I'm wondering... Average of RTK, AND PPK?

Duane Frymire

Active Member
I'm a proponent of documentation, so for shorter duration I would not want to average the rtk and ppk. I don't have documentation on the ppk process like I do with the rtk (screenshots showing the results of the analysis). I would also think the rtk process was developed with shorter duration in mind, while ppk envisioned longer occupation (at least 10 minutes or so). I have found usually rtk position is more reliable on shorter duration occupations. So, if I happen to get a solution with both on say a half hour occupation, then I would consider averaging them. But I don't know, maybe Javad could weigh in on how these algorithms/processes were developed? Is one designed to be more robust for shorter duration than the other? Still, the ppk I don't have really anything to print and show the court, so I like rtk for short duration (at least in the Javad world, don't know if any other manufacturers have anything, which is why I have Javad of course). Long duration I can point to procedures in textbooks calling for certain time on point for ppk solutions of differing reliability.
 

Bill Eggers

Active Member
I generally count on my quality control being agreement between my RTK and PPK values. However, I have had two cases recently where they did not agree by as much as several feet. Both times the DPOS choice was the RTK and that did not agree with record dimensions, but I found that the PPK did.
These have been fairly obstructed sites that were a struggle to get a fix.
There are also no statistics given for the PPK solution like there are for the RTK.
 

Matt Johnson

Well-Known Member
5PLS
I generally count on my quality control being agreement between my RTK and PPK values. However, I have had two cases recently where they did not agree by as much as several feet. Both times the DPOS choice was the RTK and that did not agree with record dimensions, but I found that the PPK did.
These have been fairly obstructed sites that were a struggle to get a fix.
There are also no statistics given for the PPK solution like there are for the RTK.

Were you using the default Boundary Action Profile? If you let it complete then there it is very unlikely produce that much error.

HRMS, VRMS, number of epochs and number of satellites are all given for the PPK solutions. What statistics are you saying are missing?

PROCESSED-POINT-INFO-SCREEN_20180820-12.14.32.png
 

Bill Eggers

Active Member
Yes, I am using the boundary profile, mainly default, but have it set to include all DGPS float/fixed and stop is at least 300 sec with stop button.
The statistics would be the base/rover. In the screen you show above, if I picked the base/rover button under the RTK solution I would get something like this:
00484_Base___Rover_Statistics_20180828-17.05.38.png

but when I pick the button under the ppk solution I get this:
00484_Base___Rover_Statistics_20180817-17.53.45.png

Here is the point info screen for one of the shots I am referring to where the RTK and PPK differ by a lot and the PPK turns out to be much better for the expected location by deed dimension:

00484_Processed_Point_Info_20180817-17.51.58.png

The HRMS and VRMS for both RTK and PPK look very good, so I would expect to have good solutions either way, but the position is different by 1.5 foot horizontal.

Here is the point info screen for another shot on the same project with good correlation between RTK and PPK:
00484_Processed_Point_Info_20180817-17.53.40.png
00484_Base___Rover_Statistics_20180828-17.05.38.png00484_Base___Rover_Statistics_20180817-17.53.45.png00484_Processed_Point_Info_20180817-17.51.58.png00484_Processed_Point_Info_20180817-17.53.40.png
 

Shawn Billings

Shawn Billings
5PLS
1136 only has 146 seconds for RTK. That's not using the boundary profile, at least not allowing it to finish. The boundary profile should go for at least 180 seconds, plus another couple of seconds for validation. More in canopy.

Also, do not use accept all (DGPS and Standalone). Use Fixed only.
 

Shawn Billings

Shawn Billings
5PLS
RMS values are helpful within so long as you don't have a blunder (bad fix). It's like having a five arc-second total station and repeating angles, but you backsighted the wrong point. The RMS of the angle set will look great, but the angle will be wrong because it includes a blunder. You must start with proving the fix is correct, and then you can start relying on statistics to tell you how good it is.
 

Matthew D. Sibole

Well-Known Member
5PLS
Bill, I recommend not accepting an RTK shot that has less than 240 seconds between fixed epochs that agree. I have seen a bad fix last for more than 240 seconds only 1 time in 4 years.

Also I highly recommend taking multiple shots on every corner. It helps identify if you have a bad fix or a blunder. It also helps increase your relative accuracy. It will take more time but I never have to go back and shoot points again.
 

Nate The Surveyor

Well-Known Member
Another question. Was this near a large tree, or such?
It's not the bad fix that bothers me. It's the bad one, that slips by, undiscovered. We all get in a hurry....
 

Matthew D. Sibole

Well-Known Member
5PLS
I keep my confidence level set at 25. Rarely do I leave it set long enough to go all the way through the collect process. If I get one group that has more than 240 seconds in it I generally accept it and start the shot again. The shot I was speaking of in my previous post I had a few fixed epochs that agreed with each other with 411 seconds between them. This was probably the worst place I have had it. That fix was off by about 0.75’. I eventually did end up get a repeatable shot on it with more than 1900 seconds between fixed epochs that agree and was able to repeat it with a shot that had more than 500 seconds.

My Pdelta at the top left was the calculated corner. This ended up being the good shot. I would have to go back through my project and pull the screen shots to show it.
 

Attachments

  • 59DD7AFA-87C3-4FB5-A812-8D5CDFE2D99B.jpeg
    59DD7AFA-87C3-4FB5-A812-8D5CDFE2D99B.jpeg
    1.1 MB · Views: 339
  • 35A10E0F-CA7E-463D-B77B-6D310F6236A4.jpeg
    35A10E0F-CA7E-463D-B77B-6D310F6236A4.jpeg
    895.6 KB · Views: 319

Darren Clemons

Well-Known Member
Bill, I recommend not accepting an RTK shot that has less than 240 seconds between fixed epochs that agree. I have seen a bad fix last for more than 240 seconds only 1 time in 4 years.

Also I highly recommend taking multiple shots on every corner. It helps identify if you have a bad fix or a blunder. It also helps increase your relative accuracy. It will take more time but I never have to go back and shoot points again.
Exactly. The issue that most of us have in the deep, dark corners of the world where we take this wonderful little machine, is that many, many times, no matter how long you stay on “one session” the LS just won’t go through all three phases. I know the “directions” say we should do that, but it just doesn’t work some places.

For that reason, as soon as “session 1” gets above my set 15 minute minimum for PPK, or slightly past I’ll stop and store that point. Hopefully there is an RTK bucket in there that has, as we’ve all said, over 240 seconds between epochs. If that’s the case, I’m 99% sure already I’ve got a good point - if not, that point just becomes a PPK option in the office.

I immediately then start another session. If I had over 240 seconds in the first session, then I’ll have DTL white box as a guide for this session. If I get any two epochs, again, over 240 seconds in this session, and DTL is showing me 0.05’ or so - it’s 100% and I’m outta there. Now, if I had no “solid” RTK on that first session, then DTL is no help. I most times though, have a good “design” point from record surveys that I can also have in one of the Pdelta boxes. Depending on how well I’m adjusted to the old survey and it’s accuracy, I can use that box to get quite a bit of feedback on any session.

Last option, for us is, session 1 and session 2 (both 15-20 minutes of PPK) neither got any “good” RTK where DTL nor Pdelta to a design point checked, then the only option is starting a third session. If that completes with still no visual confirmation of a viable RTK solution the choice is to hope two of the three PPK’s check back at the office or try a fourth session.

I know it’s been discussed quite a lot of preference of one 45 minute session or three 15 minute sessions but for me, I’ll take the three different sessions every time so as to have much more data “sets” and things to check “on the fly”. Two separate RTK solutions in two completely different sessions, each with over 240 seconds of separation that both match within 0.03’ are much more comforting to me than one session with 600 seconds (although anything with 600 seconds is money ;) ). I will often also change rod heights and/or orientation after each session to get the LS in a slightly different angle to the signals if it is really struggling to get good RTK.

That’s just our way that we’ve grown to like, but with the LS everyone can have different flavors of cakes and icing and it can all be spread and mixed however you prefer. That’s what makes this baby the most versatile equipment I’ve ever used!!
 

Bill Eggers

Active Member
The Base/Rover Statistics screen works properly for the PPK solution in the latest build of J-Field when I try it. Are you still on the WinCE OS?
The Base/Rover Statistics screen works properly for the PPK solution in the latest build of J-Field when I try it. Are you still on the WinCE OS?
Yes, have not updated yet. With the hiccups some have been reporting, I have hesitated to do so.
 

Tyler

Member
1136 only has 146 seconds for RTK. That's not using the boundary profile, at least not allowing it to finish. The boundary profile should go for at least 180 seconds, plus another couple of seconds for validation. More in canopy.

Also, do not use accept all (DGPS and Standalone). Use Fixed only.
When is it appropriate to use the DGPS and standalone options? Let's say I am in dense canopy and only getting an occasional fix that is not enough to verify in a reasonable time frame... I know I should be taking redundant measurements for 10-20 minutes each so that we can get processed results that hopefully match. In situations like this should these difficult shots only accept fixed solutions as well?
 
Top