USERS MANUAL ISL55210-ABEVAL1Z AN1831 Rev 1.00 Active Balun Evaluation Board May 7, 2013 Only +3.3V supply is required where the ISL55210 delivers Using the Active Balun Board to exceptional performance using only 34mA supply current. Test Design Options The resistors here have been snapped to 1% standard values. To get a desired input impedance matched to R , and gain (Av) This relatively simple board is intended to provide a quick S from the input of R to the differential outputs, the exact means to test the performance of the active balun g1 element values are given by these two simple equations configuration using the ISL55210. While all wideband, voltage 2 (Equations 1 and 2) . feedback (VFA), fully differential amplifiers (FDA) can perform a single ended input to differential output operation, the The two feedback resistors should be equal and set to: frequency span for acceptable input match (or low return loss) AvAv + 4 R S is greatly enhanced by the >1.5GHz common mode loop (EQ. 1) R = ------------------------------------ f 2Av +2 bandwidth internal to the ISL55210. The input and gain setting elements for this board are very simple, while the output side Then R will be set to: g1 includes 3 possible output interfaces. As delivered, the R differential output is converted to single ended through a very S (EQ. 2) ---------------- R = g1 A wideband transmission line transformer. This allows easy v 1 + ------ response shape measurements with minimal transformer 2 rolloff effects. The two other output options include a 200 And R will be set to the sum of R and the R element to load differential to single ended path for OIP3 measurements g2 g1 S get balance in the differential feedback loop. The blocking with lighter loads and a differential 50 output path for direct caps are set as necessary to pass the lowest frequency of measurement of just the ISL55210 response using a 4 port interest. Using this simple blocking cap approach places all the network analyzer. DC operating voltages on the two input and output pins at the While the board itself is completely flexible for input default internal Vcm voltage = 1.2V for the ISL55210. This impedance setting and gain by changing only 4 resistor circuit can implement a DC-coupled signal path if the supplies elements, the default circuit for the board as delivered are set to +2.5V and -1.2V to keep the I/O headroom implements a 50 input match with a 16.4dB gain to the FDA requirements satisfied but that is not supported by this board output pins. The simulation circuit for this configuration is where only AC-coupled designs can be tested. Running a set of shown in Figure 1. parametric frequency response curves vs Av to the FDA output pins in ADS, with the input impedances targeted to match an R = 50 , gives the expected parametric response curves of S The operation of this circuit is well modeled using the Spice Figure 2 where the blocking capacitors have been set to 10F 1 model for the ISL55210 within the free iSim PE simulator . in simulation to eliminate their effect. The board input is C2 while the output is the balun output pin. FIGURE 1. iSim PE SIMULATION CIRCUIT FOR THE DEFAULT CONFIGURATION OF THE ACTIVE BALUN EVM AN1831 Rev 1.00 Page 1 of 11 May 7, 2013ISL55210-ABEVAL1Z largely through feedback requiring very low input resistors to 40 achieve the match. This gives input noise figures <7dB for gains 34dB 32dB > 18dB using the low noise ISL55210 in this circuit. The closed 35 30dB 28dB loop bandwidths are also extended over typical approaches with 30 3 a resistor to ground since the noise gain will be 1 + Av/2.The 26dB bandwidths at lower gains do not strictly follow a gain bandwidth 25 product type response showing significant bandwidth extension 20 at lower gains due to reduced phase margin effects. 15 The 16.4dB gain used as the default setting for the active balun 22dB evaluation board simulates very flat through 1GHz with 10 24dB 20dB 16dB 18dB approximately 4GHz F-3dB bandwidth for the response to the 14dB 5 ISL55210 outputs. In fact the board shows a response shape set GAIN = xdB 0 by the blocking capacitors on the low end and the output 1 10 100 1000 transformer on the high end. All of these elements are well FREQUENCY (MHz) modeled and running a response simulation of Figure 1 gives the FIGURE 2. RESPONSE CURVES vs A FOR THE ACTIVE BALUN WITH V expected shape of Figure 3 where the 6dB matching loss at the 50 INPUT output is modeled but the insertion loss of the output transformer is not. That insertion loss is specified as 0.4dB TABLE 1. (hence the 16.4dB gain setting) but measures only 0.2dB. The SIMULATED measured response on the board will have a nominal midband A GAIN A GAIN BW V V gain of 10.2dB with a response shape very close to this V/V (dB) Rf Rg1 Rg2 (MHz) simulation. The simulation is predicting -0.5dB response flatness 5.00 14 160.71 14.29 64.29 3900 from approximately 2MHz to 500MHz. 6.29 16 195.31 12.06 62.06 3155 10.4 7.92 18 238.04 10.08 60.08 2037 10.2 9.98 20 291.06 8.35 58.35 1200 10.0 9.8 12.56 22 357.12 6.87 56.87 811 9.6 15.81 24 439.67 5.61 55.61 588 9.4 19.91 26 543.07 4.57 54.57 445 9.2 25.06 28 672.79 3.70 53.70 330 9.0 31.55 30 835.72 2.98 52.98 260 9.8 9.6 39.72 32 1040.51 2.40 52.40 205 9.4 1M 2M 4M 10M 20M 40M 100M 200M 400M 1G 50.00 34 1298.08 1.92 51.92 158 FREQUENCY (Hz) FIGURE 3. FOR THE CIRCUIT ON THE ACTIVE BALUN BOARD OF This sweep is stepping the gain up in 2dB steps showing the FIGURE 1 required exact resistor values and the expected F-3dB bandwidths. Since the ISL55210 is a 4GHz gain bandwidth VFA based FDA, the response bandwidth decreases with gain. The Full Evaluation Board Schematic significant benefit of using the common mode feedback loop to The schematic with power supply decoupling and the optional set the input match is the vastly reduced resistor values. The output interfaces is shown in Figure 4. action of the common mode loop develops the input impedance AN1831 Rev 1.00 Page 2 of 11 May 7, 2013 GAIN (dB) dBV at /dB VOUT