Spectrum analiser center frequency is wrong?

[norbre-hu]

Hi,

Based on the manual, the "The current VFO /Memory frequency will be the center frequency of the spectrum sweep" but I find that the set frequency is on the right edge of the spectrum, not the center.

For example if the VFO is 446,00625MHz, channels are 128x, frequency step is 1kHz, I see 445,94225 +/-64,00k under the graph.
If I transmit with another PMR radio on channel 1 (446,00625), the graph shows the transmission at the right side of the graph instead of the center.

Is this a bug or did I set something wrong?

Thank you

[armel]

Hello,

Here is a short video 😌

spectrum.mov

Armel, F4HWN

[norbre-hu]

I see.. I've tried, and as I saw if the frequency step is higher than 1kHz, the selected VFO is in center of the spectrum.
The phenomenon I described occurs when if the frequency step is equal or less than 1kHz, in this case the selected frequency is on the right side of the spectrum.

[armel]

The real question is: why choose a frequency step of 1 kHz? The frequency you gave as an example is that of the first PMR memory channel. The PMR band uses a 12.5 kHz step. On the amateur UHF and VHF bands, as well as on the AIR bands, the steps are clearly defined. It’s better to stick to them rather than listening outside of them. It will always be more efficient.

[norbre-hu]

I am a very beginner in this radio thing, thank you for this informations.
Btw I tried to listen ISS on 437.8, and as I know there is a doppler shift, what is max +/-10kHz, because of the speed.
This shift is continously changing depends on the position of ISS, so that was my idea I use the spectrum analyser with low freq steps to catch it.
For 6 sec I've heard them on the top of it's path, then I was realised the signal is on the very right side of the spectrum not the center.
This is why I tried to figure how it works with the help if an another PMR radio.

[jflinchbaugh]

FYI, I am able to reproduce this behavior (center frequency not centered when setting the step < 5khz) on 4.2. Since it's an available configuration, it should probably be made to work as expected. I've looked at this code previously, so I might have another look sometime soon.

[armel]

Compile and test with this patched spectrum.c, if you want.

[jflinchbaugh]

With this patch, it seems about the same (or maybe worse): just changing up and down through the step-sizes, even larger ones, shifts the center around and can sometimes push it the whole way to the left or right.
Here's the diff between the gist and 4.2 code:

diff --git a/app/spectrum.c b/app/spectrum.c
index b01a2bb..31d2dc8 100644
--- a/app/spectrum.c
+++ b/app/spectrum.c
@@ -369,7 +369,25 @@ static void ResetPeak()
     peak.rssi = 0;
 }
 
-bool IsCenterMode() { return settings.scanStepIndex < S_STEP_2_5kHz; }
+bool IsCenterMode() { return settings.scanStepIndex <= S_STEP_2_5kHz; }
+
+/* Added prototypes + helper to keep window centering consistent */
+uint16_t GetScanStep(void);
+uint16_t GetStepsCount(void);
+
+static inline void RecenterWindowToVfo(uint32_t f)
+{
+    if (IsCenterMode()) {
+        // In centered mode, currentFreq represents the center frequency
+        currentFreq = f;
+    } else {
+        // In edge mode, currentFreq is the window start (left edge)
+        currentFreq = f - ((GetStepsCount() / 2) * GetScanStep());
+    }
+}
+/* End added block */
+
+
 // scan step in 0.01khz
 uint16_t GetScanStep() { return scanStepValues[settings.scanStepIndex]; }
 
@@ -1209,7 +1227,7 @@ static void OnKeyDownFreqInput(uint8_t key)
             break;
         }
         SetState(previousState);
-        currentFreq = tempFreq;
+        RecenterWindowToVfo(tempFreq);
         if (currentState == SPECTRUM)
         {
             ResetBlacklist();
@@ -1676,8 +1694,8 @@ void APP_RunSpectrum()
     }
     else {
 #endif
-        currentFreq = initialFreq = gTxVfo->pRX->Frequency -
-                                    ((GetStepsCount() / 2) * GetScanStep());
+        initialFreq = gTxVfo->pRX->Frequency;
+        RecenterWindowToVfo(gTxVfo->pRX->Frequency);
         #ifdef ENABLE_FEAT_F4HWN_RESUME_STATE
             gEeprom.CURRENT_STATE = 4;
         #endif

[jflinchbaugh]

Why are there 2 different display modes (edge vs center)? I guess that could be causing some of my confusion too. I would have only expected "center", and I don't know how or when I'd want to trigger the "edge" display.

[jflinchbaugh]

Oh I see: Upon switching step sizes (when they're larger), the left boundary of the graph is staying fixed, and not adjusting to keep the chosen frequency centered as it was when we launch the spectrum.

[jflinchbaugh]

Additionally, in 4.2, it appears that switching the step from 5 to 2.5 (pressing 1 or 7, ) will cause it to uncenter. exiting, and restarting the spectrum will center it again where you expect it. Very small values, like 1khz don't center, even on restart.

[armel]

This isn’t an issue specific to version 4.2 ! You can try with older versions and get the same result. Feel free to fix it if you wish ☺️ I honestly think that some users, particularly demanding ones, sometimes place too much importance on minor details about this tiny spectrum analyzer, simply because it doesn’t work exactly the way they want.