Servo Surge Capacitors

[cyberjeff]

I'm wiring up a servo board for my Due.

I'll have a dozen servos of mostly standard hobby (MG995R and MG996R) with a few 9g micros (MG92R), mostly digital.

It seemed to me that I should have a surge cap right on the servos bus. I could do a couple of things, I could dig up some 16V 4700 uF (or so) out of my salvage stock.

Or I could go new school and order some super caps. 3 1F super capacitors in series would be .33F at 8.1 V. With a surge of 2A per servo that is a RC time constant of around 1 second. That seems like overkill but perhaps the way to go... From what I gather NiMH batteries can run ~ 1.2 times their AH capacity, so even with the 5200 mAH I have on order I would only have about 7 A available minus resistance losses...

So, super caps? How well do they match, will I need resistors across the caps to help balance the voltage or can I just run 3 2.7v caps in series and not worry?

[ProgressiveAutomations]

From my experience with motors, they spike for a very short amount of time.  You may only be getting a current spike for 0.1-0.2 seconds, so smaller caps may suffice.

What's the likely hood of every servo spiking at once?

[cyberjeff]

From my experience with motors, they spike for a very short amount of time.  You may only be getting a current spike for 0.1-0.2 seconds, so smaller caps may suffice.

What's the likely hood of every servo spiking at once?

Would seem most likely at turn on.

I think I'll isolate the microcontroller (DUE) with a diode and a cap and not worry about the servos too much. Otherwise I can attach one at a time on startup with that .1 second delay between them. ESR on small supercaps appears to be relatively high so might not be worth the effort.

[ProgressiveAutomations]

Sounds like you're making a Quadroped, or something else that stands up on power up.

Staggering the power-up would be the easiest way of doing it.

[cyberjeff]

Sounds like you're making a Quadroped, or something else that stands up on power up.

Staggering the power-up would be the easiest way of doing it.

Yes, that is exactly what I am making. A cat like quadruped.

My thinking is to to hold it by it's spine on power up so that it goes to either a "sit" or "down". Actually my thinking is a bit warped at the moment as I've been studying quaternians.

[ProgressiveAutomations]

As a final project in school, my partner and I made a hexapod.  We had 18 servos (3 DOF legs), and a LiPo battery (11.4V I believe)from an RC car.  It was happy when it started laying flat on its belly, then all the legs would push it to standing positions.

I think it depends more on your battery, if it can provide enough current.  If its a lithium battery, it should be OK, because they have high discharge rates.  With a NiMH you may have more troubles.

I think only if your battery is insufficient, would you need caps to help it.

And even if you do use caps, you'd need a delay after your robot is powered on.  It cant stand up immediately, you would need to give the caps time to charge, else you're just increasing the load.

[cyberjeff]

As a final project in school, my partner and I made a hexapod.  We had 18 servos (3 DOF legs), and a LiPo battery (11.4V I believe)from an RC car.  It was happy when it started laying flat on its belly, then all the legs would push it to standing positions.

I think it depends more on your battery, if it can provide enough current.  If its a lithium battery, it should be OK, because they have high discharge rates.  With a NiMH you may have more troubles.

I think only if your battery is insufficient, would you need caps to help it.

And even if you do use caps, you'd need a delay after your robot is powered on.  It cant stand up immediately, you would need to give the caps time to charge, else you're just increasing the load.

Thanks.

The Li Poly was out of my price range for the Watt Hours I wanted. I hope that changes!

The curious thing about DC permanent magnet motors (I believe that is what my TowerPro servos are) is that when they are drawing the most current they are not producing the most power, and efficiency is going to hell.

http://members.toast.net/joerger/pic2/motorcurve.gif

What this means to me is that the mechanical design of the robot should be such that it never needs to deliver more than half the stall torque. The robot will run faster and need far less amps. More amps is less power! If I stagger the start up sequence, like you suggest, then I won't need more current later.

BTW, most small supercaps have very low load current capacities, so the super cap idea I had won't fly.

When are you guys at Progressive Automations going to have a small linear actuator in the 30kg/cm to 100kg/cm with a short 5cm or so stroke that I can afford? I could put them to work...