CC1101/CC110L modules outdoors losing range over time

[JanErik]

Hello,
New to the forum and have this problem;
I am using this kind of radio modules with an on-board RF stage, activated with a PAC pin;
https://www.aliexpress.com/item/CC1101-CC110L-wireless-module-medium-power-600-800-meters/1576224501.html
http://www.electrodragon.com/product/cc1101-433m-wireless-module-variable-designs/

Using one of them in my cold (corrugated sheet metal) garage to measure temp/humidity/groundtemp and controlling a block heater relay.

After around a year, they start losing range, first one lasted almost two years and now I had one last 1.5 years. Would not contact indoor unit with garage door closed. Replaced module and it works.

Module is in an enclosure but antenna sticking out, sealed as good as possible with Blu-tak and hot glue. For the new node I will 3D print an enclosure that fits both module and antenna, in clear PLA for best RF performance.

I sometimes suspected the SMA antenna connector, cleaned multiple times and sometimes that helped but on this latest module I had the antenna soldered so that is not a cause.

Could it be humidity creeping into caps/resistors and then cracking them in subzero temperature?

Sometimes I suspect the AS169 RF switches that direct the signal to/from RF stage.

Could it be salt depositing on the board? I live 500meters from the Gulf of Bothnia and 200meters from a salted motorway. Tried cleaning the modules with distilled water and propanole to no help.

And to add, it is controlled by an Atmega328, Arduino and PanStamp driver, https://github.com/raimue/arduino/blob/master/libraries/panstamp/cc1101.cpp .

I have a similar module indoors that lasted many years.
Also have a CC110x module without RF stage in my basement storage and that is able to send even if the garage node is acting up so it is not the indoor base station module having problems.

[Gribo]

Do you have access to the device schematics? What you describe can be attributed to crystals drifting apart. The CC1101 has a minimum RX bandwidth of 58KHz IIRC, so it is quite tolerant to such drift, however, if the design uses 40ppm or worse 26MHz crystal, then such problems occur.
I have used the CC1101 in very harsh environments (Salt, humidity, heat, cold etc) with proper crystals and calibration, the devices ran well for years.

[coppice]

Could it be that the difference between your indoor module that lasts, and your exterior module that doesn't is the temperature of the crystal? If its not a very stable crystal, and had poor ageing characteristics, the additional effects of temperature might be pushing the transmission too far off channel.

[JanErik]

I have drawn schematics of the RF side but not of the digital. Will draw clean and post.
I tried setting the sending to adjacent "channels" (4 and 6, normally use 5), do not know the BW but receiver did not see them at all.

And I have now tested the modules without enabling the RF stage (PAC pin) and they seem to work fine with range of a non-amplified module.
More and more suspecting the transistor (or RF AS169 switches) actually is burnt. I thought it was a FET, SOT-89, but now I measure a diode (0.5V forward) from what could be emitter and base, and around 1.3V forward from what could be collector to base.
So it could actually be a PNP which is burnt from collector to base.

One issue in my software was that I turned PAC on 50ms before sending ("//let the rf stage stabilize"), and guess it does draw some amount of current (especially with NiMH AA batteries) as the transistor is connected (Vcc -> 0R22(wire-wound) -> drain (collector?) and source (emitter?) -> gnd.
Now I have lowered that to 1ms (0ms would not work) and it still works.

The bias of the transistor gate/base is around 1.3V on one broken module and around 2V on the other, with RF stage enabled.

I have two healthy modules on their way from China so I will investigate the transistor on them.

[Gribo]

It is not a transistor, it can be either a power amplifier (Most likely) or an LNA.
50msec is a very long time for it to be active.

[JanErik]

In SOT-89, three-pin?

[Gribo]

Yes
https://ca.mouser.com/Semiconductors/RF-Integrated-Circuits/RF-Amplifier/_/N-3ri04?P=1z0yydo

[JanErik]

Alright then, just thinking now as I find nothing from the marking, "H3 2341". Is this some Chinese noname?

Adding some sketches of the RF and bias stage, where the FET may be an amplifier.

[JanErik]

Got the new module now and it also has the same diode test properties of the transistor/amplifier.

[JanErik]

Now, the new module is seriously off in frequency! I see around 100 kHz with an RTL-SDR stick.
I know there is an offset setting and it automatically tries to compensate for frequency drift (enabled in driver).
Now as I have one "basestation" and two "measurement nodes", would it make sense to disable this compensation on the "basestation"?

[Gribo]

That depends on how the base station does the compensation. The CC1101 measures the received carrier's offset against its on LO and tries to compensate for any difference during TX. From my experience, it can't do it reliably if the drift is more than 120ppm and the RSSI is less than ~6dB above the sensitivity level. Your best bet would be to calibrate the frequency on your devices.

[JanErik]

Added functions in my driver for getting frequency estimation and setting frequency offset, and now it works, the new module. Got -23 in offset which seems quite much.
Hopefully my other modules will work more reliably with this. Averaging over five receptions even though it seems quite stable.
Left is to try what the "broken" modules get as offset.

[JanErik]

Reporting back now with the new module in use over half a year, still working fine. I have soldered the antenna on to avoid burning the amplifier due to bad contacts. Frequency offset is constant so far.

But I have two burnt modules that I was thinking to revive. Here you can see a closeup of a similar module at Electrodragon store: https://1436477554.rsc.cdn77.org/wp-content/uploads/2012/05/CC1101-high-output-2.jpg
It is marked "H3 2341" and this is SOT-89, right? While the slightly larger AMS1117 regulator is SOT-223?

I find only a P-channel MOSFET when searching, https://alltransistors.com/mosfet/transistor.php?transistor=23825 . It cannot be this, and certainly not at 433 MHz.
Thinking to replace it with AP112 which seems reasonable and can be ordered from Aliexpress, http://rfhic.com/prod/download.php?idx=53 . Do you think that will work?

[Gribo]

The Hitite (Analog Devices) H452 is a close match.
http://www.analog.com/en/products/hmc452st89.html#product-tools


There is also an alternative from Skyworks, with the same pinout and frequency range.

[JanErik]

OK so I replaced it with AP112 which I could source.

But it did not solve the problem! Still the same range, and same RSSI/LQI reported from basestation (I made a Ping routine that replies that), regardless if I use the RF stage enable pin or not.

Should solder an antenna directly after the RF stage to rule out the RF switches.