AN RFMD® APPLICATION NOTE Integrated Configurable Components from RFMD Multi-Market Products Group Automatic Calibration Schemes for the RFFC207x, RFFC507x, and RFMD208x Series Introduction The RFFC207x, RFFC507x, and RFMD208x integrated synthesizer products have three automatic calibration schemes. These allow high performance and very wide band devices to be manufactured with superior performance characteristics. These calibration schemes allow: • Selection of the correct voltage controlled oscillator (VCO) • Selection of the correct VCO tuning capacitance (CT calibration) • Correction for variation in the VCO tuning sensitivity (KV calibration) VCO1 VCO tuning gain (KV) calibration Loop filter Phase detector 2700-3450 MHz VCO2 3450-4450 MHz VCO autoselect and coarse tune (CT) calibration N-div VCO3 4450-5400 MHz Figure 1. Calibration Mechanisms The VCO and tuning capacitance selection (CT_CAL) is on by default, but may be switched off, if required (to minimize lock time, for example). The VCO tuning sensitivity correction (KV_CAL) is off by default and can be switched on, if required. Auto VCO selection The device achieves its wide tuning range with the use of three VCOs. An algorithm is used to automatically select the correct VCO for the desired frequency. This is performed as part of the coarse tune calibration; initially the VCO set in p1/p2vcosel [1:0] of the P1/P2_FREQ1 register is selected. This is normally set to VCO1; it can be set to another VCO dependant on the programmed frequency in order to reduce calibration time. If the coarse tune calibration reaches an end stop, the VCO is changed and the CT_CAL is performed again with the next VCO. The end stop values are set using the ctmin and ctmax values in the VCO_AUTO register. These values should be set to h00 and h3F respectively to ensure there is sufficient overlap between adjacent VCO's. Auto-VCO selection can be disabled by clearing the auto bit [15] in the VCO_AUTO register. However, this is not recommended because it will then be necessary to program the correct VCO number into the P1_FREQ1 and/or P2_FREQ1 registers. RF MICRO DEVICES®, RFMD®, Optimum Technology Matching®, Enabling Wireless Connectivity™, PowerStar®, POLARIS™ TOTAL RADIO™ and UltimateBlue™ are trademarks of RFMD, LLC. BLUETOOTH is a trademark owned by Bluetooth SIG, Inc., U.S.A. and licensed for use by RFMD. All other trade names, trademarks and registered trademarks are the property of their respective owners. ©2012, RF Micro Devices, Inc. AN120706 7628 Thorndike Road, Greensboro, NC 27409-9421 · For sales or technical support, contact RFMD at (+1) 336-678-5570 or [email protected]. 1 of 4 Automatic Calibration Schemes for the the RFFC207x/507x Series Coarse tuning calibration (CT_CAL) Each of the three VCOs within the chip consists of an integrated inductor, a switched bank of capacitors, and a voltage controlled variable capacitor, as shown in Figure 2. The switched bank of capacitors is used to center the VCO close to the correct operating frequency. The voltage variable capacitor is then used to lock the frequency of the VCO using the control voltage (VCTRL) output from the loop filter. The coarse tune calibration value is determined when the device is enabled. 2 4 8 16 32 64 C C C C C C C L Vctrl Figure 2. Voltage Controlled Oscillator It is important that the correct fixed capacitance is selected to ensure that the voltage control can cover the required range. For example, Figure 3 shows the tuning curves for a number of fixed capacitance values around a VCO frequency of 3400MHz. N+4 N+3 N+2 N+1 VCO frequency (MHz) 3420 N 3410 N-1 3400 N-2 3390 N-3 3380 Vnom 0.5 VCTRL range 1.0 VCO tuning voltage (V) 1.5 Figure 3. VCO Tuning Curves and CT_CAL Strategy A number of tuning curves can be used to lock the phase-locked loop (PLL). The internal algorithm selects the curve that locks with a VCTRL closest to VNOM, in this case curve N. It is possible to change the nominal lock voltage by programming the coarse tune values in the CT_CAL1 and CT_CAL2 registers. The recommended coarse tune value is p1/p2ctv = h0C. Coarse tune calibration can be disabled by clearing the cten bits in the CT_CAL1 and CT_CAL2 registers. If this is done, the user must program a suitable coarse tune value into the ctdef bits in the CT_CAL1 and CT_CAL2 registers otherwise the VCO will not be able to lock. AN120706 7628 Thorndike Road, Greensboro, NC 27409-9421 · For sales or technical support, contact RFMD at (+1) 336-678-5570 or [email protected]. 2 of 4 Automatic Calibration Schemes for the the RFFC207x/507x Series VCO tuning sensitivity calibration (KV_CAL) When calculating loop filter components, a number of assumptions are made regarding the phase detector gain (, A/deg), divider ratio (N), and VCO gain (KV, MHz/V) in order to determine the required capacitor and resistor values. It is usually acceptable to take the default values for and KV and a nominal value for N to do the calculation, since loop bandwidth accuracy is not critical in many applications. However, where it is critical, a calibration method has been included in the chips to correct for variation in KV. Figure 4 shows the tuning curves in the vicinity of 3400MHz. N+4 N+3 N+2 N+1 VCO frequency (MHz) 3420 N 3410 N-1 3400 ΔV Δf N-2 3390 N-3 3380 0.5 VCTRL range 1.0 VCO tuning voltage (V) 1.5 Figure 4. VCO Tuning Curves and KV_CAL Strategy The KV calibration is performed by changing the frequency of the VCO and measuring the difference in VCO tuning voltage, as shown in Figure 4. The measured voltage is then used to modify the charge pump current to correct the loop bandwidth. The following shows how this can be achieved. The PLL loop gain (and therefore loop bandwidth) is G ICP * KV/N where ICP is the charge pump current, KV is the VCO gain and N is the divider ratio, as shown in Figure 5. The ICP can be used to compensate for changes in KV to keep G (and therefore the loop bandwidth) constant. Figure 5. Simplified Phase Locked Loop (PLL) Block Diagram The internal calibration routine uses a frequency step programmed into the dn and sgn bits of the PLL_CAL registers. The dn value will be added to, or subtracted from (based on the value of sgn), the fractional part of the frequency programming word. To calculate the required frequency shift, first determine the approximate range of VCO gains that are likely to occur. The total spread in VCO gain is approximately 12-45MHz/V (nominal: 23MHz/V). However if the VCO frequency range is restricted, a narrower estimate can be made. Figure 6 shows the minimum and maximum VCO gain versus frequency of the three VCOs. AN120706 7628 Thorndike Road, Greensboro, NC 27409-9421 · For sales or technical support, contact RFMD at (+1) 336-678-5570 or [email protected]. 3 of 4 Automatic Calibration Schemes for the the RFFC207x/507x Series K V Frequency (1/mV) KV Min. and Max. versus Frequency VCO Frequency (MHz) Figure 6. VCO Gain versus Frequency To determine the optimum register settings, first determine the minimum VCO gain (KV) value, KVmin. To achieve the best phase noise the maximum charge pump current should be used (189A). To calculate the desired frequency shift take the minimum VCO gain and multiply it by 0.363V (the maximum voltage that is used to determine the VCO gain), for example: f < 0.363*KVmin. If a wide operating temperature range is desired the value of KVmin should be reduced approximately 10% to allow for variation of KVmin with temperature. The loop filter components should be calculated based upon the average (geometric mean of minimum and maximum values) charge pump current and divider ratio to center the design correctly. If the VCO frequency range is large the frequency shift can be modified to correct for the variation of N. In this case, the minimum value of KV/N should be used to determine f. Figure 7 illustrates the typical variation of KV/N with frequency. K V Frequency (1/mV) Normalized KV Range versus Frequency VCO Frequency (MHz) Figure 7. Normalized VCO Gain versus Frequency As shown in Figure 7, the minimum value of KV/N occurs near fNOM = 4450MHz. The value of KVmin is determined at this frequency (see Figure 6), and fNOM is determined from the equation above. The value of f used for dn is then set at: f = fNOM*fVCO/fNOM. AN120706 7628 Thorndike Road, Greensboro, NC 27409-9421 · For sales or technical support, contact RFMD at (+1) 336-678-5570 or [email protected]. 4 of 4