting off-track on my own thread, but I'm trying to conserve money for our school.

[Whippet]

Originally Posted by ProfessorAlekM

They also sold every team a base kit that can't even clear any of the obstacles, so that you would then have to buy another base kit or upgrades (Making them more money).

On top of that they knew what this years competition was, and they couldn't even keep Rhino Treads in stock during build. Some supply and demand could be going on.

I'm going to correct you on just this one point, since there will be plenty of people better able to address your concerns about motor controllers specifically. This year was the first time that FIRST revealed elements of the game to certain suppliers prior to kickoff. The only thing that Andymark or VEX knew prior to kickoff was that there would be obstacles for the robots to traverse and (I think) that it was a ball game. Nothing was disclosed of the size, type, or nature of the obstacles or balls. Accordingly, Andymark adjusted the Kit of Parts chassis to include 6 inch wheels rather than the 4 inch versions that have been provided for the previous two years. They did not know enough to guarantee that they would be able to move a sizable stock of Rhino modules, pneumatic wheels, etc. The same thing applies to VEX and West Coast Products.

The FRC suppliers have done an excellent job of handling supply shortages for the past few years. We recently purchased a set of hex shaft collars from Andymark and received a set of VEX collars - still in vex packaging - in the Andymark shipment. These companies do everything they can to help make sure your robot gets completed in time and I have nothing but good things to say about them.

[Kevin Sevcik]

Originally Posted by ProfessorAlekM

I was looking up the encoders on AndyMark. The encoders they sell are only $20 each when you buy them in bulk (50), and they sell them for less if buy more (it says I have to ask for a quote if more then 50), but they sell them for $40 each.

And I'm sure shipping +$5-10 in profit is not $20.

They also sold every team a base kit that can't even clear any of the obstacles, so that you would then have to buy another base kit or upgrades (Making them more money).

On top of that they knew what this years competition was, and they couldn't even keep Rhino Treads in stock during build. Some supply and demand could be going on.

I know that I'm getting off-track on my own thread, but I'm trying to conserve money for our school.

Andymark doesn't get specific game information ahead of time any more than any team does. I believe Andy said they were told there would be obstacles in this game or something like that, which inspired the rhino tracks. Makes it hard to design for specific obstacles. It also makes it hard to pre-buy enough stock to satisfy the demand of 3000 teams. It's obvious after kickoff that there will be a run on pneumatic tires, or banebot wheels, or whatever the hot item is. It's never obvious before kickoff.

On the cable, I suspect you didn't actually specify the encoder correctly on US-Digital. When you specify one spacer and centering tool per encoder, the price per 50 is $28. And then you add a $5 cable on top. Andymark and Vex don't have a huge markup. They can't because if they did, they couldn't compete with other suppliers like Robot Marketplace, Robotshop, Mouser, Digikey, various industrial supply houses, or any random Joe that sets up an EBay or Amazon shop. It's called capitalism.


Honestly, the best way to save money for your school is to not break things so you can save what's salvageable from year-to-year. After about 3-4 years of not exploding motor controllers or melting motors, you'll have plenty for whatever robot you're going to build. A few more years and you'll have plenty for two robots. The best way to save money in a particular season is to CAD and actually design your robot so you can minimize buying the wrong thing or express shipping something you need in 2 days before competition or bag.

[philso]

Originally Posted by ProfessorAlekM

You would be surprised would students can do. It's really not hard to make motor controllers, I've made them myself before and I'm only in 9th grade.

What sort of continuous output current was your motor controller capable of? How efficient is it? How durable was it, especially when abused? How long has it lasted? How does it behave during fault conditions (does it fault in a safe and benign way)?

A schematic is only a starting point for a successful piece of equipment. A schematic for a good motor controller will look a lot like the schematic for a bad motor controller. You still need an enclosure to keep the circuit board from shorting to adjacent pieces of metal and to keep the metal shavings out. You will need to have a suitable heatsink. Do you know how to select a heatsink that performs adequately yet is not too big and expensive? The higher the current in the circuit and the higher the edge rate of the switching, the more critical the track layout on the PCB is. The layout of the circuit board suitable for FRC motor controllers will not be trivial. A poorly done layout will have high inductance paths, leading to high surge voltages and possibly destruction of the switching devices. A common cure for high surge voltages is to slow the edge rate of the switching but this causes higher losses and heating and lower efficiency and possibly destruction of the switching devices.

There is already a student designed motor controller out on the market. Test results published on this forum show that it is less efficient than other motor controllers. I have been told that it uses twice as many MOSFETs as other controllers indicating that the devices are not being operated in an efficient manner. It is quite likely that the board layout was not done in an optimum way (see above).

Originally Posted by ProfessorAlekM

Although it doesn't seem like it, the point I'm trying to argue is that FRC would be much cheaper if they would give students the ability to create and buy their own motor controllers. When you calculate it, our team has spent hundreds of the dollars on motor controllers.

What is your team's time worth? Are you going to propose building your own controller too (because the processors are under $10) instead of purchasing a RoboRio ($435). This is a robot competition, not a motor controller competition.

It is likely that providing just a schematic will lead many teams to disaster. Not all of them will be able to obtain the proper parts and substitute a part that "looks close enough" (yes there is a difference between two parts with the same nominal rating). Most of the parts that will be suitable are "surface mount" types. Most people I have met (including engineers working in the field) do not have the skills needed to successfully hand solder surface mount components correctly.

There is at least one other individual who has posted in this thread who has indicated that they have to professional experience in this field to comment in an authoritative and credible way and they are all saying that there is a lot more to have a successful power electronic circuit than just having a schematic. My background is doing the design and development for motor controllers ranging from 5hp to 1200hp over the last 20+ years. I use many of the same design techniques necessary to make a successful FRC motor controller and it took me (and the others posting here) much longer to learn these techniques than a high school student would have. A few years ago, I visited a startup company that had a microprocessor guru/consultant who sold them on the idea that he could design a complete motor controller for them because he got hold of a generic schematic. I saw their prototypes (as well as the scorch marks on their equipment) and it was pretty easy to see what they were doing wrong. That start up closed down pretty soon afterward.

[GreyingJay]

There is a time and a place for saving money. DIY is sometimes the way to do that, but sometimes it is not.

2706 is a new team. We started from scratch last fall so we have had to buy everything new including all our tools, all the stuff on our robot and all the extra stuff we needed for prototyping and testing. Believe me we understand wanting to save money. There were many things we did ourselves in the interest of saving team money. Among them was judicious shopping for parts we could have bought from one popular supplier but found cheaper elsewhere. Motor controllers fall in this category, I think, especially with new controller options on the market now. You can even buy them from Amazon!

[Sperkowsky]

It takes more time to make custom motor controllers then finding a sponsor or two that will give you enough money to buy them. Its a cool learning experience but a waste of time in a 6 week season.

[GreyingJay]

Originally Posted by Sperkowsky

It takes more time to make custom motor controllers then finding a sponsor or two that will give you enough money to buy them. Its a cool learning experience but a waste of time in a 6 week season.

This is a very good point. In one Saturday of concerted effort our team can execute a community bottle drive that raises about $800-1000. You might be able to get even more by pounding the pavement and talking to shops and businesses in the area.

[zcohen]

I applaud my students who want to learn how to make custom circuits or custom anything. While our team doesn't do it much yet, my understanding is the RoboRio allows for this a lot more than past controllers (I may be wrong, programming/controls is my weak spot)
That being said, these are the things that control a 120 pound robot going fast enough to hurt someone. They also control a huge amount of amperage. I trust my students a lot, but not when it could be something where a mistake could greatly injure someone else.
Further, out of all of our costs, motor controllers are pretty small. Yes, we did dump about $2k into the Talon SRX's when they came out, and I do understand thats a lot of money, but you can easily get away with a couple hundred and still have the Talon SRX and not a cheaper option like the REV one if the options the SRX delivers are important to you
One big thing you're missing too is these are reusable. That was largely a one time cost compared to motors/gearboxes/framing/etc which far exceeds 2k a year.

It would be great to save money in FIRST, but I dont think this is where to look. If money is an issue with a team, look at the structural components as a way to save money. Reusing simple small things like hex clamps, sprockets, or bearings adds up to far more than a few speed controllers.

[Richard Wallace]

Originally Posted by zcohen

One big thing you're missing too is these are reusable. That was largely a one time cost compared to motors/gearboxes/framing/etc which far exceeds 2k a year.

Thank you for bringing this point into the discussion. FRC approved motor controllers are durable.

This year we are running several Talon SRs that we've used in two previous seasons. We also have the same number of new SPARKs, which are on our practice robot. After those have proved reliability we intend to re-use them on future robots. To date we have had no failures of Talon SRs (which are obsolete) or of SPARKs.

[Greg Needel]

I think that most people in this thread have summarized the reasons why letting teams make their own motor controllers is not a practical idea, as there are so many things that can happen when hand assembling something to handle this amount of power. Even if 99% of the motor controllers were done properly the 1% of them that fail can cause fire, robots not responding in a controlled fashion, and a bunch of other issues.


Some people might not have realized this, but we have published the schematic for the SPARK at the back of our user guide. We believe that people should be able to learn from the product itself and not just the use of the product. Here is the direct link (PDF) http://www.revrobotics.com/wp-conten...1200-UM-00.pdf

If any student wants to create their own "homebrew" version of the SPARK we encourage you to try, because reproducing is a great way of learning. For a low cost way to make your own PCB's, my friends own Dirtypcbs.com (and regardless of their name make good quality cheap boards)

[kiettyyyy]

Originally Posted by philso

The layout of the circuit board suitable for FRC motor controllers will not be trivial. A poorly done layout will have high inductance paths....

I'm sure that most of these teams when given the schematic will just arbitrarily place components onto the board layout and hit autoroute.

There's a reason for everything!

[Andrew Schreiber]

Originally Posted by kiettyyyy

I'm sure that most of these teams when given the schematic will just arbitrarily place components onto the board layout and hit autoroute.

There's a reason for everything!

How many of those issues are particular to high current designs? Does something sitting in the 15A constant 30A burst range have to worry quite as much about these issues?

[philso]

Originally Posted by Andrew Schreiber

How many of those issues are particular to high current designs? Does something sitting in the 15A constant 30A burst range have to worry quite as much about these issues?

It is the interaction of the circuit inductance AND the rate of change of the current that provokes issues (dV = L * dI/dt). Higher currents typically lead to higher values of dI/dT.

At work, I am working on improving a switching power supply where the peak primary current is less than 1 Amp. The person who originally did the PCB layout clearly didn't know how to do it properly. Neither did the person who designed the transformer. The circuit sprays copious amounts of electrical noise.

If one does not have a thorough understanding of how to control the parasitic inductances inherent in the physical construction of a fast-switching circuit, one WILL have trouble with issues such as noise and over/under-shoot. The parasitic elements are not indicated on a typical schematic since they are dependent on exactly how a circuit is physically implemented. It is up to the designer to determine where in the circuit the parasitic elements will be significant and how they can be mitigated. It is not black magic but it takes many years of hands-on experience to be able to do this effectively. Reading a book, attending a class and being smart are not enough.