Phase difference detector, 433.92MHz

Hi, I am hoping to detect the phase difference from a 433.92MHz carrier signal detected at two antennas, where one antenna is about a 1/4 wavelength further from the transmitter than the other.

The circuit looks as follows. I use PLLs to ensure phase lock on the carrier and pass the 2 signals to a phase detector:


I hope you can understand my circuit diagram. A second Bandpass filter after the BJT Amps would double the 6dB of the first to 12dB, so I will probably put those in each line aswell.

My questions are:

1) Do I really need the PLLs? Can I just put the antenna signals into the Phase difference detector without phase lock, or do I really need the VCO (powered) output?
2) If PLLs needed, what ICs do you recommend? Do I need special PLLs capable of locking onto a 433.92MHz signal? Or can I somehow use standard 4046/LM565??
3) What IC for final phase comparator? Anything special required?

I hope you can answer my questions!! If you can, then you have probably guessed that the circuit is designed to detect which antenna is closest to the transmitter source, within a wavelength as per diagram:


Antennae are about a foot apart, transmitter is off in the distance.

HOPE YOU CAN HELP!!!!!

If your intention is to find the transmitter antenna then you might also want to look at the article:
"A Companion Tracker for the Miniature UHF Fox Transmitter" by Dave Bowker K1FK; QST May 2006, page 32.

This article shows a clever way to use an antenna switch and an FM receiver to build a tracking array made of two dipoles. It is considerably easier to construct than your proposal and may end up doing the same job.

As for your idea, there are some problems to solve that are going to be tough. First of all, your idea of using a PLL on each receiver is interesting, but doesn't strike me as efficient. The main reason you are using the PLLs is to boost the signal so that you can feed the phase comparator. This is a complication you may not need. Did you know that a simple double balanced mixer can make a good phase comparator? A mixer may be a better choice for a phase comparator in this case because it can operate well on very low level signals. Let's have a look at your signal levels.

Your first amplifier, being a JFET type can have a very good noise figure if you build it right. I hope you have the means to measure noise figure (hint: the easiest way is to simply put this amp at the front end of an existing receiver and then measure the overall sensitivity). It might be wise to add a bandpass filter before the JFET amplifier or you risk overload or desensitization from other strong in-band signals. Let's assume you add such a filter.

Your filter will have a loss somewhere around 2 dB, your JFET amp will have a gain of about 14 dB (just an educated guess for now), your second band pass filter is another 2 dB, then your BJT amp has a gain of about 15 dB or so. So you have an overall gain of 25 dB. If you are working with low level signals your PLL input may only be about -70 dBm or so. That's a very low level signal, too low for the PLL to work with. You would need a fair bit more gain, which gets a bit dicey because the more gain you have, the more chances of instability (this is one good reason the superheterodyne receiver was invented).

Now, if you use a mixer as your phase comparator, you are still going to have a low level signal coming out of the mixer, but this time it will have a frequency in the low audio range. This is really good because you can build an amplifier for this fairly easily. This is a much more easily built circuit than your proposal (but not as easy as the one in QST magazine).

If you insist on going with your approach, you may find that the input sensitivity of the typical UHF PLL, such as the types used in modern synthesized radios, is about -15 dBm to achieve phase lock. If you have to amplify your input signals up to -15 dBm, you almost have enough signal to feed your phase comparator directly, so there isn't much point in having the PLLs there.

The 4046 and LM565 operate at very low frequencies and would be hopelessly inadequate for use at 433 MHz. I recommend using a double balanced mixer. Heck, I could even send you a couple if you want them. I've got lots in my junk box.

I thought of a couple more points.

If you use a PLL that doesn't have any frequency translation (in other words, if the VCO in the PLL is supposed to be the same frequency as the input) then you will find it impossible to avoid interference between VCO leakage and your input signal. At low frequencies there is no problem, like when using an LM565, but at RF frequencies like 433 MHz you will find it impossible to keep the VCO leakage from getting back to your antennas or into your receiver front ends. The only way around this problem is if the VCO operates at a different frequency and dividers are added to divide the VCO or the input signal to the comparison frequency. As you can see, its getting kind of complicated. Too much in my opinion.

There is one minor negative point about using a mixer as your phase detector, but it is manageable. That is, that it will give you a zero voltage output when the two incoming signals are exactly 90 degrees out of phase with each other rather than when they are 0 degrees out of phase. This isn't a big deal if you are prepared to add an op amp circuit that handle this.


here are some references you could read:

http://www.minicircuits.com/appnote/detectors12-2.pdf#search='mixer%20as%20a%20phase%20detector'


http://www.minicircuits.com/appnote/phase12-3.pdf#search='mixer%20as%20a%20phase%20detector'

http://rfwireless.rell.com/pdfs/TN_WJmphase.pdf#search='mixer%20as%20a%20phase%20detector'

RadioRon,

Thanks for the excellent advice! I have previously had a look at TDOA and Doppler-ish methods. They operate with a single FM receiver and switching between two antennae, as you know. Sudden phase shifts in the carrier are achieved when switching between the two antenna, inducing a frequency shift that the FM receiver translates into a pulse. The pulses arrive at the antenna switching frequency (audio freq.), and there are various ways to deduce which antenna is closest to the transmitter or if the antennae are equidistant from the transmitter. I mention this purely so that you may criticise my understanding of this.

Anyway, these audio-frequency methods are clever, but they require a FM demodulator (radio), which I do not want to include in my circuit. I (stubbornly?) want to do phase shift detection directly in the carriers detected simultaneously at the antennae, without using switching or an FM demodulator/radio. I was guessing that switching can introduce noise. I want to deal with the problem at the carrier frequency, directly on antennae outputs, if possible..

What do you reckon?

Hmm.. maybe you are right, this is a complicated method? It looked as simple as "boosting" the signals received as the antennae, and the n doing phase comparision, eliminating the need for antenna switching and FM demodulation. This is because there is no data/audio being sent over the RF signal - just the carrier at a constant freq. and amplitude - no modulations.

OK, sounds like using the PLL/VCO to 'boost' the signals would be complicated, a bad idea. Maybe I can just bandpass and amplify the signals, then do the 433MHz phase comparision with no other processing? That would be only a few ICs in a simple cct. But it probably wouldn't work, right?

I wanted to keep this simple, but I will look at all the references you have suggested to boost my knowlegde on this topic.

If you have any other ideas that don't use antenna switching or conversion to audio frequency, I would very much appreciate them!

*off to do the suggested reading*

Oh yeah.. the double balanced mixers. I'd love to play with them, but I'm in Australia so borrowing them might be an expensive task . I'll try to get my hands on some here...

Cheers!

OK, I get it. But the ultimate justification for your method depends on why you want to do it that way, and since I don't know why, I am happy to go along with your judgement.

I didn't study the principle behind the article on TDOA in great detail because it seemed intuitively obvious to me that by switching the two antennas at an audio rate and assuming there is a phase difference between the two received signals, you would phase modulate the signal at the switching rate. And since phase modulation and frequency modulation when using a single frequency modulation source are indistinguishable then you can of course hear this phase modulation on an FM receiver. I understand that when you rotate the two antennas until they are broadside to the source there is no phase difference between the two received signals so the modulation at the switching rate goes to zero since there is no phase shift when you switch between the two antennas. I think that is a cute approach, very practical.

So, the summary of this post is, let's talk about using a mixer as phase detector and implement your scheme that way. You will want a low noise DC coupled audio amp to follow the mixer, and you should plan to include some sort of DC offset correction. All done easily with good op amps.

who said anything about borrowing. I'd give them to you for keeps. heck, i've had this one box of DBM mixers for over ten years and haven't done a thing with them. You're welcome to a few.

cool on the DBMs! I'll have to contact you about that

My main motivations for looking into this is:
-to do RDF with as few ICs as possible (total, inc. all extra bits like radio, etc)
-make RDF cct with homing output
-make it as small as possible for mobile/mini applications
-with as little power consumption as possible
-RDF without buying a radio or FM demodulator
-make it as cheap as possible (as above)
-to buck the trend and do it a different way (though I'm sure someone must have done this before, I just cant find refs or reasons not to do it this way.)

So.. All I 'might' need is:

Right? Leaving me with a difficult task of picking components and designing cct!! HELP!

Just to substantiate my concerns, here is the data sheet for one popular line of PLL/synthesizer parts "suitable for 433 Mhz input".

https://www.electro-tech-online.com/custompdfs/2006/05/LMX2350.pdf

This part shares typical input sensitivity characteristics with many other models and most other manufacturers. There are two inputs available. The reference input can only tolerate up to 50MHz, so 433 Mhz input here won't work. The RF input can tolerate up to well over 433 MHz, but the input sensitivity is -10 to -15 dBm. This is the input that is supposed to hook up to the VCO, so there really is no input available for your 433 MHz signal. These PLLs all assume that you are phase locking the VCO to a crystal oscillator running below 50 Mhz, not an incoming signal at 433 Mhz.

There is no IC that is a functional equivalent to the LM565 that works at 433 Mhz as far as I know. Think double balanced mixer, that's the ticket!

We seem to be overlapping on our posts! that's a first for me on this site.

Too bad we can't chat on 20 meters about this. I haven't been on the air in about 8 years. Then again, nowadays skype works for me.

Yeah, thanks for checking the LMX2350. I'm not yet sure exactly how the DBM operates - I'll read up on that tonight. It must be a multiplier of some sort right? I'm guessing this because phase comparision is usually done (in terms of math) by multiplying the two signals togeather, and then LPF-ing away the extra higher order frequencies, leaving the sin(w(p1 - p2)) phase difference term (from memory). Again, I'm stating this to be criticised so that other kids may learn from my mistakes and the punishment I deserve! .

Yeah, we're a bit out of phase HAHAHA. Wow, This thread has achieved a 2nd page!

Your new block diagram seems OK, but I just thought of another important problem with your approach. Since it has no ability to distinguish one signal from another within the passband of your front end filters it will generate one heck of a mess of outputs from the mixer. So now I have doubts that it is viable. The basic principle will work fine if you only receive one carrier, but as you probably are well aware, the band may have many signals, and some of them stronger than the one you are working with. This circuit will not distinquish one from another, so the output of your mixer will include your desired DC signal, the results of phase detecting all the other signals on the band, plus the result of mixing every signal with every other signal. It will be a dog's breakfast, as we like to say.

The more you look into this idea, the more it becomes practical to consider using a narrowband receiver, such as with the TDOA method. Maybe that's why all the other methods use them!

I recall several years ago seeing other hams on foxhunts and they were using this very cool setup of four monopoles arranged on the roof of a vehicle in a square array. This fed a receiver setup (whose details I don't recall) which ultimately displayed a dot on an oscilloscope and the dot would be placed on the scope to show the direction of the signal source. As you drive, you simply keep the dot at the 90 degree position and you are heading straight for it. Have you every seen this method and setup? I understand it is quite popular for serious RDFing.

I've got it! Your idea will work if and only if we do indeed use a PLL to lock your incoming signal to a VCO and IF the VCO's control range is only as wide as the channel you are listening to. This provides the necessary selectivity.

Of course, this is actually EXTREMELY difficult to do. I've built some VCOs at UHF frequencies and they never stay within a 30 or 20 KHz range no matter what you do unless you lock them to a crystal. Hmmmmm. This calls for donning of the "bright idea hat".

That 4 antenna job sounds like the 'Doppler' methods. Something about switch rotating between antennas at a really fast rate - I came across this yesterday when researching for my idea, but it requires heaps of hardware and antenna.

see:
**broken link removed**

After this chat, I'm becoming more inclined to shut up and do it the way everyone else suggests - eg:

https://utaharc.org/rptr/ark_df_desc.html

?????

Anyhow, regarding band interference: The usual (legal?) transmission ranges on the 433MHz band in OZ are low, ~300m. And this is for outdoor use, so maybe interference (at least from other transmitters) is not a massive problem? Do multipath problems might crop up?

Just in the nick of time!!! That sounds like the ticket...

That's sort of how I naievly imagined it could work.. but with little knowledge of PLLs I did not realise achieving required selectivity was difficult (amongst other poor assumptions and oversights)...

Looks like we're back to a block diagram similar to the original one?? Can you clarify the issues we need to address with this method? Do you think we can still achieve a ~$10 solution?

I don't think multipath will be too big a problem. It will cause some additional components to appear on the mixer output, but hopefully these components will be partly AC (which you may be able to filter) and/or much weaker than your main signal.

I'm shutting down for today and will consider the problem while I sleep.