Ranging using radio 2

[astroclone]

Hello. I am fairly new at the whole circuits thing, I'm trying to get into using microcontrollers and such to control everyday applications and what not. I do have an engineering degree but my 3 credit class in engineering instrumentation didn't prepare me for everything I want to know.

I was wondering if it's possible to send out a radio signal that is recieved by a reciever and then sent back to the original transmitter for purposes of getting range? I know it's possible because that's sort of how RADAR works, but I want the return signal to be able to be processed in order to determine what the range is for a given object in a field of say 20 or 30 objects. I know there are addressing issues and stuff like that, that's fine. I'll deal with that later. Right now I'm just looking for the basics at sending out a pulse and having one sent back so that the reciever knows the time of flight and what object sent it.
Hope this isn't too confusing.
T

 

[IRstuff]

Not at all, BUT your thoughts have apparently not gotten to the fact that radio waves travel essentially at the speed of light in the atmosphere, so for any range less than 1km, you're talking about resolving range from a time of flight of 6.7 microseconds. In order to resolve ranges of 1m, you'll need to be able to accurately measure time of flight to 6.7 nanoseconds

This is something that requires custom electronics and counters, not really suitable for a newbie project. Your typical microcontroller would have a system clock on the order of 100 nanoseconds.


If you're stuck on doing something like this, an acoustic rangefinder may literally be more to your speed. The speed of sound is around 340 m/s. so 1m range resolution is much easier to deal with.

However, your desire to discriminate objects requires significantly more complex hardware, so you might consider simply concentrating on getting range from a single object for the time being.

TTFN

 

[astroclone]

Well, I was planning on slowing down the speed of light using a pair of sunglasses. Stupid speed of light, why does it have to be so fast?
T

Just in case you are wondering, I am kidding.

 

[Comcokid]

This bi-directional type of ranging is done for some applications. Ranging between a satellite in orbit and a ground station is performed in a manner similar to this. In a uni-direction manner, LORAN navigation times the reception of several transmitters at a receiver in order to determine position. In these cases, the accuracy is on the order of a few meters at best, and a few hundred meters at worst.
Short distance ranging can be performed with some of the impulse radar and other ultrawideband location technologies (usually referred to as MIR for micropower impulse radar). Some of the patents within this area describe applications similar to your described problem. Because signal reception with MIR is equivalent-time sampling based, resolution between signals with only picosecond differences can be performed with a timing source of microsecond resolution.

 

[astroclone]

Alright, new thought.
What about 2 or 3 transmitters transmitting unique carrier wave type of signals and a reciever that can determine what direction the individual signals are coming from. Once again a triangulation thing.
That way each individual reciever needs to worry about itself instead of one reciever having to worry about multiple inputs from different sources.
My current thought is this:
2, 3, or 4 fixed transmitters transmitting unique frequencies.
a reciever located somewhere in between the transmitters at an unknown location.
The reciever gathers direction information from the transmitters every 5 minutes or so and then sends the information of to a different source that collects and analyzes the data.
The eventual goal is a GUI that will show the user exactly where the object is within the given box.
Any suggestions?
T

 

[IRstuff]

Seems overly complicated if all you're trying do is triangulation.

1) The "receiver" will actually need two receivers to get any sort of angular accuracy, multiplied by the number of unique sources.

2) Instead, have one transmitter at the unknown location. Two dual-channel receivers located roughly orthogonal positions. Each dual-channel receiver accurately measures direction of arrival (DOA). The intersection of DOA's is the location of the transmitter.

BUT, either is quite non-trivial, since you effectively need to accurately measure signal phase across two receivers.

TTFN

 

[lcsjk]

Maybe you should investigate optical rangefinders if you are trying to differentiate objects within a short ( <1km or <1mi).
It can be done with RADAR, but for a person wanting to use UP's for a project, you are jumping into the fire in trying to work with radio tracking/location at this point in your development.

Keep in mind that if the project you choose is too hard to implement, you will never get to the UP because of all the frustration of trying to become an expert in an unrelated field.

Good Luck!
LCM

 

[Heydave]

A base station could poll individual units to respond and then at some predetermined point in the transmission, a phase shift or some other trigger event is sent and responded to by the addressed receiver. A course time window for this trigger is then determined by the base controller clock then follow-on polls are used for a fine resolution test. Here the base transmit signal is incrementally delayed using a programmable delay line such as units from Maxim/Dallas semiconductor. Resolution of a few feet are not unreasonable with common off-the-shelf hardware and a microcontroller. Switching between two antennas can give you an X,Y location and three will result in X,Y,Z if the locations if the antenna are known. The transponders need not be as complicated as the base. The higher the carrier frequency the better but look out for multipath. The first trigger received is one to use. The delay line should also delay the transmit carrier reference for consistancy.

 

[BrianR]

astroclone,

Transponder systems are not all that accurate and pulse transponders like aircraft TACAN for example, are limited to 0.1nm accuracy. You must measure time to 3ns accuracy to resolve 1m and that is generally done by triangulation using phase difference between two accurate signals of in the case of GPS, by correlating 2 or more signals modulated with maximal length pseudo random sequences. You can easily get 1m accuracy using this method. I think GPC C-code uses a 10 bit sequence. Over short distances you will not have a lot of difficulty with variations in atmospheric velocity factor (due to temp, moisture and pressure), changing the speed of lightlocally but you will, over 20km or more.

 

[AlexKrauska]

The mutiple transmitter/receiver thing may be the most economical. Actually there is a &quot;game&quot; you can play called fox hunting. Its often done on amateur radio frequencies, but can be done at fcc part-15 frequencies as well - FM radio or AM radio band, or 915MHz, 2.45GHz, etc.

The idea is this, a &quot;course&quot; can be laid out with one or more &quot;foxes&quot;, radio transmitters that keep transmitting their own ID, and all the compettitors start out at the same time to find the hidden transmitter. You can take several strategies. The first is to find the direction of the transmitter, and keep moving in that direction. Another approach is to take one direction reading (bearing) write it on a map, then move at right angles to the possible location, take another bearing, and with two postions marked ont he map with two bearings you can project a final location.

The receivers can be simple amplitude affairs with rotatable antennas, or they can be relatively fixed antennas that are omni directional, with electronic phase shifting to steer the beam. They can work by finding the direction of the strongest signal, or by finding the weakest location. Some antennas have a very sharp null only in one direction, which is convenient for direction finding.

Problem: RF signals only travel directly on a straight line in outer space. Objects, the shape of the terrain, buildings all affect the signal, and several signals can take different paths from transmitter to receiver. There is not necessarily a single, strongest path. But if the situation you immagine is simple, like flat asphalt with no obstructions, it might work.

 

[searsjames]

I have a very similar problem. I need to distance range using 3 recievers in a fixed grid and a 433 MHz pulse transmitter moving around. The objective is for the recievers to each time stamp their reception of the transmitter pulse to within a repeatability of 4 nanoseconds. Accuracy will be calibrated into the system, timing repeatabillity is the key to the solution I want. The antenna signal strength can be as low as -100 db and up to -20 db.

I have been assuming I need to make a very narrowband, very stable preamplifier, and then diode detect the output to trigger my time stamp. Is the the best way? How would I construct the preamp or where would I buy it?

Anyone got any design tips?

 

[BrianR]

Although it will be beyond home construction to make one, a modern fighter radar does what you want. The transmitter signal has a digital signature superimposed on it allowing very small portions of the transmitter pulse to be uniquely identified and in the receiver these are correlated. Sounds simple but there are a few hurdles to cross.

Get a copy of AIRBORNE RADAR by STIMSON where you'll probably find as good an explanation as you're likely to get.

 

[IRstuff]

Not quite that easy. You still need stable, accurate timing for all units, just to do the time stamping. The only simple way of doing this is to use GPS, and even then, you'd need to do some atmospheric corrections probably.

While you say that you'll time stamp the reception, how do the receivers know when the pulse was sent? You're doing the GPS problem in reverse, which means that you'll need 4 receivers to solve for the 3 position unknown and the 1 time unknown. If it's all in the same horizontal plane, it might be possible to use only three receivers.

Since you'll have to use GPS anyway, it would seem to be easier to use DGPS in the first place, which is all off-the-shelf.

TTFN

 

[BrianR]

For searsjames, try modulating your 433MHz signal with a BPSK maximal length pseudo random sequence, eg a 10 bit sequence running at 5MHz chip rate. When you correlate two received sequences you can measure difference in distance to the two transmitters to better than 1m, quite accurately. I made one, and on the bench test with a signal of about -60dBm was able to maintain and get correlation when it was buried under an interfering carrier at -10dBm and noise at -30dBm. In practive you should get correlation down to better than -150dBm. GPS gets to -170dBm. The receiver needs a matched filter for the modulation envelope which (from memory) is about 5-10MHz wide.

Sorry this is so general but it's 17 years since I did this.

 

[BrianR]

I meant to add to previous (and in support of IRSruffs comment), that the transmitter clocks have to be very accurate, like atomic clock accuracy and for absolute distance measurement, the receiver clock has to be locked to the received signals through a Kalman filter.

 

[searsjames]

First I am confined to a plane and so 3 receivers work. The relative time of reception at these 3 receivers is all i need.

The first problem is the detection circuit for triggering a 4 ns resolution time tag repeatably after the 433 MHz hits each receiver.

The second problem is the counters and data latches them selves need to work at 250 Mhz.

Any recommendations?

 

[searsjames]

One of my constraints is that I will have hundreds of transmitters to track from each pod of three recievers. Each transmitter needs to cost <$10 @50K quantity and so will only send its ID # and a few more bits.

The recievers can cost several hundres dollars or more and can be constantly time syncronized to each other to within a few ns via rf because the delay times between them are fixed and known.

Given this information, if i can just time tag the cheap transmitter bursts to within a consistant 4 ns window i will know the relative delay distances.

This system should not require any fancy clock stability, just detection delay stability over a large variation of signal strength.

 

[IRstuff]

I think that you're seriously underestimating the stability of the timebase required to maintain 4 ns stability. Even assuming a calibration every hour, a 4ns drift translates to 1 part in 10^11, which, as BrianR pointed out, requires atomic clock stability.

Even calibration every second requires stability of 2.5 parts in 10^8, which still requires oven stabilization.

TTFN

 

[BrianR]

&quot;The relative time of reception at these 3 receivers is all i need.&quot;

The relative time includes any drift in or differnces between the transmitter clocks + propogation delay + detection errors + receiver clock errors.

However, assume a perfectly accurate clock in each transmitter. Variations in radio propogation delay due to weather will give you more than 4ns shift in a 20km circuit over a day or so. This radiolocation system needs to estimate it's position based on the time difference between received signals and known transmitter locations, determine the error circle size (caused by time measurement, propogation errors and receiver clock errors etc), shift it's clock, or by some other means it's estimate, until the error circle is down below 1m in size. That's where the Kalman filter comes in.

When the transmitter clocks have uncorrelated individual errors you are not in the race. Also you will not measure range accurately over any useful distance using simple pulse technology. Some form of pulse compression is needed and PRS modulation is as good as it gets.

 

[searsjames]

I plan to send a time calibration signal to each receiver about once per second and oven stabilization of their clocks will be used if needed.

I am only ranging up to 500 meters so weather is less of a problem and yes I will have fixed, known position transmitters to perform periodic range calibration at least once per second.

The transmitter clocks or what ever encoding i send dosent matter for anything in this ranging scheme, only having a clearly defined onset of the 433 MHz carrier.

I am afraid I dont understant pulse compression and PRS modulation...How do they help me reliably timetag?