|
|
|
|
|
Valen
Experienced Roboteer
Joined: 07 Jul 2004
Posts: 4436
Location: Sydney
|
w00t
Its ALIVE
quote of the moment
"hmmm i really need to do some work on that failsafe code"
well i managed to get the radio working using timer interrupts rather than fixed delays and it has made a masive difference to the recieved signal. before the "signal error" light was on almost constantly now its back to the occasional flicker.
In operation, its twitchy as heck. I'm going to cut drive motor power to 50% and then limit the steering to 50% as well which might help.
I want to get one of these
http://www.sparkfun.com/commerce/product_info.php?products_id=741
and stick that in and use it for traction control.
with our current radio it would mean i would have to do the quackulations on the bot itself but if we can get bluetooth happening then the data rate and latency should be low enough that it can be farmed off to the controler for 32 bit calculation lurvin.
my current checksum system dosent seem like its quite up to snuff.
(or theres some logic bug)
basically the scheme is like this
code:
if (packetmode == 1) then
-- we are recieving a motor drive packet
if rxptr == 2 then seqid = rxdata end if
if rxptr == 3 then ch1 = rxdata end if
if rxptr == 4 then ch2 = rxdata end if
if rxptr == 5 then ch3 = rxdata end if
if rxptr == 6 then opts = rxdata end if
if rxptr == 7 then
-- packet finished
check = rxdata
-- check validitity
var byte doublecheck
doublecheck = 0
doublecheck = doublecheck + packetmode
doublecheck = doublecheck + seqid
doublecheck = doublecheck + ch1
doublecheck = doublecheck + ch2
doublecheck = doublecheck + ch3
doublecheck = doublecheck + opts
if (doublecheck == check) then
so basically it just sums all the bytes up (rolling over on 256) then if the controlers sum and the computers sum are the same we call it valid.
i think whats happening though is its getting mixed up if it "looses its place" in the stream and its "wrapping" then this is getting through because packetmode, seqid,ch1,ch2 all can be around 0 or so.
so i'm thinking perhaps i'll change packet mode to 127 or so and try for a 16 bit error check, if i can understand it, i'll make it a CRC, But it does seem confusing so far, lots of big scary maths words ewww.
hmmm perhaps just the inverse of the 8 bit sum would be enough.
twould be odd to get one byte then its opposite i'd think.
This thing gyro dances real good, and urithane can leave skid marks on concrete. _________________ Mechanical engineers build weapons, civil engineers build targets
|
Mon Sep 25, 2006 10:04 pm |
|
|
|
|
|
Valen
Experienced Roboteer
Joined: 07 Jul 2004
Posts: 4436
Location: Sydney
|
Event report
Status : crappy.
Physically the bot is fine. In our first fight Vs road kill we went the full 3 minutes and the battery pack was a bit warm, dad could keep his hand on it, i couldnt, we are calling that 65c ;->
During the fight though the bot was very jerky, driving in spurts but i didnt notice the failsafe or bad signal lights flashing. If anybody has vids i'd like to see them. Weapon was a little underwhelming, I think the diamiter is just too low to really get the KE we were looking for. Orbit seems about 2-3x more effective. We never really got a good hit in on roadkill though so who knows.
Second match Vs cobra, everything was going fine, wasnt too jerky, Then we got a big hit on cobra (about 20 seconds in) and it went to crap, i lost all drive )pretty much though the wep was fine. Turns out one side died and we only had forwards on the other. We had one more big hit and we were dead. When i got into the arena and killed the power the "everything's OK" light wasnt on, niether was the failsafe or the bad signal light, perhaps most worrying the "sometimes turns green" light was also not on. Anyway put it on the pit table and turned the power on and all was well. Killed my assumption that we had just popped a controller chip off the board :-<
We ruled out continiuity problems in the brain transplant cable and i got the multi meter onto the PWM pins and we could see the voltage change as i changed throttle so the PIC and everything seemed to be working so then we tried swapping the hip chips over. Turned the thing on and nothing much happened "ok" light was good, i tried driving it a bit then there was a big spark type bang that i didn't see but heard, dad said it was coming from around the area between the HIP's and the fets. Either way then we had no motor drive.
The fets seem un-explodificated but with the amount of carbon on the board i'm amased something hasnt happened before.
I cant think what could take out 1.5 H-bridges in one hit, the only thing i can think of is the metal blocks that the FET's heatsink to touched? not quite sure what that would do right now, my brain isnt working, but its the only plausiblle explination i can come up with.
Next Plan is to go sensored brushless. _________________ Mechanical engineers build weapons, civil engineers build targets
|
Sat Sep 30, 2006 10:41 pm |
|
|
|
Valen
Experienced Roboteer
Joined: 07 Jul 2004
Posts: 4436
Location: Sydney
|
Got a replacement V1.3 IBC off Jeff-f Via Glen on thursday. Way better in terms of design I must say, good ground planes etc. During the battles it hardly even got warm though i must admit i was handling it with kid gloves, partly because it was somebody elses IBC partly because the bot tended to do 180's if it got too sideways. (24V running drill cells limited to 50% drive, though in the jolt match i did add a scroll bar to vary the "nerf" level ;-> ie adjust the drive power limit)
Fight 1 was jolt, the most feared.
We went far better than I expected to lol. Had our "good" battery pack in (as opposed to the one with all the dodgy cells in it). We put the "jolt catcher" steel on the front and the undercutting blade thingie on the back which caused us to be sitting up off the ground before some minor adjustements were made. We had to go with the polycarb lid for this one because of the weight.
Basically because the bot just kept flinging sideways i just kept the throttle real low and aimed the nose at jolt then advanced slowly lol. Somehow we managed to "wedge" jolt a few times which let me open it up a bit as jolt acted as a pretty good stabiliser. Jolt caught the catcher a bunch of times making lots of pretty sparks ;-> bent it a little in a few places but all in all it just worked w00t. Jolt caught the glass a few times (only just) but that stood up well, looked like it got a bit of a hair cut was all. The poly carb "flexed" untill it shattered sending a chunk flying outta the way. Highlight was wearing jolt like a hat with the bar going madly on top ;->
Fight 2 was against cobra.
We put the drum on and were hoping to see the hard facing on the teeth actually stand up. It did w00t. Turns out the trick is to weld an inch then quench it. All was going according to plan (yes us getting picked up and thrown around the arena is part of the plan as was driving straight up the wedge and using it against your spikes ;->) with us gouging bits of glens spikes off and bending assorted peices. Untill glen must have nicked a belt which made it slip off to one side by 2 V's where it rubbed on the side of the bot, making mucho smoke. Glen however was also having intermittant drive at that point so we decided to call it a draw. (our bot was fine, could have continued running but i guess i'm just paranoid wrt smoke pouring outta the bot)
Fight 3 was roadkill.
Off to a bad start with the bot stopping all the time, I updated the controller firmware recently to include a 1 second delay on startup to help aleviate the "tick" it had when it turned on, this combined with i think a really bad cell in the pack lead to the bot basically halting whenever current was drawn or whenever a connection bounced. We pulled it apart and did the battery pack up which seemed to fix the problem somewhat but it still intermittantly froze during the match. Was fairly energetic match with lots of turning competitions, angus driving in a ring around us with me trying to turn fast enough to get the weapon on him without spinning out. All in all I think it was anguss' driving and reliability that won him the match. No complaints wrt that judgement here. _________________ Mechanical engineers build weapons, civil engineers build targets
|
Sat Nov 04, 2006 9:51 pm |
|
|
|
|
|
|