Switching Regulators and Loads >> 1 Amp

[Joe]

Since this is my 1st post I will start off with a little bit of my background. I am an Electronics Engineering student at a 2 year college (http://www.gtcc.edu) and am beginning my 3rd robot. The first 2 did not use motors (besides the ones inside the servos) and I am hoping to use DC motors for the wheels of my next project.

I have been reading this site having recently discovered it. And I have noticed much more emphasis on switching regulators than on other robotics sites (eg. lynxmotion.com, seattlerobotics.org, robotcafe.com) which seem to assume that you're just wire the motor straight to the battery. The thing that confuses me, Mr. Admin, is that most of the motors I've been looking at online with a decent amount of torque are rated at about 3.8 Amps with the shaft locked. I can only assume that this is the amount of current that the supply would have to output for each motor everytime the bot hits a wall (or gets stuck somehow). The regulator you recommend is rated at 1 Amp. They say you can use them in parallel for more current, but the thought of using 8 of them in parallel (for two motors) is slightly worrisome to me. My worries may be completely unfounded, but I would like to know if anyone has done this and how that went, or if I'm completely not understanding something altogether.

Also, Mr. Admin, you commented in one thread (http://www.societyofrobots.com/robotforum/index.php?topic=83.0) that "...you will need a switching regulator and it still wont be pretty". Care to elaborate?

[Militoy]

I was holding back comment, as your question was directed towards Admin, and I’m not sure which 1A regulator you were referring to – but I’ll jump in with a couple of general comments on linear and switching regulators, and hope it may help. Starting with the small 1 to 1.5A 3-terminal linear regulators (usually in a TO-220 package, or one of its derivatives), there are 2 basic types: fixed and adjustable. The 78xx and 79xx series (positive and negative output, respectively) are set at a fixed voltage. The last 2 digits of the part number give the output, such as a 7812, having a 12VDC fixed positive output. The raw input should be set around 3 volts higher than the output, to avoid dropout at full load. Linear regulators act kind of like a controlled resistor – that is, they regulate voltage by dropping it across their internal resistance, and turn it into heat. You put 15 VDC into a 7812, and out comes 12 VDC. At 1A of output load, the 3 volts dropped by the regulator results in 3 Watts of wasted power (P = I x V). Let’s say you want to get 5 VDC at 1 amp off the same source; 15 volts in – 5 volts out = 10 volts dropped x 1A = 10 Watts lost as heat. You can stack these regulators up in parallel for more current, and they will “droop current share” fairly equally. This is an easy way to get the voltages you need, but you can see that the heat you have to get rid of can get out of hand in a hurry if you don’t watch what you’re doing. Another type of 3-terminal regulator works the same, but allows you to set the output voltage using 2 external resistors. The LM317 is an example of this type, and their flexibility makes them much more useful to designers than the fixed type. A simple linear regulator with higher current capability can be made using a transistor as a Pass Element (variable resistor function), and a zener diode or reference IC for control. There’s some advantage to this approach, beyond a lower parts count, as it is possible to design with a lower amount of voltage drop that must be turned into wasted power, or heat. Linear regulators have the advantage of low cost, simplicity, and low noise.
Switching regulators work by converting your raw voltage to a lower level (“buck” regulator) or a higher level (“boost” or “buck-boost” regulators), by switching your raw voltage at a controlled rate and period, then rectifying and filtering the switched output pulses back into a DC level. There are many circuit topologies in use for various applications, but one of the simplest, the buck regulator, is what I imagine you would need – should you decide to go with a switcher. The disadvantages of switchmode regulators are their higher complexity, and their tendency to create high-frequency switching spikes (noise), if not prudently designed. Even some commercial off-the-shelf switchers will play havoc with your control circuits, just from their common-mode noise spikes. The advantage of switchers are their higher efficiency, resulting in less heat to get rid of, and longer battery life. For the sake of simplicity, at your current level, I would evaluate how much voltage you must drop, then explore the possibility of using a linear regulator carefully before deciding to go with a switcher.

[JonHylands]

I'll kick in something here as well - you generally don't need to regulate the voltage going to your motors. Typically you regulate the voltage going to your microcontroller, and use an h-bridge to interface the microcontroller to the motors.

Most h-bridge chips/modules connect directly to the battery, and the switches inside the h-bridge allow you to control the polarity of the voltage going to the motors, usually with a couple digital input pins per motor coming from your microcontroller.

So, for example, my mini-sumo Seeker II (http://www.huv.com/miniSumo/seeker2) uses an LM2940 low-drop-out linear regulator to switch the battery voltage from 7.2 volts down to 5 volts for the microcontroller and sensors.

http://www.huv.com/miniSumo/seeker2/Schematic.gif, although at the time I made that schematic I used a 9-volt battery, the same principle applies.

I use a 754410 h-bridge to control the two motors, so the battery plugs directly into both the voltage regulator and the h-bridge chip. The motors run off the 7.2 volts of the battery, but switched and controlled through the microcontroller.

- Jon

[Admin]

summarizing what has been said:

use linear regulators for high precision low noise voltage electronics: microcontroller, sensors

use switching regulators to coarsely drop voltages: usually actuators that will overheat if you dont, such as servos that have voltage maximum limits (for example, a 12V battery for a 4.8V servo)

dont use anything (just attach battery to the driver directly) if the battery voltage is about the same voltage you want to run the actuator (for example, a 6V battery for a 6V servo)

the schematic at the bottom is typically how i do it:
http://www.societyofrobots.com/schematics_powerregulation.shtml
im thinking of updating that schematic with maybe a ~.2F supercapacitor (to replace the 10mF cap), now that they are super cheap n all . . .

"...you will need a switching regulator and it still wont be pretty". Care to elaborate?

ehhhh basically, dont do what he was thinking. 

[Militoy]

use linear regulators for high precision low noise voltage electronics: microcontroller, sensors

use switching regulators to coarsely drop voltages: usually actuators that will overheat if you dont, such as servos that have voltage maximum limits (for example, a 12V battery for a 4.8V servo)

Good breakdown - I'll add one more thought to this philosophy though -

"Bypass the heck out of your switching regulator using some good ceramic caps". Common-mode noise, generated by the switching element(s) of the switchmode regulator, and coupled to the robot chassis through the SMPS transistor/FET heatsink, can get into your whole system and cause all kinds of "mysterious" interference issues. Bypassing the regulator output terminals to chassis ground, using a pair of 4700pF caps can save a world of trouble. A 1uF ceramic cap placed right across the output won't hurt, either.

[Joe]

use linear regulators for high precision low noise voltage electronics: microcontroller, sensors

use switching regulators to coarsely drop voltages: usually actuators that will overheat if you dont, such as servos that have voltage maximum limits (for example, a 12V battery for a 4.8V servo)

Good breakdown - I'll add one more thought to this philosophy though -

"Bypass the heck out of your switching regulator using some good ceramic caps". Common-mode noise, generated by the switching element(s) of the switchmode regulator, and coupled to the robot chassis through the SMPS transistor/FET heatsink, can get into your whole system and cause all kinds of "mysterious" interference issues. Bypassing the regulator output terminals to chassis ground, using a pair of 4700pF caps can save a world of trouble. A 1uF ceramic cap placed right across the output won't hurt, either.

So what would you recommend using as a substitute for chassis ground if my robot will be mostly plastic? (metal working will be avoided wherever possible)

Thanks for all the replies. What I was actually wondering is how many of the switching regulators http://www.dimensionengineering.com/DE-SWADJ.htm you could stack in parallel to get how much current since most DC gear motors I've seen are rated at 3 to 4 amps a piece (and I will need 2, 6 - 8 amps). I was wondering if anyone has ever stacked 6 to 8 of these and how it went. However, I've decided to go with the unregulated source but I'm still curious if this can provide 6 to 8 amps of current at a stable voltage.

[Admin]

So what would you recommend using as a substitute for chassis ground if my robot will be mostly plastic?

I could be completely/partially wrong on this . . . someone please correct/verify . . .

Grounding of your chassis just creates a common ground to:
A) prevent shorts from contact with electrically charged external objects
B) reduce/eliminate interference of signals passing through your metal frame 'antenna'
C) Some DC motors require (or already have) the outer metal frame (of the motor) grounded

If your chassis is entirely plastic, A and B will not be a problem. C depends on the motor.

What I was actually wondering is how many of the switching regulators http://www.dimensionengineering.com/DE-SWADJ.htm you could stack in parallel to get how much current since most DC gear motors I've seen are rated at 3 to 4 amps a piece (and I will need 2, 6 - 8 amps). I was wondering if anyone has ever stacked 6 to 8 of these and how it went. However, I've decided to go with the unregulated source but I'm still curious if this can provide 6 to 8 amps of current at a stable voltage.

For DC motors, you dont need to regulate them - being off by a few volts is ok.

As for your switching regulator idea, funny stuff happens when you put them in parallel (explained on the dimension engineering site) . . . a few can be put in parallel, but I dont believe anyone has tried that many, though . . . If you are dealing with currents that high, you would do better using a transformer or a lower battery supply . . .