Filter electric motor noise for video?

[rocknbil]

I've been experimenting with those cheapie video cams in a small electric heli. For what you pay for them, they work fine. When I use a separate individual power source, such as a 3.7v 160 mah lipo pack, I get little to no interference.

However this makes it too heavy. So I tested plugging it into the main pack for the heli, a 2-cell 7.4v 800 mah, and it also works fine until you wind up the motors (as I predicted.)

See bad video here (http://www.rc-resources.com/images/testvid/video_interference.wmv) (small)

I tried an uncharged capacitor across the main power to the camera, this did not help. Does anyone know of a simple way to filter out the line noise caused by the motors?

I'm guessing I'm out of luck and it's back to the lipos and figuring some way to trim weight, but was hoping one of you had an idea. Thanks for looking. :D

[Nicetie]

Try this: 1 ohm resistor from supply to camera. 1Uf and .001Uf cap at camera side of resistor. Be sure the 1Uf cap has the + to the + side of the power at the camera.

The larger cap should filter low frequency noise and the small one should
filter the higher frequencys. If the 1 ohm resistor can be increased to
10 ohms, the filtering could be better, but whether you can do this depends on how much of the DC voltage is lost across the resistor due to camera current. The lower the required current, the higher the resistor value can be.
Try raising the resistor value if you aren't getting enough filtering. There
will be some point where too much power is dissapated in the resistor or
the camera supply voltage is insufficient.

Ken K5MBV

[Comatose]

A regulator with good line regulation would solve the problem. A 5v regulator like a 7805 with capacitors on the input and output would do the trick. One of our Sw05 switching regulators would work too, but is probably overkill.

Alternatively try running it from the ESC's BEC with a .1uf bypass cap as close to the camera as possible.

[rocknbil]

You guys are awesome, thank you! :D It just so happens I have a 7805 here from a different project - the only concern here is that the nominal voltage of the cam is 6-9 volts and 5 will be too low. I will try it though, the worst that can happen is the camera just won't light up.

Nicetie, I will try this, presuming you mean something like


+ bat --------- 1 ohm ------ + cam
|.001.uf |
| | 1uf
- bat ------------------------ - cam



(Where | is cap's going from + to -)

Ordinary caps, correct, not charged ones? (or power capacitors, my terminology is rusty) What wattage of resistor? I have half-watt ones here and power resistors, but those are huge. :D

[Gary Warner]

A 10uh coil can be used in place of the 1 ohm resistor if you want (color bands brown, black, black).

Gary
--

[Mr.RC-CAM]

With a 2-cell Lipo powering the motor, it is very difficult to expect good operation with a 5V wireless video system. The problem is the motor commutation ripple, which can traverse below the minimum operating voltage on LDO linear Vregs. The fast rise times are a challenge for switching VRegs too.

It can be done, but the LM7805 is not a good pick. If it was me, I would feed a LDO Vreg's input via a schottky diode with a huge amount of bulk capacitance on it. This will act as a charge pump during the ugly commutation periods. FWIW, from my experience, a small dedicated LiPO pack for camera power weighed less than a fully filtered solution that shared the flight battery. Worked better too.

[Nicetie]

Put both caps on the camera side. They should be in parallel as close
to the camera as possible. The 10 uh coil would be a better choice
than the resistor. You won't have to worry about power or heating and the
coil will be small and light. The 1.Uf cap is an electrolytic and the
.001Uf cap is a small flat disc. You don't need a regulator to reduce
noise, but they can perform that function. A switching reg would be best
but it if you are doing elec flight it would have to be very small and light.

Ken K5MBV

[Eric Brouwer]

You can also try installing ferrite rings on the control wires of the speed controller, gyro & tail rotor servo. Three turns per lead throught the ferrite ring will normally work. This will eliminate a lot of noise/spikes, although it will not help a lot on voltage regulation.

[rocknbil]

Thx Eric, but I have no access to the ESC/etc., this is a 4-in-1 unit unfortunately.

Well I tried various resistors and capacitors yesterday and while I managed to not blow anything up, the interference didn't go away. It may have decreased a little but not enough to move into any positive direction. I also tried the 7805 and sure enough, not enough voltage to power the cam.

I could not find any coils or inductors at Radio Shack (big surprise there,) but haven't given up on it. The only thing they had were RF chokes at 10h, but they were huge and had a big ferrite core, thing must have wieghed 20 grams.

With a 2-cell Lipo powering the motor, it is very difficult to expect good operation with a 5V wireless video system.....a small dedicated LiPO pack for camera power weighed less than a fully filtered solution that shared the flight battery. Worked better too.

^^ Thank you, actually that is where I started, with two 160 mah 3.7v lipos in series and it works perfect. Just one problem. The extra 10g leaves me with about 60 seconds of flight time as opposed to 8-10 minutes. If I can get 3-5 out of it I will consider it a success. But as mentioned, the cams are 6-9v nominal.

Thank you everyone for the helpful input on this, I **KNOW** this is outside the box and the ultimate solution would be to get a GOOD camera and a craft that can carry a payload. That will come someday. But at the moment I'm intrigued with the idea of the mimimalism of a small indoor electric and getting the cam to work with it. :D I'm soo close!

I will dig around for other vregs and a light 10h coil/inductor (right? an inductor or choke would work for the experiment?) and try that next. NEVER GIVE UP! :D

[rocknbil]

A 10uh coil can be used in place of the 1 ohm resistor if you want ....

I've been hunting high and low for a coil, found lots of SMT's which I could probably implement but only a couple through-hole (in-line) ones small enough to try. The question is the amperage rating: the batteries are 800 mah, the one spec'ed below is 500 mah, will it melt down if I try it?

These things are dirty cheap, so I have a few different ones on the way.

JW Miller pn 4612 10uh 500 Phenolic (.875 lenth)

I am looking at phenolic and ceramic as opposed to ferrous for weight considerations. Bad idea?

link to part (http://www.newark.com/NewarkWebCommerce/newark/en_US/endecaSearch/partDetail.jsp?SKU=59F279&N=0)
link to full specs (pdf) (http://www.jwmiller.com/pdf2/4600.pdf)

[Mr.RC-CAM]

I think that you will find that the ripple currents you have to tame will not be solved with a uH valued inductor. Assuming I am understanding how you are using the 2-cell Lipo as a power source, this situation is not likely to be EMI/RFI related. It is more likely to be inadequate VReg input voltage.

The huge motor currents cause KHz interval'd voltage dips that have excursions much lower than 5V. If you have an o-scope, take a peek at the output of the Vreg. It will tell you all you need to know, including if my hunch it on target.

Assuming my thoughts are correct, a diode fed LDO VReg, with bulk capacitance (min 1000uF) at the Vreg Input to act as a charge pump/filter, is probably a more likely solution. Once you can maintain a voltage that does not stray below the LDO's minimum requirement you will then have a fighting chance. {note: Be sure to add low ESR ceramics decoupling caps on the VReg's input and output to tame any EMI that might be there.} Keep all leads as short as you can.

Of course I could be wrong. But, it would not hurt keep my recommendations in mind.

[rocknbil]

I'll drop by Radio Shack and see if they have anything above 5 v and give it a shot, thanks! Trying anything at this point. :D

[Mr.RC-CAM]

If you are going to take a stab at my suggestions, then Radio Shack will not have the parts you need. Their goodies are now down to the bare essentials.

[jeffs555]

RC-Cam is most likely right. Those cheap Chinese cameras are incredibly unstable, and what you are seeing is this instability, not normal electrical noise. There is probably nothing you can do to make it as stable as with a separate battery, and if you even come close, it will likely weigh more than separate Lipos. Your best bet for weight reduction would be to use smaller Lipos for the camera. Two 80mah Etec's would weigh less than 5 grams. I went through all kinds of filters, and finally went back to the 80mah Lipos. Another advantage of the lipos is that you end up with a small brick that can be easily moved from plane to plane.

Jeff

[rocknbil]

And try I did, anyway. :D Well back to the lipos I went, using 160 mah 3.7v's in series. I see a ray of hope here, see the log here (http://www.rcgroups.com/forums/showthread.php?t=495088). It's **really** raw and my build methods are crude, but now that I know it will work it can only get better.

Right? :D

Thanks again for the help, will keep hacking away at this stuff, having a lot of fun.

[Jarhead#1]

An LDO linear regulator can reduce the ripple tremendously (as Mr.RC-CAM said), typically they can reduce the noise by a factor of 60dB (which is alot).

Here is a nice small part which may work nicely for your application, but I do not know what current these little cameras draw. If one knows the actual current draw, an even more suitable part might be found. The other nice thing is that they have free samples of the part, if you sign up.

This particular part has a fairly low dropout, a bit lower than an atypical LDO, while still handling up to 10V input. The dropout may be low enough on this one that you'd not need the diode trick, which adds additional drop itself, where one needs to add the capacitor to hold up the voltage in-between motor load pulses, though, as the battery runs flat 2.5V to 2.7V per cell for Li-Poly (?), assuming you have two, the input voltage would drop below the LDO's output (say output was set for 6V), and you'd start getting noisy video at that point.

Unlike many other parts, this part will maintain ~60dB of rejection up to 10KHz, and even has over 20dB of rejection at 1MHz, while delivering 150mA at the same time.

It comes in a SOT-23 package, which is plenty large enough to work with (about 3mm by 3mm footprint), but not so large as to add too much weight.
http://focus.ti.com/docs/prod/folders/print/tps76301.html

If need be, and you cannot deal with stuff this small, if you give me a holler, with a PM (I hope they have those on this board), you might be able to talk me into building you one for a very small fee, or we can barter.

I assume your camera works fine with no noise, sync, or other issues off the separate cells. If you are still getting a little bit of something in there, one can also try wrapping the camera up in aluminum or copper foil (copper is nicer, since you can solder the seams up). It may help to bring it all the way up to and covering a portion of the power wires.

I could rig you up a custom boost regulator, which would boost the voltage up above 7.2V (think you said your camera can take 9V). It would be best to know the actual current draw of your camera to best match things, since the R/C crowd is usually big into minimal weight. For a really low noise power source, we could do a boost regulator, followed by a linear regulator, which would be minimal size impact, and minimal weight impact, depending on just how sensitive you are to weight. I'm thinking of something that would be around the size of a dime overall, and about 2/10" of an inch tall.

This is my first post over here at R/C forums, but for reference, you can check me out over at www.candlepowerformums.com . ( http://candlepowerforums.com/vb/member.php?u=4194 )

If you wanna live chat with me you can find me usually over at our live chat room, IRC server 206.41.117.133 port 7070, or go to www.candlepowerforums.com and use the java chat application.

I work as an Avionics Design Engineer (EE) for a living.