OpenMarine
pwm for clutch - Printable Version

+- OpenMarine (https://forum.openmarine.net)
+-- Forum: Pypilot (https://forum.openmarine.net/forumdisplay.php?fid=17)
+--- Forum: Feature Requests (https://forum.openmarine.net/forumdisplay.php?fid=21)
+--- Thread: pwm for clutch (/showthread.php?tid=2777)



pwm for clutch - johnm - 2020-07-23

purchased a used octopus 1212lar12.
Will not install until spring but tested to make sure it works.

I propose pwm the clutch output after an initial full power pulse for about 100ms.
Holding the clutch at full power waists a lot of power.
It's hot enough that it could be a burn hazard and possibly damage the coil.

For this clutch these are my measurements:
coil  10 ohms
full: 12.57v 1.11a
drops out: 1.27v 0.1a Needless to say want to stay well above the drop out point.
kicks in at 6.83v 0.6a

Way back when  I  serviceed equipment with large solenoids they always had a resistor and capacitor paralleled in series with the power. That will work but pwm will save even more power.

From searching a couple additional ideas I found:
Periodically give a full pulse just in case it dropped out.

Probably should have small capacitor across output to keep down rf noise.

The inductance will rise considerably when engaged. The surge current will be much less if engaged so one can sense if the clutch is engaged by measuring the current. Interesting but probably overkill and would need setup an another analog input.


RE: pwm for clutch - Styles21 - 2020-07-23

(2020-07-23, 03:32 PM)johnm Wrote: purchased a used octopus 1212lar12.
Will not install until spring but tested to make sure it works.

I propose pwm the clutch output after an initial full power pulse for about 100ms.
Holding the clutch at full power waists a lot of power.
It's hot enough that it could be a burn hazard and possibly damage the coil.

For this clutch these are my measurements:
coil  10 ohms
full: 12.57v 1.11a
drops out: 1.27v 0.1a Needless to say want to stay well above the drop out point.
kicks in at 6.83v 0.6a

Way back when  I  serviceed equipment with large solenoids they always had a resistor and capacitor paralleled in series with the power. That will work but pwm will save even more power.



From searching a couple additional ideas I found:
Periodically give a full pulse just in case it dropped out.

Probably should have small capacitor across output to keep down rf noise.

The inductance will rise considerably when engaged. The surge current will be much less if engaged so one can sense if the clutch is engaged by measuring the current. Interesting but probably overkill and would need setup an another analog input.
I second this request! The Octopus clutch isn't designed to have constant power. It just needs momentary power to engage the clutch and then the power is not needed. This is actually a safety issue. When the clutch is engaged and the power is off a quick jerk of the wheel releases the clutch. With the power on all the time the clutch will not release making manual steering impossible.


RE: pwm for clutch - johnm - 2020-07-24

(2020-07-23, 07:54 PM)Styles21 Wrote: I second this request! The Octopus clutch isn't designed to have constant power. It just needs momentary power to engage the clutch and then the power is not needed. This is actually a safety issue. When the clutch is engaged and the power is off a quick jerk of the wheel releases the clutch. With the power on all the time the clutch will not release making manual steering impossible.

I think the RS drive works that way. For mine the "clutch" is a normally open valve that bypasses the pump and cylinder.


RE: pwm for clutch - seandepagnier - 2020-07-24

The clutch output is on a pwm pin as I had considered this possibility, so the code needs to be modified to use timer0 in place of timer2 so that timer2 can be used for this, not hard to implement.

The problem is, the schematic has a rather soft start for the mosfet (arduino is driving gate directly) The gate is charged through a 560 ohm resistor which gives a softer start to avoid as hard of voltage spikes since the clutches are typically inductive. There is a diode of course, but even still... So I don't really think it will work in the ultrasonic frequency so it may make an audible sound if you didn't filter the output.

If your clutch only needs momentary power to engage, then this will need a different setting and changes to the code, but completely possible. Is it spring loaded? How does it work? Could you just use a capacitor in series to get it to turn on, or would the capacitor be too big?

The solution of using a resistor and capacitor comes to mind and is a good way to limit power, but you are trying to avoid resistor losses too, could you attach a cheap dc-dc converter to the clutch output and dial in whatever voltage you need? This is basically required already to operate a 12v clutch if the motor runs at 24 volts.

Can you find out what voltage is needed to engage the clutch and what voltage is needed to hold it?

Maybe it's possible to use the dc-dc converter to the hold voltage with diode on output, then put this in parallel with a capacitor if a higher voltage is needed to engage.

The next controller version, I'll consider a mosfet gate driver so pwm can work at high frequencies.


RE: pwm for clutch - Styles21 - 2020-07-24

I’m still trying to Learn most of this stuff. What I do know is that while testing my autopilot with the clutch output It will engage perfectly as you have designed it. But my clutch only needs a momentary activation of power and then it will stay engaged by itself possibly by a spring. It is designed so that if In an emergency I need to take control a quick jerk of the wheel it will disengage. With power constantly to the clutch it will not allow me to take control unless I shut the autopilot off first. If the clutch control powered on for 2 seconds and then off it would be perfect!


RE: pwm for clutch - seandepagnier - 2020-07-24

I'll add an option in the future to set the clutch timeout, but it will require reprogramming the motor controller.


RE: pwm for clutch - johnm - 2020-07-24

(2020-07-24, 01:24 AM)seandepagnier Wrote: The clutch output is on a pwm pin as I had considered this possibility, so the code needs to be modified to use timer0 in place of timer2 so that timer2 can be used for this, not hard to implement.

The problem is, the schematic has a rather soft start for the mosfet (arduino is driving gate directly)   The gate is charged through a 560 ohm resistor which gives a softer start to avoid as hard of voltage spikes since the clutches are typically inductive.   There is a diode of course, but even still...   So I don't really think it will work in the ultrasonic frequency so it may make an audible sound if you didn't filter the output.

If your clutch only needs momentary power to engage, then this will need a different setting and changes to the code, but completely possible.   Is it spring loaded?   How does it work?   Could you just use a capacitor in series to get it to turn on, or would the capacitor be too big?

The solution of using a resistor and capacitor comes to mind and is a good way to limit power, but you are trying to avoid resistor losses too, could you attach a cheap dc-dc converter to the clutch output and dial in whatever voltage you need?   This is basically required already to operate a 12v clutch if the motor runs at 24 volts.

Can you find out what voltage is needed to engage the clutch and what voltage is needed to hold it?

Maybe it's possible to use the dc-dc converter to the hold voltage with diode on output, then put this in parallel with a capacitor if a higher voltage is needed to engage.

The next controller version, I'll consider a mosfet gate driver so pwm can work at high frequencies.

I'd think the situation would be similar to pwm the motor?
Particularly when engaged suspect that the inductance will be significant and will draw less than expected from the resistance with short pulses.  Will probably have to experiment. Will be annoying if one can hear it.
Maybe a choke and capacitor to make it into dc? Choke may be pricey particularity for those that do not need it.

From my calculations to have .2a holding and about 100ms to pull in should have 2000uf and 50 ohms over 2 watts.
From the junk box found 3000uf and 12 ohm. Seemed to work fine and even that helps with about 500ma holding current.
For any that may want to try use wire wound and not carbon composition resistor. When a carbon resister gets hot the value goes down ultimately burning out. A hot wire wound resistors value goes up.

For Syles21 need some resistance in parallel to so the capacitor can discharge when deactivated. Make big enough that it does not hold.