Journal98 may
From Simreal
May 1998
Mon, 11 May 98
A two-week delay again... it has been slow in the software corner, but the stuff is very nearly ready. This weekend was completely consumed by ... I don't even remember what ate Saturday, but Mother's day ate Sunday. No complaints, though -- I spent some well-deserved time with my various families, before they completely forgot what I look like. I've been trying to get some quality time in with my Father, too, before I move a zillion miles away. I may be an aging coot (well, 34) but I haven't ever been away from my family! Not that I ever see them, but knowing they are only 15 minutes away is different somehow than having them 2,500 miles away.
This week, I will wire the Boris Leg up to a PIC, and watch the leg position sensors in the Brain. That will be *very* exciting... in quick succession from that point I will be adding more I/O between my MCU and the Leg, and more Brain activity. I have some software PWM code to write (for use in both Boris and my other Robot work), and some hardware stuff to get the pressure sensors to give me a useable signal. With any luck, by the end of this fortnight, I should have rudimentary control and feedback on the leg through the PC-based Brain software. I'm so excited I could burst! It has been nearly a year now since I started planning this project, and parts are finally coming to life. It's almost harder to move into the end-game with such a long term project; the stake in its success is higher.
RobotWars itself is still in hot contention -- Profile Records (the money sponsor) and Marc Thorpe (the idea guy) are playing dualing-partners with the competition itself as the big loser this year. Most unfortunate. There is a bright side though... if things go well, RobotWars will be delayed which gives me more time to finish Boris! I can only hope... the move to Texas, and the simple fact that I don't have the $4,000 to buy the full pneumatics package put some cramp in my style right now.
On the bright side, my sponsor is still very interested (heck, Boris and their system have a lot of overlap) -- they just have cash flow issues right now, since it is a young company. I'll be talking a lot more about them when the time comes -- but for now it all stays a mystery. Anyway, I'll be getting a nice radio modem-like thing in the next month or so through them; I was beginning to get worried about radio control, since the Radiometrix module I worked with didn't make me fully happy (my needs don't quite fit their product), and the Linx HP module isn't ready for prime time yet (it is hard to get their development system; they claim far more demand than supply, which is likely true. It looks to be an excellent unit).
My Machine Shop guy (Charles Davis, of Davis Technologies, in southern California -- look him up if you need machine work! A great guy, and a long-time friend) finished moving his shop, called me all happy and tired, and says I should expect parts soon. It's amazing how much time it takes to weld up something like a simple cylinder bracket -- *I* was amazed, at least. His shop time is going to free up weeks of my shop time, and probably makes the Boris project just possible for an August release.
Are there any readers out there familier with air compressors? I'm not happy with my air compressor choice yet, and was wondering if there are any "special" compressors with a higher output-per-weight ratio than the 9cfm at 125psi for the 160lb shipping weight compressor I have selected. It must, of course, be gas-powered. One friend suggested I go with a 2-cycle engine (eg: Chainsaw) instead of the standard 4-cycle engine (eg: Lawnmower). Any thoughts?
I finally have re-organized my links page... and I fixed some dead links and other lint. Let me know if there is anything else goofy out there, and I can do spot fixes.
Tue, 19 May 98
Hey, I said I was going to wire up sensors, and this week I'm actually ahead of schedule!
I rigged my PIC to manage the serial communications and A/D conversion off of interrupts, which was fun, and it makes most of the stuff run automatically while I'm not watching. Then, I moved Boris and his one leg into my office -- egads! So much for my office; Boris fills it.
Wired some potentiometers to the joints (the new brackets and couplings are very nice -- no backlash). Ran them into the onboard A/D on the PIC. Hooked up the Brain software and all that and... voila! When I move the leg, I watch the values adjust on the computer's screen! The sensors weren't getting full-scale of course, so I added some automatic scale adjustment to the Brain's program... so the leg is essentially blind until it exercises through a range-of-motion test, but then I get full scale information. Very cool.
When I get my next set of pneumatic cylinders, I'm thinking about adding the limit-sensors they have... they cost way more than they are worth, but will help when I boot the system. Maybe. Maybe not.
So, flush with excitement from my rotary sensors, I decided to test a pressure sensor. These things are a differential sensor with roughly 30mV offset and 60mV range. Well, reading about these things I find that everyone recommends a 3-op amp instrumentation amplifier. Heck with that... even using quad op-amp packages, the six sensors would require 18 op-amps, or 5 chips! So I just plugged the sucker into a single op-amp, added four resistors to give me, heck I don't even know, "some" amplification, and tested it.
Cool!
About a 1/2 volt offset, running up to about 4 volts at 100psi. I may want to adjust this some, you know, to give room for over-pressure or something... I know the op-amp won't go rail to rail, it's just from Radio Shack; it can't be that good.
So I feed that signal into one of the A/Ds on the MCU... tada! Perfection.
This weekend, I should be able to pick up a nice digital camera (Sony Mavica -- not the best resolution, but a very nice camera and the only one that stores on diskettes), so expect many more pictures in the updates to come.
Next task -- PWM valve control. This should be a breeze. Oh, I also want to hook the pressure sensors onto an external A/D chip, maybe the serial Maxim A/D. I need a total of 9 A/D channels, so I must go external. Once the A/D, pressure amplifiers and Valve controls are working together, I am going to design a circuit board and get some made. I even have a sponsor for the circuit boards, and as soon as I get his check I'll put up a "Sponsors" page and talk about him.
This board will have a total of 12 active A/D channels, some digital I/O, a nice Maxim3100 UART, RS485, plus who knows how many other features... maybe I could make extras if it sounds fun enough. Don't write me saying you want one though! Not until I have specs... it may be just a bit too odd, and expensive, to be of any value outside of Boris.
Wed, 27 May 98
I always think that these long holiday weekends will get me ahead on Boris -- instead, I seem to fall further behind! Had a bunch of stuff to do, it turns out, for Memorial weekend.
One of the things I did that took so much time was, I worked on the Boris PC board schematic and layout. It's coming along nicely, and I'm adding a few things that I may not need just because it will round out the board nicely. The tentative specs are:
PIC16C74 MCU for the brains
Onboard power regulation (including bridge rectifier; AC/DC)
8 channels of 8-bit analog input through the PIC
8 channels of 12-bit analog input through the Max147 serial A/D
8 bits of power output (MOSFET control)
8 bits of digital output (never know when you want to add an LED)
2x8 bits of digital input (never know when you want to add a switch)
1 RS232 serial port, via the Max3100 UART (it's nicer than the one on the PIC)
1 TTL level SPI/serial port to communicate with a Debugging LCD
1 Microship serial memory device (the fourth is the serial A/D mentioned above).
Of course, I could have 8 each parallel and serial ports/devices... but that seemed to be overkill.
The output ports are latched, so they are fire-and-forget from the PIC. Of course, they are multiplexed (as are the serial ports... but each serial line is given a feedback interrupt into the PIC through PortB).
The power of the MOSFET outputs will depend on the chips dropped in, but I only need about 100mA. I'm actually using the IRLD014, which gives about 1 amp. The MOSFETs will be driving 12 volts (or whatever the board is fed).
The serial memories will be useful for field programming the PIC... I intend to taking my fuzzy control language, modifying it for the onboard version (different data types, limited functions, and extra features), and then having it read the serial memories for the interpreted program code. Something like that. It should be fun!
This board isn't designed to be cheap, and it won't have prototyping space. The actual card, in 10 quantities, will probably cost me $20 each without solder masks or any frills. The parts to populate them... well, I haven't checked yet.
As for my other work this last weekend...mostly the hardware fought back. I hooked up my Motorola pressure sensors, and finally got them to talk through the Maxim serial A/D chip. The serial A/D I have is 12 bit precision, so I just ignore the low 4 (which isn't great, but works well enough for now). It's very cool -- I can tug on the legs and watch the debug graph for the pressure sensors change on the PC Brains.
I went to hook up my MOSFETs in the wierd 4-pin DIP package, and discovered that International Rectifier didn't see fit to document the pinouts. I think I figured them out, but my wimpy 9-volt power supply doesn't quite push the valves into operation. Time to hook up to the 24-volt wall wart and add 12-volt power regulation. I also wrote the software to do 10mS PWM (damn slow, in case you are wondering). Took forever (well, an hour) to figure out the Timer1 on the PIC... don't know why, but finally got a nice 10mS pattern out of the PIC. I have some more stuff to write, to hook the PWM pattern up to the fuzzy variable sent in from the Brains, but I can work on that this week.
Friday (I'm jumping around in time a bit here), I went to pick up a digital camera (Sony Mavica FD7)... but the store was out of them! Got it finally yesterday -- so more pictures next week (or, really, just more detailed pictures of what is already there... not much different on the hardware front).
No word from my machinist in California yet .... I suspect he is very busy. Gotta get those parts soon, though, to make it before the move. Also, no check from my circuit board sponsor yet... I may have to order before I get the money, since I want the boards soon. The breadboarding is getting really ugly! I now have six sensors hooked up (two angle, four pressure) plus eight lines out to the four valve solenoids! It's only going to get worse.
Speaking of worse -- there has been a lot of news on the Robot Wars forum (see my link in the Links page), and my personal opinion is that... Robot Wars '98 is NOT going to happen in mid August. There may be a substitute event, some event may happen later, SOMETHING.... but the current event doesn't look possible right now.
There are a lot of competitors, who have put a lot of money into robots over the last few years -- there will be a combative robot event in 1998 or 1999. Absolutely no doubt in my mind.
So, we build, plan... and we wait.
