UWB Transceiver Design by Ali: Antenna & Transistor Issues

[ali.cmi]

Hello all,
Its Ali here. I am a Electrical (Telecom) Engineering student and my Final Year Project (FYP) is Portable GPR in which I am designing UWB transceiver. Currently I am working on the transmitter part.

Here are a few specs of the design:
-Simulation Software: NI Multisim 12.0.1
-Its an Impulse Radar with Bandwidth of 1 GHz centered at 500 MHz (currently)
-I used the logic of NORing a clock with its inverted delayed version, giving an approximate Gaussian Pulse of duration 1 ns. Clock decides Pulse Repetition Frequency (PRF). Its frequency is set to 100 MHz currently. (Idea from the Paper I am following)
-I generated the ref using AD 8009-ARZ, a high speed Op-Amp and generated the inverted version of it using the same op-amp.
-I made the NOR gate by generic transistors (FETs). (Please guide about practical transistors I should use)
-I haven't designed the pulse amplifier yet (current output voltage)

First of all, please comment on the approach I am using i.e., is it okay to generate clock using Op-amps? Is there

any other better way? etc.

The problem(s) are as follows:

-Antenna Design (Vivaldi etc.) on frequency of 0-1 GHz yields an impractical design. I thought of designing an active modulator to ease off the antenna design but have no idea on how to do it. Please help me in this matter.
-Choice of transistors for NOR gate.

What, in your opinion is the best approach to tackle this project?

If anything I wrote above isn't clear, please say so that I can elaborate it/provide visual aid.

P.S: Please don't quote number of results of Google by just typing 'frequency multipliers' or anything else like that. I have done my homework on this.

Also, the more you elaborate, the more I can understand as I am quite an amateur in this matter. :)

Thank you in advance.

 

[Enthalpy]

0-1GHz is impossible, especially the antenna. You must reduce the relative bandwidth, for instance 1-2GHz is already difficult enough.

An op amp is the wrong choice to make a pulse. Feedback is useless to make square waves and leads to slow circuits. Use logic gates or transistors. Any broadband transistor (BFR90 or more recent ones) produces pulses with <<1ns rise time and desaturates in <<1ns. I had made a ring oscillator with them, they had <<0.5ns propagation time so the tr and tf must have been around 200ps without optimisation (I had no instrument to measure that).

To complement the signal, use a transformer, for instance with striplines, possibly aided by ferrite beads, or with 1-2 turns on a ferrite toroid. An op amp would delay it.

Why shouldn't you use a single dual-gate MOS as a NAND gate? Like BF981 if it still exists. Put identical levels on both gates, who cares. At least that will be fast and with identical delays on both inputs. Or have two Schottky diodes for microwave frequencies.

You could maybe input the same signal, delayed but not inverted, on the base and emitter (or gate and source) of the same transistor, but then said transistor must be stable with the emitter not grounded - uncommon.

Voltage comparators still existed in ECL technology 20 years ago. If you find one, bias the inputs, inject the same pulse on both but with a delay. A line receiver is a voltage comparator as well, maybe easier to find.

Instead of two inputs, you could load a single output with a propagation line in parallel, and have the opposite end shorted. The line will allow a voltage during two propagation times. Do something non-linear for the opposite transition. Alternately, put an open-ended propagation line in the source and a voltage step in the gate; do something for bias.

-----

I suppose you're not going to radiate to the open world with pulses between 0 and 1GHz, will you? Radiocomm regulations would prohibit that.

Coud you tell more about this bizarre radar? Normal people have a narrower bandwidth and achieve the bandwidth through chirps in order to get enough energy per pulse.

 

[ali.cmi]

Hey Enthalpy, how are you doing? Thank you for replying.

I think you missed the most important part of my post: I am quite an amateur in microwave electronics and where I am studying, this project is first of its kind. I am an UNDERGRAD as well. :)

I am grateful to you for providing so many ideas. They must come from a lot of hands-on experience but they just sped right past my head. :\ Apologies.

Let me explain about the project. A group of my immediate seniors developed a GPR prototype and used Agilent's VNA as their transceiver. I, with passion of electronics and every sort of it, took the daunting job to replace that VNA (not the whole of it). This is the first of its kind project in my institute and I feel now, with utter lack of guidance that I took a wrong step. :( But this doesn't stop me from doing it.

They used VNA to generate 1 GHz bandwidth centered on 2.4 GHz and used a Pyramidal Horn Antenna. Radar type was SFCW.

Now when I actually got into knowing about generating frequencies, I was totally startled. So I moved to Impulse Radar. Now when I came to know about it, generating a band of 1 GHz is a big task in itself and I couldn't accomplish that as well with much efficiency.

As I said, I used Op-Amp oscillator to generate a clock of 100 MHz, inverted it with another op-amp (in turn delayed) and NORed them, resulting in a somewhat ugly version of Gaussian Pulse. Output power is in nW, another problem. I need at least 4-5 Giga times its value to radiate and penetrate in ground.

Now my requirement is to somehow generate 1 GHz bandwidth with 4-5 Watt output. Antenna size opened my eyes and I am now thinking of modulating this baseband signal with 2.4 GHz carrier. But have no idea about how to design an active mixer. (active to have 4-5 Watts power at output)

Also, I know about SRD as well. It is most popularly used diode to generate short pulses. I indulged my group-mate to do it but we can't model it in ADS as the software doesn't has a built-in model. Please shed some light in this as well. SRD still requires an input of >100 MHz.

The idea regarding normal people sounds good but have no idea on how to do it. :)

I really appreciate your effort and time to write all those flashing ideas but I request you to kindly point me in one RIGHT direction and give practical experience of yours. (Not asking about schematics although I won't mind having them :P ) :)

 

[Enthalpy]

Your abbreviations aren't understandable, sorry. I guessed the Step Recovery Diode, not the rest. Maybe someone of exactly this background and of English language would guess them, but readers here are more varied.

Your radar is to investigate the ground, is that it?

If used in open terrain, your transmitter must comply with radiocomm rules. These prevent from covering 0-1GHz for sure, and more so using short pulses. Even if you used spread spectrum to dilute to a tiny spectral power density, radioastronomers couldn't accept it. Their incredibly sensitive receivers must be protected from interferences. In their frequency bands, the rule is just: radiate nothing.

For remote sensing in the ground, 0-1GHz is not the same as 1-2GHz nor 2.4GHz +-0.5GHz: propagation depth would probably be the information.

You must check numerically if short pulses bring you the desired sensitivity. They can be produced at 5W peak power with the proper circuits (which op amps are not, but others are) .

In case you need low frequencies in the spectrum, you absolutely need to know how low. 10-1000MHz is not at all the same as 100-1000MHz, for the transmitter-receiver and even more for the antenna.

And then, see how the desired spectrum could be made compatible with civilized practices of radio transmissions and with the corresponding regulations. In open space, you'd be allowed to use a few narrow frequency bands here and there in the best case.

More generally, I see many people here trying to make electronics using a simulation software. This come probably from the relative cost of laboratories versus software, and from the limited resources of schools and universities, but it doesn't work. Electronics must be learned in labs and books, and simulation software can't possibly let a circuit work.

 

[ali.cmi]

I apologize for this gaping hole in my conversation. I should've explained them. Thank you for pointing it out.

SFCW = Stepped-Frequency Continuous Wave (a mode of RADAR)
VNA = Vector Network Analyzer
GPR = Ground Penetrating Radar
ADS = Advance Design Systems (a simulation software)
SRD = Step Recovery Diode

Yes, the only goal is to investigate ground up to 0.5m maximum and detect a few specified sized items dumped in it.

I understand your concern about regulations of using 0-1 GHz band but don't worry, there aren't any people here to sue me. :D I understand completely now why this band is not suitable.

But, the point is I need 1 GHz band and that band centered around 2.4 GHz would be suitable as the previous work done on this project has its frequency spectrum centered around this frequency and it worked. They detected most of the metal objects like steel locks etc and even pens of people who challenged them.Yes, I am using softwares for simulation such as ADS, Microwave Office and Multisim. High frequency components aren't even available in my country's market so you can imagine how far I am from practicality.

 

[Enthalpy]

At least you have reasons not to experiment immediately. I've seen, in coutries without any kind of restriction, people and teachers stick to simulation computers just because they had ended believing that the real world had decided to behave like the simulator.

Does your radar rely on time to separate an emitted pulse from its echo? Then you'd need pulses of 1 or few ns and this is a pity, since they occupy a bandwidth which isn't available so low in the radiocomm spectrum.

Incidentally, 2.45GHz +-0.05GHz is an ISM band, where other users essentialy shouldn't be accepted. Most countries were dumb enough to leave hand-held phones and computers communicate in this band, and interferences with microwave ovens and others are getting common.

Even if, as it looks, nobody cares locally about your use of 2.45GHz, I'd like to emphasize that radioastronomy should be protected, even if they're very far from you. They have a band at 1.43GHz (neutral hydrogen frequency) and elsewhere to be kept perfectly silent.

Oscillators at 2.45GHz exist in cell phones, Wi-fi computer transmissions... Maybe you can get one from there. But then you probably want to amplify a bit, and modulate with a pulse (2ns long?). Not trivial.

 

[ali.cmi]

I appreciate your concern about the radio astronomers. What do you suggest I should do?

Yes, my radar will distinguish emitted pulse from its echo relying on time. Luckily today, we had some good results with Step Recovery Diode (SRD). Pulse durations were < 1ns in simulations, of course. And yes, my resources don't allow me to work on practical part right now. Hopefully, I pray that actual results are closer to what I am seeing in simulations. My group mate is working on impedance matching to make results better.

Lets say I somehow got the oscillator, how to I design an active mixer? That's the real question. As I said, I am a total newbie in this area so please be as practically descriptive as you can. Considering your expertise, I believe that you can provide me with a solution.

Also, please tell me about your time zone so that we may avert this replying gap. :)

 

[Enthalpy]

Ultra-wide band pulses can't be emitted legally in the open, and shouldn't neither, even if detecting mines in Afghanistan for instance, because it interferes with radioastronomy and with satellites and airplanes - be sure you'd get trouble even in remote locations.

For remote detection, you must use the corresponding frequency bands. Difficulty: you need short pulses, like 4ns maximum, which corresponds to a wide bandwidth.

So either you go to a frequency band high enough to obtain the needed bandwidth, which means many GHz central frequency and difficult technology, or - I'd prefer it - you give up the pulses completely.

Put a continuous wave transmitter in a horn antenna, a receiver at its side in its own horn antenna, shield the receiver from the transmitter efficiently, and observe the echoes from the ground.

The horns should have corrugated rims to minimize side lobes in the directivity patterns.

This one won't discriminate easily echoes from the surface and the sub-surface.

-----

Industrial radars exist to measure liquid levels. You could check how they works and at what frequency and pulse duration, and possibly buy one instead of building it.

 

[ali.cmi]

Sir, my task is to 'design' the radar, not buy it. Of course making oscillators is like re-inventing the wheel.

If you were me, how would you design a continuous wave transmitter?

 

[Enthalpy]

First task is to find the frequency bands allowed for that activity, which could be "radiolocation" in the radiocomm regulation language.

Then, in the CW option, I'd choose a rather low band like 1GHz as apparently it worked, build a crystal oscillator for stability - hence around 100MHz - and just double or triple the frequency a few times to reach the desired one. Few mW are enough for your purpose.

Crystal suppliers make for little money the frequency you want. The other option is a standard frequency and a PLL.

Before building a transmitter and a receiver, I suggest to build the antennas, connect a generator and a spectrum analyser, put all on wheels, and check if this method works as desired.

Horn antennas may be better suited to you short distance with a lens in the aperture. It's often done with metal plates.

Or maybe a pair of cigar antennas are a solution, with some shielding around the launchers.

 

[ali.cmi]

First task is to find the frequency bands allowed for that activity, which could be "radiolocation" in the radiocomm regulation language.

Then, in the CW option, I'd choose a rather low band like 1GHz as apparently it worked, build a crystal oscillator for stability - hence around 100MHz - and just double or triple the frequency a few times to reach the desired one. Few mW are enough for your purpose.

Crystal suppliers make for little money the frequency you want. The other option is a standard frequency and a PLL.

Before building a transmitter and a receiver, I suggest to build the antennas, connect a generator and a spectrum analyser, put all on wheels, and check if this method works as desired.

Horn antennas may be better suited to you short distance with a lens in the aperture. It's often done with metal plates.

Or maybe a pair of cigar antennas are a solution, with some shielding around the launchers.

I have a horn antenna (centered at 2.4 GHz). It was used previously by the group who made the first prototype.

What do you suggest about buying broadband oscillator ICs, amplifying their signal and feeding it direct to the antenna?

 

[Enthalpy]

2.4 GHz is not allowed for remote sensing nor radiolocation. Finding allowed bands should be the first task.

Horns for 2.4 GHz will work at a higher frequency, and maybe at a somewhat lower one - only the launcher at the apex needs to be replaced. Or you can make a horn from a rolled and soldered sheet of metal.

You probaby don't need to amplify. 0dBm looks like a reasonable power for an echo at 2m and is typical from an oscillator. A regulating administration would ask why you need any stronger power.

Before building or buying, you should check with a generator and a spectrum analyser if the general idea works. The receiver will be more difficult to build than the oscillator.

To detect materials shallow in the ground, I'd really like to focus the waves. First, this will give you a stronger signal, and second, the user will be able to estimate the target's depth by moving the apparatus vertically and discriminate the buried echo source from surface reflection.