How to filter out noise in the supply line which affects the PLL and PPM output
http://home.nordnet.fr/~fthobois/RX23.htm
I am experiencing servo jitter which I suppose is due to the servo directly connected to the receiver and introducing noise in the supply line which affects the PLL and PPM output. If there is any other reason which I am unable to understand for that noise... please identify and help.
Is there a way to filter out noise from the supply line other than using a low drop voltage regulator?
R-C filter, with a big C (100 uF).
Agreed a big R-C filter is needed - also use a voltage regulator as that will also filter out a lot of rubbish
Voltage regulators work well at removing power supply noise in the 0 to 10 KHz region. But after that, it is mostly the capacitor that does the work. The control loop in the regulator runs out of open loop gain at the higher frequency noise ripples.
Also, not that I said "R-C" filter. A few ohms of resistance before you hit the C give you another 20 dB of rejection.
Hi, Use a low noice regulator like the MIC5205 from micrel (used by me in a PLL system works perfect, very very low noise) or LM2905. Check the diagrams for optimum capacitors for your frequency range.
In my design I used a switching mode power supply at 260KHz and the power supply is used for the PLL too, I pass the SMPS through the MIC5205 and supply the PLL, loopfilter etc.. I can detect any residue of the 260 KHz on my analyzer at - 120 dbm ! impressive I think.
Paul.
Added after 2 minutes:
Hi, you can also use a common mode transformer and pass + and - lines through them before you go to the servo !. capacitors at the beginning and end. I doing wonders.
Paul.
Yeah, I always get a chuckle over what people consider to be "ultra low noise regulators" nowadays. That MIC5205 has 260 nV/√hz of output noise. 30 years ago you would have instead used an LM723 with 25 nV/√hz!
They really have forgotten how to do low noise analog today! (Hint, the new regulators do not allow you to capacitively filter the voltage reference anymore. I guess the extra 0.2 cents for another pin is too expensive!)
Also, Micrel saying that CMOS is a "feature" of the regulator shows some ignorance, it is more like a curse if you seek "ultra low noise"!
True, LM723 has impressive performance, but MIC5202 is SOT23 , term ultra low is a bit marketing but OK... Tubes are better than transistors also. Much less noise.
regards,
Paul.
PS: Not all that is old is better: LM741 a work horse but outdated !.
Actually, there are a few smt regulators out there that allow you to filter the internal reference, such as:
analogictech aat3258 in a tsop package
Linear Technologies LT1761 in an sot23-5, lt1762 in an msop-8, lt1936 in an SOT223-3.
You have to search around for an LDO with a "reference bypass". There are probably plenty of others.