ripples in dc line modulating locked VCO in PLL circuit
I have a PLL at 1.8GHz,the dc-dc converter which provides
15 volts supply to PLL board have ripples of 50mVpp at its switching frequency 125kHz.
I am observing spurious at 125kHz and its multiples at locked VCO output.The spectrum thus appears to be a mudulated signal at 125kHz.
Since such a spectrum is undesirable as the level of these spurious components is as high as -30dBc.
I need your expert opinion to sort out this problem,what kind of filtering do you suggest at 15 volt line.
Hi,
simple R-C.
R needs to be samll enough not to cause too high voltage drop. (DC current of your load)
And the C forms with the R a low pass filter.
As a rule of thumb: If you want to attenuate the ripple to be only 1/10 then you need to calculate cut off frequency to be 1/10 of 125Hz --> about 12kHz.
Klaus
I recommend you to look at Murata's special SMD Supply Line Filters portfolio.There are really interesting products to suppress such dirty supply problems.
Note :50mVpp ripple is huge even for switching power supply, is there any design or implementation error ?
This dc-dc converter is made by some new engineers,who were asked to reduce dc-dc converter size.I think they have compromised a lot while reducing the box size.
What about a LC filter?
In RC ,we will lose some voltage in drop.
What could be the recommended values to filter out 125kHz components?
Hi,
RC filters have no resonance frequency.
LC filters have a resonance frequency. I recommend to dampen this resonance with extra RC. But it is more critical.
In worst case you attenuate 125kHz ripple but create a ripple (higher?) at another frequency..
Klaus
Meanwhile, what is the comparison frequency of the PLL ? It might also be Charge Pump spurs @ 125kHz.
Here are few examples of low-noise supply line filters:
Thanks Vfone&Klaus for your views and suggestions.
@Bigboss fc is 2MHz in this PLL
This PLL uses differential outputs,PD_up and PD_down,not the charge pump.
OK, if PLL+VCO does not consume much current, you can use RC filter to suppress this noise by using 1/2*pi*R*C formulae in order to calculate cut-off frequency.You can also use cascaded R-C filters.
Or, you can also obtain high valued inductors with ferrite components because you need high valued inductors and low ESR capacitors to implement this Low Pass Filter.Feedtrough Capacitors are also useful in such problems.
Thanks Bigboss,for your suggestions
Yes, the PLL+VCO+TCXO draws 70mA on +15V line.I want
to restrict the voltage drop to 0.5V .
So,if I want to reduce the ripple by 1/10 of the present value,
I need to choose filter cut-off as 12.5KHz or 10KHz approx to suppress
125KHz.
The values, thus I get using 1/2piRC formula are R=4.7ohms
and C=3.3UF.
Can we use polarized capacitors in these RC filters?
Sure but not electrolytic caps. because their ESR is quite high except some low ESR models.
I suggest tantalum caps, their characteristics are much better in such applications.
Hi,
Isn't it quite unusual to drive a 1.8GHz VCO and PLL directely from 15V?
I expected 3.3V or less...
Klaus
You are right Klaus, The PLL IC indeed requires 3.0 V,but I am referring
to the PCB ,which houses other components VCO and TCXO as well ,which require
direct 15V,I am regulating supply to 3.0V for PLL IC in the PCB.
I suggest you look for a LDO that specifies a good HF PSRR.
They are out there, for this purpose (killing switcher ripple).
Many LDOs however have good DC PSRR specs but roll off
to soon to help with HF ripple.
But be sure that you don't neglect ground ripple relative
to any tune voltage & reference, VCO supply branch, etc.
Might inspect tune voltage for ripple esp. if using an
op amp to integrate charge pump or U/D outputs from
the PLL; this amp might be fed from raw 15V and its
own HF PSRR could then be an issue. Might need an RC
filter from there to VCO tune pin, with C returned to
VCO ref gnd.
a phase locked loop serves to lock a free running VCO to a reference (lower) frequency. If your free running VCO has noise caused by ripples or spikes on the power supply line, the PLL will TRY to fix any phase perturbations those ripples/spikes cause. However, a PLL has a limited amount of OPEN LOOP GAIN at the frequency of the ripple, so it can only reduce the ripple's effect by that open loop gain. If for instance, there is 23 dB of open loop gain in the PLL control circuit, then that ripple will be reduced by that same 23 dB. Unfortunately, if your ripple starts off HUGE, that open loop gain reducing the ripple will not be enough!
SO as some have said above, linear voltage regulator, coupled with capacitive passive filtering, is very good to start with. Additionally, opening up your control loop bandwidth so that there is MORE open loop gain at the ripple's 125 KHz frequency, will also help.
Thanks dick_freebird and biff44 for your suggestions.
The above PCB uses +15V secondary supply from dc-dc converter,I have regulated the voltages like 3.0 V and 5.0V
required by other components like PLL chip etc. using 3T regulator. But the VCO and Op-amp require +15V to function,so I have used LC filters(L=100nH,C=1nF,100nF,1uF) in the supply line,and using the unregulated supply line for the purpose.
Now,the PLL circuit being used is for carrier generation and the loop BW is chosen to get the best phase noise performance ,(LPBW= 70KHz).
Since,the supply spurious(125KHz) are falling out of loopBW,PLL is not able to suppress them as one desires.
The Only ,way to reduce these are by using addition RC filter in the PLL 15V supply line.
How does common mode filters can reduce the power supply ripples in the better way?
These common mode filters are connected between DC Live line and chasis(ground) and also DC return lines and chasis(ground).
Hi,
This more sounds like an EMI filter, for very high frequencies.
It doesn't attenuate in the range of 125kHz.
--> Use higher value inductors and higher value capacitors.
Klaus
You are right Klaus.
I did not anticipated this problem and decided to go for small
size SMDs to make this filter while making PCB layout,which is
not enough to attenuate 125KHz.
I need to put either RC or LC filter in the supply line externally
,to deal with this problem now.