View previous topic :: View next topic |
Author |
Message |
zamzam23
Joined: 25 Aug 2010 Posts: 47
|
I want to learn about door jamming algorithm and formulas |
Posted: Wed Nov 30, 2011 7:32 am |
|
|
my project is automate lift door controller with 18f.
I have 24V,100W dc motor. I have made this device with 4 mosfet h-bridge.
I can detect current with 0R1 ohm resistor
but my problem is:
when the door is jamming with some pressure, I want to detect instantly jamming with current sensing and stop running and run the door back to the start.
but there is different current at different position's the door.
how can I detect this stuation clearly. I can use velocity and current together but how? is anybody teach me this? |
|
|
Ttelmah
Joined: 11 Mar 2010 Posts: 19541
|
|
Posted: Wed Nov 30, 2011 9:09 am |
|
|
'Instantly'. No. Not physically possible.
At reasonable time, with a current sensor. Maybe. This is the system used on things like old car windows. However, the window _has_ to push with significant force, before an obstruction is detected.
If you look at things like modern train doors & lift doors, they may have pressure limit stop and reverse systems, but only as the 'backup' to the primary obstruction detection. Normally they use a variety of things like:
1) Optical reflection detection on the edge of the door.
2) Proximity detection on the edge of the door.
3) Optical 'curtain' detection just in front of the door.
4) Optical obstruction detection between the doors.
5) Gas pressure detection on the edge of the door (tube).
Now normally a 'lift door', would be moved by a PWM controlled servo algorithm, with optical encoder feedback of the door position. The system uses a PID algorithm to accelerate, move, and decelerate the door, and there is a learning cycle, used when the door is installed, or setup at the factory according to the weight of the door. Reversal is triggered if the power required moves more than a small fraction away from the expected parameters, once they have been 'learnt'. You then bring the door together at a slow speed once more. The speed variation needed, and the change in power needed over time, means that something rather more complex than a simple on/off control is required.
Then the second problem for this sort of detection is that if the door is to handle things like dirt, there has to be some margin in the allowed figures, and this means time/pressure before responce. Hence the preference for using things like proximity detection...
Best Wishes |
|
|
zamzam23
Joined: 25 Aug 2010 Posts: 47
|
|
Posted: Wed Nov 30, 2011 9:28 am |
|
|
I think I can do something using velocity of door, and current.
I have made something but there is no instantly. I think I need good algorithm for this.
so for example, the way is 90 cm.
when the door is at 10.cm, then the v=10 and current is 1A.
when the door is at 30.cm, then the v=50 and current is 4A.
(...)
when the door is at 70.cm, then the v=10 and current is 1.5A.
when the door is at 80.cm, then the v=5 and current is 1A.
I have pid and smooth acceleration.like this(look the shape only):
-----------------
I have search this from other forums and found this formula:
pressure=kfi*I-damping*velocity-lassitude*acceleration
what is this formula mean? |
|
|
Lemosek
Joined: 15 Oct 2011 Posts: 36
|
|
Posted: Wed Nov 30, 2011 3:53 pm |
|
|
Hello,
Dear zamzam23 I also project lift door drive. But I don't have hardware jet. My proposition is do it this like others. Implementing function teaching and store distance and average current for specific distance.
Can You show Your H bridge solution because I have many problem with this ??
Best regards |
|
|
zamzam23
Joined: 25 Aug 2010 Posts: 47
|
|
Posted: Sun Dec 04, 2011 1:51 am |
|
|
Any ideas for my problem? |
|
|
Lemosek
Joined: 15 Oct 2011 Posts: 36
|
|
Posted: Sun Dec 04, 2011 4:38 am |
|
|
Hello,
You writes is max current is 4A. So can you set overload to 4.2A and check what is a force at all positions of doors. Max for lifts is 150N. Also you can parallel detect move of doors: distance[n]< or > distance[n+1], sampling this 1/100ms.
P.S. 4A is strange large current. What type of doors you use EI90??
In my experience max current for normal doors 900 is about 2A. I use 24V dc motor with gear.
Best regards |
|
|
zamzam23
Joined: 25 Aug 2010 Posts: 47
|
|
Posted: Sun Dec 04, 2011 3:40 pm |
|
|
My motor is 100W 4A without gear. brushed motor.
I have already make an algorithm that is check the current as over 4A and detect the force on motor but this is not good algoritm. when motor is high velocity, the current is 3A and I am checking 4.2A and the motor is slow, current is 1A and I am also checking 4.2A.
I want to calculate real pressure so it must be max 150N as you told. and I am seaching a formulas that I can do it. or nearly solution for it.I want to make motor more sensivity for force. |
|
|
Lemosek
Joined: 15 Oct 2011 Posts: 36
|
|
Posted: Mon Dec 05, 2011 8:48 am |
|
|
Hello,
I have idea. You have something like this:
[img] http://msdl.pl/ramp.bmp [/img]
So If you set current limit like this:
if(distance<distance_acc)
current_limit=4A
if((distance>distance_acc )&& (distance<distance_dec))
current_limit=3A
if(distance>distance_dec)
current_limit=1A
Maybe this help You.
Best regards |
|
|
zamzam23
Joined: 25 Aug 2010 Posts: 47
|
|
Posted: Mon Dec 05, 2011 9:08 am |
|
|
Thanks for your effort.
Maybe the door is 700,800,900 or more. I need a function that works for all of these doors. Just I will change the parameter of door 'pressure'. Then it will work all of doors.
I think the best way is save current and velocity for all every 50mm. Then recalculate the average current and velocity every 50mm and compare with the set point in the memory.
How is your project going? Which level do you choked? |
|
|
Lemosek
Joined: 15 Oct 2011 Posts: 36
|
|
Posted: Mon Dec 05, 2011 9:55 am |
|
|
Hello,
This is simple
You must implement function "teach in". After install this door drive adjuster must press the learn button.
In function teach in:
1. learn door width - low speed and if distance[n+1]=distance[n] then you have end door way.
2. you must learn the cam path.
3. learn distance for mode: acc, normal, dec.
So if you have this algorithm then you don't need set new parameters.
Best regards |
|
|
zamzam23
Joined: 25 Aug 2010 Posts: 47
|
|
Posted: Tue Dec 06, 2011 1:33 am |
|
|
I know all of these.
My project is completed but I want to rewrite the door pressure part of my software.
Anyway maybe I must want help in another forums. Tthanks everybody who care my subject. |
|
|
|