Jump to content
Lgbeno

imp.guru Droplet and 915MHz Long Range Radio to WiFi

Recommended Posts

That makes sense though there is a HAM test question that seems to conflict with that: "T3A02-2014: Why are UHF signals often more effective from inside buildings than VHF signals? ", answer: "The shorter wavelength allows them to more easily penetrate the structure of buildings "

Wouldn't it be opposite of that?

This is wrong.  I won't get started about HAM and ARRL, and how they destroy innovation in the modern era.  But at least you can figure out my opinion.

 

Lower wavelengths represent smaller particles via the wave-particle duality, more smaller particles can be produced with a given energy level, and therefore lower wavelengths propagate better through any medium.

 

What I assume this question and answer refer-to is practice rather than theory.  In a concrete bunker, VHF will propagate through the bunker much better than UHF.  In a building with windows and doorways, there are situations where the diffractive losses of a longer wave signal will result in greater attenuation than the bouncing of shorter wave signals.  This can be very experimental and it depends a lot on physical geometry, construction materials, etc.

 

UHF is basically 300-1000 MHz.  Planet Earth and humans are scaled geometrically such than UHF often propagates well in "real-world" environments.  But does it really propagate better than a 30 MHz signal?  That's a stretch.

Share this post


Link to post
Share on other sites

I'm happy to report that the bee hives are officially up and running along with a new feature on imp.guru called the "view" which allows users to over lay multiple charts from multiple streams. Check it out: imp.guru/v2

Share this post


Link to post
Share on other sites

I'm happy to report that the bee hives are officially up and running along with a new feature on imp.guru called the "view" which allows users to over lay multiple charts from multiple streams. Check it out: imp.guru/v2

Great work with your IoT devices!

Share this post


Link to post
Share on other sites

So all of this recent discussion about cortex m0+ and alternative transceivers, I'm contemplating playing around with a different chips, my primary goals are:

 

Better Range

Lower Cost

Same form Factor

Lower Power

 

Pretty much in that order.  I know that @@jpnorair was talking about better antenna design as well, I'm still looking forward to some ideas there.  I wish that I had a better understanding of solving the antenna problem, I'm just there yet in the learning process.

 

There are 3 alternative transceivers that I'm considering

ST Micro Spirit1 <- seems to be the cost leader (and good performance)

Semtech LoRa <- seems to be the range leader, I know nothing about the cost

Silicon Labs Si1064 <- integrated micro which is kinda nice, for this reason it could end up as a cost leader

 

For Micros, I'm contemplating:

EFM32ZG110F32-QFN24 <- cheap and looks slick with the pre-programmed bootloader

One of the ST CM0 parts <- Good low cost eval tools and a big user base which I like for support

 

Any opinions?

 

 

 

 

Share this post


Link to post
Share on other sites

There are 3 alternative transceivers that I'm considering

ST Micro Spirit1 <- seems to be the cost leader (and good performance)

Semtech LoRa <- seems to be the range leader, I know nothing about the cost

Silicon Labs Si1064 <- integrated micro which is kinda nice, for this reason it could end up as a cost leader

 

For Micros, I'm contemplating:

EFM32ZG110F32-QFN24 <- cheap and looks slick with the pre-programmed bootloader

One of the ST CM0 parts <- Good low cost eval tools and a big user base which I like for support

 

Any opinions?

 

 

 

There is some information leaked on the internet about CC13xx, which appears to be a Cortex M3 + CC1200 SoC.  I'm not sure when it is due, but it is an interesting option.

 

If you're going with discrete MCU + RF, I have good support for STM32L1 + SPIRIT1 via OpenTag, and there is some experimental work for the L0 support.  In any case, the code is open sourced on my GitHub, so feel free to take a look.  If it is between STM32L0 and Zero-Gecko, I would go with the STM32L0 simply because it is more versatile.  This way, you can have a single platform for both the endpoint and, presumably, a USB-connected gateway.

 

LoRa has a maximum data rate around 30kbps.  I have achieved 5 km range using the SPIRIT1 with a custom, highly-tuned configuration, in an outdoor line-of-sight environment (433 MHz).  LoRa, using a similar data rate, frequency, and transceiver power has the same outdoor range.  In environments with more interference it performs much better, though.  In any case, if your range requirement is less than, say, 200m, LoRa is overkill.  I'm happy to start a thread discussing optimal setup for basic FSK/MSK transceivers.  If you're interested, maybe send me a bunch of questions and I will try to answer them.  This is easier than coming up with my own questions for myself.

Share this post


Link to post
Share on other sites

Thanks @@jpnorair, I am definitely interested in OpenTag.  I know that it is open source, I'm just not sure how you would feel about using it in a product that I may potentially sell.  

 

The other item is hardware, I'm perfectly comfortable spinning my own but I get this worry that I wouldn't be able to bring up the software on whatever I create.  It feels like you may have something that is a little more "proven" already that I could test and get my confidence going.

 

Ultimately, my long term goal is to have nodes that would get a mile or greater range, but still cost $5-$10 per node.  I'm fine with a very slow data rate, I mostly envision sensors that will transmit 10 or 20 bytes of data every 1-5 minutes.  There is an incredible amount of applications for such sensors.  For me, I think that the range/cost ratio is the killer application.

 

I've become quite happy with the convenience of the WiFi gateway, it is very flexible in the fact that it can be placed and requires minimal external infrastructure.  I find as well that most people do not keep a PC running in their homes as before.  Also, very often, it is a portable laptop where a dongle is very inconvenient.

 

My highest priority is to minimize the amount of firmware that I need to write, this is where most of my projects typically lose steam.  This is one reason why the first revision of the droplet took shape, so much of the firmware was plug and play courtesy of energia.

Share this post


Link to post
Share on other sites

Thanks @@jpnorair, I am definitely interested in OpenTag.  I know that it is open source, I'm just not sure how you would feel about using it in a product that I may potentially sell.

It uses a BSD-style license, so do what you want.  It actually uses the "OpenTag License" which gives you some patent rights as well.

 

The other item is hardware, I'm perfectly comfortable spinning my own but I get this worry that I wouldn't be able to bring up the software on whatever I create.  It feels like you may have something that is a little more "proven" already that I could test and get my confidence going.

Yeah, I need to decide what I want to do about my dev kit HW. I don't make money on dev kits, so I'd like to find a clever way to open source it, while still allowing me to collect small royalties for commercial use of the design.

 

Ultimately, my long term goal is to have nodes that would get a mile or greater range, but still cost $5-$10 per node. I'm fine with a very slow data rate, I mostly envision sensors that will transmit 10 or 20 bytes of data every 1-5 minutes. There is an incredible amount of applications for such sensors. For me, I think that the range/cost ratio is the killer application.

CC1200 is better than SPIRIT1 for this. It is a tad more sensitive, but more importantly it has some DSSS features (and a more low-tolerance sync-word setting). To get the advantage of low data rates, you need to do extreme narrowbanding with a TCXO, or you need a way to do spread spectrum with some processing gain (e.g. DSSS). So, CC13xx could be a very interesting option. LoRa is a really great device, but it is expensive.

 

I've become quite happy with the convenience of the WiFi gateway, it is very flexible in the fact that it can be placed and requires minimal external infrastructure. I find as well that most people do not keep a PC running in their homes as before. Also, very often, it is a portable laptop where a dongle is very inconvenient.

Get an OpenWRT box and plug the USB gateway dongle into it.

Share this post


Link to post
Share on other sites

Very cool (and generous) on the license terms.  I know that you have an extraordinary amount of work and expertise in building that.  

 

About the dev kits, just to throw out an option.  It sounds like the antenna design and optimization is your core IP.  If I was producing a commercial product, I wouldn't have a problem paying 50 cents to a dollar per unit in royalty in order to use your antenna design.  That is what the chip antenna costs in low volume anyway...  Obviously if anyone hits a home run and starts shipping 100,000's of units, the price would need to go down but that is a good problem to have and I would say it is way off in the distance.

 

For the more open source option, would it be possible to offer a detuned or offer a purely SMA style solution.  At that point the hardware design is pretty predictable but still very capable of evaluation.

 

I would say that if you are getting 5km of range with Spirit1, I'm more than pleased with >1mile and Spirit1 is cheaper than CC1200.  We'll see how things unfold with CC13xx, a single chip solution would be quite nice but it would need to fit the cost target.

 

Would you be willing to sell a few of your hw kits without the source?  It would be cool to try this out.

Share this post


Link to post
Share on other sites

@@jpnorair, I have a thought about how you could monetize your antenna design.

 

Sell or license a DXF or gerber file to the Integrator.

 

They can import it into their design files and you can have a passive income stream.

Share this post


Link to post
Share on other sites

This looks very similar to what I'm trying to accomplish with the droplet!  Very cool, thanks for pointing it out.  It seems like the founders have a lot of different add on sensors as well which was something that I was also planning to do.  I might need to buy one to try out when it goes live.

 

How did you find it?

Share this post


Link to post
Share on other sites

I would say that if you are getting 5km of range with Spirit1, I'm more than pleased with >1mile and Spirit1 is cheaper than CC1200.  We'll see how things unfold with CC13xx, a single chip solution would be quite nice but it would need to fit the cost target.

5km is outdoor line-of-sight at 433 MHz with 11dBm, chip rate 13.24 kHz (actual data rate about 0.4 times that), good antennas (omni dipoles), and using my implementation of the Voyager Code's Reed Solomon error correction in addition to the built-in convolutional coding of the SPIRIT1. I have not actually released the RS code implementation into open source yet. Without the RS, though, 2km outdoor line-of-sight is still quite feasible. In difficult environments, it's more like 300m.

 

This solution will decode packets down to about -118 dBM -- with the SMPS on -- however it doesn't do tremendously well in environments with lots of multipath. Keep in mind that ground-to-ground links, even if you can see the other device, are not line-of-sight because the frequency of the RF is very low whereas the frequency of visible light is very high. So, at street level there is a ton of multipath. Having a spread-spectrum modulation can help a lot. I've experimented doing this in SW with the SPIRIT1, and it works OK, but the sync-word threshold can only go down to -3 on the SPIRIT1 whereas it can go down to -8 on the CC1200. Assuming you have a high-performing sync word (I have a list of a bunch of them), this makes it more feasible to do DSSS in SW. And DSSS is great for mitigating multipath interference.

Share this post


Link to post
Share on other sites

5km is outdoor line-of-sight at 433 MHz with 11dBm, chip rate 13.24 kHz (actual data rate about 0.4 times that), good antennas (omni dipoles), and using my implementation of the Voyager Code's Reed Solomon error correction in addition to the built-in convolutional coding of the SPIRIT1. I have not actually released the RS code implementation into open source yet. Without the RS, though, 2km outdoor line-of-sight is still quite feasible. In difficult environments, it's more like 300m.

 

This solution will decode packets down to about -118 dBM -- with the SMPS on -- however it doesn't do tremendously well in environments with lots of multipath. Keep in mind that ground-to-ground links, even if you can see the other device, are not line-of-sight because the frequency of the RF is very low whereas the frequency of visible light is very high. So, at street level there is a ton of multipath. Having a spread-spectrum modulation can help a lot. I've experimented doing this in SW with the SPIRIT1, and it works OK, but the sync-word threshold can only go down to -3 on the SPIRIT1 whereas it can go down to -8 on the CC1200. Assuming you have a high-performing sync word (I have a list of a bunch of them), this makes it more feasible to do DSSS in SW. And DSSS is great for mitigating multipath interference.

Very good points. So is it coding that gives LoRa it's longer range, DSSS or both.

 

The environment that I'm working in is that the transmitters and embedded in the middle of a bee hive surrounded by wax, bees and wood, then the signal goes through the woods and then there is my receiver. Thinking about it, multi path may be a big part of my losses. I'm at 915mhz currently.

Share this post


Link to post
Share on other sites

Very good points. So is it coding that gives LoRa it's longer range, DSSS or both?

LoRa doesn't use DSSS (Direct Sequence Spread Spectrum), it uses CSS (Chirp Spread Spectrum). If you are curious, you can look these up on Wikipedia, the entries there are good. At the end of the day, they have basically the same performance. Spread-Spectrum modulations don't generally improve maximum sensitivity, however they do improve decodability of a message in interference. An optimal spread-spectrum modulation has equivalent SNR to a narrowband modulation. In the real world, it often works better, though. Additionally, for very low data rates, narrowband modulation becomes increasingly vulnerable to interference, and it is very difficult to ensure that both transmitter and receiver are set-up using the same exact frequency. For ultra-narrowband modulations, usually TCXOs are needed. Spread Spectrum techniques can be applied to allow very low data rates to utilize the most efficient raw modulation (often QPSK), without succumbing to narrowband interference or oscillator mismatch.

 

The major reasons why LoRa works is because it uses a good CSS scheme in combination with a very sensitive receiver. It also has a forward-error-correction scheme built-in, although it is less sophisticated than the Conv-Code + Reed Solomon that I use with the SPIRIT1. I would be curious to see if my reed-solomon codec could improve the LoRa performance to the next level.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...