While using the ADL5375, carrier feedthrough is present in the RF output. what are the optimization techniques are there to nulling the carrier feedthrough? and one time optimization can be OK for broadband application?
The attached application note will, I think, address the questions you are asking.
Thanks for your reply. I understand the optimization technique. but my concern is that, In a broadband real time application optimization is required everytime at every frequency or only one time optimization is sufficient? and what will be the values of the resistors for optimization mentioned in the attachement?
This is an interesting question that is not that easy to answer. As I understand it, you would like to do a one-time calibration at one frequency and then operate the system across a broad frequency range. We often refer to this as set-and-forget. Whether you can do this depends on how good you want the post-calibration LO leakage to be across frequency. As you can see from Figure 7 in the ap note, after you move off the nulling frequency of 1910 MHz, you quickly see LO Leakage and sideband suppression degrade. While they do degrade, they still are still better than the uncalibrated performance (although I would admit that in the case of the sideband suppression, at the band edges, it almost goes back to the un-calibrated level).
If you were willing to calibrate in 5 MHz steps, then you could probably hold the LO leakage below -60 dBm. However, if you band is, say, 100 MHz wide, that is 20 cal points which is going to drive up your calibration time.
In general, I find that the quality of your post-calibration performance will depend a lot on your pre-calibration performance. By that, I mean that if you have poor LO leakage at the outset, you can certainly calibrate it away. However, you will find that it comes back very fast as you move off the calibration frequency.
The elephant in the room here may be temperature. Take a look at Figure 8 in the application note. In this case we calibrate LO Leakage at 1842 MHz. As you change the frequency, the LO leakage degrades back to -60 dBm (at two frequencies). However, notice what happens over temperature. At +85 degC and -40 degC, LO leakage degrades to the -50 to -55 dBm range and you can no longer even tell at what frequency you calibrated.
As you note, another alternative is real-time compensation where you have some kind of an observation receiver which is continuously monitoring and adjusting the quality of the transmitted signal.
Regarding your second question, the resistors that connect the Aux DACs to the main signal path set the range or adjustment. They don't perform the calibration per-se; they just define the maximum amount of compensation voltage that you have at your disposal.
Using AN-1039 as a reference, what is the recommended configuration if an AD9963 (instead of the AD9788) is driving an ADL5375 and instead of a passive BB filter, an ADRF6516 programmable BB filter is used? Can the AUX DAC be used in the same way for carrier nulling but with the resistors tapped into the junctions of the ADRF6516 outputs/ADL5375 inputs? What are the recommended gain setting for the ADRF6516? What is the recommended DAC output resistor network?
As a starting point, figure 64 of the AD9963 data sheet shows AUXIO connections at the Tx output needed to do LO feedthrough cancellation.
Variable Gain Amplifiers like the ADRF6516 are typically placed between the output of a QMOD like the ADL5375 and used to drive a Power Amplifier. The baseband inputs of teh ADL5375-05 need to be kept in the 0 to 1 volt range.
ADL6516 is indeed intended for use as a baseband VGA/Filter, not at the modulator output. It is specifically designed to be placed between the output of an IQ Demodulator and the input of a dual baseband ADC. However, I guess that you could also use it between a DAC and an IQ Mod as you are suggesting. However, as Larry notes, the common mode voltage going into the IQ mod should ideally be 0.5 V. The common mode output voltage of ADRF6516 can go down to 0.7 V. I think that the ADL5375-05 will still work with an input common mode voltage of 0.7 V but it's not optimal. Among other things, supply current will increase. Another option would be the ADL5375-15 IQ Modulator which needs an input common mode voltage of 1.5V.
Retrieving data ...