Hi, I have a few questions about the reference signal requirements:
Can the synthesizer accept logic type (CMOS/LVPECL/LVDS, etc…) square waveform reference signal or does it have to be clipped sine wave as the datasheet suggests?
I am using multiple RFFC507x ICs on the same PCB. What is the best distribution strategy to provide the reference signal to each IC (power divider or fanout buffer or something else) if I have a single input to the board?
What is the input impedance at the reference input pin (REF_IN)? Is it high impedance or 50 ohms?
A clipped sine wave is recommended as it has the best trade off of fast edges for phase noise with lower harmonics than a square wave. The level is important, as going above the max of 1500mVp-p in the datasheet will create higher clock and clock harmonic leakage at the mixer output. So yes you could use a square waveform, as long as the level is within the range specified in the datasheet, and taking out harmonics may help.
No particular recommendation for distribution network. It would depend on the clock frequency and level available. A resistive splitting network may be OK, or you could use some active buffers. Good to have isolation between the RFFC507x devices to prevent any unwanted interactions. Active buffers would be the lower risk approach.
The REF_IN pin is high impedance. It was designed to be used with TCXOs that can drive load of 10KOhm//10pF. It is more like several 100KOhms//1-2pF.
Note that if you have multiple RFFC507x devices on one PCB it is important to keep them isolated. As well as isolating clock, digital, and supply lines, this includes use of shielding to prevent any radiated coupling. This is especially important if any of the devices are tuned to the same LO frequency (or multiples of) to prevent injection pulling effects.
If switching between mixers (eg TDD operation) then the two mixers can be set for different LO frequencies, there are two sets of PLL frequency register banks. If operating in ‘Full Duplex’ mode where both mixers are on simultaneously then they will share the same LO frequency, since there is only a single PLL/VCO in the device.
Do you have any simulation results that show how the parasitic capacitance and inductance at the balanced input and output pins of RFFE5072 vary as a function of frequency? My operating frequency range can go up to 6 GHz?
Is there a significant difference in performance between RFFC5071 and RFFC5072 especially in terms of spurs and conversion loss? Given the fact that there is only one PLL in the IC, I don’t have a need for the 2nd mixer. So, I’ll be using RFFC5072 single mixer IC in my design.
What are some strategies to lower spur levels? Initial measurements (on the original eval board with no hardware modifications) show high LO feedthrough, as well as 2LO or 4 LO (i.e., VCO freq) on mixer 2 output.
No don’t have this information. Qorvo has just tested the device in various configurations up to 6GHz. See app notes on the web-site ref1 and ref2.
They use the exact same die. The RFFC5072 has one mixer bonded out, RFFC5071 has both mixers bonded out. Performance should be the same.
Unfortunately not much you can do to lower the 2LO and 4LO levels at the mixer output, apart from trying to filter them out. Mixer bias current has a small effect. The baluns and matching on the mixer output will also have an effect, dependent on the CMRR at the harmonic frequency. See typical mixer spur table to show levels to expect. RFFC207x_507x Example Mixer Table.pdf (42.7 KB)
Hi Chris, a follow-up question on #1: What value of inductance do you suggest at mixer 2 output if my output frequency range is 2.5 to 3.5 GHz? In one of your links, mixer output uses a 1:4 balun and inductor of 9.1 nH (at 2.45 GHz). In the other link, mixer output uses a 1:2 balun and inductance of 33 nH (for 3.1 to 4.8 GHz frequency range). I am planning to use the 1:2 balun. Should I put any inductance on the balanced side?
The 33nH is used for IF output frequency of 1000MHz. For 2.5 to 3.5GHz that’s quite a wide bandwidth for this resonant match with a 4:1 balun, but you could try around 7.5nH or 8.2nH as a starting point. If you drop the output balun down to 2:1 then this improves bandwidth, at expense of conversion gain.
Hi Chris, thanks for the quick response. In my application, amplitude and phase variation of conversion gain over frequency, spur levels, and phase noise are the most important specs. Absolute conversion gain is not as important as long as it is not ridiculously high (I would say anything greater than 12 dB loss will be problematic). Given these requirements, what value of inductance would you suggest? Please keep in mind that we don’t have in-house capability to solder PCB components at this time. Rather we are sending the board to an outside shop, that does the rework and charge us for every job. So, if possible, I would like to minimize trips to the outside shop. With the above requirements in mind, what inductance value would you suggest? We can experiment a few times, but not too many times. Thank you so much for your understanding and support. I appreciate it a lot.
It’s a bit difficult to be sure of the inductor value without trying it out. The layout and balun used will also have an effect. To resonate in the center of your band at 3GHz the inductor needed should be around 5nH. For your wide bandwidth you have a couple of approaches; use a 2:1 balun and shunt inductor and hope you get the bandwidth, or use a 1:1 or 2:1 balun and just accept the natural roll off. I found that for higher frequencies just using the balun on the output unmatched worked fairly well, but a bit of a compromise. This may be best if you need to minimize amplitude and phase variation.
Hi Chris, the RFFC507x datasheet says that the LO frequency can be set from 85 MHz to 4.2 GHz. I need to set an LO frequency of 4.5 GHz. Can this be done with this part? The reason I ask is that the datasheet shows a phase noise plot for a VCO frequency of 5.2 GHz (page 15). With the LO divider set to 1, can’t we set an LO frequency higher than 4.2 GHz?
Hi Chris, just wanted to follow-up on this. The 4.2GHz max LO is becoming a major hurdle for me to use this mixer in my design. Can you suggest some ways in which I can overcome this problem?
Slight correction from the previous post => I actually need to go up to 5GHz LO.