Making chirped radar signals (any radio guys around?)

[Sternkucker]

Hey guys,

There are these weird new signals in astronomy called "fast radio bursts". Lately there was some discussion whether they are cosmological or man-made interference. I'd like to try and see how difficult it is to generate such a signal with standard electronics. But I need your help!

The radio signal I like to produce is in the microwave L-Band at 1200-1600MHz. Power requirements are very low. Most important: The signal needs to have an exponential time-frequency delay, so that lower frequencies are emited later. The time delta between 1600MHz and 1200MHz should be about half a second.

I attach an image of the prototypical burst I like to reproduce.

I understand that I probably need a "waveform generator"? The thing it must make is then called an "exponential frequency modulation (EFM) down-chirp"?

Perhaps you can guide me a bit how to start. Are there perhaps any emiters that I can readily buy, connect to the USB port of my PC and start generating pulses? With the help of some software which I can program to generate a specific profile?

Thanks for your help!
Michael

[Fank1]

Not too tough.
You need a 1.2 GHz to 1.6 GHz oscillator with varactor tuning.
Put a large cap in parallel with the tuning voltage.
Then put a resistor to a mosfet drain in parallel with the cap.
Charge the cap to full voltage, 1.6 GHz.
Remove the charging voltage.
Turn on the oscillator and at the same time turn on the mosfet.
The cap voltage will decay, exponentially, and the frequency will drop exponentially.
The exact time and curve will depend on the varactor curve and the values of the cap and resistor.

[w2aew]

Not too tough.
You need a 1.2 GHz to 1.6 GHz oscillator with varactor tuning.
Put a large cap in parallel with the tuning voltage.
Then put a resistor to a mosfet drain in parallel with the cap.
Charge the cap to full voltage, 1.6 GHz.
Remove the charging voltage.
Turn on the oscillator and at the same time turn on the mosfet.
The cap voltage will decay, exponentially, and the frequency will drop exponentially.
The exact time and curve will depend on the varactor curve and the values of the cap and resistor.

And if designing a varactor tuned oscillator at that frequency range, with that kind of tuning range, isn't practical for you, then you could do it at a lower frequency with a narrower chirp, and use a multiplier and filter to bring it up...

[Marco]

The only thing you can buy off the shelf is a sweep generator AFAICS. Plenty on ebay if you have 1000+$ to spend.

Not necessary to make your own varactor tuned VCO BTW. You can just buy one for 30$, it won't be very linear with voltage though.

If you need linearity, the ADF435X can do sweeps I think and you can buy devices with them really cheaply on ebay (just look for 4.4 GHz generator). So you might be able to sweep the ADF435X and use a PLL to control the VCO.

[Sternkucker]

Thank you so much for your help!

The ADF4351 looks very good. I downloaded the software and it can generate sweeps, although they are linear and not exponential. But this might be a software issue and with other tools it might be possible (ADIsimPLL)? Or are there products from other manufacturer that can do it off-the-shelf?

I did not understand why I would need a PLL to control the VCO?

When I have a sweep generator, what do I need to emit the signals? Can I connect any antenna/wire to the device, directly? Or would I need another device in between?
Sorry for the basic questions, I promise to learn quickly 

Finally, if I want to test the output, what commercial device can receive and visualize the signal? Preferably also for the computer (and not stand-alone oscilloscope). The device would need to generate a time-frequency plot as in my figure, with time resolution ~1millisecond and frequency resolution ~10MHz (or better, but that would be sufficient).

[Dago]

Thank you so much for your help!

The ADF4351 looks very good. I downloaded the software and it can generate sweeps, although they are linear and not exponential. But this might be a software issue and with other tools it might be possible (ADIsimPLL)? Or are there products from other manufacturer that can do it off-the-shelf?

I did not understand why I would need a PLL to control the VCO?

When I have a sweep generator, what do I need to emit the signals? Can I connect any antenna/wire to the device, directly? Or would I need another device in between?
Sorry for the basic questions, I promise to learn quickly 

Finally, if I want to test the output, what commercial device can receive and visualize the signal? Preferably also for the computer (and not stand-alone oscilloscope). The device would need to generate a time-frequency plot as in my figure, with time resolution ~1millisecond and frequency resolution ~10MHz (or better, but that would be sufficient).

If you want non-linear sweeps then you just command the chip to change frequencies in a non-linear fashion. I doubt the whatever software you downloaded has too good support for sweeps. AdisimPLL is a software for designing PLL loop filters.

VCOs usually respond non-linearly to control voltage changes, which is the reason why you use a PLL to linearize the response. PLL also provides other benefits such as phase-locking, better phase noise within loop filter bandwidth etc.

For emitting it totally depends on what you want to do. If you need very low range you could connect the generator straight to an antenna. Do note that transmitting a signal like this is totally illegal, even if you have a ham radio license (which you need to transmit basically anything).

For visualising the signals you would probably need to use a spectrum analyzer (or maybe SDR-radio if you do not have thousands for a spectrum analyzer).

[Sternkucker]

Mhm, what do you think about doing it all in software, like HackRF? Would that be good enough?

[tggzzz]

The ADF4351 looks very good. I downloaded the software and it can generate sweeps, although they are linear and not exponential. But this might be a software issue and with other tools it might be possible (ADIsimPLL)? Or are there products from other manufacturer that can do it off-the-shelf?

When I have a sweep generator, what do I need to emit the signals? Can I connect any antenna/wire to the device, directly? Or would I need another device in between?

I don't know the details of that software, but "linear" will mean that if you change the control voltage linearly then the frequency will change linearly. Hence if you change the control voltage exponentially, then the frequency will vary exponentially (within the range and rate-of-change of the device, of course).

Why do you actually want to emit the signals? Usually the output of any test equipment will be connected to the unit-under-test by a cable. If you want to test the front end of a very low noise radio then you will also need to have an electrically very quiet environment, which can be difficult to achieve.

In addition, be very careful that your signal cannot possibly interfere with any legal radio signal. I suspect ensuring that requires significantly more RF experience than you possess.

[Sternkucker]

The ADF4351 software allows to set the frequency (in MHz) and steps (in KHz) per time, i.e. to make the device send at 1400MHz at t=0, and 10 milliseconds later change to 1399.9MHz. I guess the calculation of voltages is invisible to the end user. But the steps and times are linear. I'm perfectly fine with calculating the exponential time/frequency curve myself (I do this every day), but I would need to feed the curve into the software.

Be assured that I do NOT plan to do anything illegal! I'm in astrophysics research, and we will only perform experiments in a lab environment, after it has been authorized by the authorities. This is to find out possible locations of interference. In radio observations at Parkes Radio Telescope (Australia), Effelsberg (Germany) and Arecibo (Puerto Rico), bursts like the one in my picture are regularly received. We want to understand possible locations of man-made interference. This is to find out if the signals are in fact man-made or come from space.
For example, the microwave oven at the visitor centre of the Parkes radio telescope caused similar signals which were received in the telescope, through reflection of the metal strut structure. We like to explore other sources and their origins.

My problem is that I'm more of a software guy, although I have done some basic electronics and soldering before. So I'd love the adventure of doing this with my own hands, instead of transfering the idea to some "proper" research group which would probably have no time at the moment.

Is this HackRF stuff useful for my purpose?

[Marco]

I did not understand why I would need a PLL to control the VCO?

The ADF4351 is a square wave generator, you could throw a low pass filter over it, but it will never be super clean.

If you use a normal sine VCO to generate the signal you can use a PLL to lock it to the ADF4351's frequency, this gives you accurate closed loop control over frequency while still having a good quality sine.

[Dago]

The ADF4351 software allows to set the frequency (in MHz) and steps (in KHz) per time, i.e. to make the device send at 1400MHz at t=0, and 10 milliseconds later change to 1399.9MHz. I guess the calculation of voltages is invisible to the end user. But the steps and times are linear. I'm perfectly fine with calculating the exponential time/frequency curve myself (I do this every day), but I would need to feed the curve into the software.

Be assured that I do NOT plan to do anything illegal! I'm in astrophysics research, and we will only perform experiments in a lab environment, after it has been authorized by the authorities. This is to find out possible locations of interference. In radio observations at Parkes Radio Telescope (Australia), Effelsberg (Germany) and Arecibo (Puerto Rico), bursts like the one in my picture are regularly received. We want to understand possible locations of man-made interference. This is to find out if the signals are in fact man-made or come from space.
For example, the microwave oven at the visitor centre of the Parkes radio telescope caused similar signals which were received in the telescope, through reflection of the metal strut structure. We like to explore other sources and their origins.

My problem is that I'm more of a software guy, although I have done some basic electronics and soldering before. So I'd love the adventure of doing this with my own hands, instead of transfering the idea to some "proper" research group which would probably have no time at the moment.

Is this HackRF stuff useful for my purpose?

I think we would need more details on what kind of signal you want to produce to be able to more accurately say what you need.

If you have no more specific requirements (phase-locking, phase noise, spectral purity, transmit power, very fast sweep times etc.) other than "half a second slow exponential sweep from 1200 to 1600 MHz" then you can create that kind of signal in many many different ways (multiply low-frequency PLL sweep signal, use a synthesizer chip like ADF4351, use a sweep generator, modify some kind of radio transmitter for the job etc.) and I'm sure you could use a hackrf for it too. Those requirements are not special or hard to achieve in any way.

[Sternkucker]

Other requirements are only optional, but it would be cool to include:

• Pulse width (at any given frequency): 1..5ms
• Pulse broadening, so that the pulse width is wider at lower frequencies. How much does't matter much, something like 2ms at 1500 MHz and 3ms at 1200 MHz would do
• Pulse profile at a given frequency can be a rectangle on the rising side, but should decay exponentially on the falling side
• Transmit power can be (very) low, as these tests would be observed with a 100m dish in close distance (500m)
• Sweep times: 100ms to 1000ms adjustable for the range 1600 to 1200 MHz would be perfect

[Marco]

The HackRF does look really nice, I didn't remember it being that cheap ... can't really bodge anything for that price, unless your time is truly cheap. I see some python code out there to just directly write to the frequency generator registers, that should suffice to script your sweep.

[G0HZU]

You could try using a MiniCircuits JTOS1650 VCO and something like an Arduino board with some DAC outputs.  i.e. store the tuning slope and tuning rate in the Arduino board so it outputs a linearised (or a custom/ragged/noisy) tuning ramp from a DAC driving the VCO.

To refine it you could include an IQ modulator inline to give you some ability to refine the signature of the signal to make it more realistic.
i.e. drive the IQ modulator with the Arduino DACs to play around with phase or amplitude (or OOK pulsing) as it chirps. There are loads of low cost IQ modulator chips available cheaply from AD or RFMD or Hittite etc etc...