Designer’s Guide Community

ic_engr

Senior Member

Offline

Posts: 135

PLL Phase Noise
Jan 31^st, 2006, 11:18am

Hello,

I have a PLL with VCO using a varactor and an inductor. The VCO covers the range of 480MHz to 630MHz using 64 curves. Each curve spans approx 10MHz with a control voltage 0.2V to 0.8V. The VCO curves ensure continuity by overlapping frequencies.
The PLL input Freq=66.6kHz and I am using a feedback divider of 9280 decimal to give me the correct coversion.

A measurement showed us that the phase noise dropped by 10dB when the curve was switched although the frequency was maintained at 630MHz which means the control voltage from the charge pump changed.

I suspected the VCO and therefore I performed the open loop VCO simulation with the control voltage forced by a voltage source. Same frequency at two different curve settings gave me a 3dB difference in Phase NOise.

However, I also suspect the charge pump up/dn pulses contributing some noise on the control voltage. The control voltage directly controls the mosvaractors, and therefore could contribute to phase noise shift.

The issue is the charge pump uses a boosted supply which is generated by a regulator. The supply of charge pump has this inherent 21mV step every 112usec (9kHz). The nominal value is 1.5V then the level drops slowly by 21mV every 110usec.

Now if I plan to use the pss-pnoise, how can I model the effect of charge pump supply (sort of rampling voltage with a repetition rate of 110usec and level of 21mV step). How can the pss-pnoise simulation evaluated a periodic solutions.

And would any guru comment if this 21mV step every 110usec contrbute to phase noise, especially the fact that the up/dn pulse from the PFD are coming at 66.6kHz.

Please help

ic_engr

Back to top

IP Logged

ACWWong

Community Fellow

Offline

Posts: 539
Oxford, UK

Re: PLL Phase Noise
Reply #1 - Feb 3^rd, 2006, 3:33pm

I guess what you are interested in is the psrr for the charge-pump. although the boost circuit repeats at ~9kHz, i'm guessing that it isn't always EXACTLY the same rate (assuming your boost circuit is based upon simple k*Vref trip points), so variations in the exact vboost refresh rate can cause havoc to your close-in phase noise.... can you see/measure 9kHZ spurs/humps on the VCO output ?
the psrr of your charge-pump is most likely to be Vtune dependant, so you can simlulate the effect (without the need for VCO to be included, so it'll be quick). you might see that the psrr is worse when the vtune is at the voltage which gave you very bad phase noise.

Back to top

IP Logged

wave3x

Junior Member

Offline

Posts: 10
Shang Hai, China.

Re: PLL Phase Noise
Reply #2 - Feb 23^rd, 2006, 11:36am

hi, ACWWong
Why you said "variations in the exact vboost refresh rate can cause havoc to your close-in phase noise.... ", I think even a periodial refresh rate would cause spur on the VCO's output. and also I wondered why this "cause havoc" to the close-in phase noise but not the out-of-band phase noise?
Anyone bring me a clear answer?
wave3x,

ACWWong wrote on Feb 3^rd, 2006, 3:33pm:

I guess what you are interested in is the psrr for the charge-pump. although the boost circuit repeats at ~9kHz, i'm guessing that it isn't always EXACTLY the same rate (assuming your boost circuit is based upon simple k*Vref trip points), so variations in the exact vboost refresh rate can cause havoc to your close-in phase noise.... can you see/measure 9kHZ spurs/humps on the VCO output ?
the psrr of your charge-pump is most likely to be Vtune dependant, so you can simlulate the effect (without the need for VCO to be included, so it'll be quick). you might see that the psrr is worse when the vtune is at the voltage which gave you very bad phase noise.

Back to top

IP Logged

ACWWong

Community Fellow

Offline

Posts: 539
Oxford, UK

Re: PLL Phase Noise
Reply #3 - Feb 24^th, 2006, 3:21am

hi wave3x,

what i meant to suggest is that the periodical refresh rate (and what happens within this cycle) isn't of a high spectral purity, so the spur @9kHz offset will look more like a wider hump.
yes "close-in" wasn't clear... given the loop bandwidth is likely to be quite narrow (7kHz ?) this 9kHz hump isn't quite in band, although remnants maybe.
sorry, force of habit of working with wider bandwidth plls means anything in kHz is close-in for me Tongue

my experience with boosted power supply chargepumps, is that the refresh rate was variable as it depended on the exact discharge characteristic, whcih was although dominately in terms of time constant by a decoupling capaictor (the 9kHz in this case). what happens in the boosted supply cycle is important on discharge and recharge (pump back-up up phase) and can cause poor phase noise, in this case as you rightly suggest, alot of this is out of the loop bandwidth.

Back to top

IP Logged