Journal98 dec
From Simreal
December 1998
Thu, 3 Dec 98
I honestly expected Boris to be mechanically complete by now, but I simply haven't received the parts I need yet.
Speaking of parts, though, I finally got the surface mount logic I needed to test the Atmel bus management... about time, too. Digi-Key came through right nicely, and those samples never did show up. I expect the sent them to the wrong address, or some paperwork got snagged.
Those companies that cater to the online shopper seem to be far more organized and responsive than the so-called distributors and representatives that dot the landscape. In my search for online sources of parts, I've found Digi-Key is very nice for the basics... and Atmel chips can only be found through Marshal and Insight, both of which have online shopping! My new slogan -- "If it's not on the Web, it doesn't exist."
Anyway, I'll have the Boris Brains Board tested this weekend, and by the next journal entry I will have made a determination of whether I will use this design or go back to a simpler, serial bus design. I may have feedback from Randy Eubanks, too, since he received his board this week and may have a chance to mess with it too.
Thu, 10 Dec 98
Whoo-hoo! Did that testing on the board.... the Static RAM worked, the interface from the MCU to the I/O board is good, the A/D works... lots of fun! I picked the wrong chip for my data latch (level triggered vs. edge triggered), but that is easily rectified. Also, in some places I seriously failed the KISS principle.
All in all, a good testing run. I went back and did some heavy re-design to fix the flaws, and simplify the system. I'm still cutting the board to 4" x 4" but I could probably take an inch off one side if I worked at it... or I could add more functionality... or... or... <whack>
But I think I'll keep it simple.
The only real loss of functionality in my simplification is... I dropped the external UART, and I'll be using the built-in UART in the Atmel chip.
Tomorrow I put in the parts order, and if things go well, the board order too. It will take a couple of weeks for the parts to arrive (says my supplier), but that's okay. The electronics order will cover the entire set of brain boards -- a $1,200 order! Oof. But that is almost it for purchases; just a few boards left.
I haven't put up the new schematics and board layout yet... I'll do that for the next update. I think the Eagle format I used last time won't work for their free version, so I'll put up Postscript too (or maybe only).
It's good to be making progress again.
Thu, 17 Dec 98
Things are crankin' now. Got the new boards in (they look nice, already noticed one mistake: I put in the DIN in-circuit programming connector upside-down). Today I got the Atmel 8515's, which means I have ALL THE PARTS except Maxim's. You know, $1,000 worth of robot electronics just doesn't make a big enough pile on my desk.
If the Maxim stuff comes in tomorrow, I'll be set for progress this weekend. Even if it doesn't, I could (in theory) un-solder the RS232 chip off of the current prototype and use it for communication and just ignore the A/D until it shows up. I love Maxim parts; it's their high price and low availability that causes me trouble. But man, they do know how to build a nifty chip.
My son (9 years old) is flying in tomorrow and I have the day off work. Cool! I get to show off.
Saturday, it looks good for me moving my mechanical workshop down from Georgetown and into the Robot Group's workspace... so I can finish the Boris mechanicals more better. The Robot Group is just down the street from me, a most happy accident.
I'll probably call Shane and have him send me whatever he has completed early January, and find ways to build the rest by hand. The change in plans came about when I called the builder of my home. They informed me it won't be done until MARCH... seems Austin is having lots of labor problems, what with the construction boom. I just hope the INS stays away from the builders until my house is done...
Today I'm bone tired... but things are good. I'm still delaying putting up the plans, etc... things are so busy, and I'm so tired... but soon... soon. Vacation time is coming.
Tue, 29 Dec 98
If you missed last week's update, there is just one thing I have to say... get a life. Seriously, I was going to put something up on Thursday but the spirit of the season engulfed me and I simply forgot. I had a great Christmas, by the way.
My robot shop tools, parts, and supplies are almost all at the Robot Group warehouse now; tagged and piled in a corner, waiting for me to organize them into a living, breathing workshop again in the new year.
I have all of my parts installed in the latest prototype boards, except for the Maxim A/D, and they all seem to work (and the A/D worked on the last prototype, so I expect it to work on this one too). It was almost a letdown -- solder parts, test, repeat. Works, no problems. That's it. What's next?
"Next" was a few days of seriously violating the KISS principle ("violate"... there's a word for you. Reminds me of the evil clown from the live-action Spawn movie). I have all this room on the I/O board, just wasted space. So I crammed in some digital ports and a handful of really cute SMD LEDs for indicators on this and that. Yup. I violated that sucker. I'm tidying up the layout right now.
The MCU board has lots of spare space, too -- but it's just the brain board, there isn't much I can add to it without violating its charter. The size of these cards is mostly determined by the number of connectors they are required to have for operation... so I'll just have to live with a less than crammed-full MCU board.
I'm actually adding a third, undersized, card that expands the A/D ports out to the many devices and lines that use them. It provides living space for the pressure transducers plus sockets for the potentiometer hookups.
The entire leg computer will cost around $200 each (I think; I haven't checked recently) but this gives me (once more with the laundry list) 16 high-speed A/D in, 8 digital out, 8 digital in, 8 12-volt 1-amp power out, in-circuit programming, RS232 communication, an SPI port, a couple of input buttons for debugging, about 20 indicator lights for debugging, and the ability to expand to a zillion cards per computer if desired. Running the entire thing from memory-mapped I/O with a parallel bus, while causing me endless grief and uncertainty during the design, gives me endless flexibility now.
I finally got around to updating the schematic pages, too...
Now for some programming and construction.
