Low Frequency Radio Trilateration

[RadioEng]

Here's a calculator that shows you're wrong:

http://www.cdt21.com/resources/siryo3_01.asp

Enter 1000 for TX Power, 65000 for the distance and 0 for the gain. If you now look at the Power density Wr, it agrees with my figure at 1.8834905E-8.

 

[Studiot]

Yes I can now agree with the arithmetic in your post #66.

I have edited my earlier forumula.

 

[Studiot]

OK, so now we are agreed on the very small number that would have to be measured to tell the difference between 65Km and 65.1 Km, can you calculate the change (drop) in transmitter power output that would be required to make the receiver at 65Km think it was at 65.1Km?

Is it possible for a transmitter to maintain the stability of its power output to bette than this, and by how much better does it need to be to make such a source of error insignificant?

What does this imply in terms of the stability of the voltage supplies to the transmitter?

 

[RadioEng]

OK, so now we are agreed on the very small number that would have to be measured to tell the difference between 65Km and 65.1 Km, can you calculate the change (drop) in transmitter power output that would be required to make the receiver at 65Km think it was at 65.1Km?

A quick check with the calculator showed 3 Watts or 0.003%.

Is it possible for a transmitter to maintain the stability of its power output to bette than this, and by how much better does it need to be to make such a source of error insignificant?

Yes, this is very broad, there should be no reason even with moderate hardware that this could be controlled to milliWatt level or 0.000003%. I would say that no source of error is ever insignificant as it will be compounded by further sources of error.

What does this imply in terms of the stability of the voltage supplies to the transmitter?

It needs to be very clean. Expensive, but achievable.

 

[sophiecentaur]

It might be worth pointing out that an error of only 1% in the measurement of transmitted power would produce an error of 2km in distance estimation at 100km. This 1% corresponds to a 0.04dB measurement accuracy, which is round about the limit of modern RF Power measurement accuracy. Absolute power accuracy may not be essential but decreasing the error by a factor of ten (this corresponds to 0.004dB and I don’t think you can buy a meter that will go anywhere near this accuracy) would still give an error of 200m. Still, producing a transmitter with this level of power stability would require a power supply with voltage fluctuations of better than 0.005%. Not quite sure what sort of high power supply could achieve that stability. “Expensive but achievable”? What evidence is there of this?

The radiation pattern of transmitting and receiving arrays would also need to be known to within a similar accuracy. The pattern could not be calculated but would need to be measured under all possible local conditions of weather, temperature and humidity. Depending upon what frequency the system operated on, the radiation pattern could be affected significantly (when you’re talking in terms of tiny fractions of a dB) by rain on a nearby flat roof, a passing truck or a sudden, local cloudburst. I was once involved with a series of measurements to characterise a small number of HF transmitting Arrays. It involved Helicopter measurement over a limited range of azimuths and elevations. It cost a fortune, took two weeks of flying time and yielded answers with accuracies of significantly more than 1dB. RF Engineers are more chuffed if the predicted field strength measurements are achieved within a quite a few dB.

Furthermore, an antenna at ground level and its feeder system would be subject to actual physical disturbance by wind (and even by vibrations from passing traffic). This will affect its pointing angle and the match it presents to the transmitter – hence the actual level of transmitted power. Again, a variation in reflection coefficient from 0.1 to 0.09 would represent a change in transmitted power in the order of 0.05dB. Who could rely on better than that? Can this be irrelevant and / or eliminated?
If spread spectrum is used, then the bandwidth of the system would introduce matching variations over the band. Can we be sure that the system characterisation could eliminate such variations reliably?
Many of these factors could, of course, be averaged out (ultra ultra low bandwidth measurement, effectively) but just how long would you want to wait for the measurement result to emerge? And then, how are you going to characterise the receiving equipment, in every possible location it may arrive at, to the same degree of accuracy?

Any Engineer or Physicist would know that there is a limit to how much money and time you can throw at any measurement problem. Even CERN acknowledge that there are limits to what can be achieved in Geneva.

 

[RadioEng]

It might be worth pointing out that an error of only 1% in the measurement of transmitted power would produce an error of 2km in distance estimation at 100km. This 1% corresponds to a 0.04dB measurement accuracy, which is round about the limit of modern RF Power measurement accuracy.

That's about 10W of instability, that's outrageous. Where are you getting the dB figure from?

These figures include the dBW value:

1000W Transmitter at 65Km
PDs = 1000.0
PDr = 0.000000018834904507916607783299854
dBW = -77.250365773078073927642982876853

1000 W Transmitter at 65.1Km
PDs = 1000.0
PDr = 0.000000018777084420741732059254669
dbW = -77.263718411584801401266580854133

dbW = 0.013352638506727473623597977276642


1000W Transmitter at 500Km
PDs = 1000.0
PDr = 0.000000000318309886183790671537768
dBW = -94.971498726941338543523154836328

1000 W Transmitter at 500.1Km
PDs = 1000.0
PDr = 0.000000000318182600416320126834229
dbW = -94.973235731174317688093801267624

dbW = 0.0017370042329791445706464313035054

Absolute power accuracy may not be essential but decreasing the error by a factor of ten (this corresponds to 0.004dB and I don’t think you can buy a meter that will go anywhere near this accuracy) would still give an error of 200m. Still, producing a transmitter with this level of power stability would require a power supply with voltage fluctuations of better than 0.005%. Not quite sure what sort of high power supply could achieve that stability. “Expensive but achievable”? What evidence is there of this?

Whilst a fair question, its not really the point of this thread. So, I will leave this for the moment.

The radiation pattern of transmitting and receiving arrays would also need to be known to within a similar accuracy. The pattern could not be calculated but would need to be measured under all possible local conditions of weather, temperature and humidity. Depending upon what frequency the system operated on, the radiation pattern could be affected significantly (when you’re talking in terms of tiny fractions of a dB) by rain on a nearby flat roof, a passing truck or a sudden, local cloudburst.

True, but these are transient errors and there are methods to disregard such wide deviations through comparative analysis.

Furthermore, an antenna at ground level and its feeder system would be subject to actual physical disturbance by wind (and even by vibrations from passing traffic). This will affect its pointing angle and the match it presents to the transmitter – hence the actual level of transmitted power. Again, a variation in reflection coefficient from 0.1 to 0.09 would represent a change in transmitted power in the order of 0.05dB. Who could rely on better than that? Can this be irrelevant and / or eliminated?

In this scenario, it would be a transient phenomenon. There would be enough data to compensate for this at the processing stage.

Rather than focusing on instantaneous issues, try to think of a more dynamic system continuously in motion capturing snapshots of data.

If spread spectrum is used, then the bandwidth of the system would introduce matching variations over the band. Can we be sure that the system characterisation could eliminate such variations reliably?

Can you be more specific?

Many of these factors could, of course, be averaged out (ultra ultra low bandwidth measurement, effectively) but just how long would you want to wait for the measurement result to emerge? And then, how are you going to characterise the receiving equipment, in every possible location it may arrive at, to the same degree of accuracy?

I'm unconcerned with time at this point. Its really a function of data and computational power.

Any Engineer or Physicist would know that there is a limit to how much money and time you can throw at any measurement problem. Even CERN acknowledge that there are limits to what can be achieved in Geneva.

True, but we're not hitting those limits yet.

 

[Studiot]

One thing that puzzles me.

At the outset the question was about atmospheric propagation and I understood the transmitter and receiver to be terrestrial.

Somewhere along the line discussion of space vehicles crept in.

For terrestrial propagation there is a big difference in propagation characteristics over land and sea.

So where, in gneneral terms, are the TX and Rx located?

 

[RadioEng]

At the outset the question was about atmospheric propagation and I understood the transmitter and receiver to be terrestrial.

Somewhere along the line discussion of space vehicles crept in.

Go back to post #15, its explained there.

For terrestrial propagation there is a big difference in propagation characteristics over land and sea.

So where, in gneneral terms, are the TX and Rx located?

The transmitter will be located close to, or on, the Earth to a maximum altitude of about 20Km and maximum depth of 200m-300m.

 

[sophiecentaur]

"That's about 10W of instability, that's outrageous. Where are you getting the dB figure from?
A back of a fag packet calculation tells you that, if there is 0.04dB error (that is about 1%), the resulting error in distance measurement, at 100km, of 2km. This assumes an inverse square law i.e. free space loss.
Received power is proportional to 1/distance ².
Say x is the error in distance measurement.
Ratio of power at 100km for 1 unit of transmitted power will equal power at (100+x)km for 1.01 units of transmitted power.
Solve the equation for x:
(1/(100+x)²/(1/100)² = 1.01
And x = about 2, implying about 2km of error in 100km if your power measurement is 0.04dB wrong.
Analytic calculations tend to be more straightforward than numerical ones; it is very easy to check it but it seems right to me.
You don't spell it out but your figures seem to imply that a power measurement error of 0.013dB wold give an error of 0.1 km in 65km (yes?). Is that so very different from my simple result?
Why does 10W of uncertainty in 1kW of RF power upset you? Power, as with many quantities has measurement error proportional to its value.

Your description of errors as being "transient" assumes that you are measuring over at least a year, if you want to attempt to cancel seasonal fluctuations. Is that really the serious intention? A GPS satellite receiver will give you 10m accuracy about a minute after switch on. Who can supply all this data for analysis at the "processing stage"? You can't monitor all these variables - and the actual accuracy of monitoring is again limited. If you have no evidence of actual figures and cannot justify your claims of possible improvement in all the areas so far mentioned then the system cannot work.
To improve simple SNR by a factor of 3dB, you need to analyse,in broad terms, for twice as long. Powers of two will rapidly build up and give you a ridiculous required time for your analysis.

A transmitter needs to be matched to its load. This is very difficult to achieve over a wide bandwidth. The result is always a 'frequency response' with undulations in the order of 1dB for most applications. This is adequate for most applications (digital and analogue). I'm not sure where a breed of transmitter/ matching network / feeder / matching network / antenna will come from for which the frequency response is much better than that. Don't tell me - it's been accounted for or it's a trivial problem. It is a relevant factor.

A transmitter at an altitude of 20km would be on a plane or balloon. They both move about a lot. How would the variations be measured in order to eliminate them? GPS, perhaps. Why not cut out the middle man and just use GPS?
200m - 300m underground?! What's all that about. If you mean under water then your available frequency bands are a bit limited. Submarines can use just a few tens of kHz whilst submerged.

Also, you don't say where the receiver will be. There would be even more problems in characterising the conditions at the receiver - which will be changing, presumable, as it moves about, over ground of varying conductivity, air of varying temperature, past obstacles that will cause multipath propagation / reflections / diffraction. Are these all going to be "accounted for"?

"Its really a function of data and computational power"
Yes - DATA is what you need and, to get enough data to average out all the effects mentioned and others requires a long time and a lot of monitoring points to gather. Even if you could process it all instantly, you still have to wait for it to build up. As has been said many times, it boils down to bandwidth / time. What time do you think you would need in order to reduce inherent variations, some of them of 'several' dB to a total of what would have to be in the order of 0.001dB?
One grouse you have had is the lack of numbers in the objections. Well, now you have some but you still say that any problem can be overcome.

You still haven't told the forum whether you have any practical Engineering (or Physics) experience which can qualify you to decide on the relevance of the many practical implications. Are you, in fact, any more than a software developer? Some of my best friends (and family) are software developers but they would not make wild assertions on engineering matters.)

 

[RadioEng]

A back of a fag packet calculation tells you that, if there is 0.04dB error (that is about 1%), the resulting error in distance measurement, at 100km, of 2km.

That's fascinating, but I meant this statement:

which is round about the limit of modern RF Power measurement accuracy.

How did you work out that 0.04dB is the limit of modern RF Power measurement accuracy?

I'll think you will find this is orders of magnitude off, even by tech of the late 80's.

Analytic calculations tend to be more straightforward than numerical ones; it is very easy to check it but it seems right to me.
You don't spell it out but your figures seem to imply that a power measurement error of 0.013dB wold give an error of 0.1 km in 65km (yes?). Is that so very different from my simple result?

Its best to get accurate.

Why does 10W of uncertainty in 1kW of RF power upset you? Power, as with many quantities has measurement error proportional to its value.

We're talking about high precision engineering, this level of power instability would be associated with catastrophic failure, rather than normal operations. Its just a misleading figure.

Anyway, we are unconcerned with the transmitter at this stage.

Your description of errors as being "transient" assumes that you are measuring over at least a year, if you want to attempt to cancel seasonal fluctuations. Is that really the serious intention?

Seasonal fluctuations can be mapped with a sweep from a network of known locations running around the clock. This is really a resources issue.

A GPS satellite receiver will give you 10m accuracy about a minute after switch on. Who can supply all this data for analysis at the "processing stage"? You can't monitor all these variables - and the actual accuracy of monitoring is again limited.

Of course you can, it would take a high performance computing lab and networks of dedicated hardware. Its not impossible, just expensive.

If you have no evidence of actual figures and cannot justify your claims of possible improvement in all the areas so far mentioned then the system cannot work.

Again with the feasibility study...no one has asked you answer this, nor would you be capable of answering it given that you have never seen the hardware.

To improve simple SNR by a factor of 3dB, you need to analyse,in broad terms, for twice as long. Powers of two will rapidly build up and give you a ridiculous required time for your analysis.

The wonders of supercomputing...its faster than you think.

A transmitter needs to be matched to its load. This is very difficult to achieve over a wide bandwidth. The result is always a 'frequency response' with undulations in the order of 1dB for most applications. This is adequate for most applications (digital and analogue). I'm not sure where a breed of transmitter/ matching network / feeder / matching network / antenna will come from for which the frequency response is much better than that. Don't tell me - it's been accounted for or it's a trivial problem. It is a relevant factor.

These are engineering issues, it comes down to the bandwidth of the antenna and accounting for the signal loss the circuitry. This would probably be one area with the most accurate experimental data and with in-built sensors real-time information could augment that.

Again, its a matter of information, accounting for it and processing overhead. Its not unsolvable.

A transmitter at an altitude of 20km would be on a plane or balloon. They both move about a lot. How would the variations be measured in order to eliminate them? GPS, perhaps. Why not cut out the middle man and just use GPS?

The whole system is moving anyway, the satellites are in orbit and the Earth is rotating. Being in a balloon or airplane won't make much difference.

200m - 300m underground?! What's all that about. If you mean under water then your available frequency bands are a bit limited. Submarines can use just a few tens of kHz whilst submerged.

Its just a requirement. This is why the sub-1000Hz range was chosen, very deep penetration, little attenuation and little signal loss.

Also, you don't say where the receiver will be. There would be even more problems in characterising the conditions at the receiver - which will be changing, presumable, as it moves about, over ground of varying conductivity, air of varying temperature, past obstacles that will cause multipath propagation / reflections / diffraction. Are these all going to be "accounted for"?

See post #15. Given the patterns, echos can be disregarded for trilateration purposes, but could be used to fill in gaps due to noise. So, they may prove more useful than a hindrance at the processing stage.

Yes - DATA is what you need and, to get enough data to average out all the effects mentioned and others requires a long time and a lot of monitoring points to gather. Even if you could process it all instantly, you still have to wait for it to build up. As has been said many times, it boils down to bandwidth / time. What time do you think you would need in order to reduce inherent variations, some of them of 'several' dB to a total of what would have to be in the order of 0.001dB?

It doesn't matter at this point. Its not as long as you think though.

One grouse you have had is the lack of numbers in the objections. Well, now you have some but you still say that any problem can be overcome.

Any problem can be quantified, broken down into manageable units and solved. Its the basis of science. So far, whilst you have listed some problems, they all have clear resolutions and associated costs.

You still haven't told the forum whether you have any practical Engineering (or Physics) experience which can qualify you to decide on the relevance of the many practical implications. Are you, in fact, any more than a software developer? Some of my best friends (and family) are software developers but they would not make wild assertions on engineering matters.)

Ending on an ad hominem is a sign of a weak argument. This line of questioning will not change the facts.

 

[Studiot]

... limit of modern RF Power measurement accuracy...?

Do I understand your scheme to be that you are continually measuring the Tx power and somehow communicating this to the receiver, which your statement seems to imply?

And by Tx power I mean the actual radiated power, not the input power to the Tx system.

This is a fundamental point.

 

[RadioEng]

Do I understand your scheme to be that you are continually measuring the Tx power and somehow communicating this to the receiver, which your statement seems to imply?

No. The power would be recorded for each pulse as it passes through the filters, or perhaps pre-amp, stage. This is coupled with an array specifying which frequencies were detected. This is relayed to ground for analysis, correction and integration into an existing dataset.

And by Tx power I mean the actual radiated power, not the input power to the Tx system.

A highly controlled ERP through experimental mapping.

 

[sophiecentaur]

I get it now.
It really is a huge software simulation we're discussion, in which "resources" can be increased by adding a couple of zeros to the end of the value of a variable.

We can filter out long term fluctuations with a filter having microsecond impulse response and measure all the parameters we could think of with the same speed.
You should have mentioned that the "facts" are not changeable because you have written them down.
It sounds fun but it has wasted quite a lot of my time as I thought it was a serious, physical, application we were discussing. No wonder you were shy about declaring your qualifications situation.

So as not to and on an ad hominem, I might ask what this statement is supposed to mean:
"Given the patterns, echos can be disregarded for trilateration purposes, but could be used to fill in gaps due to noise. So, they may prove more useful than a hindrance at the processing stage."
afik, multipath propagation has never been a positive aid to any form of radio transmission system. Unless you include Radar, of course, and even there, clutter is a damned nuisance.
But, in a simulation / game anything is possible.

 

[RadioEng]

It really is a huge software simulation we're discussion, in which "resources" can be increased by adding a couple of zeros to the end of the value of a variable.

Partially, more HPC-DSP networked with dedicated hardware to handle the grunt work.

We can filter out long term fluctuations with a filter having microsecond impulse response and measure all the parameters we could think of with the same speed.

It obvious from this statement that you cannot begin to formulate a design to compensate. As you have not worked with this type of hardware, you are not in a position to attempt sarcasm.

It sounds fun but it has wasted quite a lot of my time as I thought it was a serious, physical, application we were discussing. No wonder you were shy about declaring your qualifications situation.

So, you went for the complete ad hominem this time? Who do you think this reflects poorly on, me or you?

So as not to and on an ad hominem, I might ask what this statement is supposed to mean:
"Given the patterns, echos can be disregarded for trilateration purposes, but could be used to fill in gaps due to noise. So, they may prove more useful than a hindrance at the processing stage."
afik, multipath propagation has never been a positive aid to any form of radio transmission system. Unless you include Radar, of course, and even there, clutter is a damned nuisance.
But, in a simulation / game anything is possible.

Looks like you did end on an ad hominem after all. Anyway, its for an unrelated function.

 

[sophiecentaur]

"So, you went for the complete ad hominem this time? Who do you think this reflects poorly on, me or you?"
It reflects very poorly on the project, which is totally unsubstantiated without some backup of referenced facts or track record. Do you think you could actually sell it to anyone without either of those two?

 

[sophiecentaur]

Mods
Please lock this.
I need saving from myself. :)

 

[RadioEng]

It reflects very poorly on the project, which is totally unsubstantiated without some backup of referenced facts or track record. Do you think you could actually sell it to anyone without either of those two?

Once again, no one has asked for your opinion on feasibility. Its clear you do not have the experience to make such assessments.

Allow me to refer you to post #60:

What more do you need than trilateration of a spread spectrum signal in the sub-1000Hz range?

Two questions:

1. What are the potential sources of loss?
2. What is the expected accuracy?

Stop making up your own questions.