Problem BAP beta or Vista or?

kurte · February 4, 2008, 4:30pm

I am trying to get my Brat back up and working after I updated which servos were hooked up to which pin to help another member debug his brat. After I completed this my brats walking sequences were not working well at all and I think I may have one or more servos connected to the wrong pins. So I thought I would use my Brat Sequencing code that communicates with the PC running a VB app to control each servo to verify that right joint was moving.

Well after having the Brat enter the appropriate mode (hitting Record button on the Remote), I tried having the VB app send out a TRY STEP command, which I have programmed a value of -1 for the number of steps to download. The brat displayed “Sequence Loaded” on my LCD which normally implies that I was saving a new sequence instead of trying to run one step of a sequence. I changed my code on the brat to display on the LCD the number of steps that it was instructed to download and when I hit the try button my display software shows the value of 2 for the number of steps.

I am not sure yet if this is a VB/Vista problem or a problem with the BAP beta software properly handling the SDEC…

The Basic Code on the PC looks like:

[code] Private Sub Try1_Click(ByVal sender As System.Object, ByVal e As System.EventArgs) Handles Try1.Click
’ Send strings to a serial port.
Dim si As SEQ_ITEM
Using com1 As IO.Ports.SerialPort = _
My.Computer.Ports.OpenSerialPort(ComLB.Items(ComLB.SelectedIndex), BaudCB.Items(BaudCB.SelectedIndex))
com1.WriteLine(-1)
si = SEQS.Items.Item(SEQS.SelectedIndex)
com1.WriteLine(si.ToString)
End Using

End Sub

[/code]

The code to receive it on the brat:

[code]
…
GHCCnt var sword
GHCVals var sbyte[6]
GHCTime var sword
GetCommandsFromHost
gosub LCD_ClearAndReturn
LCD_String = “Host Command”, 0
gosub LCD_Print[1,0]
pause 250
command=0xFF
GCHMain
if command<>0xFF then
if command = 29 then ; until we get another record command
gosub LCD_ClearAndReturn
sound 9,[50\3960]
return
elseif command = 25 ; Got a pause button. try to display what is recorded…
gosub PlayRecordedSeq[0]
elseif command = 26 ; Got a play button. try to play what is recorded…
gosub PlayRecordedSeq[1]
endif
endif

Command=0xFF 

;
; now lets try to get some serial input from our Host PC
;
serin s_in, i2400, 250, GCHTimeout, [sdec GHCCnt]


; Ok we got a count.  If it is -1 then it is a do immediate!
gosub LCDConvertNumToString[GHCCnt]
gosub LCD_Print[1,0]
if GHCCnt = -1 then
	; now lets try to get some serial input from our Host PC
	serin s_in, i2400, 5000, GCHTimeout, [sdec GHCVals(0), sdec GHCVals(1),sdec GHCVals(2),sdec GHCVals(3),sdec GHCVals(4),sdec GHCVals(5),sdec GHCTime]

	; We got a command!
	gosub movement[tofloat GHCVals(0), tofloat GHCVals(1), tofloat GHCVals(2), tofloat GHCVals(3), tofloat GHCVals(4), tofloat GHCVals(5), tofloat GHCTime]				
elseif GHCCnt = -42
	; The pc has asked us to download the loaded sequence back to them.

…[/code]

It appears like either the PC is not generating the value correctly or the SDEC modifier did not work.

I will try to debug more later, but I got to run…

Suggestions?

kurte · February 4, 2008, 6:03pm

I have done some investigations and I think it might be a BAP problem. I hooked up my primitive USB Oscilloscope to the S_IN pin on the BAP and updated the code on the pc to loop so many times outputing the -1 until I could capture what the BAP was seeing. Here is one such capture:
http://img223.imageshack.us/img223/5999/minus1onsinro1.png
Now If I read my ASCII properly, with the first high being the start, and LSB to MSB with a stop bit and Start bit between three bytes, I believe this decoded to:
2d 31 0a or (-1(lf))

Maybe I should transfter this information to the BETA bug list?

acidtech · February 4, 2008, 6:44pm

Again too much code and not enough code. You give a section of your program that you think is where the problem is happening. That just isn’t going to help us figure this out.

You either need to narrow it down to a simple program that shows the problem or you need to post your entire program source code so I can simplify it down to something that I can work with. My suggestion would be to test if negative values are being received properly in a simple program.

something along the lines of:

[code]value var sword

main
serin s_in,i9600,[sdec value]
if value < 0 then
serout s_out,i9600,“Negative”,13]
else
serout s_out,i9600,“Positive”,13]
endif
goto main[/code]

kurte · February 4, 2008, 9:34pm

Sorry, I was first localizing down to know if the problem was on the PC side or on the BAP side. So Step 1 was to see what the BAP was receiving on S_IN. From my early debugging it appeared like the BAP source code line: serin s_in, i2400, 250, GCHTimeout, [sdec GHCCnt]
Received the appropriate ascii input characters of "-1 ", lf
But GHCCnt did not have the numeric value of -1.

So I think your sample program would be sufficient to test this out. Although I would have a tendancy to also dump out the actual value maybe in hex to see what might have happened…

If I have a chance I will run this as suggested or you should probably just run your test suite for the serin and/or SDEC commands.
EDIT: I have run a modified version of the simplfied loop as you had and most of the time it appears to get the appropriate value. I may play around a bit more and find out if maybe having HSERVO and the simple timer interrupt running may cause it to possibly get the wrong values…

kurte · February 4, 2008, 10:28pm

FYI - Still debugging. On it’s own, receiving the string -1 from the debug terminal appears to be handled properly. So probably not a generic SDEC problem. Maybe it is a streaming problem between the PC program and the BRAT program on the BAP. Will try to figure out more later. It will probably end up being something like there was an extra character in the buffer that the BAP was not expecting.

But this is not critical at this point as this was simply a diversion to try to see if my Brats servos are on the right pins and why the sequences don’t look right…

acidtech · February 5, 2008, 1:50am

I would be leaning towards interrupts(HSERVOs and your own) causing timing problems with serin. HSERVO doesn’t use much but it does use some processing time and basic interrupts can eat quite a bit as well. It doesn’t take much to mess up the bitbanged serial commands.

kurte · February 5, 2008, 2:44pm

Thanks, that is one strong possibility. The other may be that I am just getting unlucky timing wise in my code and when the -1 is being sent, the atom pro may not be waiting the full time in the serin. When I get back to it, I may try to add more handshaking, maybe in the loop the bap outputs a prompt to the PC and the PC outputs something right after that. Or some such thing.

But currently trying to see why the HSERVO being generated by the TV brat does not appear to making the servos move the way I expect them to. There may be two servos reversed (an ankle with a hip maybe), but for example the step function is moving real slow through the functions and the angles and the like do not look at all right. Like it is trying to put a foot through the BAP board that is hung off the back of the brat and the max angle of any of the servos should only be 20 degrees, but it is trying to move a lot more, until it hits the limits of the brackets…

The requested moves (calls to movement) look like:

[code]Movement: -7.0000000000 20.0000000000 20.0000000000 7.0000000000 -20.0000000000 -20.0000000000 500.0000000000
-1166(46) 3331(133) 3331(133) -1166(46) 3331(133) 3331(133)

Movement: 7.0000000000 20.0000000000 20.0000000000 -7.0000000000 -20.0000000000 -20.0000000000 500.0000000000
1166(93) 3331(0) 3331(0) 1166(93) 3331(0) 3331(0)

Movement: 7.0000000000 -20.0000000000 -20.0000000000 -7.0000000000 20.0000000000 20.0000000000 500.0000000000
1166(0) -3331(266) -3331(266) 1166(0) -3331(266) -3331(266)

Movement: -7.0000000000 -20.0000000000 -20.0000000000 7.0000000000 20.0000000000 20.0000000000 500.0000000000
-1166(93) -3331(0) -3331(0) -1166(93) -3331(0) -3331(0)

[/code]
Note these tables are Right Ankle, Knee, Hip and then Left Ankle, Knee, hip and speed

The first line is the degrees. The second lines are the actual values passed to hservo with the number in parentheses the speeds. Each movement takes several seconds and appear to go through several gyrations. I will try to convert this output to a simple test program.

The real question is, is this issolated to my BRAT?

robotdude · February 5, 2008, 3:09pm

James has two BRATs set up for the TV remote control. He can test your code as a reality check.

You didn’t happen to use the same configuration as the original article did you?

righthip con p15
rightknee con p14
rightankle con p13
lefthip con p12
leftknee con p11
leftankle con p10

kurte · February 5, 2008, 3:24pm

Thanks Jim, I actually moved off of the default servos as I was trying to free of the hardware serial port. Actually it would be good for James to try your current brat code with the current beta and see if it is working properly or not. IE if you hit button 2 on the remote does it make a reasonable step.

Here is an extracted code that simply calls movement with the calculated, values for speed and servo location. He should easily be able to change the servo pin number constants and the number of servo groups to make it work on the default pins…

Note, I think I am still using all of the default calculations. I was wondering about the calculations from Angle to HServo value that used a constant of 166.6 steps per degree as the HSERVO value in the manual goes from -12000 to 12000 so if I wanted 90 degrees it would give a value of 14994… But that could simply be a documentation issue…

Now back to the code:

[code]debug con 1
; Try TV Remote Brat taking one step forward with HSERVO. Nothing else in the program

;System variables
righthip con p11
rightknee con p10
rightankle con p5 ; was p9 but that is speaker
lefthip con p8
leftknee con p7
leftankle con p6

;Interrupt init
USE_HSERVO con 1
HSERVO_GROUPS con 3
ENABLEHSERVO

; We simply want to emulate the gosub movement command for one step forward
; we will precalculate all of the speeds and the like
; if(command = 1 or command = 116) then ;(Button 2 or joystick up) Walk Forwards
; RA RK RH LA LK LH SPEED
; gosub movement -7.0, 20.0, 20.0, 7.0,-20.0,-20.0, 500.0]
; gosub movement 7.0, 20.0, 20.0, -7.0,-20.0,-20.0, 500.0]
; gosub movement 7.0,-20.0,-20.0, -7.0, 20.0, 20.0, 500.0]
; gosub movement -7.0,-20.0,-20.0, 7.0, 20.0, 20.0, 500.0]

;Movement: 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000 0.0000000000
gosub movement [0,0, 0,0, 0,0, 0,0, 0,0, 0,0]
pause 2000

;Movement: -7.0000000000 20.0000000000 20.0000000000 7.0000000000 -20.0000000000 -20.0000000000 500.0000000000
gosub movement -1166,46, 3331,133, 3331,133, -1166,46, 3331,133, 3331,133]

;Movement: 7.0000000000 20.0000000000 20.0000000000 -7.0000000000 -20.0000000000 -20.0000000000 500.0000000000
gosub movement [1166,93, 3331,0, 3331,0, 1166,93, 3331,0, 3331,0]

;Movement: 7.0000000000 -20.0000000000 -20.0000000000 -7.0000000000 20.0000000000 20.0000000000 500.0000000000
gosub movement [1166,0, -3331,266, -3331,266, 1166,0, -3331,266, -3331,266]

;Movement: -7.0000000000 -20.0000000000 -20.0000000000 7.0000000000 20.0000000000 20.0000000000 500.0000000000
gosub movement -1166,93, -3331,0, -3331,0, -1166,93, -3331,0, -3331,0]


pause 2000
gosub movement [0,0, 0,0, 0,0, 0,0, 0,0, 0,0]

loop
goto loop

;Should never need to edit anything below this line. Add user subroutines above this and below main.
lefthippos var sword
leftkneepos var sword
leftanklepos var sword
righthippos var sword
rightkneepos var sword
rightanklepos var sword
lhspeed var word
lkspeed var word
laspeed var word
rhspeed var word
rkspeed var word
raspeed var word
movement [rightanklepos,raspeed, rightkneepos,rkspeed, righthippos,rhspeed, leftanklepos,laspeed, leftkneepos, lkspeed,lefthippos,lhspeed]

#ifdef DEBUG
serout S_OUT, i2400, |
sdec (rightanklepos), “(”, dec raspeed, ") ",|
sdec (rightkneepos), “(”, dec rkspeed, ") " , |
sdec (righthippos), “(”, dec rhspeed, ") ", |
sdec (-leftanklepos), “(”, dec laspeed, ") ", |
sdec (-leftkneepos), “(”, dec lkspeed, “) “, |
sdec (-lefthippos),”(”, dec lhspeed, “)”, 13, 10]
#endif
hservo [lefthip\lefthippos\lhspeed, |
righthip\righthippos\rhspeed, |
leftknee\leftkneepos\lkspeed, |
rightknee\rightkneepos\rkspeed, |
leftankle\leftanklepos\laspeed, |
rightankle\rightanklepos\raspeed]

hservowait [lefthip,righthip,leftknee,rightknee,leftankle,rightankle]

return
[/code]

fish-5 · February 5, 2008, 10:09pm

I used the current beta, and the BRAT code, and all the commands work

I used your code, and the bot takes a step forward. I’m not having any problems with either bit of code, or the Atom-Pro IDE beta

kurte · February 5, 2008, 10:19pm

Great. Did you try it with the same Servo numbers or I am guessing that you updated it to your servo numbers.

So it is my assumption that it is either:
a) My hardware/software has become flaky. I will try downloading the beta software again.

b) Maybe which servos are used. I will rearrange to the original brats configuration and see if things get happy. I will probably try this first as it is the easiest…

Thanks again

fish-5 · February 5, 2008, 10:26pm

Yeah… I changed them around, forgot to mention that…

Oh yeah, I also commented out “HSERVO_GROUPS con 3” which prevented the program from working… (at least for me)

That probably should have been in the original reply…

kurte · February 5, 2008, 10:50pm

Which version of the IDE are you using?

If I use 8.0.1.6 I get the results I mentioned. If I use 8.0.1.6exp, my test program appears to work, but when I go to the brat, I get the beginning of the first step and then it is like it hangs as also servos go limp.

EDIT: If I run 8.0.0.0 which I still have loaded on my machine. They both work fine. So my guess is still problems with HSERVO command

fish-5 · February 5, 2008, 10:57pm

Well… I thought I was using 8.0.1.6exp, but it seems as though it didn’t install even though it said it did…
It says I’m still using 8.0.1.3…

I’ll give it another shot tomorrow, sorry about that, my mistake

acidtech · February 6, 2008, 6:23am

You have to be real careful when you install multiple version of the IDE. You could end up with a mish mash of different versions of different files which could be explaining a lot of your problems. Note that 8.0.1.6Exp is for people to test there already working code on to see if the expiremental version of HSERVO in it works. If you code doesn’t work with 8.0.1.6 it is not going to work with 8.0.1.6Exp.

I also tested the brat code in the IRPD thread and everything worked as it was posted(remote and servos functioned as it should have).

I HIGHLY recommend you completely remove all versions of the AtomPro IDE. Then install just one version at a time until the problems you are seen have been correctly diagnosed.

acidtech · February 6, 2008, 6:26am

P.S. I hate Install Shield but it’s pretty much the only game in town.

You could install each IDE. After each is installed move all the files installed to the AtomPro IDE directory in the ProgramFiles folder somewere else(or just rename them appropriately). No files are installed anywere else so you don’t have to worry about that. You just want to make sure that other installs don’t install over top of each other leaving random versions of files. For some reason Install Shield(even those I set it to always overwrite older versions) still insists on sometimes leaving older files. Thats why I always recommend uninstall the old version before installing the new one.

kurte · February 6, 2008, 3:05pm

In between each of these runs I did a complete uninstall of the Atom Pro IDE. I have 8.0.0.0 still up there as you used a completely different directory. (I also keep 8.0.0.0 around as HSERVO and serial output worked well on the earlier beta)

I do have the downloads for different versions in my download directory, so I can switch to different releases. I wish you had an option in your install/uninistall that would allow you to save your personal preferences. Things like history, comm settings and the like, but that is a different story.

I have sent you an email with a video of the brat trying to run the simple program earlier in this thread, which is the same as pressing the 2 on the remote (or take one small step forward). I have verified that the same code works as expected on 8.0.0.0. James(Fish) said it worked on 8.0.1.3, but I believe that this was a version that hservo trashed registers which one symptom was serouts hanging the machine.

It would be good if you could try running the simple test program on 8.0.1.6 so you can debug it… Note: last night I did reconfigure my brat again to be back to the default pins that Jim posted and commented out the HSERVO_GROUPS, unplugged the other servo and I2C components and it still acts very wierd!!! I guess I will probably move on to something else until this is resolved.

Thanks

fish-5 · February 6, 2008, 6:51pm

Ok, now that I actually have 8.0.1.6exp installed, I can say that I couldn’t get either program to work…

When I use your code, it looks like the robot has a temper tantrum and then goes limp

and when I use the BRAT code, it goes into seizure mode when you press a button

kurte · February 6, 2008, 7:12pm

Thanks, that is the same as I was seeing! Yes it sure was having a temper tantrum. I sent Jim an email last night with a poor quality video of it. I had a higher quality one but did not want to sent 10+MBs. When I tried it on 8.0.1.6 it did not go limp but stayed in the tantrum for awhile…

Hopefully AcidTech will have a chance to look at it soon as it is confirmed on a couple of machines.

Kurt

robotdude · February 6, 2008, 7:23pm

I’ve seen the video and I’ve seen James results. They are not acting the same. We are going back to 8.0.1.6 and testing more with it.