Voltage Sensitivity of Servo Motors
sjenny
Posts: 25
I am a high school teacher who recently incorporated Boe-Bots into my curriculum for the first time. Overall my classes had a wonderful experience with the Boe-Bots but we ran into a bit of frustration in the final project. I built a maze and each group was charged with programming the Boe-Bot to follow the maze using precision navigation with or without whisker wires. Each Boe-Bot was powered by 4 alkaline batteries and the Boe-Bot performance seemed to be extremely sensitive to battery age. The effect was noticeable over a fairly short period of time and I am wondering how others have coped with servo motors whose output is so dependent upon exact voltage levels.
Thanks for any advice or suggestions,
Susan
Thanks for any advice or suggestions,
Susan
Comments
photoman's voltage regulator suggestion has merit. Duane Degn (here in the forums) achieves good consistency by powering his CR servos through a 5V regulator, which likely also requires adding extra batteries.
I've found the navigation repeatability of the stock Boe Bot (Mine is named Bob) to be a little lacking. Different surfaces have a significant effect on how the 'Bot navigates, tho' I don't think that is a factor with your maze.
I've started running my 'Bots with Stamp BOE boards from a 9.6 volt rechargeable r/c car battery.
Not sure if this is financially feasible, but it does work.
Amanda
I switched from four NiMH of about 5V to four Li-Ion cells of 16V. The regulator dropped the 16V to a nice consistent 5V level. It didn't take long to figure out the loop timing once I added the regulator. I have bill of material in post #4 of the cheap bot thread which includes a link the regulator and batteries I used.
I assume this maze has walls high enough that whisker wires will touch them? I think the solution then is to run the bot till it touches a wall directly in front of it at every corner. This will let you re-align the bot's position to the maze at each corner. Done correctly, falling battery voltage shouldn't matter. (it will be rather slow though)
Lawson
First, you have an electronics board internal that decodes the PWM. And then, you have an H-bridge to drive the motor. And finally, you have the motor itself.
There have been a wide array of manufacturers that had all chosen different options in different combinations, so it is really hard to offer up one good answer.
With internal PWM decoders, some might become damaged by 6 volts or more; while the rest of the device remains intact. This requiring 6 volts or less seems to be a trend in newer servos.
With h-bridges, the older ones were transistors and had significant voltage drops (1.4 to 2.0 volts) that made users deisre to run higher voltages to compensate. Of coruse, newer devices might have MOSfets which have almost no internal voltage drop.
With small DC reversible motors, there are a wide variety of ways to build these. Some must be better than others, but who really knows without a lot of investigation.
~~~~~~~~~~~~
What I have found... is that one can always remove all the internal electronics from a fried servo and deploy and external h-bridge (those tiny IC h-bridge chips or make your own) in an attempt to revive.
Some motors will never allow continuous rotation, but the ones that do are then good to go for a long, long time. And just maybe less sensitive to voltage variation.
Actually, the kind of battery chemistry might be the easiest solution to the observed problems.
Alkaline cells have a slanted voltage discharge curve; whereas NiMH and Lithium ion cells have very flat discharge curves.
In other words, the NiMH 5 cell pack or using either a 9.6 volt or 7.2 volt Lithium ion pack would hold more closely to one discharge voltage without the need for fancy regulation.
I am overwhelmed with your support and very grateful for all of the suggestions!
Best regards,
Susan
You can download a series of charts for various battery chemistries.... Lead Acid, Alkaline, and the NiMH. And then compare.
These would be Voltage x Discharge time (the X and Y axis) or maybe a Voltage x Charge time, if you can't get discharge curves.
All have steep tails at both ends, but the sweet spot is the mostly straight line in the middle. You will see an obviously flatness in the NiMH line compared to older batteries.
The implications are
A. Output stays very near to one voltage for most of the discharge.
B. Using a voltage meter to measure how full the battery is won't work. You need more information to determine the per-cent charged. So, when in doubt recharge everything to start out will optimum power availability.
This may be advanced for some kids, but it is the kind of real-world electronics problem that we have to live with today. If someone is headed for a future in electronics or robotics, they will appreciate the added depth. ... A battery is not a set voltage, like a wall plug. And there are a lot of different batteries with different strengths and weaknesses.
Good luck.
Switching to 5-cell NiMH would certainly power your servos for a longer duration before dropping off but on a fresh charge that pack is gonna put out slightly over 7 volts. While this will not hurt the servos, they are gonna operate at higher speed. The bottom line is, if you want consistant performance from a servo, you need a regulator and a good voltage source.
Why 4 cells instead of 2? I know you're not using a linear regulator.
B. This is likely to be a tight school budget.
Yes, the ideal might be to use Parallax's Lithium ion boards with 2 Lithium ion cells that have built-in charger and low voltage protection. And to add in a voltage regulation scheme. But do you really think the funding will be easily gotten for all of this?
@Susan
Get a quote for a bulk buy with educational discount for the 5 AA cell holders from Parallax. You are entitled to a good price. This isn't a perfect solution, but I suspect your electronics parts budget might be better spent on other items that expand the range of what is taught into more sensors and such. Register with and use the Educational Forum to develop a good cost effective syllabus.
Just maybe you should try one 5 AA pack before you commit to a bunch.
Because my battery holder holds four cells and I figured the extra cells would give a longer run time. Since I was using a switching regulator, I didn't really see a down side to using four cells.
And yes, I do still have those dummy cells I could use as shunts when just using two cells.
Going excessively high with the voltage supplied to the regulator won't necessarily give you longer run times. The extra voltage has to go somewhere and when using a regulator the extra voltage is converted to heat. If a longer run time is your goal you would be much better off wiring 2 cells in series (2 sets) and then connect those sets in parallel. You'll end up with 4 cells that put out 8 volts instead of 16.
A linear voltage regulator will consume somewhere between 20-30% of the power; a switching regulator will likely consume somewhere between 15-20% of the power.
There are better ways to monitor and control motor speed and position than to regulate the voltage to the motor. But it might be best to use a conventional motor with a good h-brdige, and an encoder. The little hobby servo motors are like a bicycle with training wheels. There comes a day when one want to move on to something better.
I was using a switching regulator which doesn't dump the excess voltage as heat so I should have almost double the run time using four cells rather than using two cells.
I think a linear regulator can dump much more than 30% of the power as heat. In this example, I think going from 16V to 5V, I'd loose over 50% of the power as heat if I had used a linear regulator. I think I would have had a hard time dumping that much heat. I think your 20% to 30% range is correct for the voltage differences commonly used with linear regulators.
I agree there are better ways to control motor speed than regulating the voltage. I'm a fan of encoders myself. But it's hard to bet the simplicity of a CR servo for someone new to robotics.