Gateway behaviour switching from Rx to Tx

Most gateways have two radios, of which only one is used for transmit.

When the radio is in receive mode, it can receive multiple signals simultaneously across several channels and data rates.

I am assuming that this work is actually being done one level up in the SX1301/1308 chip rather than in the radio chip itself.

The question is: what happens to packets that are in the process of being received when the radio is switched to transmit mode for a downlink packet?

I am guessing that any packets in the process of being received via that particular radio are, in effect, lost.

Or is there some magic happening?

A given SX125x radio chip has distinct RX and TX RF ports.

In the typically design, there is an RF switch that connects the antenna to either the RX ports of both radio chips, or the TX path which is the TX port of one radio chip followed by an amplifier.

So it’s not the receivers are so much “disabled” during transmit as that they loses their antenna.

I’m not 100% sure, but I think that when I had gateways transmitting occasional packets on uplink frequencies (with the idea that their peers would pick these up), they would occasionally receive their own transmissions - something that would be logical because even with the switch set the other way, there’s still going to be some energy leakage, and LoRa is designed to work with weak signals.

Unless you node is basically sitting next to the gateway, reception would cease and the low level packet CRC would fail even before the LoRaWan MIC.

Seeing if it is possible to get a node close enough for the received signal to be coupled even with the antenna switched the wrong way would be interesting, but relatively meaningless.

There’s an additional problem, too: buildable radios do not have perfect output on a narrow channel, but rather there’s some weak leakage, not only at spurious frequencies related to the design, but at close-in adjacent frequencies as well. I’ve found that if I have a distant node, and also a node very close to the gateway, if the close one transmits on the same time at a different frequency it will still manage to blank out the distant one. So even if it weren’t for the T/R switch, if your downlink frequencies are close to your uplink frequencies (iirc not all band plans keep them distinct) and even if the receiver had sufficient blocking dynamic range, this bleedover could still make reception during transmission not work.

Very useful information. Thanks.

Where I am going with this is that I notice from the global_conf.json files for my region (Australia), that the radio_0 is usually used for the 500kHz downlink channel (Tx). But it is also the radio used for listening on the 500kHz uplink channel for the sub band. It would make sense to me to split this so that radio_1 does half the 500kHz work.

With ADR enabled, nodes should migrate (given favourable conditions) to the fastest possible data rate which is DR6 here on the SF8/500kHz uplink channel. This could hypothetically result in one radio handling all traffic in both directions.

Given that I don’t know whether radio_1 circuitry even has a valid Tx circuit path, is there a case for switching the receive channel configurations between the two radios?

Radio_0, Rx channels 4,5,6,7 + Tx
Radio_1, Rx channels 0,1,2,3,64

How would “splitting the work” be beneficial? I don’t believe the SX125x really “knows” if it is receiving anything, it is just downconverting RF to the special format the SX1301 wants, only the SX1301 actually has any ability to distinguish LoRa from FSK from noise. So it’s not like SX125x power consumption or heat load would change by having the SX1301’s full-bandwidth chains look at the other radio’s output instead, the radio’s role in reception would remain unchanged.

Nor can the radio tolerate the other radio transmitting into it, nevermind not being deafened while it happens - to be able to receive while transmitting, you’d need to build a true diplexer, or use a distinct antenna a little ways away.

While this is possible in practice not all networks implement it. Remember while you can get on and off the air faster, you also only get one frequency for all such nodes to share. For SF8 I think the 8x reduction in channels is worse than the 2x improvement in speed. FSK might be another story since it’s quite a bit faster. If you really wanted to use these fast rates, I’d augment the gateway with additional node radios sitting on particular frequencies.