SlimLoRa library

I am fine with both names :slight_smile: brevity vs precision

Not to start a discussion here but not being (fully) compliant is usually easier.
(e.g. mastering only half of the theory required for driving a car in public traffic is easier but will make you a poor driver.)

No event handling

This library does not implement any kind of event handling. A call to the Join and SendData methods will block for several seconds until the end of the second receive window.

If there is no support for event handling then how do you handle (sequence of) downlinks?

That will also trigger the Lora Alliance as it infringes on their rights. Nothing wrong with the current name.

4 Likes

If I remember correctly, when there’s pending data for a class A device, the device has to send another uplink packet to receive the pending data no? While there’s currently no pending data support it would be easy to notify the user of the pending data and ask him to queue for a second uplink packet.

How this exactly happens on protocol level I’m not sure.
When using LMIC and the node receives a sequence of queued downlinks, this will happen automatically without intervention of the application code. This is done by the library automatically (like handling of MAC command like Status request).
The application will handle al downlink messages (payload data) via an event handler. Therefore I was interested in how SimpleLoRa handles downlink messages without supporting event handling.

The author said the code “will block for several seconds until the end of the second receive window.”.

Thus it handles downlinks by not doing anything else while they are possible.

Since TTN only supports class A, a downlink is only possible in direct response to an uplink. Even the FPending bit is just an invitation to send another uplink so that an additional downlink can follow, the additional downlink won’t be sent on its own.

4 Likes

With (MCCI) LMIC this is handled automatically by the LMIC library.

It’s presumably handled by this library too. Though not handling it would not really cause an issue, because the network can’t do anything until the node sends another uplink, and its free to make that minutes, hours, or days later at its discretion.

Perhaps some of the confusion is coming from the way some other codebases try to create a sort of interface between library and application (LMiC does so in a very unclear way, LoRaMAC does so not just there but between layers of itself in a way abstracted nearly to the point of absurdity); I suspect this one is flatter and more directly coupled.

2 Likes

After changing the sleep commands to the correct ATTiny84 version and selecting the correct port for tne RFM95 NSS I only had 1 error during compiling regarding the timer TIMSK in the timing.cpp file.

I made the following change in timing.cpp file
TIMSK |= (1 << OCIE0A); ====> TIMSK0 |= (1 << OCIE0A);

After that I was able to compile but no result when trying to connect to TTN

Than I changed
TCCR0B |= (0 << CS02) | (0 << CS01) | (1 << CS00); ===> TCCR0B |= (1 << CS01) | (1 << CS00);

This resulted into 1 activation message as per below. So I believe it has something to do with how the timer has been set-up. However I am not familiar with how to correctly set-up the timer for the ATTiny84 using registers. Maybe you can help me setting up this part of the code correctly.

Appreciate any help you can give me.

image

1 Like

Seems likely the timer issues may be preventing firing up the receiver at the correct time to catch the join accept message

On my fork / branch: try-to-compile I adapted some code and compiles with arduino IDE.

Only the first time I had a successfull join (!) and afterwards an uplink. I was excited and moved forward and added SetPower, SetDataRate, debugging options, but after that I never joined again. Tested with V3. I abandoned OTAA and tried ABP. It works since it does need downlinks, but it seems that also the downlinks are not working.

In other words, it seems that the timing creates problem with downlinks. I didn’t investigate it further. I have the impression that instead of 5 seconds it opens the window at approx 1 sec.

The code is here.

@novag thanks for sharing.

1 Like

TTN V3 absolutely requires proper downlink and MAC command support, even for ABP.

The reality is that trying to make a minimal LoRaWan implementation which actually works correctly is extremely hard, because LoRaWan is not a simple protocol at all.

There’s been talk / an effective feature request that TTN have a way to register nodes with only partial LoRaWan support, but so far it does not - unless or until such a thing is added “incomplete” nodes must not be registered on TTN. because they will trick the network into sending unanswered downlinks over and over and over again, each of which blocks the gateway’s ability to receive other properly functioning nodes.

I know. That’s the reason for my report and my effort to port it to arduino. To enable support for downlinks. At current state it’s the same with TinyLoRa.

Today I saw two examples for timing. I hope I will try those days.

Forgot to add, that maybe the problem is with my GW. That’s one more reason I want people to test SlimLoRa. Maybe it’s already listening for downlinks at right time (default 5 secs to my understanding).

2 Likes

Still no functional library for Arduino?

What do you mean?

Of course there is a very functional LoRaWAN library for the Arduino framework: MCCI LMIC. Just search the forum for more information.

I mean easy to use straightforward library for Atmega328P (for example Arduino Uno, Mini). Tinylora was perfect goal. I have made well tuned fw.

Why to use veeery old and limited Atmega328P? Because can run on 1.8V, rfm95W can run on 1.8V too. Can be used with LiSOCl battery, 2x AA (rechargeable) battery etc. No need to step up, sted down circuit… Low power consumption max 2mA (mcu) working, 5uA sleeping (mcu, transceiver and sensor).

I have not find simillar micro controller that has these features, can be easy programmed and has good avaiability in Czech Republic (or to Czech Republic from Ali etc). Probably the closest mcu is ESP32 underclocked to 10MHz.

I have try LMIC-node | One example to rule them all

Add support for Arduino Uno and I get warning, that used library is deprecated. I don’t exactly known where to add power saving features. I have ended with join loop and high battery drain (poor signal). With good signal example works fine, but fw is close to fill all avaiable resources and maybe I’m not sure if is possible to add BME280 sensor.

You mean this library?

Looks like it is only for 32b mcu’s. I got lost in all the *mic; *cci libraries - all have same name.

Yes that is the LMIC library to go with. This is actively maintained and is the most LoRaWAN compliant of available open source LoRaWAN libraries for the Arduino framework.

You can also use it with 8-bit MCU like ATmega328 and ATmega32u4.
The problem with these ATmega32x MCUs however is that they have very limited memory resources compared to more modern 32-bit MCUs.
LoRaWAN has progressed and the newer versions have more requirements. The MCCI LMIC library has grown in size accordingly. It can still be used on ATmega32x but you will not be able to run very large sketches in combination with MCCI LMIC on these MCU’s.

Check the ttn-otaa.ino example included with the MCCI LMIC library. It will run on ATmega32x and it should also be possible to add support for BME280 in addition.
Be aware that when using MCCI LMIC with ATmega32x you will have to:

  • Disable PING and BEACONS to save memory (and they are not needed for LoRaWAN Class A devices anyway).

  • Adjust the LMIC ClockError (otherwise downlinks and joins will fail).

Search the forum for more information about these subjects. Check the BSF for the Pro Mini board in LMIC-node for how to set ClockError with an approriate value.
LMIC-node does not directly support the Uno but you can use the Pro Mini as example which will not be much different (except for some pin mappings maybe).

Be aware that LMIC-node uses the predecessor of MCCI LMIC which is called ‘Classic LMIC’. It uses Classic LMIC for ATmega32x because there is too little memory to use LMIC-node with the MCCI LMIC library on these boards. But Classic LMIC is now deprecated because it is not LoRaWAN compliant and can cause certain issues with TTN (TTS CE V3).

LMIC-node is a rather extensive example application that uses many text messages to make clear to the user what happens. It supports multiple MCU platforms and was to certain extent optimized for use on ATmega32x. However, over time the MCCI LMIC library has grown in size and modern LoRaWAN applications just require/prefer use of MCUs with more memory resources.

LMIC-node uses MCCI LMIC for all supported boards except for the ATmega32x based boards. For these boards LMIC-node uses the older Classic LMIC which is now deprecated (for valid reasons) but it still works and therefore LMIC-node uses it to make it possible to run on these ATmega32x MCUs so you can still have an example that ‘works out of the box’ experience.

When LMIC-node uses Classic LMIC for these MCUs it gives a warning that the Classic LMIC library is deprecated, but it will still run.

If you still want to use ATmega32x my advice is to start with the ttn-otaa.ino example and build from there. You can look at the code of LMIC-node for code examples of how certain things are implemented/solved that you don’t find in ttn-otaa.ino.
Another option is to start with LMIC-node and strip it from everything you don’t need to make it fit on ATmega32x while using MCCI LMIC but that requires (much) more knowledge.

Also see: How to connect Adafruit Feather 32u4 LoRa to TTN? - #14 by bluejedi

Power saving and LoRaWAN is a more complex topic and LMIC-node currently does not provide support for it (because different per MCU platform and the MCCI LMIC library currently still does not support save/restore LMIC-state features).

Look at Atmega328 / ATmega32u4 / Arduino Pro Mini and deep sleep elsewhere on the forum for more information.

1 Like

It is indeed a very good little MCU which I use for very simple non-critical nodes running off a couple of AA batteries and it works fine with the MCCI LMIC 4.1 if you turn off the pings & beacons as suggested. I use decent Pro Mini’s so haven’t had a need to adjust the clock as yet.

I have a work-in-progress board that has an ATmega4808 48KB flash, 6KB RAM) and RFM95 on it which will give me far more latitude for cheap & cheerful devices because …

the only place with good availability for anything right now is somewhere over the rainbow, way up high. I’m struggling to get electrolytic capacitors and my Raspberry PI CM4 gateway project is dead in the water as pretty much anything with a Raspberry is Unobtainium.

But I can usually get a few ATmega4808’s here and there and I’ve got other boards in the works to give me options.

In some random ideal world, probably over the same rainbow, we could port MCCI LMIC to being a bit less reliant on the Arduino bits and a little less bloated in areas (OS Jobs, I’m looking at you) so we all have more options.

You may find this repro useful as a starting point. I may get around to updating it to LMIC 4.1 shortly.

1 Like

I’m not afraid to use another mcu (then atmega328P), but I must be able to get them. And it is not problem of Covid etc, it is problem somewhere else. And good community around these mcu’s is important too.

My problem with new libraries will be maybe, that I underclocked Arduino Mini to 1MHz to save some power ;). LMIC-Node was tested on regular Arduino Uno 16Mhz.

Thanks for tips guys, I will look on them.