PWM Library using a PCA9695?

[rob...@hotmail.com]

Hello all,

I sometimes see projects using servo's using the JAL servo llibraries. I have not yet used any servo's but I regularly have a project with LEDs that uses PWM (like the servo library). 

The number of PWM channels is limited on a PIC and it may be that the timers on the PIC are needed for something else.

A solution would be to use a PCA9685. There are modules available to control up to 16 servo's (or LEDs) using this module.

I was wondering if there are JAL users that have plans to use this device. If so I might make a library for it in some near future.

Thanks.

Kind regards,

Rob

[hans]

I've been using Matt's servo lib for years. The maximum number of servos in one project was 12.( see https://youtu.be/ujfMYD5zXNU)

 

  The only problem I had with the lib was that it resets the servos itself int a fixed basic position and that can conflict with the initial position in the program. So when you start up, you get a different base position for a while and that can cause damage.

A second thing for me is the speed of movement from one position to another. I now do this in a procedure within the program.

If this sometimes doesn't work out, I'll take another pic, even the small 12F683 works perfectly.

regards

Hans

Op zondag 16 januari 2022 om 10:35:02 UTC+1 schreef rob...@hotmail.com:

[Rob CJ]

Hi Hans,

I watched some of your video's. Impressive.  I did not know that the movement of the arms could be so fast with a servo. 

You shoul sell it to The Efteling 😊.

Kind regards,

Rob

---

Van: jal...@googlegroups.com <jal...@googlegroups.com> namens hans <hansvanve...@gmail.com>
Verzonden: zondag 16 januari 2022 15:05
Aan: jallib <jal...@googlegroups.com>
Onderwerp: [jallib] Re: PWM Library using a PCA9695?

 

--
You received this message because you are subscribed to the Google Groups "jallib" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jallib+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jallib/5cded4d9-1760-45a0-a8b0-d1f79718afaan%40googlegroups.com.

[vsurducan]

Impressive! Congrats Hans!

Perhaps using an encoder to "see" the servo position?  The less expensive it's a japanese amazing potmeter which can be rolled over ( 360 degree movement, 270 degree cursor). I've used such devices, no failure...

[hans]

Here is a program part which i use for reduce 4 servo's speed :

,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

-- basic position all servo's
servo_move(124,1)
servo_move(124,2)
servo_move(129,3)
servo_move(121,4)

 ---------------------------------------------------
var byte old_pos[4] = {124,124,129,121} -- start up beiden in centrum
var byte new_pos[4]
var byte tell =0
-----------------------------------------------------

---------------------------------------------------

procedure set_servo_pack (byte in x) is
var byte flag = 0

  var byte tell = 0
  while flag < 4 loop
    flag = 0
    tell = 0

   for 4 loop
       if old_pos[tell] < new_pos[tell] then
          servo_move((old_pos[tell] + 1),tell+1)
          old_pos[tell] = old_pos[tell] + 1
          delay_1mS(x)
       end if

       if old_pos[tell] > new_pos[tell] then
          servo_move((old_pos[tell] - 1),tell+1)
          old_pos[tell] = old_pos[tell] - 1
          delay_1mS(x)
       end if

      if old_pos[tell] == new_pos[tell] then
         flag = flag + 1
      end if
        tell = tell +1

   end loop
end loop
end procedure

Op zondag 16 januari 2022 om 19:40:07 UTC+1 schreef vasile:

[Oliver Seitz]

Hi hans,

heel mooie video's :-)

Your program fragment can be optimized a little, with no loss of function but saving code and data space and also execution time (which is not so important, as the purpose is to delay...)

tell=0

for 4 loop

  [...]

  servo_move((old_pos[tell] + 1),tell+1)

  [...]

  tell=tell+1

end loop

This can be done like

--tell=0

for 4 using tell loop

  [...]

  servo_move((old_pos[tell] + 1),tell+1)

  [...]

--  tell=tell+1

end loop

This way an invisible loop variable is omitted. In that special case, there's another way to improve:

tell=0

for 4 loop

  tell=tell+1

  [...]

  servo_move((old_pos[tell] + 1),tell)

  [...]

end loop

Now, "tell" is always one count ahead. In each servo_move call, you're saving the "+1", which takes code and possibly data space. So, why didn't I just state "tell=1" and leave the increment at the bottom of the loop? Very curious PIC assembler details... Setting a variable to zero is cheaper than setting to any other number. The increment is done four times anyway, so there's no extra cost to move it. What about combining both optimizations, like "for 4 using counter loop; tell=counter+1 ;[...]"? Terrible. Incrementing a byte "tell=tell+1" is a single assembler instruction. "tell=counter+1" is quite a complex calculation which takes code and data space.

</smartassing>

The purpose of a servo is to move to a certain angle as fast as possible. How fast, depends on the power of the servo and to the mass (or rotational inertia) connected to it. It is not the responsibility of a library to slow the movement down, although I think such functions wouldn't hurt as an extra.

Unfortunately, the communication to the servo is one-way, so, when switching on, the controller can not know in which position the servos are. Thus, IMHO, a library shoudn't move to any "initialisation position" on it's own. It's the responsibility of the application program to move to a starting position. There may be a certain procedure that works from all possible starting positions, like e.g.

1. bend elbows

2. wait for elbows are bent

3. move shoulders outward

4. ...

That example is clearly wrong for the linked video, as when shoulders are up and inward, bending elbows might have the hands collide. But there might be another order, which works from any starting position.

Greets,

Kiste

(Jij kunt me ook "Doos" noemen :-)

Am Montag, 17. Januar 2022, 12:43:56 MEZ hat hans <hansvanve...@gmail.com> Folgendes geschrieben:

https://groups.google.com/d/msgid/jallib/c72da766-937f-4e05-9971-7f50f2ac858fn%40googlegroups.com

.

[hans]

Thank you DOOS, once I have a working program that does what I want for me, I usually don't have the courage or the will to optimize it further. I will definitely apply your recommendation.
That Matt's LIB contains the commands of the basic servo positions is difficult as I mentioned.
You can disagree about the speed of the servos. When I use them in my dolls without speed reduction, the movements are much too wooden.

Op maandag 17 januari 2022 om 13:49:51 UTC+1 schreef Kiste:

[Oliver Seitz]

I'm a theatre and show technitian, you're absolutely right that movement can be to fast ;-)

What I meant is only, that it is not the servo's job to limit the speed. The program should limit the speed like yours does, or a library can have a slow-down adjustment.

I did not even look at the servo lib(s) yet, but I have an idea why Matt did initialize the servos. When you're starting to transmit data to the servos, you need to give them a position. A probably better, but more complicated way would be just NOT to transmit anything to a servo which has not been assigned a position yet. That way, the servo would usually just do not move, until it receives the first position data.

Greets,

Kiste

Am Montag, 17. Januar 2022, 19:25:56 MEZ hat hans <hansvanve...@gmail.com> Folgendes geschrieben:

https://groups.google.com/d/msgid/jallib/bae15549-0e90-45f7-a3e5-76442463f6edn%40googlegroups.com

.

[rob...@hotmail.com]

Hi all,

Although not requested I made a JAL library for the PCA9685 16 channel 12-bit PWM IIC LED controller. You can also use it to control your servo.

I had no experience with servos so I thought let's give it a try using this library.  The library including some sample files have been uploaded to GitHub and will be in the next bee-package.

In order to promote JAL, I made a video and posted it on Youtube: https://www.youtube.com/watch?v=3R_2mupvRjM

Kind regards,

Rob

Op maandag 17 januari 2022 om 22:02:56 UTC+1 schreef Kiste: