Timer controlled movement of RC servo strange

autoguider · Joined: 24 Sep 2007 Posts: 82 Location: Aachen

Dear All, I kindly ask for your advice.
I want to control 2 RC servos (KUMAN , 270°) named servo_2 and servo_3.
Length of pulses is between 500µs and 2500 µs.
As movement needs to be slow I do an update of the pulse length every 40ms. With 20ms it was the same.
Servo 3 runs smoothly the whole way. Only Servo 2 is stuttering in the final 30% of the movement in and from the working position.

Update: I get the first hickups at a value of 215 for timer2. This correponds to 2.24 ms pulse length.

Changing the servo hardware did not help. So it seems to be in the software. Either of the servo or in the Atmel.

Pulse generation is done via 2 timers.
Timer0 generates the 20ms frame.
The ISR of timer0 calculates the new values for the interrupt generation of timer 2 and stores them into the array B_Impulsdauer.

For each servo there is a reserved time frame determined by the value of pulseframe.
I set pulsframe to 253. This means 2640 µs for each pulse which is sufficient for the 2500µs and allows to extend the ON- time of the pulse in case of tolerances with the servos.
So the 20ms frame is kept for each of the servos.
Impulsdauer(3) contains the value for the ON –time for the pulse for servo2
Impulsdauer(4) contains the value for the OFF –time for the pulse for servo2
Impulsdauer(5); Impulsdauer(6) does the same for servo 3.

How the new values are changed depends on the value of B_Fahrzustand.
These values are:
1 ( forward), 0 (standstill), &HFF (backward). So timer0 can run permanently.
Timer0 triggers timer2. Timer2 generates the pulses for both servos. Servo 2 is served first. The ISR Servopuls_isr_t2 is executed 5 times and then shuts off timer2. Timer2 remains inactive until it is re-triggered by timer0.
B_Servonr is a counter to address the correct time period. Values range from 2 to 7. When value 7 is reached B_Servonr is reset to 2 (see label Reset_servo_count)

Parameters for the Timers:

hgrueneis · Joined: 04 Apr 2009 Posts: 902 Location: A-4786 Brunnenthal

If you use RC-servos, then you are driving these servos out of range!
Most Servos can accept pulses from 0.9 milliseconds to 2.1 milliseconds.
If you need more than 60 degrees of movement, then you have to modify the servo with additional resistors at the potentiometer.
In practical applications it can be extended to about 180 degrees of servo movement.
Slow movement can be done by loops or certain interrupt routines, which can be called at fixed times.
Reverse movement requires to switch sides on the potentiometer.
If you do not write your own routine, then you can use Bascom commands.
I control 18 servos with just AT168 with a resolution of 1 microsecond. Proven with a scope.

Regards
Hubert

laborratte · Joined: 27 Jul 2005 Posts: 299 Location: Berlin

Hi,
I'm not familiar with RC Servos, but here some notes to your code:

- I don't see any overflow control in your on-off-time calculation. How do you avoid values >255 or <0?

- When writing asm code in "no save" interrupt routines you should use the $notransform compiler directive. Depending to your MCU Bascom may change instructions like CBI, SBI, OUT with use of R23 (see help: mixing ASM/Bascom) - which you are not saving to stack. This can produce very nasty and difficult to find errors.

- I try to avoid the use of Y pointer - it is too important to Bascom. I the case shown above it is OK, but if you somehow later reuse the code in mainloop this can cause serious problems.

- Instead of using

Duval JP · Joined: 22 Jun 2004 Posts: 1161 Location: France

Mark does a lot for us :

see https://wiki.mcselec.com/bavr/CONFIG_SERVOS

JP Wink

hgrueneis · Joined: 04 Apr 2009 Posts: 902 Location: A-4786 Brunnenthal

JP,
yes, that is right, but Bascom , as good as it is,
can not handle 18 Rc-Servos at a resolution of one microsecond.
In order to have a smooth movement for a high number of servos, you need to write your own routines....in Bascom.
At least, if you want a reasonable repeat time of 40 times a second.
With my system and change of processor I could expand to over 30 servos per controller at the same resolution.
It is not rocket building knowledge.

Best regards
Hubert

hgrueneis · Joined: 04 Apr 2009 Posts: 902 Location: A-4786 Brunnenthal

I did not go through your program, but you need to check the limits of your servos first,
and I am sure that you are out of range. Like I wrote before, most servos can handle 0.9 to 2.1 milliseconds, but there are some that can only handle a lot less. If a servo starts to jitter above 2.1 milliseconds or 1.9, this means that it is at its limits. Driving these servos further can result in damage and loss of servo and much higher power consumption (the end is near).

Best regards
Hubert

Evert :-) · Joined: 18 Feb 2005 Posts: 2156

autoguider · Joined: 24 Sep 2007 Posts: 82 Location: Aachen

Dear All,
many thanks for your replies and advice.
@ Hubert: The servo is designed for 270 control angle. According to the information from the supplier the length of the control pulse can vary between 500µs and 2500 µs. So hacking of these servos is not required. That is the reason for me to choose them.
@ laborratte: I was not aware of the $notransform directive and I will put it into my code. Also the lds command is very useful. I will implement it as well
I did not see how to use the PWM capabilities of the timer. I need to serve 2 servos, maybe 3 in a later stage. For timer 0 I have OC0A and OCOB. Changing the trigger values to execute the ISR is essential for each movement. I you can help me with a code snippet or a link?
@JP: Tkanks for the link. I do not understand how I can run the feature that Mark has provided for a 270° control angle servo. How is the pulse generation synchronized to the 20ms frame for all servos?
How can I do the slow movement ?
Also the question of overflows arised.
To avoid overflow of the trigger values for timer2 is control them in the sub that coordinates the movement.

autoguider · Joined: 24 Sep 2007 Posts: 82 Location: Aachen

Update.
When I disconnect the 2nd, smooth running, servo from the circuit board I have no hiccups in the movement of servo 1.
I do not see bursts, spikes or voltage drops on the suppy voltage for the servos.

Now it looks that most probably the software is not the issue but some interference.
control pulse of servo 1 is over before the pulse for servo 2 starts.
It remains strange.

hgrueneis · Joined: 04 Apr 2009 Posts: 902 Location: A-4786 Brunnenthal

How many Kg/cm do the servos do?
Regards
Hubert

autoguider · Joined: 24 Sep 2007 Posts: 82 Location: Aachen

It is 17 kg cm. KUMAN KY72-1
I think that they are oversized.
When plugging in I measure 30mA at the entry of the 7805 on 12V level per servo.
In movement it takes less than 100mA.
The ripples in the movement do not occure when the movement starts but in between.

I will do a measurement of the 5V level that I use for the supply of the servos.

The control voltage is generated by a separate 7805 that supplies the Atmel.
The strange behaviour is that the output voltage at PORTC.1 / C.2 is 5V when servos are disconnected.
As soon as I plug in one servo this voltage drops down to 1V. Pull up rsisitors did not help.
I contacted KUMAN to check on the impedance.
It also may be possible that these servos need 6V or 7V for operating properly.
Then I will install a 7806.
To my feeling the impedance of the servos has to be in the kOhms area at least. I am not the 1st one to connect servos to an Atmel.

best regards
Christian

hgrueneis · Joined: 04 Apr 2009 Posts: 902 Location: A-4786 Brunnenthal

I use 10K servos, but on a 5 Amp. supply for the 18 servos.
Your problem might be the control voltage output from the ports.
or one servo input might be defective.
One way to find out, put an mA-meter in the control line or better a 1 or 10 ohm resistor and scope the voltage to find out the drain.
It should be less than 20mA and at least show 4,5V at the port.
If it is less than 20 mA, then your Atmel is shot with the port at 1V.
You might also try to switch the servos on the ports.

Best regards
Hubert

autoguider · Joined: 24 Sep 2007 Posts: 82 Location: Aachen

Dear Hubert,
this is what I expected. Max. 20 mA.
I have the voltage drop on both servos.
To avoid interference from the motors to the Atmel I installed the separate 7805 + free wheeling diode on the 5V side.
My strong feeling is too that the 1Vpulse level is much too low for the servos.

This morning I got a contact to the supplier and asked for information about the impedance.
Maybe the servos' control circuit is damaged.
I also found performance data based on 6V and 7V operation voltage.

Can servos run on this high voltage. To my experience 5V is the normal case.

If 6V or 7V is nominal voltage then it may happen that the servos' control circuits doe not operate properly.
Did you experience such voltage issues on servos?

best regards
Christian

MWS · Joined: 22 Aug 2009 Posts: 2262

hgrueneis · Joined: 04 Apr 2009 Posts: 902 Location: A-4786 Brunnenthal

The minimum voltage for your Servo is obviously 6V, max. 7V.
No mention of the control voltage drive (Voltage/Amp.)at WMS's link.
Regular servos need 4.8V to 5V. and the control pulse is within TTL specs.
I would ask sales for a valid datasheet per email.
Good luck.

Regards Hubert