Any RF wideband transceiver which supports fast frequency hopping of 1000 hops/S

[sangar]

Hello All,

I am looking for RF wideband transceiver with the interest of band of 4-5 GHz and it should be capable to achieve 1000 frequencies hopping in one second. Initially, I analyzed AD9361 and AD9371 for this purpose. But, after discussed with Analog devices, it seems to be difficult to achieve with them. So, any of you know the fast frequency hopping RF transceiver, please share with me...



Thanks,
Sangar

[rhb]

You might want to consider using a pair of ADF4351s.  A brief perusal of the datasheet suggests that you can change frequency in <1 mS. So if you heterodyne a pair of them you should be able to cover the range of interest.

Another possibility is to use an AD9851 and multiply the output frequency before mixing with an ADF4351.  That will certainly meet your hop requirement easily.  But at the price of more difficult filtering to suppress harmonics.

Otherwise, an FPGA based AWG with multiple outputs multiplied to span 1 GHz and then mixed to the target band.

After considering it a bit more, 3-4 ADF4351s and some RF relays.  Use one at a fixed frequency and hop the others.  Switch between them if you can't get just one to lock fast enough after a hop.  Because they are square wave outputs in the 2-4 GHz range, it's easy to filter out the harmonics.

[sangar]

Hi,

AD4351 is the solution with AD9361?

[rhb]

What did AD suggest doing?  What is the goal?  Why this combination of specs? 

The AD9361 will take an external LO and the AD9910 will cover 400 MHz and hop much faster.

Fast hopping is easy over narrower ranges and broad ranges are easy hopping more slowly.   It's hard to see a justification for such slow hopping over such a wide range.  What do you expect to gain?

You should be able to get very fast hopping over 1 GHz range by PIN diode switching among 3 ADF4351 + AD9910 LOs into the AD9361 external LO input with each ADF4351 + AD9910 set to hop a different segment of the band. But that's going to get very difficult to implement.

I suggest using the ADIsimPLL and ADIsimRF programs to simulate what you want to do.

[David Hess]

Is the limitation on how quickly the local oscillator frequency can be changed due to the frequency synthesizer or the digital interface?

Details are scarce in the datasheets but they appear to both use a DDS to drive a fractional-N PLL frequency multiplier which should inherently have a fast settling time because the high PLL reference frequency allows for a wide PLL bandwidth.

Using an external synthesizer with a fast settling time as described above for the local oscillator should solve this problem.  This is especially the case if the frequency step size is sacrificed allowing for a higher reference frequency for the PLL frequency multiplier.

[sangar]

Hi,

Our system should be capable to hop between 6000 channels in 4.4 - 5 GHz band at 1000 hops/sec rate, here, the channels are separated by 100 KHz. Hopping can happen from 1st channel (at 4.4 GHz)to the 6000th (at 5 GHz) channel and we are expecting that the settling time of the device for the new channel should not exceed more than 20us...
 
AD9361 have fast lock profile method feature by which we can create synthesizer coefficients for 6000 channels. During the  application is running, the user program can load the respective coefficient to the synthesizer register in AD9361 when we want to hop to that frequency.

As per AD suggestion, one can achieve this less than 20us settling time by the above method only if frequency separation between the new hopping frequency (let us say 4400.1 MHz) and last calibrated frequency( calibrated at device initialization, let us say 4400 MHz) is less than 100 MHz. If it is above 100 MHz (that is the last calibrated frequency is 4400 MHz and the new hopping frequency is 5000 MHz ), the device has to perform DC offset calibration, TX quadrature calibration and Rx quadrature for that frequency(5000 MHz). Since these calibrations take at least 30 to 40ms, we are violating our 1000 hops/sec.

Please suggest...

Thanks,
Muthu

[David Hess]

Oh, well, that explains that.  I did not know direct conversion receivers had to recalibrate DC offset over such a small frequency range but it does make sense.

What occurs to me is to keep the AD9361 or AD9371 but operate it at a fixed intermediate frequency which is convenient for bandpass filtering while using an external synthesizer and mixer to frequency hop.  I think some WiFi transceivers do this for improved performance.

[sangar]

Hi,

What occurs to me is to keep the AD9361 or AD9371 but operate it at a fixed intermediate frequency which is convenient for bandpass filtering while using an external synthesizer and mixer to frequency hop.

I don't understand the above. Could you explain?

Thanks,
Muthu

[David Hess]

What occurs to me is to keep the AD9361 or AD9371 but operate it at a fixed intermediate frequency which is convenient for bandpass filtering while using an external synthesizer and mixer to frequency hop.

I don't understand the above. Could you explain?

The idea here is to make a superheterodyne receiver with the AD9361 or AD9371 acting as everything past the first mixer replacing all of the IF and demodulation circuits.  The synthesized local oscillator for the first mixer now controls the input frequency and can be made as fast as necessary to support the frequency hopping rate while the AD9361 or AD9371 operates at a constant input frequency.

I am not sure what your channel bandwidth is but if an IF filter is used between the mixer and AD9361 or AD9371 to remove the image produced by the mixer and increase selectivity and the right IF frequency is selected, then a common and inexpensive IF filter may be available.  Given your input frequency range, you probably want an IF frequency higher than 600 MHz to prevent image problems.  A SAW filter intended for cell phones seems suitable.

[hendorog]

For the LO, if you are open to an old benchtop system, then this family of synths will hop fast enough:

https://www.ebay.com/itm/Comstron-Corporation-FS2000-Frequency-Synthesizer-W-OPT-4-105/142443634796?hash=item212a4d606c:g:ds4AAOSwRvdZZk6k

I believe it has BCD digital inputs for frequency selection.

You will need a doubler of course.

[hendorog]

Another option which might work - using the frequency ramp function and RampDIR/RampCLK inputs:

http://www.ti.com/lit/ds/symlink/lmx2572.pdf

[rhb]

Test the ADF4351 and see if it will hop over the primary 2-4 GHz range fast enough with a comfortable margin (i.e. 1500 hops/s or faster).

If so, then pick an IF of say 2 GHz for the 9361 input, hop the 4351 from 2.4-3 GHz with a 3 GHz lowpass filter between the 4351 and the mixer.  You'll need to do this on both Rx & Tx.

If the ADF4351 won't hop fast enough then design a frequency plan so that you can mix the output of a DDS with and ADF4351.  Hop a few frequencies with one 4351, then a few frequencies with the other.

However, if the AD9361 will hop fast enough over 100 MHz, you could use two ADF4351s and restrict your hops to a 100 MHz range while the ADF4351s settle and leave out the external DDS.

The ADF4351s have several advantages.  Their primary range is in the 2-4 GHz region and they put out a square wave, so the significant harmonics start at 6-12 GHz.  They're also cheap.

[sangar]

Hi,

Thank you all for your input...I have one more question about AD9361 function...

My over the air data rate will be 2,4,6,8,10 and 12 Mbps. For these data rates, how to calculate the digital baseband rate of IQ samples which will be fed at AD9361 input...We will use 16-QAM for our application. The datasheet says that Min and Max digital baseband rate of IQ samples is 0.4 to 61.44 MSPS and RF bandwidth is 200 KHz to 56 MHz. 


Thanks,
Sangar

[sangar]

Hi rhb,

Out of 1ms, I have only 20us to hop to a new frequency and get settle in that frequency. Remaining time for data transmission and reception. So, Does AD9361 hop to a new frequency and ready for transmission/reception by 20us?


Thanks,
Muthu

[coppercone2]

don't you need a YIG to filter after the multiplier if you use a slow DDS chip?

[sangar]

Hi,

As per rhb said, I can use my AD9361 at a fixed intermediate frequency says 2GHz...Then, AD9361 will give a modulated output at 2GHz...This modulated output will be mixed with external ADF4351 synthesizer's output which ranges from 2.4 - 3 GHz so that I will get my band of 4.4 GHz- 5 GHz...This is fine and I can understand it at all...

If the ADF4351 won't hop fast enough then design a frequency plan so that you can mix the output of a DDS with and ADF4351.  Hop a few frequencies with one 4351, then a few frequencies with the other.

Reminder:
Could anyone answer for the above?

However, if the AD9361 will hop fast enough over 100 MHz, you could use two ADF4351s and restrict your hops to a 100 MHz range while the ADF4351s settle and leave out the external DDS.

In the second choice, he said the above...I don't understand the purpose of DDS here? Could you tell me why DDS and the purpose of this here?

[sangar]

Hi,

@David Hess

Could you answer the above?


Thanks,
Muthu

[sangar]

Hi,

I have seen the following lines in Analog Devices forum where he was tried the same approach for fast frequency hopping...

During frequency hopping our LO will settle within <1uS, but still there will be phase dis-continuities due to frequency change.

I don't know, there is an LO which can settle within <1uS for frequency hopping...As of now, the concern is that phase discontinuity...Does it matter for my application of frequency hopping in which Our LO might take 60us to settle?...

Please suggest...

Thanks,
Muthu

[David Hess]

Hmm, the PLL loop bandwidth which controls the settling time is limited by the reference frequency so a channel spacing of 100kHz will be a problem for a settling time of 1 microsecond.  (I am not sure how this works with fractional divider PLLs.) There is an  easy solution though.

Use two local oscillator sources and switch between them while alternately adjusting their frequency.  Now the settling time is only limited by the hop time.

Or alternatively use a DDS to generate a higher reference frequency for the PLL. The fine tuning resolution of the DDS allows a small channel spacing even with the high PLL reference frequency.

[sangar]

Hi,

In the transmitter side, I need to put IF filter between AD9361 and Mixer to pass our 2GHz IF modulated signal from AD9361... Wha should it be? Low pass filter (Or) Bandpass filter?...


Please suggest...

Thanks,
Muthu

[dazz1]

Hi
Why do you need to hop?
Do you have design control of the other end, or is that a given?
If you are trying to track hopping at the other end? How do they do it?
Are you trying to link with multiple TX/RX devices at the other end?

Is it for security? If so can the signal be encrypted?
Is it for signal path continuity (avoiding noisy or fading channels).  If so, could you use frequency (channel) diversity.  Simultaneously Tx/RX the same data across multiple channels.
Is it for stealth?  Could you use spread-spectrum?

[lugaw]

Transmitting at all channel at the same time and filtering it on the output by YIG or diode switch.  Has this been done before?

[dazz1]

Hi
Simultaneously transmitting the signal across multiple channels (frequency diversity) has been used for a long time, particularly with HF.  There are modern alternatives.

The problem you will need to solve is the signal fade and phase shift with different frequencies following different paths. 
Likely the most significant constraint you will have is legal.  If you are an individual, it is unlikely that you have purchased the rights to the frequency band you are looking at.   

I don't think you have said why you want to do all of this.