time constant of AGC?

[Greg Troxel]

I am troubleshooting an odd problem, which is:

I have a good dongle running rtl_433 at 915, receiving a few Ecowitt
devices. It's been fine for months.

Today it stopped receiving one of them (air quality), but the Ecowitt
receiver display was fine. The rest were still received.

I restarted and tightened connectors.

The signal strengths seem to move around, so I suspect interference.


I used a second dongle on a second computer in a second location. It
gets all devices, but with differnet signal strength patterns -- no
surprise.

Looking at a plot of snr vs time (with device id as color), I see one
device right at 0 for signal strength, and ohers lower. Noise is
uniformly low.

So:

How long does it take AGC to recover? Should I expect that the
receiver permanently adapts to the strongest transmitter, even if
once/minute?

Or is it simply that the others are weak enough that there isn't
enough gain to bring them to zero - or perhaps the interference is
louder.


Plot legend:
4 colors, 4 emitters
boxes: signal strength
x: snr
+: noise

[Benjamin Larsson]

On 30/08/2025 02:45, Greg Troxel wrote:
> How long does it take AGC to recover?

There are several AGCs in play. The tuner has it, the demod/usb bridge
has it and rtl_433 uses it. You need to dump some signals and look at
the start and end to figure out the time constants. After the signal
stops there should be a clear range when the noise is amplified and when
it is not.

To figure out why it works in one place but not the other you need to
look at the raw signal. My guess is that you need to change the
frequency slightly.

MvH
Benjamin Larsson

[Greg Troxel]

"'Benjamin Larsson' via rtl_433" <rtl...@googlegroups.com> writes:

> On 30/08/2025 02:45, Greg Troxel wrote:
>> How long does it take AGC to recover?
>
> There are several AGCs in play. The tuner has it, the demod/usb bridge
> has it and rtl_433 uses it. You need to dump some signals and look at
> the start and end to figure out the time constants. After the signal
> stops there should be a clear range when the noise is amplified and
> when it is not.

Thanks; good point about looking at it in scope mode.

I should have asked a crisper question: is there any reason to think AGC
would persist more than a second or two, in any dongle?

> To figure out why it works in one place but not the other you need to
> look at the raw signal. My guess is that you need to change the
> frequency slightly.

I suspect it's about local noise, but agreed that I need to figure it
out and then I'll know -- while now I do not!

[Benjamin Larsson]

Hi.

On 31/08/2025 22:18, Greg Troxel wrote:
> "'Benjamin Larsson' via rtl_433" <rtl...@googlegroups.com> writes:
>
>> On 30/08/2025 02:45, Greg Troxel wrote:
>>> How long does it take AGC to recover?
>>
>> There are several AGCs in play. The tuner has it, the demod/usb bridge
>> has it and rtl_433 uses it. You need to dump some signals and look at
>> the start and end to figure out the time constants. After the signal
>> stops there should be a clear range when the noise is amplified and
>> when it is not.
>
> Thanks; good point about looking at it in scope mode.
>
> I should have asked a crisper question: is there any reason to think AGC
> would persist more than a second or two, in any dongle?
>

No.

>
>> To figure out why it works in one place but not the other you need to
>> look at the raw signal. My guess is that you need to change the
>> frequency slightly.
>
> I suspect it's about local noise, but agreed that I need to figure it
> out and then I'll know -- while now I do not!
>

Could be the DC-frequency gap also.

MvH
Benjamin Larsson