New FrSkyX protocol

The interpretation of the RSSI telemetry byte was fixed with this pull request as discussed above . Nothing changed about the protocol itself. I've attached a graph showing the difference between the old and new displayed telemetry values.

RSSI is in my case very jittery moving around a lot....
I can maybe get it at 100 but not above certainly not 125, and tried indeed if is there swamping etc. RSSI it drops very quickly to 80 some meter away. In the past implementation max was around 110, what I could achieve. tested x4r, XsR

With quad static on table and tx static RSSI is very jittery and moving fluctuating around al lot.

Just would like to know someone did flights using lastest nightly.

Have you adjusted fine frequency based on LQI telemetry value? Should be able to get down to about 50.

Based on the graph if you were seeing a max RSSI of about 110 then the max you'll see now is about 90. The new RSSI will drop off more quickly as the received value changes which could make it look more jittery.

Don't have anything flying frskyx at the moment but just did a bench test with the latest nightly and everything was fine. Saw the expected RSSI values. Using a Devo 10 with 4in1 module.

hexfet wrote: Have you adjusted fine frequency based on LQI telemetry value? Should be able to get down to about 50.

Based on the graph if you were seeing a max RSSI of about 110 then the max you'll see now is about 90. The new RSSI will drop off more quickly as the received value changes which could make it look more jittery.

That's right, the signal fall from 110 to 90 near the model.

By now my AD2GAIN was set to 100 and the freq-Fine to 0, the xsr green light is solid with this setup.
Can u tell me how to adjust the fine frequency ?

Fine frequency adjustment is in the protocol options - click on the protocol name to see the options. Change it in 5-10 increments.

Use the LQI telemetry value on the telemetry test page to evaluate the changes. Should be able to get LQI down to around 50.

My 4in1 module works best with a fine value of about -15, while the single module likes -40. I set power to 1mW while doing these tests.

Hi Hexfet, maybe it could be the issue, as indeed did not tune that....
but no sure what is the effect of off center towards range, but.

The Rx with the failsaves today, I found now that it binds over range -22 to +105 so I set fine variable to +42
www.banggood.com/IRangeX-RX16-16CH-Recei...-X12S-p-1090120.html

Original Frsky X4R, I have my center fine +12.


Howver I I watch LQI it is indeed around 49-50, but it is that over very wide range of freqency shift, (I turn off on tx after every change) So not so clear how to find the center using that.

One more question shall I set my Out of range warning at 43? Same what using with d series ? Any recommendation, maybe will test at the flying field tomorrow.

I'm sorry, but in my case, it doesn't work well. With the last nightly, my range with 1mw doesn't exceed 3 meters against before more than 25 meters in my house with big thick walls.
You can see it in the footage.

New build until 1:30 and after the build before.

hexfet,

It seems problems here are from finding correct fine variable.
Maybe you can introduce changing fine variable on the fly whitout switching TX off.
This is already implemented in multiprotocol.It is easier find it like this.
''
Option=fine frequency tuning(in multiprotocol is called option). This value is different for each board. To determine the option value, find the two limits(min and max of fine variable) where the RX loses connection(or the LQI change consistently to bigger values than 50) then set the option value to half way between them.''

so the fine variable = (fine max lost -fine min lost)/2.

@Sfersystem, I can confirm your finding, I do not trust Frsky X protocol currently....
Previously I had good range with X protocol, but now it seems to fail early range....

The precise fine tune of the signal likely makes a difference to but I expect that it is more no 1500m instead of 500m.... not 50m?
Also stock Frsky there is some tolerances in X tal parts etc, I assume that when +/-10 should be ok? How Frsky deal with it?

midelic wrote: (or the LQI change consistently to bigger values than 50).. .

Speaking of LQI, the lowest I can manage to get it is 49, with the threshold being 50 am I safe?

Thanks for the suggestion midelic. It makes adjusting fine freq much easier. I've pushed the change and it will be in the next nightly build. With this change just select the fine freq protocol option and hold down R/L to increase/decrease the setting quickly until bind is lost.

I tested the change in two Devo 10s, one with 4in1 and the other with banggood cc2500. The 4in1 stayed bound for freq settings about -111 to +7. The BG module range was -80 to 50.

The LQI stays around 49-50 for the entire bind range, with a sharp transition upward around the bind loss freq values. This matches the LQI graph midelic referenced earlier. Don't think it ever gets much below 50.

As for range the 4in1 and BG module performance was the same. The most I could test tonight was about 35 meters (through an exterior and two interior walls). At 1mW power the bind was solid and RSSI was bouncing around 50.

Fernandez, you could try the easier freq adjustment with tomorrow's nightly build but my guess is your range issue is some other problem. It would be good to hear from some others using FrskyX with the latest nightly but it's working well for me.

sfersystem wrote: I'm sorry, but in my case, it doesn't work well. With the last nightly, my range with 1mw doesn't exceed 3 meters against before more than 25 meters in my house with big thick walls.
You can see it in the footage.

The video shows the RSSI value causing an alarm, not bind being lost. The reported RSSI value has recently been fixed to correctly interpret the value sent by the receiver. Use the graph in this post to adjust RSSI alarms accordingly.

Unless it is losing bind at short range...

Guys what was changed was the RSSI interpretation.
The previous one was wrong,With that you cannot make connection fail in base of RSSI,That would stay on higher values even you cover completely tx antenna as you can see on the graphic above.
That was very dangerous because it can cause false security that you have good connection and cause some problem in the field or hurt somebody.
I would not have noticed before because I don't have Devo TX but some Devo TX users complained to me about that in private so for that I let you know.
Now you have RSSI alarm before not.What you need is to adjust RSSI alarm properly not based on previous one.

Another thing about range ,In my experience with testing.I found many times problems with antennas even new ones.
It looks the same 2dB antenna rubber duck but they are not .I have 4 antennas scrap now ,I tested them with RF explorer could not believe.The signal power dropped fast in few meters compared with good one.
So when you do testing make sure you have a spare antenna to get the right conclusions.
Also antennas might get damage when tx fall on antenna or when you bent it may times horizontal/vertical., the wire inside may get broken.

I'll test that build too. I had no range issue with the one from Jan 20th; hopefully that will be the case with the new one as well.

hexfet wrote: Thanks for the suggestion midelic. It makes adjusting fine freq much easier. I've pushed the change and it will be in the next nightly build. With this change just select the fine freq protocol option and hold down R/L to increase/decrease the setting quickly until bind is lost.

I tested the change in two Devo 10s, one with 4in1 and the other with banggood cc2500. The 4in1 stayed bound for freq settings about -111 to +7. The BG module range was -80 to 50.

The LQI stays around 49-50 for the entire bind range, with a sharp transition upward around the bind loss freq values. This matches the LQI graph midelic referenced earlier. Don't think it ever gets much below 50.

As for range the 4in1 and BG module performance was the same. The most I could test tonight was about 35 meters (through an exterior and two interior walls). At 1mW power the bind was solid and RSSI was bouncing around 50.

Fernandez, you could try the easier freq adjustment with tomorrow's nightly build but my guess is your range issue is some other problem. It would be good to hear from some others using FrskyX with the latest nightly but it's working well for me.

Idk how to explain better. why is not understood.
You may have a range of Lqi with values around 50/49. this is the range of peak values of the channel where is the maximum power output of the channel.The channel ma thematic representation is like a Gauss curve.
The Lqi grapic representation is like an inverted Gauss curve so is taken into account smaller values.
Because the fine variable adjust the frequency is small steps(fine) it appears the range is large but frequency related is small, the top of the curve.

What you need to do is to find the middle of that range and the value of the fine variable in that point.
That is your center of the channel with maximum output.You are safe in the middle of that range.

Cereal_Killer wrote:

midelic wrote: (or the LQI change consistently to bigger values than 50).. .

Speaking of LQI, the lowest I can manage to get it is 49, with the threshold being 50 am I safe?

Just to be practical, what is the margin you need to tune the, fine freq? How many steps off would be acceptable?
I assume there are tolerances in X-tal even in original Frsky TX and RX modules?

Anyway need to do more testing, how do we explain the input of SFERSYSTEM above?

@ SFERSYSTEM, can you confirm you use the same TX-RX combination with your test? As it seem between different FRSKY RX fine tune could be +/- 50 steps off, was it tuned?.

Fernandez wrote: @ SFERSYSTEM, can you confirm you use the same TX-RX combination with your test? As it seem between different FRSKY RX fine tune could be +/- 50 steps off, was it tuned?.

The test has been done with the same tx-rx combination at the same time. I flashed the new nightly, done the test, record pause, re flash the old nightly, record play and done the test.
Playing the fine freq seem not to change something to the range.

Edit : just try to find the best Freq-fine, and i can bind between -60 to 60 so at 0, this is the optimal value for me.

Edit 14:40 : Since i'm using frskyx protocol, i've always stange thing like in the video from last week (distance 100 to 150m)

Hmm how can we explain that, guys?
Are we sure not some unexpected bug sneaked into X protocol ?

None of the recent changes would affect protocol performance.

sfersystem your first video just shows the new RSSI values opposed to the old ones. The RSSI interpretation has changed.

For the second video you say you've always seen that behavior with FrskyX, so it's nothing new? Do you have any data logging from the quad or deviation to see if RSSI is dropping when it happens?

hexfet wrote: None of the recent changes would affect protocol performance.
sfersystem your first video just shows the new RSSI values opposed to the old ones. The RSSI interpretation has changed.

Right, but i loose the signal (number in black) after 3 meters instead having 85d 25meters in the house with the old nighlty

hexfet wrote: For the second video you say you've always seen that behavior with FrskyX, so it's nothing new? Do you have any data logging from the quad or deviation to see if RSSI is dropping when it happens?

No, i haven't any data, i try to fly not too far waiting a better release (I am aware that this is a nigthly and not an official release) but i can try to take some if u need.
I don't know if a lot of people have problem like these with the XSR, can't find range test on youtube or in forums.
XSR is not a long range receiver but 150 meters is not enough i think. Maybe should i buy an X4R-SB that frsky says it's a full range receiver.

Another strange thing is when i flash the last nightly, i got several times, message with missing multi-module or CYRF module is missing.

Thanks for help and the good job.