[greggold]

My Edge 3d plane has a single band R9 900mhz rx, that I use with a frsky tandem tx. It has a 3S lipo. Measurements were taken at 0.5 sec intervals, and with frsky logging the RSSI is a percentage.

At 207 sec the valid frame rate fell to 64%, one measurement only, with a level below 70 considered an incident. The RSSI remained above 70%. I looked for any anomaly in the power supply. With frsky logging you can simultaneously monitor 3 lipo cells. At 207 sec there is nothing happening, the rx voltage remains at 5.0v for the entire flight, and the lipo cells are all together, in the expected range.

I conclude there is no problem here, the R9 rx has been without a glitch for many months, and others have found R9 rxs, either ACCST or ACCESS,  work great at PRCAC.

I want to log the rx voltage on my planes with elrs rxs, but so far I can only monitor either the lipo voltage, or the rx voltage, but not both. I use the lipo voltage for deciding when to land, and am reluctant to give it up. So I need to find a way to monitor lipo and rx voltage simultaneously, and I do not want to add a flight controller to achieve this.

[greggold]

Another incident!

My slick is a 90cm span 3d plane with a elrs matek R24P rx; it has a 2S lipo that I connect to the vbat plug signal pin (in row7 of pwm pins) to monitor lipo voltage. In the following plots from monitoring a flight yesterday a burst of throttle led to a sharp decline in the rx signal-to-noise ratio (RSNR). As pilot I was not aware of any change in signal link quality, although I got the warning about the jump in tx power to 25 mw. 

Up to 129.5 sec into the flight RSNR was 10 or 11, and TPWR (tx power) was 10 mw. At 130s RSNR fell to 6 and TPWR was 10. At 130.5 s RSNR fell to 3 and TPWR increased to 25. TPWR remained at 25mw until 137 sec into the flight, when it returned to 10mw.   

Throttle setting has a range of -1000 to + 649 in the flight log. I divided values by 100, providing in the plot a throttle range -10 to the flight maximum throttle of 6.49 which occurred at 128.5 sec.
Lipo v. dropped from 7.1v at 127 s to 6.8 at 128 s; this is normally plenty of power for the esc.
It is clear from the plots that an burst of throttle at around 128 s into the flight triggered the fall in RSNR and the jump in tx power (TPWR).
What does this mean? A fall in RSNR caused a jump in power, as expected, however  a RSNR that fell to 3db means the signal is still 3 times stronger than the noise, and it seemed this was enough to avoid a loss of RSSI (rx signal strength indicator) and RQly/VFR (rx valid frame rate). It is unclear whether the incident was caused by a loss of signal-to-noise due to interference, or a drop in rx voltage, or something else.
 
What I need to do is to monitor the rx voltage. Unfortunately I am using the vbat on the r24P for the lipo voltage, and I have now decided to modify the slick and other planes to allow rx voltage monitoring instead. I use the lipo voltage for deciding when to land, and I am reluctant to give it up. However I now think rx voltage monitoring has priority.

[caffeine]

Interesting results. I have never seen a drop in LiPo voltage cause issues with the RSNR or cause the TX power to increase.

Having said that, I'm on 3S and only discharge my batteries to 11.2V (from 12.6V fully charged), I'm usually only using ~300mAh of my 850mAh packs. This is equivalent to 3.7V per cell, or 7.4V on a 2S setup.

BEC setups on ESCs usually need a bit of headroom to properly regulate the voltage. A good linear BEC with a low drop out (LDO) regulator needs around 300mV of headroom (so at least 5.3V to regulate to 5V) but older style linear regulators needed a few V of headroom. This may be why I haven't seen any issues running at 3S (using both linear and switching regulators) as I will always have heaps of headroom.

[greggold]

Interesting results. I have never seen a drop in LiPo voltage cause issues with the RSNR or cause the TX power to increase.

Having said that, I'm on 3S and only discharge my batteries to 11.2V (from 12.6V fully charged), I'm usually only using ~300mAh of my 850mAh packs. This is equivalent to 3.7V per cell, or 7.4V on a 2S setup.

BEC setups on ESCs usually need a bit of headroom to properly regulate the voltage. A good linear BEC with a low drop out (LDO) regulator needs around 300mV of headroom (so at least 5.3V to regulate to 5V) but older style linear regulators needed a few V of headroom. This may be why I haven't seen any issues running at 3S (using both linear and switching regulators) as I will always have heaps of headroom.

re: I have never seen a drop in LiPo voltage cause issues with the RSNR or cause the TX power to increase.

Good point. It seems a big coincidence that the large drop in RSNR happened at the time the throttle was increased. I thought that a drop in RSNR usually needed an increase in noise (interference), but here it looks like somehow the increased motor current led to the drop. My esc is a good Dualsky 12A unit, it is not near the rx to cause interference, and the drop in lipo voltage was small, though as you say I was running with each cell at 3.4v. But when I land this plane the voltage usually increases to 3.6 or 3.7 in each cell. I know this plane setup has been solid/safe over months of flying, so I plan to only add the rx voltage monitoring and no other change. If the rx voltage drops I will replace the esc with a switching unit.

The reason the tx power increases is apparently (from my reading) an algorithm in the tx that considers RSSI, RQly and RSNR. How the values lead to a decision to increase, and how much, remains mysterious. 

I have only started data logging a week ago, but now I know how easy it is to do I will keep going and hopefully use it to improve the setups.  My impression is drops in RSNR occur several times in each flying session.

[caffeine]

How the values lead to a decision to increase, and how much, remains mysterious. 

There's a reasonable explanation on the website, but I haven't sat down and tried to calculate it yet to see if it's working as intended

[greggold]

Here is a data log conundrum. As the pilot I experienced no radio signal problem.

This plot of data logging from a flight this morning shows several unexplained drops in receiver signal to noise (RSNR). The plane was a Flash 1m 3d profile with a 3S lipo, controlled by a RM Tx16 tx with elrs internal module and Edgetx firmware,  and a matek R24P rx. I set dynamic power on, which means a rx algorithm decides when/whether there is a radio signal issue and increases tx power whenever needed.

Each data series is modified so that all 6 series can be clearly seen in 1 plot.

The RXBat+20 shows that the rx voltage was fine at 5v throughout - none of the approx 600 measurements at 0.5 sec intervals deviated from 4.9 or 5v.

The throttle setting (THR) goes from -20 (no throttle) to +20 (full throttle).

From the THR you can locate the 2 blenders in this flight: at the start of a blender there is a vertical climb requiring full throttle, at the top I cut the THR to zero  to commence the vertical plunge and spin. For the first blender the THR was cut at 50 sec, the second at 225 sec.

The RSSI (signal strength) and the RQly (rx quality aka VFR) were within normal limits throughout the flight - not a single measurement was abnormal.

However the TPWR (transmitter power) which was at 10 mw throughout had 4 episodes of an increase to 25 mw. Each tx power increase was associated (maybe triggered) by a significant decline in RSNR (receiver signal to noise ratio), from 10 or 11 down to 4 db. A RSNR of 4 means the signal is still much stronger than the noise, and RSSI and RQly were not affected, but what caused the fall in RSNR?

The increase in TPWR means the rx algorithm thought there was a signal problem. Two of the episodes of TPWR increase were associated with a blender, the other 2 also seem linked to high THR.

Interference from the 5G tower or a nearby factory would not be in time with my blenders, so cannot account for the episodes of reduced RSNR.

Maybe the increased power in the vertical climb for the blender means interference from the esc reduces the RSNR, and that triggers the tx power increase? But my esc is not mounted very near by the rx or antenna.

[caffeine]

So if your signal strength (RSSI) is still stong, a drop in RSNR would seem to indicate that the noise is increasing, so the ELRS link increases power to ensure a signal is stilll stronger than the noise.

Perhaps this RF noise is what's causing issues with other RF links?

[greggold]

So if your signal strength (RSSI) is still stong, a drop in RSNR would seem to indicate that the noise is increasing, so the ELRS link increases power to ensure a signal is stilll stronger than the noise.

Perhaps this RF noise is what's causing issues with other RF links?

I don't think its a noise/interference issue.

I think I am closing in on a solution to the problem of blender-induced anomalies (rapid spins) in the telemetry that I describe in 2 posts to RCG. The said anomalies occur both with my X20 R10 system in my CY, as well as my edge-elrs system in my Flash.
1.
https://www.rcgroups.com/forums/showthre...ad/page326 
post #4890
2.
https://www.rcgroups.com/forums/showthre...S/page1648   
post #24717

I have done the posts because I don't fully understand the problem and experience shows there is a chance on RCG an expert will explain. Bardwell provided a clue:
https://www.youtube.com/watch?v=wzf0yQ8VAVA
He explains "telemetry lost" messages occur without loss of control input because there is a competition between telemetry and control. The control signal link is bidirectional, so elrs allows the user to vary the allocation of power between control and telemetry with the telemetry ratio.

It means if I set my Flash to a lower packet rate and different telemetry ratio it may solve the problem. My packet rate is 333 (full) and telem ratio 1:128, whereas Florian who does not have my telemetry issues uses 100 (full) and 1:32. So I plan to lower the packet rate and increase telem in the ratio.

Remarkably, the Frsky-ethos telemetry looks to have the same problem as elrs. I have never had a manifest control issue on the Crack Yak with Frsky R10 rx, but that may be because there is dual band in play, at 2.4G, and 900 mhz.

Why does the spin affect the telemetry? I suspect its the rapid variation in signal strength RSSI that some manoeuvres cause, and other telemetry values that are partially based on RSSI eg rx SNR do not have time to keep up at high packet rates. The telemetry needs more resources/higher telem ratio.

[caffeine]

The ELRS website does recommend a telemetry ratio of at least 1:32 with the 100Hz packet rate.

What does your TQLY reading look like at those points where the TX power is increased?

If the system misses a few telemetry packets it will also raise the TX power level.

To be proactive when telemetry is not received, Dynamic Power will also increase power one level for each missed telemetry packet, starting when two are missed back to back.

[greggold]

just to clarify, the experts on RCG I referred to in the previous post are people who actually wrote the software for frsky/ethos, and edgetx/elrs, they hang out in the above RCG threads and in the past they have answered my posts on RCG. I had no intention to bypass experts on PRCAC.

[greggold]

[quote="caffeine" pid='32092' dateline='1686622618']
What does your TQLY reading look like at those points where the TX power is increased?

If the system misses a few telemetry packets it will also raise the TX power level.

To be proactive when telemetry is not received, Dynamic Power will also increase power one level for each missed telemetry packet, starting when two are missed back to back.

re TQLY : I have not studied this yet! However for all my flights with logs TQly is present in the spreadsheet. Presumably the gaps and low readings in the rx telemetry in incidents when TPWR is raised will also appear in the
TQLY.

[greggold]

What does your TQLY reading look like at those points where the TX power is increased?

If the system misses a few telemetry packets it will also raise the TX power level.

To be proactive when telemetry is not received, Dynamic Power will also increase power one level for each missed telemetry packet, starting when two are missed back to back.

I just looked at TQly (see attached) and TSNR (attached). Neither show any relation to the 4 incidents when TPWR moved from 10mw to 25mw, associated with a reduced RSNR, and a reduction in RQly.

Its a good point you made about missed packets, I need to look into this as my impression is missed packets show in the csv file as blanks, not zero. I have a few logged flights with the crack yak (frsky_ethos RC system) that have quite a few blanks when the RSSI falls during a blender.   However during this particular flight there are no blanks. So even though the RQLY fell to 84% on several occasions, meaning a valid packet rate of 84% there were no missed packets.

[caffeine]

Interestingly I went for a flight today with my Radiomaster TX16S II with internal ELRS module, connected to a Matek R24-P6 receiver. Both running 3.2.1 firmware.

I have the radio set to 100Hz packet rate, 1:32 telemetry, 8 channel mode. The dynamic power is set to a maximum of 100mW.

In the flight, the RSNR dropped as low as 3 and the TX power never changed (from 10mW)

   

[caffeine]

Today I saw my first increase in TX power (momentarily) in response to a drop in RQly/RSNR.

There was no loss of RF signal, and no issues with control of the plane, ELRS smoothly transitioned to 50mW TX power and then back to 10mW immediately.

I didn't graph 1RSS as there was nothing of note in it.

   

[greggold]

On my Flash with a matek r24P I had drops in RSNR to a concerning level (like 2-4 including once to 0) on a number of occasions with no effect on RQly or RSSI, and I have moved the rx about 10 cm back to a different location which has resolved the situation. I have been logging many flights and I am preparing a set of graphs which may help define a standard for what log features are OK and what are not OK. Its become a bit time consuming.

I think maybe you might consider to do something if RSNR is dropping from 12 or 14 down to below 5. I would say in your plot the RSNR is dropping to 4, should be OK but I would probably consider trying a different location for the rx, if that can be done easily.

I think logging will change the way we assess when models are safe to fly, once there is sufficient experience.

[greggold]

I have 2 plots that are similar to your plot, on the Slick with a matek T24P6.
In Slick #201, RSNR dropped to 3 at 86 sec and at the same time TPWR increased to 25 where it remained for 1 second (2 green dots); at 242 sec RSNR dropped to 1, with no effect on TPWR and no problem with RSSI or RQly.

In Slick #113, RSNR dropped to 2 at 122 sec; there are a few other low values of 3 and 4. But no TPWR increase was triggered, and again there is no problem with RSSI or RQly.

I had a theory that blenders were affecting TSNR, but that is not really supported in these plots. Throttle cut is -20 on the plots and always indicates the top of the climb, followed by the plunge and spin.