DPRG List  

[DPRG] Servos and Arduduinos

Subject: [DPRG] Servos and Arduduinos
From: Quinones, Jose jquinones at ti.com
Date: Wed Dec 12 16:05:14 CST 2012


Not certain how much equipment is available for a troubleshooting session of this sort, but first thing I would do is see what kind of voltage ripple the power supply is offering. I would look at this, as well as other signals, with a scope. However, my impression is the PS is fine.

After reading the description, it seems the problem is the massive amount of current the system was asked of. I don't think this fried the power supply, but something can get fried, or at least adversely affected, nonetheless.

The possibly problematic notion here is that GND is this magical entity at which there is zero potential and everything is gorgeously fine (AKA 0V). If designed properly, you can approach this state (note: approach!). But deviate from a super solid GND and what you get is a resistor (or even worst, a bunch of them!). Now, you have 22 motors asking 2A a pop at the same time, and you have 44A worth of current flowing through said resistor(s). What this means is you do not have 0V at GND as you thought. This is called GND bounce and to give you an example, if you had an estimated series resistance of 100 milli ohms, at 44A you would have had a 4.4V voltage drop across said resistive path. And we are not even considering parasitic inductance which is kind of benign at most low current levels, but certainly not at 44A. BTW, the PS may be limited at 22A but for a brief period of time it may have been able to deliver the 44A.

Now, I am not saying this is necessarily a problem as I have no idea where on the circuitry such a voltage would have been present. I also do not know for how long was the voltage present. But sometimes these voltages do not need to be present for a long period of time to cause all sorts of issues. If the logic gets whacked with this GND bounce (and if there is GND bounce, the logic will be the first to feel its wrath), you may start getting stuff enabled which you didn't want to enable in the first place. Illegal situations may arise, which are the ones that can in fact cause the cooking!

So why is all of this important? Well, because the experimenter is thinking the problem is with the power supply and has decided to go for batteries. If the problem is with GND bouncing, batteries will do no good. In fact, it will be worst! Because chances are that the 22A power supply would have tanked after the 22A were asked for, in which case the GND bounce is semi-contained. A battery is going to deliver pretty much whatever you ask of it until the internal resistance limits it, or until it discharges. And to circuits, either will look like a lifetime.

I would check into this GND. If there is room for improvement, I would make sure it sits as close to zero ohms as possible. This is only obtainable by making it with a humongous piece of copper so the resistance is as small as possible. And even then, you will see some effects when you ask for 44A of current all at once. The real solution here would be to schedule motion so that it almost seems like it is all happening at the same time, although the motors are getting enabled at different periods of time. This way you can disperse the inrush current across time and not just ask for all of it at once.

PS: At 22 motors and 44A, it may also start to make sense to look into some form of isolation.

Hope the info helps!

Best regards,

From: dprglist-bounces at dprg.org [mailto:dprglist-bounces at dprg.org] On Behalf Of Brian A. Stott
Sent: Tuesday, December 11, 2012 1:04 PM
To: dprglist at dprg.org
Subject: [DPRG] Servos and Arduduinos


Would someone with extra cycles of grey matter and experience help out a cool guy who is creating and learning to do super cool things?

You've all seen InMoov the 3D printed robot project by hairgael? He has a servo control jitter issue. I know you guys do have the answers.

His project InMoov, posted dilemma, and diagram is: here<http://inmoov.blogspot.com/2012/12/power-supply-robot-head-printed.html?utm_source=feedburner&utm_medium=email&utm_campaign=Feed%3A+Inmoov+%28InMoov%29>


Pittsburgh, PA

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://list.dprg.org/pipermail/dprglist/attachments/20121212/1fd83884/attachment.html 

More information about the DPRG mailing list