Actions
Forum Thread Details
  • State Not Answered
  • Replies 3 replies
  • Subscribers 391 subscribers
  • Views 552 views
  • Users 0 members are here
Related

Arduino for CVT transmission management?

Former Member

I'm looking into the capabilities of an Arduino Uno or Due to control an electromagnetic clutch on a CVT transmission. I'm leaning towards a Due, if only because of the increased speed. The board would need to take multiple inputs, and determine through the programming if it was time to change the gearing to a different ratio.

 

A little background, I goto an college that participates in an international competition for go carts of sorts, we currently run a CVT that uses weights with springs on a hub assembly to determine when to shift the gearing ratio to a higher or lower gear, based on the rotational mass.

 

The inputs would be; Engine RPM, Wheel Speed, CVT Clutch Speed, rear axle speed, throttle sensor. Mostly using hall sensors. I realize that this is a big undertaking, not something that could be completed in a week, but I think that the college has the resources available to complete such a task. I'm just looking for some insight as to the capabilities of the Arduino, if a Uno or Due would be upto such a task. Is there a better board that could undertake the rigors of being run in a endurance event, better suited than an arduino?  Which board? Would the inputs be able to be output to a graph, or have a logging capability?

Parents
  • 0

    For straight sensor input the main things to keep in mind are you are looking at 8-bit vs 32-bit microcontrollers (uC).  My first three questions:

    1. What are your sampling intervals?

    2. Are you looking for something that can directly act on programmed logic via sensor values, or something that can factor algorithms to derive PID control?

    3. How many analog inputs will be used, and will you use smoothing that requires X values per stored value?  (OK, technically four questions).

     

    The Atmega328 in the Uno runs at 16MHz with 6 analog ports, vs. the ARM-M3 AT91SAM3X8E Due at 84MHz with 12 analog pins and two DAC pins.  Since you'll be using registers IMHO the Due is much more responsive, as would be expected.  That said, if I could use an Uno I'd likely prefer it; simply based on simplicity of design, but I'm not certain that's possible for you.  There may be some automotive embedded systems engineers able to give you a more detailed response.

     

    As far as ruggedness, I've put several SMD R3 Uno's in an industrial environment where 60-400Hz vibration is very high, power is very noisy, and ambient temps range from -10F to +120F.  I can't quote you a MTBF, but replacements have been one board out of 12, in a period of 2.5 years, with systems that run 24/7 with no sleep period,  I've not deployed Due systems in that same environment, but I've pushed them in environmental conditions that would be considered "rugged".

     

    The Arduino IDE just included graphing via Serial.println(analogRead(A0)) (obviously A0-A5 can be used in the Uno) using the serial monitor.  There are also excellent graphing packages for Arduino (Earth People's Serial Graph Link, MakerPlot (my favorite), and others), as well as being able to use Labview.

     

    If you're looking for something to give you PID (like perhaps the CVT transmission) you might want to use the Arduino just for I/O management, and feed a stream to a single-board computer (SBC) for more complex algorithms; like a Raspberry Pi or a BeagleBone Black.  There are plenty of others - Those are just two that are well-known.  Just a thought.

  • 0 in reply to Former Member

    Brian,

     

    For straight sensor input the main things to keep in mind are you are looking at 8-bit vs 32-bit microcontrollers (uC).  My first three questions:

    1. What are your sampling intervals?

    2. Are you looking for something that can directly act on programmed logic via sensor values, or something that can factor algorithms to derive PID control?

    3. How many analog inputs will be used, and will you use smoothing that requires X values per stored value?

     

    Honestly, I'm trying to figure out a board first, 8 bit vs 32 bit, no idea. Electronics are out of my realm, beyond fixing basic circuits and replacing/re flowing/re soldering components. I'm trying to start with a simple circuit of at X rpm to transition to a higher gear, show that the principle works, and then go from there.

     

    The sensors will mainly be for turning the basic system into something far more advanced. I'm not thinking that this will work on programmed logic atleast initially, but more on sensor values, though that might not happen for awhile either. I'm not even sure if an electromagnetic clutch is the way to go, car manufacturers use an electro-hydraulic system, but because be are competing against other teams, and have such a small engine, parasitic loss is a real concern.

     

    I need to contact a teacher that can help me find a student that is advanced enough in his/her C programming that can help out on that end as well, I think I'm more of the bend the rulebook or idea person, as well as a mechanic then being into the technical electronics side.

     

    I appreciate the quick feedback.

Reply
  • 0 in reply to Former Member

    Brian,

     

    For straight sensor input the main things to keep in mind are you are looking at 8-bit vs 32-bit microcontrollers (uC).  My first three questions:

    1. What are your sampling intervals?

    2. Are you looking for something that can directly act on programmed logic via sensor values, or something that can factor algorithms to derive PID control?

    3. How many analog inputs will be used, and will you use smoothing that requires X values per stored value?

     

    Honestly, I'm trying to figure out a board first, 8 bit vs 32 bit, no idea. Electronics are out of my realm, beyond fixing basic circuits and replacing/re flowing/re soldering components. I'm trying to start with a simple circuit of at X rpm to transition to a higher gear, show that the principle works, and then go from there.

     

    The sensors will mainly be for turning the basic system into something far more advanced. I'm not thinking that this will work on programmed logic atleast initially, but more on sensor values, though that might not happen for awhile either. I'm not even sure if an electromagnetic clutch is the way to go, car manufacturers use an electro-hydraulic system, but because be are competing against other teams, and have such a small engine, parasitic loss is a real concern.

     

    I need to contact a teacher that can help me find a student that is advanced enough in his/her C programming that can help out on that end as well, I think I'm more of the bend the rulebook or idea person, as well as a mechanic then being into the technical electronics side.

     

    I appreciate the quick feedback.

Children
  • 0 in reply to Former Member

    Understood Andrew;

     

    You generally want the board to fit your end goal - Not build the system based on a board you want to use,  If you need to start with the board then your inclination to go with the Due is a good one.  That will give you the best chance of having all the input/output pins you need; as well as being able to use a chip that's 6.5 times faster than the one on the Uno.  The caveat with that is to look closely at any "Shields" you might want to use.  Shields are piggy-backed boards that provide easier interfacing between the main board and whatever peripherals you're using (sensors. motors, etc.).  The Due is a 3.3VDC board, rather than 5.0VDC as with the Uno (Mega, Micro, etc.).