As per your datasheet ADRV9008-1 Image Rejection spec is given as 75dB with QEC active, within 200 MHz receiver bandwidth. Also the RECEIVER DC OFFSET (dBFS) value over frequency range is given as >-90dBFS over temperature. Now when we use ADRV9008-1 hopping requirement and we need to get Receiver Image Rejection spec > 70dB and RECEIVER DC OFFSET (dBFS) >-80 dBFS. 1. What is the QEC calibration approximate settling time required for the frequency range from 100MHz to 3GHz range? 3. We know PLL settling time is around 70uS then what is the approximate minimum settling time need to be given for QEC calibration to achieve above spec over our specified range.Note:Our Hopping may be anywhere from 100 to 3000 MHz, But we will use 200MHz bandwidth after every hopping frequency.
What is the frequency by which you want to hop?
If you are hoping by more than 100MHz then you need to perform QEC and RFDC offset init calibrations which will take approximately 6msec.
If your hoping frequency is less than 100MHz then only tracking calibrations needs to be run(time required for tracking calibrations is not fixed as it runs continuously to minimize the gain and phase error) for better performance(if you want to achieve an image rejection of 75dBc).
Thanks for your detailed response.
Please confirm our below additional queries,
1. Since the receiver support instantaneous bandwidth of 200MHz, Then why your are limiting hop frequency above 100MHz to perform QEC and RFDC offset init calibrations?
2. If QEC and RFDC offset init calibrations will take approximately 6msec to achieve image rejection of 75dBc, Then if we hop after 3mSec instead of waiting upto 6msec, Then how much image rejection (in dBc) is possible in our 3msec calibration time approximately?
1. Yes. DC Offset calibration treats any time-invariant signal as DC and corrects for it. With the modulated signal to reduce the impact of DC, one option is to Shifting the LO slightly, filtering out the DC leakage and shifting the carrier back using NCO’s in Baseband/FPGA . Again depends upon the bandwidth of your signal.
2. What is sweep rate/ dwell time?. We manually swept from a signal generator and not seeing any issue.
If the timing for QEC tracking cals are not met image level will degrade.
3. Can you please share more details on this? Can you please share a screenshot of receive capture and your config. What's your input signal and level? We are trying to recreate the issue in our lab?
1. DC offset we will try to avoid as per your suggestion.
2. We are not sweeping at very fast, It was manually changed from one frequency to other frequency in signal generator and changing time may be considered more than 1 second and kept at same frequency for long time.
We have attached the 999MHz RF input from sig gen for 1000MHz LO and our fund is -1MHz whereas image +1MHz is <50dBc only. If we change 998MHz the image is improved above 75dBc. This is occurring at randomly not continuously at 999MHz. Also attached our GUI configuration image.
3. 3rd Harmonics issues we will comeback with screenshots in details by soon.
2. How long this phenomenon is seen? Is it stuck with 50 dBc for long?
When you sweep frequency sometimes the image level is 50 dBc and when the QEC converges it goes to 75 dBc.
Yes, It is stucked to 50dBc alone and not improving after few minutes until I change it to new frequency in signal generator.
Also this phenomenon is random not occuring at same spot repeatedly. In total 5 sweep one sweep approximately we are seeing this issue
This looks strange.
I hope while programming the EVB you are turning RF OFF in your signal generator.
When it gets stuck, can you try whether turning RF OFF and ON improves?
I am unable to understand your above points.
1. Do you meant to say during device programming or LO programming the input signal to be feed continuously from signal generator for proper QEC calibration?
2. If yes, How this kind of scenario expected in real-time? during programming input signal should present from antenna is not practical case for ideal QEC calibrations?
Please try to check in your eval-boards, This issue is coming randomly in many LO frequencies while sweeping the RF signal manually (very slowly).
1. During init cal, there should not be any tone/signal seen at RF Input port. I was asking whether sig gen RF was turned off during init call.
We tested in our lab and we are not seeing any stuck issue.
I hope you are using the latest GUI from our web.
Can you give the procedure of your testing to reproduce the issue in our lab?
We have once again verified the issue and we observed this issue is happening only -5dBFS or above input power levels. When we reduce the power level below -8dBFs we are not seeing this issues.
Also as per your recommendation we have check the device performance by signal generator OFF condition during device programming or calibration. But still the issue is there when we give signals above -5dBFS. Also this is not suppressed after Power OFF & ON the signal generator output.
Now please check this phenomenon in your Lab and confirm.
What is the RX gain that you are setting when the signal at RX input is given at -5dBm?With full gain, the maximum input level that the RX path can handle without saturation is -8dBm.
We tested the same thing in lab and the image levels are always below 75dB.