ADL5920: About Calculation of Return Loss

Hello Kakehi-San,

This takes a bit of explaining. Firstly, it's important to say that the ADL5920 measures the forward power (in dBm) and the reverse power, also in dBm.  The return loss voltage is the difference between these two values and has a unit of Volts per dB.  

To get the reading of return loss, there are two possible approaches.  You can digitize VRMSF and VRMSR and subtract them in the digital domain. Or you can digitize the differential VDIFF voltage (VDIFF is VRMSF-VRMSR). 

The two detectors that measure the forward and reverse power have a finite input range. This input range is approximately -20 dBm to +30 dBm.  But this range will get worse with increasing frequency (see Figure 9; VRMSR will have the same characteristic). 

In order to accurately measure the return loss of the load, the magnitude of the forward and reverse signals need to be within this operating range. In the plots your provided, as the input power is increasing, (VRMSF-VRMSR) and (VDIFFP-VDIFFN) are increasing even though the return loss of your load is not changing. In order to be able to observe the return loss accurately, you need to drive the amplifier harder.  If you were to keep increasing the output power, at som point (VDIFFP-VDIFFN) would flatten out.  VRMSF would continue to increase and VRMSR would start to increase such that (VRMSF-VRMSR) would become a fixed value. So this is why your data is suggesting the the return loss of the load is changing. In reality the return loss is fixed and it's a sensitivity problem. 

Please review this with the customer. Once this is clear, we can talk about how to convert the voltages into return loss in dB.

Best Regards

Eamon

Hi Tomohiro-San,

now that we have covered the operation of the device and limitations (caused by the finite  range of the forward and reverse power detectors, we can talk about how to calculate the return loss in dB. This is discussed on page 21 of the datasheet. The first thing that you do is calibrate the part. With the RFOUT port terminated with a 50 ohm load, apply two known power levels to the input (e.g. +10 dBm and -20 dBm). Measure the corresponding voltages on VRMSF. Calculate Slope and Intercept using equations (9) and (10) on page 21 of the ADL5920 datasheet.  You can then back-calculate the forward power, the reverse power and the return loss using equations 12, 13 and 14. For the Insertion Loss value that you need in these equations, you can get this from the datasheet (Figure 3). 

You can also calculate the return loss without calibration. In this case, you use the Slope and Intercept data from the specification table (e.g. 61mV/dB and -29.8dBm respectively at 100 MHz). This will however give you less accurate result because of part to part variations. Best Regards

Eamon 

Hi Tomohiro-San,

in a separate post I attached a spreadsheet. I took the data from your last post and used it to back-calculate the return loss. First we have to calculate the slope and intercept. I picked two points within the linear operating range of ADL5920. These points are +4.37 dBm and -10.5 dBm (pin). Note that below -10.5 dBm, the behavior of the forward power detector starts to become non linear (see chart2). Then we calculate the intercept. Now we have a mathematical equation for the average transfer function of the detector (i.e. VRMSF = Slope x (Pin-Intercept). We use this convert all of the  Vrmsf and Vrmsr voltages into equivalent forward and reverse power. Finally, the return loss is the difference between the two (Chart 3). 

So why is the measured return loss varying vs input power when the actual return loss is constant. There are two reasons. The forward and reverse detectors have finite range and you are also trying to measure a very good return loss. If you repeat the measurement with a return loss of 6 dB applied, you would get a much flatter result. Also, if you sweep the input power up to +30 dBm, you will see that the measured return loss starts to flatten out. 

Best Regards

Eamon