Reduce BEC Noise Using Capacitor at RX Terminal

[rcBluebird]

FYI,

I've been asked by many folks who would like to know an easy way to reduce BEC noise, so here it is. I have tested in my laboratory, and captured images for your review. However, this is just FYI, and I am not responsible for your equipments what so ever stated or non-stated.

If your BEC uses linear regulator, the noise is not that much. The on-board capacitor is enough to reduce its noise. However, if you have a switching BEC, the noise is much higher (i.e. 28mVpp - figure 1). The most easy way would be using an extra big capacitor (more than 500 micro Farads) to plug it into your extra channel, and the noise is much lower (i.e. 16.8mVpp - figure 2). That means I dropped about 4.68dBm noise in my switching BEC.

You can use many large size capacitors in your old radio or at local electronic surplus store. Cut its leads as short as possible, and solder its leads directly to the socket of an old servo connector. Please pay attention to the polarity of the capacitor (White line or black line on the side of the capacitor can indicates its negative lead). Please see my drawing for more clarifications.

I tested with 1000uF, 16V aluminum capacitor that plugged in CH7.

I hope this helps to reduce noise in your noisy switching BEC.

Happy Thanksgivings 2007

[lazy-b]

RCBlueBird, Can you Please try using a Tantalum Capacitor......this has much lower E.S.R. ( Effective series Resistance). Tantalum is just more expensive.....I've often see this Tantalum inside a Servo.

[jeffs555]

Interference is not a simple matter, and I would have to question what effect this would have on actual glitching at the receiver. You have reduced the ripple voltage on the power input to the receiver, but have also increased the ripple current in the wires from the BEC to the receiver. It would all depend on how much interference was from conduction through the receivers power inputs and how much was from radiation through the wiring. Also, you would probably need a spectrum analyzer to judge the effectiveness of any filter. It is normally the harmonics that cause the interference. At RF frequencies, a small ceramic capacitor will have lower impedance than a large electrolytic.

[rcBluebird]

lazy-b,

Sorry, I just got back from the trip, so I could not access my equipments to perform the test with tantalum cap. Anyway, here I am. I did not have a large size tantalum cap, so I used three shunt tan caps (each has 100uF, 16V from Kemet - TPSE107K016R0100), and it yields almost the same results as aluminum capacitor on the same voltage scale. Since Tantalum cap has much lower ESR than the aluminum one, so it should has better result. However, I did not have enough of them to get the same capacitance as the aluminum, so the results are slightly different.

jeffs555,

EMI has two main matters, and they are conduction and radiation. The way that I described above is only used to reduce the ripple voltage on the switching power line before entering to the RX. However, it also directly effects the conduction and indirectly effect the radiation of the switching power supply (i.e. BEC). I agreed that interference is not a simple matter to solve because you have to play mix and match with chokes (ferrite ring or bead) and capacitor. For conduction, a choke can be placed between BEC and RX. If BEC causes glitch to RX, put the BEC in aluminum foil would also help. Yeah, spectrum and network analyzers are the least and required equipments to diagnose the interference. Different BECs have different switching frequencies, and depending on the RX's freq, we would have to analyze the harmonic at that freq. However, adding a capacitor at the RX is an easiest way to reduce ripple voltage of the BEC.

[vintage1]

In general a ceramic capacitor of around 10-100nF is better at suppressing RF/IF noise than an electrolytic. At the frequencies of interest the coiled nature of a tubular capacitor and the resistance of the electrolyte severely degrade HF performance.

Standard practice in the days of ECL/Schottky TTL was to put a tantalum and about 100nF in parallel across any supply rails.

Schottky TTL put current spikes on the lines well into UHF and beyond frequencies.

Stray coupling meant it was extremely hard to eradicate..Any RF in the vicinity and the logic itself demanded a PCB layout designer with a feel for RF design, as well as the ability to lay tracks between pins artisitcally :D


I had just such a design..synthesised FM/AM broadcast receiver. I went through 13 evolutions of PCB..and never got the last 2-3nV of noise out of the FM system.

[jeffs555]

However, adding a capacitor at the RX is an easiest way to reduce ripple voltage of the BEC.
Again, I agree you have reduced the ripple voltage on the power supply to the receiver, but I just don't think that is going to have any effect on receiver performance. On most modern receivers the power input is regulated and filtered so that the ripple voltage you show would most likely never make it to the actual receiver circuits even without adding a capacitor. To show real effect of your capacitor, you would have to scope the output of the regulator and filter in the receiver. You would probably see little to no effect there from adding your capacitor on a spare servo voltage pin.

[rcBluebird]

vintage1,

I agreed with you about a good design practice that having a 100nF across the power line, but that should use ceramic capacitor (not tantalum). I usually place at least one 100nF ceramic capacitor at each power pin of every active component to the ground. That would help to reduce voltage noise from the power supply. A combination of Schottky diode with switching BEC would claim the voltage noise very well, but a typical Schottky alone cannot withstand a large voltage drop (i.e. from 11.1V to 5V to get 3 Amps or above).

jeffs555,

Yeah, RX would have to select and filter out noise before using the signal, but if the noise floor is too high, the filter does not do any good for the RX. A good example of the rising of noise floor will distort the desired signal would be driving a car under or along power line and listen to the AM radio. Audio will be distorted badly when the car run across the high voltage power line.

Actually, switching BEC has to run at a certain frequency (i.e. 100 KHz up to several mega Hertz), and that freq should be claimed down at the output port. It is not as smooth as a DC voltage as in battery. Please don't misunderstand that this is only help to reduce the ripple voltage to the receiver, and it may also help to reduce glitches. It is an easiest way to do that. However, glitch in RX can also be observed even using a pure DC supply such as NiCd or Lipo battery, and glitch is the whole new game for RC hobbyist.