Older blog entries for ROB.T. (starting at number 21)

I performed some experiments on the motors I have and came up with some interesting results. Instead of a stall torque of 50 lbs. at 12V 3.5 Amps, the stall is more like 100 lbs. at ~6 amps. This is good and bad. Good because my robot is going to have plenty o' power to move around (maybe even with a payload) bad because the L298 motor driver I was planning to use ain't enough. My solution is to make my own motor driver using FETs .

After blowing up a few "test units" (10 so far) learning to work with them properly, I learned two things - A.) You've gotta tie your gates down when you aren't using them. B.) Don't play around with blowing up FETs, get a FET driver with a high-side charge pump. I did some research and determined the HIP4081A fits my purposes well - I also found out (via the internet) that a few of my TCRG hommies have already gone down this path and are using the HIP4081As now. I just placed an order with Newark for a few drivers and I'm waiting for them now.

While playing the FET game, my supposedly good gel-cell 12V 17-AH batteries from Mendelson Electronics didn't perform very well - like at all. Maybe I didn't take proper care of them, maybe I got jacked, whatever. My next move was to stop playing around with gel cells and get a decent deep-cycle 12V marine battery. I found a swell battery that put out 105 amps and at 58 lbs would be a freakin' mongo lump in the middle of my `bot. I got 2 of them and I'm going to put them both in the frame - aesthetics be damned.

I put (mostly) together what amounts to the new proto-type leg and immediately determined I need bearings. I've been hunting for them all day and finally stumbled across a bearing source on the net - I order a couple pairs to play with. Along with the bearings I need to lathe a decent aluminum axle (here is my future aluminum source ) for each leg joint (I'm up to 5 per leg). There have been many little changes to this leg design along the way and I expect there will be many more.

I came up with this modular method of prototyping boards I was going to use across the board on my robot. I've decided that would be a bad idea, and will only be using prototype boards to figure out what the standard boards will be. When I get it right I'll make a more economical board to put in the robot.

This robot project is spinning off sub-projects out of necessity. One thing I want to do is automate Kenneth Maxon's oven-soldering method. Another thing I want to do is morph my Sherline lathe into a stand-alone lathe/mill (this is mandatory). I'm trying not to lose focus and get distracted by these sub-projects, but I might take a month and bang them out. Luckily the same technologies used in this robot project are easily morphed into these sub projects. (If I say this often enough I might start to believe it).

I hope you guys don't mind my embedding the links into this entry - I sometimes use these entries to figure out what the hell I was thinking when I did/bought whatever it was that is in question. Someday I'll actually sit down and do a web site -

you know, when I have time...

27 Jan 2003 (updated 28 Jan 2003 at 07:19 UTC) »

I've finally moved past the problems with my ICE200 emulator -

Problem #1: The emulator produced random noises on the lines and wouldn't output the data to the ports in a regular fashion.

Fix: I was using the 9VAC 200-ma power supply I used on my tiny ICE200 emulator, I switched to a Radio Shack 9VDC 800- ma power supply.

Problem#2: While port B, C, and D functioned properly, port A only sort-of functioned with some lines responding properly and some lines not turning on at all.

Fix: I read the manual (finally - this is my second edit of this post). It seems I neglected to hook AVCC to VCC which must be done when the A/D stuff isn't being used. Who knew.

The board review and testing continues. I hope to have a second proto-board set ordered early February.

On the hardware side I've produced a decent coupler for the type of motor I'm using, but I still have to test it. I've also received a gear-rack set I want to test with the idea of using gear racks to extend the legs.

Life lessons -

Your microcontroller tools will change without warning. (Let me back up) Atmel has two different types of ICE200 emulators - the ICE200 and the tiny ICE, both of which have an ICE200 instruction manual. Who knew? I spent a good couple of hours trying to figure out why the ICE200 was producing random signals on the ports (actually, I might have played a little quake somewhere in there). I switched back to my old emulator (the tiny ICE) on a whim, and all was well. Another problem is the new AVR studio doesn't see my ICE200.

Frankly testing new boards is not the time to find out your tools don't play well together. I'm this close >| |< to finding a friendlier microcontroller...

OK, I'm done venting. Here is an update on my boards -

I'm avoiding programming the controller board as long as possible (really I'm avoiding the learning curve) by operating off the emulator board and the expansion board. Since the whole objective of plugging this brand new stuff together is to test it, you can imagine the problems I'm running into. Problems like placing pass-through connectors incorrectly aligned (damn!), ordering the wrong parts, receiving the wrong parts, not receiving parts until next month, leaving stuff out, etc. My boards are so bad I might be forced to order and populate a new set of ``trial'' boards. This is a huge hit in time and money, but the trade off is I get my boards closer to solid operation before I lay down real money for the Predecessor-platform boards.

One thing I have to keep in mind - once the problems with these boards are smoothed out and I know they work, then the circuitry component of my robotics infrastructure is secured (for the time being) and I can move on to other things.

Other things like welding aluminum parts - which I haven't been doing. I'm waiting on a part for my Sherline Lathe - a 4-jaw self-centering chuck. I think this will make the whole mechanical motor interface thing go a lot smoother because I can center one of the holes so much easier.

On the brighter side - I figured out (for the most part) how I want my legs to extend from the body. It's hella- cool, I'm going to use gear racks (more on this later).

What else? Oh yeah, my last class is tomorrow. I somehow managed to pass all the courses for my CS degree, but I don't have enough credits to graduate from this particular University, so I get to take this last class. Yeah, my life is like that.

Whatever, as long as I get this robot to work.

11 Jan 2003 (updated 11 Jan 2003 at 14:56 UTC) »

I got my boards back from Olimex the other day, they look nice, real professional. I'll be hard pressed to go back to etching my own PCBs. The only real problem I had was it took almost 3 weeks to get them. Olimex had them done and shipped in two days, but it took a few weeks to get to me. Shipping was probably delayed because of Christmas.

As expected I'm finding errors in my schematics and PCB layouts, so I'm documenting them with the intent of fixing before I order the next set. I'm also having a slight problem getting some of the connectors, they won't be available until 1/22/03. I haven't tested the boards yet, but most are at least partially populated and they look good and fit together great.

I've also been working on a leg axis design that includes a mounted motor and a potentiometer for position sensing. While welding two separate pieces of aluminum, I've been having problems due to one piece being made of a different grade of aluminum than the other and melts faster. Another problem is getting all the pieces to lined up properly. I'm thinking up some tools and methods to get me past these difficulties and speed up development time.

Design wise, I have finally gotten the number of robot- size options to two, the difference being the us of 8 batteries or 4 batteries. I'll probably build for 8.

I promised my robot club that I would submit the UML diagrams I'm using during the design phase of my robot, so today I'm working on that.

10 Dec 2002 (updated 10 Dec 2002 at 00:45 UTC) »

Finished the first set of 11 boards! I'll be sending them into Olimex this Tuesday (12/10/02) - I have to check them against my materials list first. These are all the different types of modular boards I will need to get my robot moving (I'm sure I've said that).

I just got a tank of Argon for my MIG welder. I need to get a couple more parts and then I can teach myself how to weld Aluminum.

The platform is mostly designed, I should be testing circuits by January - I feel it's important to establish time lines so you can blow them ;-)

Well, gotta go study, I've got a final I need to pass.

23 Nov 2002 (updated 23 Nov 2002 at 10:50 UTC) »

Naturally my University extended my degree program to include 1 more semester, they tell me if I pass this class and the next I'll be done by summer. So I've had to deal with more class time and homework, but that doesn't mean I've stopped working on Predecessor...

Since I couldn't find the motors I need anymore, I upgraded to a car window motor, in fact I bought 50 of them. I also bought (8) 12V-17Ah batteries to power this beast, and I upgraded the drawer slides I'll be using to extend the legs.

I've determined that this robot has gotten to a size that requires some welds, so I'm collecting MIG welding equipment as well.

But wait there's more -

The following proto-boards have been completed and are waiting to be sent out for a test run - Controller Board AT90S8535 (Atmel), L298 Driver Board, Sensor Board - Strain Gauge Amplifier, Analog Board, Test Board - Switch, Test Board - LED, Emulator Board, Expansion Board, and a Motor interface Board. I still need to make the RF modem Board, Communication Board - SPI, and the Remote control Interface Board. Hopefully all these boards will be sent to Olimex by December. Fellow TCRG member Alan Kilian had these boards made there.

The cool thing about these boards is they are very modular so I can use them to make other robots. Part of the Core Technologies concept the Twin Cities Robotics Group has been developing.

The president of TCRG, Jeff Sampson, has working on a robot-qualifying concept called the ``Science Museum Challenge'' or SMC. The basic idea is that your robot maneuvers from the parking lot of the Science Museum of Minnesota, up a couple levels and through the lobby, up a couple more levels and to the room where the robot meeting is held. The current rough draft has the SMC divided into different degrees of achievement, the first being to guide your robot remotely, and higher SMC levels can be reached by the addition of various layers of AI to your robot (Jeff is still working on this part). Jeff has been working on his own robot for SMC.

I want to have the Predecessor platform ready to take on the SMC - Remote by summer. Well, that's my intent. As my wife continues to say - "Show me the Robot".

Steve I lost my login password.

School has started, it's going to be hard, but hey by December it will all be over.

I'm going to try and finish all the boards in that period of time and when I get out of class in August I'll be able to assemble the platform and test it.

It's going to be a long time before my next entry.

9 May 2002 (updated 11 May 2002 at 12:35 UTC) »

I laid out the board I call the controller. This board simply has an AT90S8535, support components, and 8 connectors one for each port on each side of the board. I'm going to give it a once over and send it off to a board house and see how it goes. Since I have to build 8 of these 3'' boards, It's going to cost me $327 just for the controller boards and I still have many boards left to lay out. Here is a list of the boards I need to assemble just to get the platform to walk -

8 controller boards

8 motor boards

8 leg-stack test boards

8 strain gauge amplifying boards

1 SPI communication board

1 communication controller (probably the same as the leg

controller boards)

1 RF modem board

1 communication test board

1 power board

1 Emulator board

I haven't worked out the single boards yet, and I still need to lay out the motor boards and the stack-test boards - they're next.

I've completed the prototype board that amplifies the output of a strain gauge for A to D processing by the microcontroller. Time to start working on the next version of the boards.

I'm looking at making the boards modular with the microcontroller and it's support hardware being the only thing on one board. Another board will have the motor drivers, and a third smaller board will host the amplifier for the strain gauge located in the ``toe'' of the leg. These boards will stack, I need 8 of each and a controller board.

12 older entries...

X
Share this page