Group Actions
Engagement
  • Author Author: dougw
  • Date Created: Date Created
  • Views 1486 views
  • Likes 11 likes
  • Comments 9 comments
Related
Recommended

The Spice of Pi - Blog Doug 3

dougw

This week the Spice of Pi has been all about designing hardware and ordering enough components to get it working.

I had ordered a Google AIY Voice Hat to implement Google Assistant and voice recognition. It has arrived, but it uses a cardboard box for packaging, so I want to design packaging that is a little more appropriate for use in a kitchen.

The Pi 3 and touch screen are going in a bamboo chassis - I thought it was appropriate for a kitchen:

image

Next up is the speaker - I spent some time designing a nice speaker enclosure for the speaker that came with the Voice Hat and my 3D printer did a great job printing it out:

Then I tackled the mechanism to lift a selected spice jar. After the carousel rotates to bring the correct spice jar into position, a lifter mechanism raises the selected spice jar so it may be taken from the carousel. I went through numerous design ideas trying to figure out a good way to lift spice jars, ranging from very complex machines to simple levers. The main issue was how to translate the rotary motion of the servo motor into straight vertical linear motion for the lift. Ultimately the design involved several iterations of design and 3-D printing of about 10 components before I was happy with the result. As time consuming as it was, this type of design would have been very difficult to pull off without a 3-D printer. The end result may look obvious and simple, but from my perspective it was anything but simple - there are a lot of moving parts that could bind up or jam, and some of the prototypes did.

The next item on the menu was the servo drive circuit - we can't have an element14 design project without a circuit design.

Most of the electronics are purchased modules, and fortuitously the Voice Hat has servo drivers on it. However one thing about servo motors is they can consume power when stationary and they can also sit there buzzing away hunting for a specific position. The Spice of Pi mechanisms will not move if power is removed from the servo motors, so I want to make a circuit that allows power to the servos to be turned off when they are not needed. To do this, I want to use extra servo channels as digital outputs to power switches. Connector wise I will use 2 adjacent channels where the second channel is just controlling power to the first channel. The circuit is pretty simple - a logic level N-channel FET will turn a P-channel power FET on or off. This allows the power FET to control power at a higher voltage than the Raspberry Pi supply. The power FET has a built -in diode that will snub any motor power spikes. The cute part about the design is it will allow 2 servo motors to plug into these power & servo control cards which in turn plug seamlessly into the voice hat without interfering with each other or anything else.

I need to send this out for manufacture right away as it is already going to be in a race with Chinese New Year - which will freeze all manufacturing activity in its tracks and backlog all shipping activity.

Here is what the circuit looks like so far:

image

The power switch is serious overkill - capable of switching 74 amps, but they are cheap and I can get them. They have 0.02 ohms on resistance - I have pieces of wire with higher resistance.

Next I will work on the carousel design. I have worked out which spice jars I will be using (spice jar prices can be shocking) and have figured out how the jars will sit in the platter. I have ordered and received quite a few parts - not sure yet if I have enough parts, but the design is taking shape.

 

Design Challenge Links:

Pi Chef Design Challenge

About the challenge

The other challengers

The kit

Terms & Conditions

 

Project Links:

Blog Doug 1 - The Concept

Blog Glenn 1 - AIY Voice Kit Unboxing

Blog Doug 2 - The Block Diagram and Bill of Materials

  • in reply to mcb1

    That is similar to what I was thinking, except I was trying to have both cams on the same cam shaft - just rotating clockwise or counter clockwise depending on which cam was to lift.

  • in reply to dougw

    I considered using cams

    I thought that when I saw your design, so I'm pleased that you had considered it.

     

    My thoughts were to use it in a fixed position and rotate the platter until the right spice is over the cam then operate it.

    Since there are two rows, it would need two independant cams.

     

    The space required could be halved by using a horizonal cam (more like a rotating ramp), bu the platter would need to have space, so it adds to that part of the design.

     

    Regardless of which system used there are always trade-offs between complexity and the number of parts.

     

     

     

    Well done, and thanks for expaining the various options.

     

    Mark

  • in reply to aspork42

    It will raise either jar up, but only one moves at a time - the second jar sits stationary on a lip while the other jar goes up and the piston under the stationary jar goes down.

    As you guessed, I considered lots of mechanisms and geometries before getting this working. Unfortunately I didn't capture the failures on camera - like the prototype that jammed up, or the one that couldn't raise a jar high enough, or the one where the piston popped out, or the one where the pistons wouldn't fit the cylinders, or the one where the connecting rods were too short, or the one where the jars tended to tip over.

    I considered using cams, but it was tough to get enough lift in the space where the cam would access the jar and the precision axle would be hard to implement with bearings. Moving the cams down to get more room requires linkages which is essentially what the connecting rods do in a simpler way.

    I spent a long time trying to work out a simple way to implement trapezoidal arms that keep the upper surface horizontal as they rotate upward. However, every time the solution got close to looking like it would work well, the complexity increased unacceptably. To start with, each arm has 4 hinges.

    I looked at using a scissors lift type of mechanism, but they have a lot of hinges to implement and the drive mechanism would also be expensive or complex.

    I thought about using a vertical lead screw, but it was an expensive method, including the fact it would require 2 separate motors and mechanisms.

    I thought about pneumatic pistons and magnetic levitation for a while, but chickened out.

    I considered raising the whole platter to rotate it, then lowering the platter leaving just the selected jar high, but it still needed a mechanism to select one of 2 jars at that angle.

    I also considered just putting a bump in the rotation mechanism so whenever a jar passed the selection point it would be forced upward, but I didn't like the aesthetics of having every jar bob upwards as the platter rotates.

    I tried using just a servo with a long horn arm and no connecting rods, but couldn't get enough lift height.

    The bottom line is designing a successful machine can take significant effort, but it is fun when it works.

  • Nice job on the mechanism! So it will raise one up and lower the other?

    i wonder about using a rack gear with a cam follower. Then you drive one way to raise one and drive the other way to raise the other. When you raise one, the opposite one stays put since the cam is in the lower position.

    could you do a quick post about the iterations you went through?

  • in reply to DAB

    Good point. The PCBs come in batches of 10 so I will have some left over for future projects. The footprint is generic, lower current FETs will fit on the card as well.