LMX2487E National Semiconductor Corporation, LMX2487E Datasheet - Page 22

• Manufacturer Part Number: LMX2487E
• Description: 7.5 Ghz High Performance Delta-sigma Low Power Dual Pllatinum? Frequency Synthesizers With 3.0 Ghz Integer Pll
• Manufacturer: National Semiconductor Corporation

www.national.com
1.8 CYCLE SLIP REDUCTION AND FASTLOCK
The LMX2487E offers both cycle slip reduction (CSR) and
Fastlock with timeout counter support. This means that it re-
quires no additional programming overhead to use them. It is
generally recommended that the charge pump current in the
steady state be 8X or less in order to use cycle slip reduction,
and 4X or less in steady state in order to use Fastlock. The
next step is to decide between using Fastlock or CSR. This
determination can be made based on the ratio of the com-
parison frequency ( f
Cycle Slip Reduction (CSR)
Cycle slip reduction works by reducing the comparison fre-
quency during frequency acquisition while keeping the same
loop bandwidth, thereby reducing the ratio of the comparison
frequency to the loop bandwidth. In cases where the ratio of
the comparison frequency exceeds about 100 times the loop
bandwidth, cycle slipping can occur and significantly degrade
lock times. The greater this ratio, the greater the benefit of
CSR. This is typically the case of high comparison frequen-
cies. In circumstances where there is not a problem with cycle
slipping, CSR provides no benefit. There is a glitch when CSR
is disengaged, but since CSR should be disengaged long be-
fore the PLL is actually in lock, this glitch is not an issue. A
good rule of thumb for CSR disengagement is to do this at the
peak time of the transient response. Because this time is typ-
ically much sooner than Fastlock should be disengaged, it
does not make sense to use CSR and Fastlock in combina-
tion.
Fastlock
Fastlock works by increasing the loop bandwidth only during
frequency acquisition. In circumstances where the compari-
son frequency is less than or equal to 2 MHz, Fastlock may
provide a benefit beyond what CSR can offer. Since Fastlock
also reduces the ratio of the comparison frequency to the loop
bandwidth, it may provide a significant benefit in cases where
the comparison frequency is above 2 MHz. However, CSR
can usually provide an equal or larger benefit in these cases,
and can be implemented without using an additional resistor.
The reason for this restriction on frequency is that Fastlock
has a glitch when it is disengaged. As the time of engagement
for Fastlock decreases and becomes on the order of the fast
lock time, this glitch grows and limits the benefits of Fastlock.
This effect becomes worse at higher comparison frequencies.
There is always the option of reducing the comparison fre-
quency at the expense of phase noise in order to satisfy this
constraint on comparison frequency. Despite this glitch, there
is still a net improvement in lock time using Fastlock in these
circumstances. When using Fastlock, it is also recommended
that the steady state charge pump state be 4X or less. Also,
Fastlock was originally intended only for second order filters,
so when implementing it with higher order filters, the third and
fourth poles can not be too close in, or it will not be possible
to keep the loop filter well optimized when the higher charge
pump current and Fastlock resistor are engaged.
1.25 MHz < f
f
COMP
f
COMP
Comparison
Frequency
( f
2 MHz
≤
COMP
> 2 MHz
1.25 MHz
COMP
)
≤
COMP
Noticeable better
than CSR
Marginally better
than CSR
Same or worse
than CSR
) to loop bandwidth ( BW ).
Fastlock
Likely to provide a
benefit, provided
that
f
COMP
Reduction
Cycle Slip
( CSR )
> 100 X BW
22
1.8.1 Using Cycle Slip Reduction (CSR) to Avoid Cycle
Slipping
Once it is decided that CSR is to be used, the cycle slip re-
duction factor needs to be chosen. The available factors are
1/2, 1/4, and 1/16. In order to preserve the same loop char-
acteristics, it is recommended that the following constraint be
satisfied:
(Fastlock Charge Pump Current) / (Steady State Charge
Pump Current) = CSR
In order to satisfy this constraint, the maximum charge pump
current in steady state is 8X for a CSR of 1/2, 4X for a CSR
of 1/4, and 1X for a CSR of 1/16. Because the PLL phase
noise is better for higher charge pump currents, it makes
sense to choose CSR only as large as necessary to prevent
cycle slipping. Choosing it larger than this will not improve lock
time, and will result in worse phase noise.
Consider an example where the desired loop bandwidth in
steady state is 100 kHz and the comparison frequency is 20
MHz. This yields a ratio of 200. Cycle slipping may be present,
but would not be too severe if it was there. If a CSR factor of
1/2 is used, this would reduce the ratio to 100 during frequen-
cy acquisition, which is probably sufficient. A charge pump
current of 8X could be used in steady state, and a factor of
16X could be used during frequency acquisition. This yields
a ratio of 1/2, which is equal to the CSR factor and this satis-
fies the above constraint. In this circumstance, it could also
be decided to just use 16X charge pump current all the time,
since it would probably have better phase noise, and the
degradation in lock time would not be too severe.
1.8.2 Using Fastlock to Improve Lock Times
Once it is decided that Fastlock is to be used, the loop band-
width multiplier, K, is needed in order to determine the theo-
retical impact of Fastlock on the loop bandwidth and the
resistor value, R2p, that is switched in parallel during Fast-
lock. This ratio is calculated as follows:
K = ( Fastlock Charge Pump Current ) / ( Steady State
Charge Pump Current )
The above table shows how to calculate the fastlock resistor
and theoretical lock time improvement, once the ratio , K, is
known. This all assumes a second order filter (not counting
the pole at 0 Hz). However, it is generally recommended that
the loop filter order be one greater than the order of the delta
sigma modulator, which means that a second order filter is
16
K
1
2
3
4
8
9
Bandwidth
1.00 X
1.41 X
1.73 X
2.00 X
2.83 X
3.00 X
4.00 X
Loop
R2p Value
R2/0.41
R2/0.73
R2/1.83
30013940
Open
R2/2
R2/3
R2
Lock Time
100 %
71 %
58%
50%
35%
33%
25%