Observations about Tetrix Kit parts
Team 2901 - [GEARS] (That's Purple Gears for those of you who don't play World of Warcraft) had a lot of fun with the Tetrix kit for FTC but in the process learned a few things that we feel could improve the kit overall. We have tried to keep in mind mechanisms to ensure attention to engineering design while also trying to take into consideration the overall cost of components. We break the observations and suggestions into these basic areas
Axles
The difference between the size of the axles and the motor shaft was the source of much challenge. The team initially built a conveyor belt configuration for their robot using a large number of tubes, tube caps, axles and rubber bands as you can see in the image. Ultimately they opted for a different configuration due to the challenges in axle sizes. These challenges stem from three basic areas.
- The TETRIX™ Split Clamps do not provide sufficient compression to lock onto the axles. While one would expect that a clamping collar to be stronger than a set screw collar, the four screw holes actually mitigate that effect. As a result even though the compression screw was completely tightened down, it would still slip on the axle.
- Attempting to use the TETRIX™ Motor Hubs which do
have a set screw to address this problem solves the slipping problem but introduces the problem with the shaft being off center. It is this off center problem that results in the stripping of gears and introduces additional friction to the rotation. Attempting to put a brass collar in the center of the gear to center it might work, except that there is no effective way to secure the brass collar. The team finally hit on a combination of a split clamp on either side of the gear with the motor hob in the center. While this does solve the centering problem with the set screw to provide sufficient locking, there is no screw to secure all four elements (including the gear)
- Additionally, the split clamps extend past the edge of the tube such as to prevent free rotation. The Motor Hubs do not have that challenge.
Suggestion: Based on looking at the various parts, we came to the conclusion that the kit really had two different sizes for the shafts. One for the motors (Motor Hubs, Gears, Shaft Encoder and another for the axles (Brass bushings, set collars and split clamps). Considering that the larger diameter shaft of the motor provides greater gripping power and taking into consideration the relative price of the elements, it would seem that by replacing the axles with one that is the same size as the motor you could eliminate the split clamps in favor of the motor hubs and only have to replace the brass bushings and set collars. This has a secondary benefit outlined below in eliminating one of the hex wrench sizes.
Motors and Servos
In The process of building our robot, we broke multiple motors as can be seen in the picture to the right. There were two different problems which occurred here. First the choice of the tabs for attaching to the motors is extremely problematic on two fronts. The tabs do not have any horizontal flexibility and the metal used for making them is brittle. With the wires hanging an additional inch perpendicular to the edge of the motor, it is easy to brush against them and break the tab off. Even if a tab gets bent without breaking, attempting to straighten the tab can result in a broken tab. As the tab is not repairable, the motor is effectively useless at that point in time.
Suggestion: Build a cap for the end of the motor which clips onto the two tabs and provides a connector for power. Such a cap can retrofit existing motors and would provide a bonus of eliminating connecting the +/- backwards. One connector used for robotics is the PowerPole connector which can handle the amps necessary to drive the motor.
- Secondarily, the brass collar on the shaft of the motor to which the shaft encoder is attached can break loose pretty easily. This happens with the gear next to the shaft encoder causing rotational pressure on the collar. Again, once this happens, there is no way to repair the motor. Given the relative short shaft length and the size of the elements to secure to the shaft - particularly when a shaft encoder is mounted, this is easy to have happen. Fortunately the two parts can be separated and one working motor can be constructed from two broken ones :-). We believe that this breakage may be a manufacturing quality control problem and not a design problem.
- The placement of the shaft encoder on the motor introduces an additional challenge. The wire going to the shaft encoder is fragile and unfortunately is right around a lot of rotating gears with not a lot of options to allow you to secure it The solid face of the motor mount presents a nice wall leaving no options to secure the encoder wire.
Suggestion: Drill a single hole in the non-split side of the motor mount to provide a clear place to route the encoder wire.
The biggest challenge comes from the choice of an off center shaft for the motor as you can see in the image. Using the standard 16mm spacing of the holes on the Tetrix system, there are many configurations where the existing gears will naturally mesh with one another. However the motor is 37.5mm with a shaft that is 7mm off-center precluding taking advantage of the natural Tetrix 16mm spacing. While we recognize that this does allow for adjustment of the motor for "tightness" we believe that this leads to sloppy designs which results in the gear slippage and grinding often heard at matches.
Another argument made here for the rotating motor to tighten is that it makes it easier for students to put on the gears, but it is worth pointing out that with the exiting 16mm spacing, it is very easy to just slip gears on the end of the shaft and have a solid connection with no chance for slippage.
Suggestion: This is probably the biggest change of all. There are two options to consider here. First is to switch from an off center gear box to a centered gear box and ensure the motor mount aligns to the 16mm Tetrix system spacing. Alternatively switch to a gear box which is 1mm further out from center (i.e. 8mm from center - ideally with some detent to mark those points) to allow natural alignment with Tetrix at the 4 poles yet still allow for people to take advantage of slop in their design. Given that the gear box appears to be a standard industry size, we are guessing that a center mounted shaft may be a more viable option.
Additionally the design of the motor mount is another source of trouble. Given where the motors are often placed, getting to the nut holding the long screw which locks down the motor is challenging at best when the team needs to change out a motor.
Suggestion: Make the the motor mount two pieces. One which is solid comprising the bottom half with the existing screws and a second for the top which is not as wide as the based and screws into the base.
-
The alignment problem with the 16mm Tetrix system also affects the servos as can be seen in the image to the right. This makes it difficult to use a servo to drive a gear which is mounted on the structural frame.
Suggestion: Move the holes on the servo mount to align the center of the servo shaft with the 16mm normal Tetrix spacing.
Structural
The 1/2" and 5/16" socket head screws provided with the Tetrix kit are reliable, solid and easy to use. Unfortunately the same can not be said for the button head cap screws. These screws appear to be made of a much cheaper metal than the socket head screws and are unique to the kit in requiring a special sized allen wrench. The combination of the small size, the quality of the metal and the situations where one needs to use this screw cause it to be easily stripped. We have had to drill out three such screws so far and end up throwing away about a third of the button head cap screws that we take out. With our previous experience with the stainless steel screws supplied with the Vex kits we have never stripped one of their screws.
Suggestion: replace the button head cap screws with a better metal and a larger sized hole. Ideally one which is the same size as the socket head screws although switching to even a flathead screw would be an alternate choice.
- The kit uses four different size allen wrenches in the process of assembly. Two of those wrenches are for unique elements. The next to the largest is only for the Motor Hub and the next to the smallest is only used for the button head cap screws while the smallest is used for axle set collars and the motor encoder.
Suggestion: Eliminate two of the wrench sizes. Depending on other choices taken this would likely be the smallest and the next to largest size but other options are possible.
- There appear to be two different sized screws provided with the motor mount and the gear hub spacers. We have found that there were several times where the 1/2" socket head screws were not long enough and would have used the longer screws except for the fact that they are not fully threaded.
Suggestion: Provide a 3/4" and even a 1" version of the socket head screws which are fully threaded.
We find that the flat brackets are an absolute necessity in building our designs. However we have found numerous times where we needed holes in the center. While we recognize that it could affect the overall strength (we have our share of bent plates) we find that we have had to drill into these plates more often then we would like.
Suggestion: Provide a version of the flat brackets with holes in the center (indicated by the green) and possibly additional holes indicated by the purple to allow for 45 degree mounting of them.
- The Tetrix Channel aluminum design is both light and strong making it a solid base to build upon. However in order to achieve any structure we find that you have to end up building something that is double high with one U
facing up and the other facing down. This leads to unnecessary height and makes it difficult to have a nice plane upon which multiple elements are placed. We have experimented with many different options to address this.
- One of the first was simply the use of the L brackets to join them on the inside as shown in picture A. This unfortunately results in a gap between the two channels and the holes for the end of one channel are not lined up with the plane of the other channel.
-
Picture B has us using the trusty flat brackets to lock them together, but this leaves nothing to secure the bottom of the channel resulting in a weak structure.
- Switching to an L brackets inside the end in picture C to address this problem looks like a good start... but ... As you can see in picture D, the flat brackets do not line up. In fact, there are numerous cases where using the L brackets results in the joined pieces not being aligned along the normal 16mm boundaries.
Suggestion: create a new element which is similar to the 32mm Tetrix Channel but is smaller by the width of the aliuminum such that it fits snugly into the end of the Channel and allows for solid bonding with the edge of the other channel while still maintaining the 16mm spacing.
Suggestion: Modify the L brackets so that when they are used such as in picture A they result in maintaining the 16mm spacing.
Electronics
We very much like the approach of using the Lego® Mindstorms® brick for programming as the base of the the robot. The plug and play capabilities of the cables to connect the motors, motor controllers, and sensors eliminate some of the frustrating aspects of the robot design and allow concentration on the real problems to solve. Unfortunately those good attributes seem to stop at the HiTechnic DC Motor controller. This design uses not one but three different types of connection. We have the Lego Mindstorms cables for daisychaining the I2C interface from the brick, wires that you screw in to get power from the battery and provide power to the motor, and a keyed molex style connector for the shaft encoders from the motor. If you consider the servo controller, you get a forth connector for the servos which is not keyed. Having different connectors for different functions is reasonable and expected in the industry. This provides a good learning experience for the students, however providing a flat head screw connector for the power for the motors introduces another unique tool in the set (and one thing very common asked for to borrow at competitions). It is also worth noting that the prototype version of the motor controller on the HiTechnic site actually uses a molex style connector for powering the motor.
Suggestion: Use a molex style connector or PowerPole connector to connect the motor to the motor controller. Ideally you can use one connector for power and a separate 6 conductor one for the motor combining the power for the motor with the shaft encoder output.
Suggestion: Since the batteries already have a molex style connector on them, why not put the same connector on the motor controller and servo controllers. It would completely eliminate an additional wire.
- One of the advantages of the Vex kits is that out of the box the controller can drive the robot with no programming required. After looking through the samples and having to help out two different teams, we didn't find one which exactly drove the motors out of the box. Ideally providing a ready to go program which controlled 2 motors and 2 servos using the standard controller would go a long way to getting that first level of satisfaction with making a robot run.
- The current configuration of the HiTechnic Touch Sensor Multiplexer allows multiple touch sensors to be used, but with only four input ports on the NXT it means the most that a team can use in building a robot is the daisy chained motor/servo controllers, 4 touch sensors and only 2 additional sensors. We were hoping to use two range sensors, a light sensor and the compass in order to get accurate location of the robot but found that there was insufficient ports to be able to accomplish that. Since all of these go off the I2C interface from the brick it would be useful to allow for more sensors on the robot.
Suggestion: Create a multiplexer like the Touch Sensor Multiplexer which allows for multiple sensors to be plugged in.
- Bluetooth... well what can we say here? The speed of pairing combined with the interference from every cell phone in the area was a constant source of pain. Since the NXT is a given here, What should be done is to eliminate the use of Bluetooth and go to a customized wireless solution (USB over 802.11) to connect the bricks. There was a wireless USB standard which used UWB but apparently many of the companies have gone out of business. I'm still researching alternate solutions for this.