My indoor gateway stopped working. Only very nearby nodes can be received with a RSSI of -100. So RSSI dropped 40dBm.
Would it help to replace the PCI-e concentrator card?
I modified it already to support on exernal antenna so warranty is probably void.
Maybe someone knowns what LNA is used and where, or what the type of concentrator is?
If still in warrant I would simply arrange RMA/Return…to vendor you got it from.
Thanks for the reply. No warranty. I modded it already. (I’ve added this info to my original post for clarity)
So I would really like to know if there is some technical knowhow about its internals.
I understand it’s a rather special ESP32 + ??? design that helps keep its cost down.
And probably beyond economic repair - ie, cheaper to buy at new one.
Very true but I’ve got some tools here and some time left for the hobby.
What happened between it having reasonable reception and then poor reception?
What mod’s have you done?
I made an external antenna connection, connected with an external diamond antenna on my roof. This worked fine for a few weeks. Yesterday I overloaded the LNA in the gateway, I assume, at the moment I switched on a +20dBm LoRa node module, too close perhaps to the gateway (+/- 8 meters). Bad idea.
Despite the dummy load connected to the Lora node antenna that measured the output power.
So, more likely to be a component is feeling less healthy, rather than having been dropped and a connection broken.
Whilst I’ve just purchased one to do some modding on, I’ve not yet cracked the case open, but looking at this page, about half way down:
I think you’re going to have to prise the metal cover off to see if there is an RF switch chip that you can replace.
First suspect would be a disconnected antenna. Next damage from having the antenna exposed up there - lighting, etc.
Quite unlikely to have caused damage at 8 meters or even sitting next to it, but not a great idea if for no other reason than that it will swap out the receive leading to misleading results and blanking out any transmissions from more distant nodes occurring at that time.
Unless you are saying you transmitted from something connected to the gateway’s radio by an actual wire?
SX127x devices are have a stated maximum RF input power of +10dBm, which is very difficult to achieve unless you actually connect the TX of one device to the RX of another.
Not sure of the limit for the TIG RF device, similar you would expect.
LoRa devices will go very low power, and its happened to me too often, if they are operated as in transmitting without an antenna or a faulty antenna connection.
Seen that blog! This is the mod I did. Indeed. Lets get the metal cover off and find the Low Noise Amplifier (LNA).
The gateway was functioning perfectly up to the last second just before I started sending with 18.5dBm (so the node was sending. I also switched off the over current protection in the node). Since then, the gateway stopped receiving (-40dBm).
I totally agree it is unlikely that this broke the gateway, because to much distance (reception <10dBm)
May be it’s just bad luck somehow. About the dummyload/power meter on the lora node: even more unlikely that it has damaged the gateway.
Anyway, its broken and I will start looking for the circuitry. Will be very challenging without a schematic diagram.
Thanks all for your replies!
Update: I found a reference implementation of a concentrator.
The schematic describes a LNA with the IC of type SPF 5043Z. Which amplifies up to 18dB.
The LNA chip on MY TTIG concentrator board uses a SOT343-4.
Also the input, output and power pins are at the same location.
If this LNA is broken then 18dB gets lost or even more due to malfunctioning of the chip.
I would be surprised if the LNA added 18dBm of effective signal improvement. The LNA will also add noise, so the realised improvement will be a lot less.
There are some tests on the effectivness of LNAs (one cost £400) on LoRa signals here;
https://stuartsprojects.github.io/2018/01/09/can-we-improve-LoRa-performance.html
Just a thought but is this your only node? or Only node the GW was seeing in the area?
Is it poss that the GW’s low RSSI value is down to fact u have blown the node Tx vs the GW Rx strength? Just looking through the other end of the telescope 
…Are any other received nodes seeing equivalent scale of Rx RSSI reduction? What was typical Rx value from your node before loss of strength…are you seeing the 18dbm drop or somewhere close to that scale of reduction?
Anything that added 18 dBm would be generating it’s own signal
But the SPF5043Z actually is specified as having 18.2 dB of gain.
18 dB is a ratio such as the gain of an amplifier, 10^1.8 = 63 times.
But 18 dBm is a power level expressed relative to a milliwatt. 18 dBm is 63 milliwatts
Several nodes were seen by this GW. Some more than 20 kms away.
Since it is broken no single device was seen.
Exept a node within say 4 meters from the antenna (my own nodes that is).
I saw my mistake. 18dB is the gain of the LNA. 18dBm was the output of my node. Easy to mix up.
“Bloody hell,” said Majikthise, “now that is what I call thinking. Here Vroomfondel, why do we never think of things like that?” “Dunno,” said Vroomfondel in an awed whisper, “think our brains must be too highly trained Majikthise.”
It’s fixed!
I had some spare MGA-81563-TR1G chips laying around.
(The pins are really incompatible but I combined the GND 4 pins to 2 GND pins. And because the pins were mirrored (out/in), I bend the pins from the bottom to the top. Soldered it upside down and I now gained at least 18dB!!)
O M G !
I have ‘spare chips lying around’ but this has to be a new record. Well done!
Now go and make a lightening arrestor!