This tutorial uses the Blackbox log of the same four-axis machine to compare the RC command spectrum of the Jumper T15 and the modified FrSky X9 Lite, explaining how to find the high-frequency noise of a certain rocker axis. It provides a controlled comparison method, not a judgment on the merits of all remote controls of the same model.
First make sure the two logs are comparable
Before testing, back up the radio and flight-controller configurations. Bind both radios to the same quadcopter and keep the receiver link, ExpressLRS packet rate, flight-controller settings, and Blackbox logging rate unchanged. The video uses 1000 Hz ExpressLRS and 2 kHz Blackbox sampling to give the log enough sampling resolution to observe the 1 kHz command stream.
Both flights should include similar smooth stick movements and several comparable turns. Label the logs immediately after recording. Do not directly compare a steady hover with aggressive racing, because the spectral differences may come from the pilot’s actions.
Inspect RC commands in FPVSIM Blackbox Analyzer
Load the first log, open Input and response, select RC commands, then switch to Spectrum. Inspect roll, pitch, yaw, and throttle in turn, looking for an elevated high-frequency noise floor, narrow spikes, or one axis that looks noticeably noisier than the others.
Keep the same view and spectrum scale when loading the second log. The modified X9 Lite in the video has a higher noise floor, particularly on pitch. The Jumper T15 is cleaner in this matched test, but that conclusion applies only to the two devices tested.
Interpret high-frequency spectral content cautiously
Deliberate human stick movements are concentrated mainly at lower frequencies. In this comparison, the creator regarded substantial content above about 50 Hz as suspected noise and content above 100 Hz as undesirable. These are observations used in this diagnosis, not universal pass/fail criteria for every radio.
Narrow lines near 1000 Hz may also relate to link timing or sampling behavior; they do not by themselves prove a damaged gimbal. Compare all four channels, the two matched flights, and actual control feel before deciding to dismantle the radio for inspection.
Find the physical cause first, then retest
After finding an anomaly, first recalibrate the radio and inspect the gimbal sensor area, connectors, ribbon cables, dust, wear, impact damage, and firmware. Feedforward smoothing can reduce the amount of command noise entering the control loop, but extra smoothing also adds response latency. Do not use it to hide a real hardware fault.
After every cleaning, repair, reseating of a connection, or calibration, record another matched flight with the same quadcopter and settings. Attribute the change to the repair only when both high-frequency behavior and actual control feel improve.
Operation steps
- Match the two test configurations
Keep the quadcopter, receiver link, packet rate, flight-controller configuration, and Blackbox logging rate unchanged; change only the radio.
- Record two repeatable flights
Use similar smooth inputs and turns with both radios and label the logs clearly.
- Open the RC commands spectrum
In FPVSIM Blackbox Analyzer, open Input and response and select RC commands and Spectrum.
- Compare high-frequency behavior on all four axes
Use the same scale to inspect noise floors and spikes on roll, pitch, yaw, and throttle.
- Load the second matched log
Keep the analysis view unchanged and compare the other radio. In the video, the modified X9 Lite’s pitch axis is particularly suspect.
- Inspect the hardware and test again
Calibrate and inspect the gimbals, ribbon cables, connectors, and firmware. After each change, record another flight under the same conditions.
FAQ
Does the video prove that all Jumper T15 radios have cleaner signals than X9 Lite radios?
No. It compares two specific devices, including an older modified X9 Lite whose pitch gimbal may have a problem. The method is reusable, but the model-level conclusion cannot be generalized.
Why use 2 kHz Blackbox logging with 1000 Hz ExpressLRS?
The video uses a higher sampling rate to represent the 1 kHz command stream and perform spectrum analysis. An insufficient logging rate may miss or alias high-frequency content.
Is everything above 50 Hz necessarily gimbal noise?
No. The 50 Hz and 100 Hz figures are the creator’s diagnostic references for these two logs. Link timing, sampling, filtering, and other electronic factors can also produce high-frequency features.
Can Feedforward smoothing fix a noisy radio?
It can reduce the effect of noise, but may add latency and hide a hardware issue. If the source can be inspected, calibrated, or repaired, address it first.
Why compare all four channels?
An abnormal axis is easier to spot when compared with the others, helping narrow the problem to a specific gimbal sensor, connection, or calibration.
Full timeline transcript
Transcripts are arranged according to video time, making it easy to quickly locate the explanation content. Transcript language: English.
Hello everyone, Jay from FPVSIM here. Today we're going to take a look at two radios. Now from the build quality or the economics, which are of course still very important, but those things are not one good ad. And you probably have seen a lot of review videos about them already. So instead, what we're going to take a look at is the signals, the quality of the signals. So just a brief introduction of the two radios we have.
This one is my DIY radio, which has the X9 Lite Core with the AG-01 Gimbal. It's one of the famous gimbals in the research community. And the shell or ergonomics are the classic JR radio, which is very similar to Terranis, I think. Let's consider it our control group. The other radio is the one I got from Jumper recently, which is
There is the flagship, T15.
What I did is essentially fly the quad with one radio, collect the Blackbox data, and then fly it with the other radio, and also collect the Blackbox data. And then we take a look at the Blackbox data points. So both radios run ExpressLRS 1000Hz. In order to capture signals slash noise up to 1000Hz, we have to set the Blackbox sampling rate to K by the Shannon theorem.
Don't ask why, just set it to 2K so that you can actually sample 1K signals accurately. Once we done that, here I recorded some Blackbox. Let's take a look at the Blackbox together. I'm using the FPVSIM Blackbox Analyzer to do the analysis. So here you can see I recorded some Blackbox here. One is flying the same quad with this radio, my DIY radio with H01 gimbal. And then the other two are flying the same drone with the
Let's first take a look at the Jumper T15. Specifically, I want to look at the RC commands. We actually have one view for the RC commands, so just click the drop-down here, go to input and response, and here we only care about RC commands. Specifically, we're going to use the spectrum graph for the analysis. If it's the first time you hear about Spectrogram, here's some explanation from Gemini with pictures.
So what is the spectrum graph? Imagine you have a magic paintbrush that can paint sounds. So you have a sound here, right? We all know that, you know, with a sound, you have bass, your higher pitch sound, right? The spectrum graph essentially tells you how many of those voice falls into, you know, the bass part, how much of your voice falls into the higher pitch part. So imagine, you know, if I just speak like this, right? This part of the voice will probably have a very high
If I speak like this, you'll see higher numbers from the pitch part. OK, that's just my layman's explanation of spectrum graph for voice. But essentially everything else is similar to voice. In here, the RC commands just jumps around, jumps around. Essentially it's also like a voice, just at a different frequency. Since the signal overall is actually very clean, let's actually zoom in a little bit.
Let's actually drag it to the maximum. So essentially here you can see you know majority of the signals which are you know we move the we move the gimbals majority of the signals actually falls under let's see 20 you know 25-inch right majority of the signals those are probably you know my movements probably won't even be
But you definitely don't want to see anything above 100 Hz. There's also a little spike here at 1,000. Those are probably the jitter, since we're running 1,000 Hz, right? These are the jitter values. So this is the roll axis. Remember, we are comparing four channels, roll, pitch, yaw, and throttle. And look at the pitch as well. Pitch is similar to roll, very similar. And then yaw, even less movement. And then throttle.
Throttle, I guess we moved a lot, so it has more data here. Overall, that's the spectrum graph of the Jumper D15. Next, let's take a look at the data of our DIY AG0 video. Okay, instantly you can see that it has actually much higher spikes above like 100Hz here. Remember, let's compare it back to the Jumper D15. Compared to that, right, this is a lot cleaner. Let's come back to the AG01.
This is a lot more signal on the higher frequency side, arguably a lot noisier too. You can see these little spikes, like those spiky hairs here. Those are all noises. That's the pitch axis, and if we take a look at the roll axis, it's slightly better, but still...
You can see all these hairy spikes. Those are not good. You definitely don't want to see those in your radio signals. Coming back to pitch. Pitch is especially bad. Compared to roll, pitch is so much worse. It almost feels like my hole sensor is 40 on the pitch axis. Whereas the roll is actually kind of okay compared to pitch. And let's take a yaw as well. Yaw is also okay. Roll and yaw is actually
OK, although the roll is actually hairier and yaw is a little better. And then throttle, also similar to yaw. So it seems like one of my, the pitch axis on my right gimbal is 40. I don't know whether it's a quality issue or I dropped it and broke something or some dust got in there and broke it. No idea. But from the signal you can see that it has a very big difference.
Especially the difference between these two axes. That's what worries me. And then you might ask, Jay, do you feel anything when you fly? I do actually. So with this radio, if I set my FeedForward boost to too high or FeedForward smoothing to too low, I would definitely get uncontrollable drone jitters. That there's just no way you can fuel it out. As a result, I have to use very high FeedForward smoothing.
I think at least 80 to 85 with 1000Hz. But you know, a side effect of that is with high smoothing you get more latency. If more latency, what's the point of getting to 1000Hz? Why not just use 500Hz, right? But with this video, I can actually set my FeedForward smoothing to be a lot more aggressive. Just to show you what my current settings are. Okay, got my quad-plugging.
Here's my current settings. My FeedForward smoothness is 60. Instead of, you know, with this video, I have to set it to at least 80. And then my averaging is actually turned off, boost is 8, gyro reduction is set to 0. And then my receiver, I've actually experimented with this number recently. I was at 25. For this Blackbox, I actually recorded at auto factor at 25, which is actually very low.
But it seems to fly fine for me, at least in my art. With this video, I do feel I get a lot faster too, with these more aggressive settings, which has something to do with ergonomics for sure, but the signal quality definitely helped as well. Cool, hopefully you're finding it useful. If it does, please let me know and leave a like and subscribe. This is Jay from FPVSIM. Thanks for watching and I'll see you next time.
