RSSI and SNR on SX1276

[simon-jouet]

Hi!

So I'm trying to get a simple sketch to connect to a gateway, join using OTAA and send confirmed messages at interval to measure the RSSI and SNR. Following the examples I got all of that working relatively easily on a Lilygo TTGO V2 1.6.1 (SX1276) but I really can't make sense of the RSSI and SNR values.

I did check around the issues (and google) before asking here, and I can see related discussions such as, #493 and #369 but i'm far from convinced with the calculated values (using #369 (comment))

Few meters away from the gateway (I know it shouldn't be too close) I get a RSSI of ~-130/-140 and an SNR of 5 at the microcontroller. Checking on chirpstack the gateway has a RSSI of -55 and a SNR of ~12, I know it's 2 different radios so I'm not expecting a match but that's a very very significant difference in calculated rssi/snr (and I think it's more a calculation discrepancy)

Does anyone know if the following code is correct (from #369) or if another calculation is necessary?

      uint8_t packetRssi = (uint8_t)LMIC.rssi;
      int8_t packetSnr = LMIC.snr;
      int rssi;
      int snr = LMIC.snr / 4;

      if (packetSnr > 0)
      {
        // -157 + 16/15 * PacketRssi, rounded
        rssi = (-157 + (16 * packetRssi + 7) / 15);
      }
      else
      {
        // -157 + PacketRssi + PacketSnr * 0.25
        rssi = (-157 + (packetRssi * 4 + packetSnr + 2) / 4);
      }

      Serial.println(rssi);
      Serial.println(snr);

Thanks

[terrillmoore]

I seem to have dropped this (like a lot of other things) when the pandemic hit.

Did you check this comment?
#369 (comment)

I think I cryptically indicated the answer ("need to subtract RSSI_OFF") but unfortunately i did not note exactly what needs to be done. If you're willing to do a few tests, we should be able to figure this out.

What are your values (close to the gateway) of packetSnr and packetRssi. snr looks reasonable. Since you're getting rssi of -130, and snr of +12, I can guess:. With a few calculations, I can see that you're probably getting packetRssi == 25; adding 64 (RSSI_OFF) to packetRssi in the calculation leads to a rssi of around -66, which is physically reasonable (as long as you widen packetRssi to uint16_t so that adding RSSI_OFF won't overflow). So I suggest the following adjustment.

      uint16_t packetRssi = ((uint8_t)LMIC.rssi) + RSSI_OFF;
      int8_t packetSnr = LMIC.snr;
      int rssi;
      int snr = LMIC.snr / 4;

      if (packetSnr > 0)
      {
        // -157 + 16/15 * PacketRssi, rounded
        rssi = (-157 + (16 * packetRssi + 7) / 15);
      }
      else
      {
        // -157 + PacketRssi + PacketSnr * 0.25
        rssi = (-157 + (packetRssi * 4 + packetSnr + 2) / 4);
      }

      Serial.println(rssi);
      Serial.println(snr);

[simon-jouet]

Hi Terry,

That's brilliant thanks, sorry I didn't get the suggestion in the previous post.

I don't know if this is correct but the values do looks to be quite inline with what I would expect, I will do more tests in the coming days at different range to see how it goes.

With the device close to the gateway I get the following (used floats for rssi and snr) the first value is the rssi and second is the SNR.

Received ack
-56.27
5.00

1122958: EV_TXSTART
1192016: EV_TXCOMPLETE (includes waiting for RX windows)
-56.27
5.25

1443935: EV_TXSTART
1512668: EV_TXCOMPLETE (includes waiting for RX windows)
-56.27
5.00

1732910: EV_TXSTART
1875100: EV_TXCOMPLETE (includes waiting for RX windows)
-56.27
5.00

5626739: EV_TXSTART
Packet queued
5695797: EV_TXCOMPLETE (includes waiting for RX windows)
Received ack
-60.53
5.50

9447440: EV_TXSTART
Packet queued
9516500: EV_TXCOMPLETE (includes waiting for RX windows)
Received ack
-59.47
6.00

and the code (pretty much exactly yours)

      uint16_t packetRssi = ((uint8_t)LMIC.rssi) + RSSI_OFF;
      int8_t packetSnr = LMIC.snr;
      float rssi;
      float snr = LMIC.snr / 4.0f;

      if (packetSnr > 0)
      {
        // -157 + 16/15 * PacketRssi, rounded
        rssi = (-157 + (16 * packetRssi + 7) / 15.0f);
      }
      else
      {
        // -157 + PacketRssi + PacketSnr * 0.25
        rssi = (-157 + (packetRssi * 4 + packetSnr + 2) / 4.0f);
      }

      Serial.println(rssi);
      Serial.println(snr);

EDIT: looks like it doesn't behave well when LMIC.rssi is negative. I will do more tests/checks tomorrow

[terrillmoore]

I didn't look at the code. Since LMIC.rssi is an int8_t, you should modify the code as follows. (There was some integer arithmetic rounding stuff in there that doesn't need to be used if you're using floats. I've fixed that, too.)

      int16_t packetRssi = LMIC.rssi + RSSI_OFF;
      int8_t packetSnr = LMIC.snr;
      float rssi;
      float snr = LMIC.snr / 4.0f;

      if (packetSnr > 0)
      {
        // -157 + 16/15 * PacketRssi, rounded
        rssi = (-157 + (16.0f * packetRssi) / 15.0f);
      }
      else
      {
        // -157 + PacketRssi + PacketSnr * 0.25
        rssi = (-157 + (packetRssi * 4 + packetSnr)  / 4.0f);
      }

      Serial.println(rssi);
      Serial.println(snr);

Or for the original integer approach:

      int16_t packetRssi = LMIC.rssi + RSSI_OFF;
      int8_t packetSnr = LMIC.snr;
      int rssi;
      int snr = LMIC.snr / 4;

      if (packetSnr > 0)
      {
        // -157 + 16/15 * PacketRssi, rounded
        rssi = (-157 + (16 * packetRssi + 7) / 15);
      }
      else
      {
        // -157 + PacketRssi + PacketSnr * 0.25
        rssi = (-157 + (packetRssi * 4 + packetSnr + 2) / 4);
      }

[simon-jouet]

Thanks Terry,

The values look more or less consistent now. I will get a battery for my circuit and have a walk around the block over the weekend to see what I get.