Backlight inteference / further FM bandwith options

[Richard-741]

[Richard-741]

Backlight interference: I've found this most noticable when listening to weaker stations across all bands (including VHF FM),
With the backlight swithed off, stations are 'cleaner' and not swamped out with background noise.

Firmware version 2.10 April 18th 2025 - using the supplied SMA telescopic rod antenna:

To reproduce the symptom, I have found tuning to MW1 shows it up the best. Distinct harmonics are present every 18khz. These can be heard at 200, 218, 236 and 255khz etc. with the backlight on (regardless of the brightness setting). These harmonics dissapear completely when the backlight is switched off.
Going to the other extreme, weak VHF FM stations can be heard much more clearly with backlight off.

I'm not exactly sure at which revison this became an issue, surfice to say that the G8PTN firmware does not have this problem.
It would be really interesting to hear if anyone else can reproduce these symptons.

Secondly, would the addition of further FM bandwidths be possible?

It would be really nice to have lower FM bandwidth options, not just for discriminating between fringe stations on the VHF broadcast band, but also to listen to NBFM radio transmissions from 64Mhz to 88Mhz - say bandwidths of 20, 10 and 5khz

Thanks very much.

[max-arnold]

According to AN332, the minimal FM bandwidth is 40kHz:

And the NBFM mode is not supported by SI4732 because the corresponding patch (like for SSB) is not publicly available.

Regarding the noise, it is somewhat expected that the display noise will depend on what is drawn on the display and how often, so it is definitely firmware-dependent and also hard to control. With that said, it would be helpful if you try to find the firmware version where it started to appear (please reset the receiver settings each time after you flash it).

The full history is available on the releases page https://github.com/esp32-si4732/ats-mini/releases You don't have to try each of them:

1. You wrote that the noise exists in 2.10
2. Try version 1.01 (it is almost unmodified G8PTN code, just built with a newer toolchain and libraries)
3. If my build of 1.01 is also noisy, it means something is different with how it was compiled
4. If 1.01 is fine, then pick a version in the middle (between 1.01 and 2.10, e.g 1.06) and try it
5. Repeat this "divide-by-two" (bisection) process until you find the version where it started to appear

In total you'll have to do 4-5 attempts to find the FW where the noise was introduced. Please report your findings here.

[max-arnold]

For a reference, here is a very loud noise produced in one of my experiments by a simple blinking animation on the display:

IMG_3518.mov

[trent-lane]

Also, very distinctive noise can be heard at 8 MHz and 16 MHz in AM mode. Moreover, themes with light background tend to produce more noise on 8 MHz then themes with dark background. Found in 2.11 and probably persists.

[max-arnold]

That is why there is Unlocked sleep mode - you can turn the display off and still tune the frequency and control the volume

[Richard-741]

I can confirm that the display noise caused by a firmware issue? occoured at the release of 1.02 (2025-03-11)

This version uses the same GUI as the G8PTN firmware with the following additions and changes:

Added

    Display timeout has been added (Settings > Sleep), a value of zero disables the timeout
    Turning the display off/on by long pressing the encoder (2 seconds). This reduces QRM from the display
    Added "O" and "o" commands (display off and on) to the serial port protocol
    AM/FM station search by scrolling with the encoder pressed (the search can also be stopped by rotating the encoder)
    Pressing the encoder for more than 0.5 seconds triggers volume adjustment

Changed

    The menu has been reorganized, and the Spare items have been removed
    A Settings menu has been added, and the display brightness adjustment option has been moved there
    The position in the main menu and settings menu is remembered (until power is turned off)
    The startup splash screen has been moved to Settings > About
    The Seek Up/Down menu items have been removed
    Changed the charging indicator (lightning icon instead of the EXT text)

Fixed

Fixed the duplication of the MW2 band name

I hope this helps in identifying a possible cause.

Many thanks, Richard

[max-arnold]

So am I correct that the following build has less noise https://github.com/esp32-si4732/ats-mini/releases/tag/v1.01 (compared to this one https://github.com/esp32-si4732/ats-mini/releases/tag/v1.02)?

[Richard-741]

That's correct - firmware 1.02 and above exhibits the symptoms. I'll post a SBS clip of two identcal receivers (one running 1.01 and the other 1.02) to demonstatrate the difference a bit later on..

Many thanks.

[Hamberthm]

I'm closely following this issue, I'm yet waiting for my radio and I'm really interested in helping.

For me, it is a firmware related issue, as the interference supposedly is "no brightness-setting related." That tells me it's not dependant on the backlight modulation duty cycle, which is a good thing.

I would bet it is something on code related to the main program loop, as it happens all the time. Since it is not the modulation, it needs to be some change induced on the backlight circuit during idle time. This could be like disabling and enabling the backlight very fast due to some bug. You would not see it, but you would generate other kinds of interference. Also, repeatedly enabling the backlight could reset the modulation duty cycle and induce interference due to phasing problems.

It would be interesting to watch the modulation on my oscilloscope, but as I said, I'm still waiting for my radio. Hope this gives a heads up to the devs!

[max-arnold]

@Richard-741 Could you try a couple of additional builds? Let's start with this one:

ats-mini-test-build-v1.01-76ffbd0.zip

[Richard-741]

Thanks for the v1.01 test build.

This build now exhibits the same display noise issues as found in v1.02 and above - it seems you're onto something!

[max-arnold]

Great! How about this one?

ats-mini-test-build-v1.01-76ffbd0-240.zip

[Richard-741]

Thanks again for another test build of v1.01. This unfortunately exhibits the same behavior and noise as the first test build.

Last night, I decided to try the ats-mini with the display disconnected. This conclusively fixed the problem - no noise or harmonic interference on any band. It seems that this could actually be a circuit design / hardware issue after all..

There are two areas worthy of investigation. Firstly, refering to the schematic diagram: the ESP32, LCD module and the LCD backlight driver transistor all share their power from the same VCC_3.3v regulator. I intend to lift PIN 2 of the LCD header and connect it to the VCC_RADIO_3.3v regulator instead. If seperating the LCD supply from that used by of the ESP32 resolves the issue, great.

The second area of interest are the data bus lines between the ESP32 and the LCD module. No filtering components or inductors are used, with all data lines are directly connected. This is bad practice. It might be worth trying the addition of a small filter / inductor in series wth LCD R_W (read/write line) as this is the only line where data is sent back to the ESP32.

I'll report back with any findings.

[Hamberthm]

Thanks again for another test build of v1.01. This unfortunately exhibits the same behavior and noise as the first test build.

Last night, I decided to try the ats-mini with the display disconnected. This conclusively fixed the problem - no noise or harmonic interference on any band. It seems that this could actually be a circuit design / hardware issue after all..

There are two areas worthy of investigation. Firstly, refering to the schematic diagram: the ESP32, LCD module and the LCD backlight driver transistor all share their power from the same VCC_3.3v regulator. I intend to lift PIN 2 of the LCD header and connect it to the VCC_RADIO_3.3v regulator instead. If seperating the LCD supply from that used by of the ESP32 resolves the issue, great.

The second area of interest are the data bus lines between the ESP32 and the LCD module. No filtering components or inductors are used, with all data lines are directly connected. This is bad practice. It might be worth trying the addition of a small filter / inductor in series wth LCD R_W (read/write line) as this is the only line where data is sent back to the ESP32.

I'll report back with any findings.

But is performance equally great with v1.01 (no test build, original) vs. screen disconnected? Are you saying v1.01 has noise issues anyways?

I think we need to focus first on getting the new builds to perform equally to the less-noisy v1.01. If something on firmware changed and made things worse, then it's worth fixing it! We can worry about hardware later.

[Richard-741]

You're correct - the original v1.01 firmware is by far the best when it comes to lowering the noise caused by the LCD display,

Some low level harmonics are still present, but at a much reduced level. Saying that, without the LCD display attached, all noise and the harmonics dissapear completely.
I tried to upload a video here without success, so here's a Vimeo link showing one of @max-arnold's test builds vs no display:

https://vimeo.com/1083492780

[Hamberthm]

As far as I understand, the backlight circuit is isolated from the main screen controller/interface

Hi, Although there is no direct interaction between the backlight and the screen, when the backlight is turned off, the screen is also set to the off state by the firmware and there is no subsequent activity. 73, Dave

Oh! If that's the case, then it makes sense it could be the screen itself after all.

Sorry, I still don't have the radio (still in post) and I thought the only thing controlling the backlight was the PWM modulation on this line:

ats-mini/ats-mini/Menu.cpp

Line 445 in 1260db9

if(!sleepOn()) ledcWrite(PIN_LCD_BL, currentBrt);

But if there's something else on code (or hardware) sensing the backlight is off and thus also disabling the screen, then the screen is likely the problem.

[Richard-741]

I've a few points to address. @max-arnold the v.1.01 firmware you sugggested I try, sadly behaves the same as the others :(

With regard to @Hamberthm's and @G8PTN's comments on how the backlight is 'switched off' I can confirm the noise is not from the 16khz PWM squarewave signal. I removed Q1 (the SO8050 baclkight driver transistor) completely and the noise is still there.
By my reckoning, this squarely makes the source of the noise the LCD display. Not by EMI radiation, but back along the data bus..

[Hamberthm]

Removing the transistor could make no difference if the PWM modulation is the problem, as the ESP32 is still modulating the output, and thus, radiating along the track up to the transistor.

I've used the PWM modulation before as a transmitter by itself with no other components inline to make a WWV time signal emulator and sync a Casio watch nearby... 😁

[G8PTN]

But if there's something else on code (or hardware) sensing the backlight is off and thus also disabling the screen, then the screen is likely the problem.

Hi,
In V1.02 the code for display off was as shown below.

In the latest code it is incorporated in the sleepOn() function in Utils.cpp
73, Dave

[G8PTN]

Hi Richard,
I expect the interference is from the display as discussed above. This could be conducted or radiated.
If it was the PWM signal is would stop when the backlight is set to the maximum brightness.
73, Dave

[Richard-741]

@Hamberthm I understand what you're saying, but that doesn't explain that with the LCD unplugged from the mainboard, all the noise, harmonics and interference issues vanish regardless of which firmware build is used..

[Richard-741]

Hi Richard, I expect the interference is from the display as discussed above. This could be conducted or radiated. If it was the PWM signal is would stop when the backlight is set to the maximum brightness. 73, Dave

Exactly!

[G8PTN]

Hi,
The noise may be from the display itself and/or the SPI signals, I think most likely the display.

When the display was disconnected the noise stopped, but the SPI signals from the ESP32 would probably still be activating.
Obviously, when the display is disconnected the trace lengths are different and reflections will change, so it does not completely rule out the SPI signals as part of the noise source.

As per Steve's findings, screening may help, but I think if you are want to negate any interference associated with the display the best option is to turn it off.

73, Dave

[max-arnold]

I've a few points to address. @max-arnold the v.1.01 firmware you sugggested I try, sadly behaves the same as the others :(

This is interesting! It means that Dave's v1.01 build and mine v1.01 have some differences that result in more noise.

That could be a different display library version, esp32 arduino core version (Dave uses 2.x, I use 3.x), or some compilation options, like CPU speed.

[G8PTN]

That could be a different display library version, esp32 arduino core version (Dave uses 2.x, I use 3.x), or some compilation options, like CPU speed.

Hi Max,
The TFT_eSPI library init sequence used for the display may make some subtle differences, but either way, I don't think the issues can be prevented by firmware alone.
73, Dave

[max-arnold]

Hi Dave!

Do you use the latest TFT_eSPI 2.5.43 library, or TFT_eSPI_Dynamicspeed (used by the Chinese firmware)? How is your User_Setup.h looks like?

[G8PTN]

Hi Max,
Neither of these.
I used the original Lilygo instructions.
https://github.com/ralphxavier/SI4735/tree/master/Lilygo_T-Display_S3/ALL_IN_ONE_T-Display_S3
This pointed to the following library.
https://github.com/Xinyuan-LilyGO/T-Display-S3/tree/main/lib/TFT_eSPI
The driver is selected in the User_Setup_Select.h file
I patched the rotation file as discussed previously.

At some point I will switch to the latest versions!!
73, Dave

[max-arnold]

@G8PTN Dave, could you please post a screenshot of your board options in Arduino IDE?

[G8PTN]

Hi Max,
Please see attached Arduino IDE board options.
Note: I may have had the Core Debug Level: “None” on some builds, otherwise no changes.
73, Dave

[DG9BFC]

in the nanovna v2 we have (had) some noise from the displays "current" rattling on the power rail ... some caps here and there and the "noise at and above 3ghz" was way lower
so .. maybe it is the current when something is written on the display causing some noise on the power rail (and creeping backwards in the receiver) ??? just thinking