SNAS665 Data sheet

SNAS665 May 2015 LMX2581E

PRODUCTION DATA.

Package Options

Mechanical Data (Package|Pins)

RTV|32
- MPQF166B

Thermal pad, mechanical data (Package|Pins)

RTV|32
- QFND448B

Orderable Information

9 Application and Implementation

NOTE

Information in the following applications sections is not part of the TI component specification, and TI does not warrant its accuracy or completeness. TI’s customers are responsible for determining suitability of components for their purposes. Customers should validate and test their design implementation to confirm system functionality.

9.1 Application Information

The LMX2581E can be used in a broad class of applications. In general, they tend to fall in the categories where the output frequency is a nicely related input frequency and those that require fractional mode. The following schematic generally applies to most applications.

9.2 Typical Applications

Figure 21. Typical Schematic

9.2.1 Clocking Application

When the output and input frequencies are nicely related, the LMX2581E can often achieve this in integer mode. In integer mode, fractional spurs and noise are much less of a concern, so higher phase detector frequency and wider loop bandwidth are typically used for optimal phase noise performance.

9.2.1.1 Design Requirements

For this example, consider a design for a fixed 1500 MHz output clock to be generated from a 100 MHz input clock. Good close in phase noise and maximizing the output power are desired in this particular example

9.2.1.2 Detailed Design Procedure

For this kind of application, the design goal is typically to minimize the jitter.

PARAMETER	VALUE	REASON for CHOOSING
Fout	1500 MHz	This parameter was given.
Fosc	100 MHz	This parameter was given.
Fpd	200 MHz	Choose a highest possible phase detector frequency. There are no fractional spurs and this increases the value of C1
Fvco	3000 MHz	The VCO needs to be a multiple of 1500 MHz, which restricts it to be 3000 MHz.
Kpd	31x	This maximizes the C1 capacitor and also the phase noise
Loop Bandwidth	256 kHz	Theoretically, optimal jitter is obtained by choosing the loop bandwidth to the frequency where the open loop PLL and closed loop VCO noise are equal, which would be about 250 kHz. The phase margin is typically chosen around 70 degrees, but is chosen to be 50 degrees to increase the value of the C1 capacitor to be at least 1 nF to reduce VCO phase noise degradation.
Phase Margin	50 deg
OUT_A_PWR	45	This yields the maximum output power.
C1	1 nF	Calculated with TI clock design software
C2	6.8 nF
R2	270 Ω
Pullup Component	18 nH Inductor	This gives maximum output power.

9.2.1.3 Application Curves

Figure 22 is an example of the performance that one might see for an application like this. The achieved results show an output power of about 14 dBm (single-ended) and a jitter from 100 Hz to 10 MHz of 100 fs. Note that the output power is higher than +12 dBm as claimed in the electrical specifications because this is at a lower frequency than 2.7 GHz.

Figure 22. Measured Data and Loop Bandwidth Choice

Figure 23. Measured Plot

9.2.2 Fractional PLL Application

For applications where the output frequency is not always related nicely to the input frequency, lowering the loop bandwidth and reducing the phase detector frequency can often improve spurs at the cost of in-band phase noise.

9.2.2.1 Design Requirements

Consider generating 1880 to 3800 MHz from a 100 MHz input frequency with a channel spacing of 200 kHz. This is the situation similar that was used for the EVM board.

9.2.2.2 Detailed Design Procedure

PARAMETER	VALUE	REASON for CHOOSING
Fout	1880 - 3800 MHz	This parameter was given.
Fosc	100 MHz	This parameter was given.
Fpd	25 MHz	By trial and error and experimenting with the clock design tool, we see that this gives a good trade-off between the integer boundary spur and phase noise.
Loop Bandwidth	28.7 KHz	This is around where the PLL and VCO noise meet. The VCO is at 2700 MHz
Kpd	31x	Choose the highest charge pump gain to maximize the capacitor next to the VCO.
C1_LF	1.8 nF	The loop filter can be calculated with the clock design tool. Note that we need to keep the loop bandwidth not too wide so that the capacitor next to the VCO is larger. Also, it is put in C4_LF spot, not C3_LF spot. Both are electrically equivalent, but layoutwise, C4_LF makes more sense. See the board layout in sections to come.
C2_LF	56 nF
C3_LF	Open
C4_LF	3.3 nF
R2_LF	390 Ω
R3_LF	270 Ω
R4_LF	0 Ω
OUT_A_PWR	30	This combination of pullup component and output power settings yields optimal noise floor.
Pullup Component	18 nH Inductor

9.2.2.3 Application Curves

LMX2581E ta02_EVMplot_integer_channel_inverted_nopanel.png

Figure 24. Integer Channel

$LMX2581E ta02_spurplot_fractional_channel_inverted_nopanel.png$

Figure 25. Fractional Channel 2703 MHz

9.3 Do's and Don'ts

CATEGORY	DO	DON'T	WHY
Output Pullup Components	Place pullup components close to RFoutA and RFoutB	Go through a Via before getting to the pullup component.	The output impedance is determined by this component and if it is far away, there will be loss in output power.
Fractional Spurs	Take advantage of TI tools that can simulate these. Read the section on spurs to better understand them. Use a systematic process to optimize them	Assume that raising the phase detector frequency always improves the integer boundary spur. Assume that changing the loop bandwidth will always impact integer boundary spurs.	Fractional spurs can have more than one mechanism, especially the integer boundary spur.
Dithering	Understand the trade-offs and when it is appropriate to use. Combine with larger equivalent fractions.	Use on simple fractions .	Dithering is very effective in eliminating some spurs, but useless for eliminating others. Dithering adds PLL phase noise, so it should be only used for appropriate situations.
VbiasCOMP and VbiasVCO	Put as much capacitance as possible, up to 32 µF	Use less than 10 µF of capacitance Ignore capacitor de-rating factors.	This capacitance impacts the VCO phase noise.