Motor Noise, Decoupling techniques

Measure the battery voltage with the motor stalled, and check if the 5 V regulator drops out (because the raw feed is not enough for it to be able to control).

My guess is that you cannot get that kind of current from so small batteries.

That is correct - hence I'm thinking about dual battery, one for the analog system, one for the digital system. I'm just wondering if this is a standard technique, and what kind of problems will arise.

Regards,

PQ

It's a very standard technique to have separate power supplies for the control system vs. locomotion.

The current drain you're talking about is hard to sustain on alkalines; if you were using a NiMH pack you'd find it supplies this demand better, due to lower (and more consistent) internal resistance.

I suspect that the 1A current drain is due to the PWM period being too low (800Hz). The motor has to restart every cycle. I will increase the PWM period into the 10kHz range and see if there is any different.

Thanks for the suggestion!

Regards,

PQ

If that is really the case, then your PWM sounds like it's under- filtered. What does the output look like (under load)?

Yes, and you should certainly develop that way.

If the 9V has size/weight impact, then when it's all working, and field proven, then you could check the Lower Rail profile, to see the sag-times, vs battery age, and see if it is possible/worthwhile to replace the 9V with an Up Converter/Supercap type combination to emulate the 9V battery. You will need to 'ride-thru' some serious sags on the Motor Rail.

-jg

On Sun, 24 Feb 2008 11:34:39 -0800 (PST), I said, "Pick a card, any card" and "Peter Q." instead replied:

It is, indeed, a standard technique and a recommended practice. Also, consider using opto-isolators in front of the H-Bridge to completely separate the motor circuit from the logic side. At that point you only need worry about any physical shielding that may be required. I also recommend that you use decoupling caps on the motor just to be safe. Ask your motor manufacturer for their recommendations as far as the value of the caps.

-- Ray

But you shouldn't be doing it based on a guess. You should take your voltmeter and _measure_, to make sure that the effort you're spending is going to do any good.

Isolating your processor power rail from your drive power rail is a good thing.

If you're using a 7805 regulator then you don't have much, if any, overhead from your input to output voltages -- the 7805 needs between 1.5 and 2V more on the input than the output to stay in regulation. Even if you are already using a low drop-out regulator, if your battery voltage is bouncing around from the motor then the regulator may have problems.

Consider also that both dry cells and NiXX batteries aren't really fully discharged until you're down to about 0.9V/cell at moderately high current. For your batteries this means that you can expect them to be fully discharged when you're getting 4.5V out of them -- and that's lower than what your computer needs. With five cells your computer will go on the fritz long before you've squeezed the last coloumb out of the batteries.

For a prototype system it's not a bad idea to power the signal processing and power electronics from different batteries. This is probably the least-engineering way to get the job done. One battery pack is certainly cleaner from a product design point of view, but I see a lot of RC toys that separate the two, so I can't say it's not an accepted method.

If you do want to decouple them while running off the one battery pack then you're going to have trouble with the number of cells. Taking the processor power through a Schottkey diode into a capacitor will prevent the processor power from dropping suddenly when a motor comes on, but it won't keep the processor rail from dropping below 5V when the batteries permanently sag. Using six batteries will help a bit there, but then your motor drive gets excessive. Using a buck/boost switching regulator would be the bees knees, but it may be more expensive than a 9V battery, and would certainly be more trouble to design.

... snip ...

Definitely consider a totally isolated power supply for the motor. You should also carefully isolate all the logic from ground return drops and noise. Brush systems are inherently noisy, and will also radiate. The brushs are interrupting inductive circuits.

...

As others have pointed out, those small alkaline batteries might not be able to deliver such currents. NiCd accumulators (if you can still get hold of them) have a low internal resistance, reducing the voltage drop.

With such loads, the PCB layout is also important. Make sure that there are separate wide tracks from the motor battery to the H-bridge. Use a separate ground system for the control system and run it from a separate 9 V battery. Add a single connection from the control system ground to the ground side of the H-bridge for the return path of the gate drive signals only. This should keep the ground noise out of the control system.

If you later want to get rid of the 9 V battery, install a small inverter, driving a transformer followed by a floating rectifier and regulator to run the control logic. With the isolated control system power supply, the only galvanic connections between the control system and the high power system is the gate drive signals and the single return path between the control system ground and the high power system, in which only the gate signal flows.

This should be sufficient to keep any resistive coupled interference from the control system, but of course, there could still be capacitively or inductively coupled interference running through the single ground coupling between the systems (and in the logic system ground). If this a problem, use optoisolators to drive the H-bridge and remove the connections between the grounds, so that the control system can float freely and no interference currents will flow in the control system ground.

Paul

In addition to what others have written.

Motor brushes generate a lot of noise, standard practice is to add a small capacitor directly between the motor terminals on the back of the motor. Do you motors have this? Other confgurations are capacitors from each motor terminal to the motor housing or a combination of these two (so you have 3 caps in total). Value can be something like 100pF-10nF or even larger, but be carefull with higher values if your PWM frequency goes up.

Such noise suppression capacitors should be in series with a resistor. The resistor size will depend on the actual motor, but generally small values (such as 47 ohms) will be suitable. Not critical, but this provides a place to dissipate the noise energy.

They never are, atleast I have never seen it. I know of one motor that can be supplied with a rather large capacitor (220nF on a 5W motor) across the brushes then 2 inductors and 2 smaller capacitors to the body after those.

All other motors I have seen use the 1, 2 or 3 cap method without resistors. Open up any toy/appliance with a DC motor and I expect you to find caps only, 1 cap being the most common.