Well, I finally "Broke" my LS today!

[Nate The Surveyor]

I was running with the following settings:
Verify, With V6 Reset,
At least 2 engines
Confidence, 10
Consistency, 10
Validate, with at least 2 engines.
Stop, AFTER 45 Epochs. (Epochs, not seconds) I have thought about it for a while. I'm not sure how long it took, to get this shot, but, as I recall, it was about a minute, or so.

I was staking a corner. I got a pretty good fix, 10.6' ft away, in a pretty open area. Moved closer, to the corner, (Within 0.12') re-observed, and it said go south, some 4.5 ft.
I just rejected it. I did not save the bad shot.

My theory has been with BEAST mode, it ALLOWS you to get IN and OUT real fast, and the net result of this, is a SHORT time, from start, to finish. Somebody else mentioned, that VALIDATE at the end of an observation, allows as much time as possible, from beginning, to end of observation. This clued me in, that a SHORT observation time, start to finish, could bust it. So, I tried it!

Then, I asked, what are your favorite settings, and I have gotten a number of answers.

I knew I was pressing it. It was in a cedar tree, when it busted. When a GPS system works as good as the LS, well, first thing, is to ride it hard, and find out HOW to break it. I FINALLY found it.

Short observation times, beast mode. That got it.

(I was staking a corner, so I was doing my own checking.)

I think it is awesome, that it took me over 3 mos of owning this thing, to FINALLY figure out how to bust it!
I have managed to get it to bust 0.12' ft before. but, that was my best to date, before today.

How can you not like a JAVAD LS?

I'm excited about the next release, that's coming.

Some have mentioned, that they turn Validate off, set confidence to 20, and observe it multiple times.



Nate

 

[Matt Johnson]

Now you see I recommend using Verify and Validate with 600 epochs while receiving corrections at 5 Hz in areas of multipath: https://support.javad.com/index.php?threads/data-collection-settings-collect-points.1320/#post-11206

I always like to see fixed initiations in agreement separated by at least 2 minutes. With only 45 epochs, the initiations might only be separated by about 15 seconds.

Beast mode does allow you to get in and out fast, not because of the fast collection rate but rather because it reduces the the time for the RTK engines to fix and complete Verify and Validate.

 

[Nate The Surveyor]

I'm playing with it, to find out what is better.... One shot with 600 epochs, or 5 shots, of 120 epochs, and use the average. Which one will yield the MORE accurate result?
Will there be a material difference?
What do those of you think, who do long studies, of things? Ie, those who leave it in automatic mode, to work all night? or, for 24 hrs?
Is there a way to hook 2 LS's to ONE antenna, so that they are both using and getting the SAME data, so we could compare a single 600 epoch shot, with 5, 120 epoch shots?
Or, some other configuration? I get the feeling, that things were well tested, before beast mode, but, maybe not as much testing, afterward. Matt, what are your thoughts, on the above?

Thanks!
N

 

[Matt Johnson]

I'm playing with it, to find out what is better.... One shot with 600 epochs, or 5 shots, of 120 epochs, and use the average. Which one will yield the MORE accurate result?
Will there be a material difference?

I would not expect there to be much difference if the points are all collected during the same time period. If the 5 shots are collected at different times throughout the day then I would expect the averaged point to be more accurate.

What do those of you think, who do long studies, of things? Ie, those who leave it in automatic mode, to work all night? or, for 24 hrs? Is there a way to hook 2 LS's to ONE antenna, so that they are both using and getting the SAME data, so we could compare a single 600 epoch shot, with 5, 120 epoch shots?

Yes, setting it automatically collect points for a long period of time is a good way to conduct test. It is possible that 2 LS's could be connected to the same external antenna.

 

[Adam]

I was running with the following settings:
Verify, With V6 Reset,
At least 2 engines
Confidence, 10
Consistency, 10
Validate, with at least 2 engines.
Stop, AFTER 45 Epochs. (Epochs, not seconds) I have thought about it for a while. I'm not sure how long it took, to get this shot, but, as I recall, it was about a minute, or so.

I was staking a corner. I got a pretty good fix, 10.6' ft away, in a pretty open area. Moved closer, to the corner, (Within 0.12') re-observed, and it said go south, some 4.5 ft.
I just rejected it. I did not save the bad shot.

My theory has been with BEAST mode, it ALLOWS you to get IN and OUT real fast, and the net result of this, is a SHORT time, from start, to finish. Somebody else mentioned, that VALIDATE at the end of an observation, allows as much time as possible, from beginning, to end of observation. This clued me in, that a SHORT observation time, start to finish, could bust it. So, I tried it!

Then, I asked, what are your favorite settings, and I have gotten a number of answers.

I knew I was pressing it. It was in a cedar tree, when it busted. When a GPS system works as good as the LS, well, first thing, is to ride it hard, and find out HOW to break it. I FINALLY found it.

Short observation times, beast mode. That got it.

(I was staking a corner, so I was doing my own checking.)

I think it is awesome, that it took me over 3 mos of owning this thing, to FINALLY figure out how to bust it!
I have managed to get it to bust 0.12' ft before. but, that was my best to date, before today.

How can you not like a JAVAD LS?

I'm excited about the next release, that's coming.

Some have mentioned, that they turn Validate off, set confidence to 20, and observe it multiple times.



Nate

Nate, let er cook for a while and take some more of the pic's you been posting.

 

[Nate The Surveyor]

Thanks Adam. I got some pics this week. I was thinking, what about a "Mode" where the following occurs:
I set the unit up, (with Bipod) on a corner. It is set to 600 epochs. (As per Matt's suggestion) and I walk some 100' away, and dig with the metal detector, and it takes me WAY longer than I expected. (The client walked up, and..... you know, it took a bit of time)
So, it just sits there, at the end of 600, waiting for me to accept the shot. Doing nothing.
When, it COULD have highly refined that shot. In other words, IF it sits for over a X amount of time, WAITING for you to accept a shot, and you don't do anything, it kicks over, and starts an INDEFINITE observation.
The indefinite observation has settable perimeters, and gets a super shot.
Today, I set the LS to 3000 epochs, at 5 hz, out in an open shot. I was gone for maybe 30 mins. It COULD have come to the end of that 10 minute obs, and started another called "Indefinite Observation" when you get back to your unit, it tells you it went indefinite. That it got 9000 epochs, and you just hit STOP, and you get a pop up, that says VERIFY? Y/N. If you hit Y, it finishes with a VERIFY. IF you hit N then it acts just like you had hit STOP, before an observation was finished.
Wouldn't that be cool?
Ok, I gotta download pics.
Nate

 

[Adam]

What about the auto restart function Nate?

 

[Adam]

I will set it up for 600 and let it restart until I get back to it.

 

[Matt Johnson]

You can always set it too stop with the stop button and do a manual reset of the engines once you get back to it.

 

[Shawn Billings]

From what we've been told, there is a maximum number of epochs that can be averaged. I think it is in the 10,000+ range. I would probably go with auto accept, auto restart. If something goes wrong with the long observation, you lose it all. Better, in my opinion, to set it to Matt's 600 epochs with verify and validate, then set it to auto accept and auto restart. Then it will automatically accept the first shot if the RMS values meet your tolerance, store the shot, then start a new one and repeat the process. Then you can use cluster average to average all of the points into one.

 

[Nate The Surveyor]

Well, Adam, and Shawn, what happens, is we "expect" to be back in 5 mins. It takes 10, and the client takes 20. It could be busy for that time. You don't anticipate it. If there was a setting, that lets is just get a real good observation, when left un attended, for over say, 5 mins. When you get back, and find a long solid shot, on a useless point, you just REJECT it. No biggie.
Some have complained that the T2 always stores a point for DPOS. WELL delete it, if you don't want it.
Same here. It just gets a good shot, when left unattended, for too long. It's just a feature that would get used. And, occasionally, it would save time.

 

[Nate The Surveyor]

Stop with stop button seems to be my answer. .... Unless that 10k limit gets me....
Another thing that occurs to me, is that deal of manual rtk reset, as Adam above said.... Or... Start another observation, and watch the little white button, that tells you how far you are from the previous shot... With verify on, of course. This could give you that last verify, with that long time separation needed, to really get it right.
I'm rambling a little. It's excitement borne. It's exciting to use cutting edge technology.

 

[Nate The Surveyor]

I have been using the LS now for about 3 mos+ straight now. I have been wanting a setting that gives me a shot, in a very poor environment, that's not very accurate, but is for certain. I want it fast, sloppy, and very high confidence. (NOT a bad fix)
If it's off by 0.3 feet, that's fine. But, I want to KNOW if it is a foot off. Or more.
It seems that the answer is sort of "Blowing in the wind". The 2 are related, ie, STRONG confidence, and hundredth level confidence.... or a tenth or so.
What do you think about a confidence level of 30, and only 30 epochs?
We are talking about a shot on a fence corner, or on a well, or something NOT close to the boundary, but could later be significant.
I am multi-tasking alot... learning, WHILE performing on the job. To give my self "Learning time" (try new things) I have to "Work it in" while I work.
The user profile pre sets that are coming, will be handy.
Nate

 

[Shawn Billings]

I have been wanting a setting that gives me a shot, in a very poor environment, that's not very accurate, but is for certain. I want it fast, sloppy, and very high confidence. (NOT a bad fix)

I think that because of the way precision GNSS works (double difference carrier phase), that the solutions are sort of binary (one of two options): right or wrong. Think of an ever expanding sphere coming from the satellite (this is the wave front). This sphere passes the base. Then it continues to expand and passes the rover. We never really know how far the satellite is from the base. We know approximately, but not precisely. What we are interested in is the distance this sphere expanded from the time it passes the base to the time it passes the rover (or vice versa the distance the sphere expands from the time it passes the rover to the time it passes the base). How do we do this. Well imagine the sphere is blinking at a rate of 1575.42 MHz (1575.42 x 1,000,000 times per second). At this rate, each time the sphere flashes it expands 0.190294m (about 0.62'). Imagine now that the sphere doesn't instantly blink, but gradually passes from on to off. This is the fractional part of the wave (the phase). This is commonly figured at about 1% of the wave length. So the fractional part of a 0.19m wave length can be determined to about 2mm. I could be wrong about the 1%, but that's fairly close. So we add the cycles (blinks) and the fractional part of the wave (what the graduation is for the on/off for each blink) and now we have a distance from the base to the rover generally in the direction from the satellite to the base (not counting atmospheric between base and rover). We do this same process with all satellites in common view. How much distance was there from the base to the rover as generally measured in the direction of each satellite to the base. When a statistically significant number of these distances fall intersect within the space of a wavelength, you have a "fixed" solution. The ambiguities have been resolved, or, for our example, we have some confidence that the number of blinks for the sphere of energy coming from the satellites have been accurately counted between the base and the rover. A float solution is caused when the intersection of blinks could be in more than one place. This is caused by a miscounting of the blinks. What happens if the objects around the rover, reflect like a mirror the expanding sphere? So the ground reflects the signal back to the rover. How many additional blinks are there? If the signal is overhead, and the antenna is 2m above the ground, there could be 2m x 2 (antenna to ground, ground to antenna round trip) / 0.19m per cycle = 20 cycles (blinks).

With this in mind, I have never had much confidence in a float solution. There are processing techniques that rely more heavily on the longer code signal and use carrier to improve the accuracy. This is where submeter positioning comes from. But for fix vs float, it's a yes or no, right or wrong proposition, in my opinion. Having said that, for challenging environments I don't think the question is a matter of 0.1' vs. 0.3' usually. The question is fixed or float. If it matters, this is the object of your pursuits. Is this point based on an accurate determination of the cycles between the base and rover or not. Once this question is confidently answered, time on site and repeat occupations will improve the precision as needed.

 

[Nate The Surveyor]

"Once this question is confidently answered, time on site and repeat occupations will improve the precision as needed."

So, they are linked, somewhat permanently, by nature of their natural mechanics. Accuracy = time.

Nate

 

[Nate The Surveyor]

Thanks Shawn, for your detailed answer.

 

[Nate The Surveyor]

Shawn, is there a detailed answer, (Like the one above, that you provided) for the CONSISTENCY COUNTER? Thanks!
N

 

[Shawn Billings]

Shawn, is there a detailed answer, (Like the one above, that you provided) for the CONSISTENCY COUNTER? Thanks!
N

Yes, in the manual and in the help screens of the LS.

Consistency is an arbitrary number that weights each epoch based on the number of fixed engines. You've probably noticed that in open areas it's easy to get six fixed engines and in wooded areas it can be difficult to get two engines fixed. Based on this, if the engine count is low, the consistency counter advances very little. If the engine count is high, then the consistency counter advances more rapidly. The consistency level advances:

• 0 for 1 engine
• 0.1 for 2 engines
• 0.25 for 3 engines
• 0.5 for 4 engines
• 1.0 for 5 engines
• 1.5 for 6 engines

If you have a Consistency level set to 10, and you have 2 engines fixed, then it will require 10/0.1 = 100 epochs. At 5Hz, this is only 20 seconds. With 5 fixed engines, it will only require 10/1=10 epochs (at 5Hz = 2 seconds). With 6 fixed engines, it will only require 7 epochs 10/1.5 = 6.7 (at 5Hz = 1.3 seconds).

As I consider this, I will probably be using a much higher consistency counter now for control at 5Hz. Probably 50. So with 2 engines this will require 500 epochs (at 5Hz = 100 seconds).

Note that all counters must be satisfied. If I have the Consistency set to 50 and I have the time set to 600 epochs, then collect a point and at 500 epochs have a consistency of 50, the LS will continue to collect data until I have 600 epochs. Conversely, if I have the consistency set to 50 but only have the epoch count set to 50, then the LS will continue to collect data after 50 epochs have been collected until the Consistency has been acquired.

Basically consistency estimates the quality of the collection environment based on engine count and adjusts the time on site accordingly.

 

[Nate The Surveyor]

This really makes me think.
I have been running 10/10.
10 confidence.
10 consistency.
I can see raising confidence counter, would make it take longer. At the beginning.
I'm wondering part of this time spent, would count towards the 600 epoch (2 minutes) needed, to "never gotten a bad fix with, as per Matt Johnston's suggestion.
Or, would it work good, to use the consistency counter for this?
It really makes me wonder.
I for sure want to have a lot of presets, for this. I hope they are user namable, and designable.
I can see somebody in west virginia, or kentucky, wanting a different palate of settings, than somebody in arizona, or Indiana, or Colorado.
Like on the radio. I have gone back to a 1/2 wave antenna. Penetrates the hills, valleys, and timber of Arkansas better. 5/8 wave gave long range, but not as good penetration.