What Is Your Interest in Roadtesting Cellular IoT Dev Boards?

Hi Kas,

Right, they appear to be near Vancouver. I'm slightly more West than the West Coast: I'm in Comox on Vancouver Island, but in a pinch an airplane or floatplane gets me to Vancouver in just over half an hour.

Thanks for the offer to collaborate! At this point I'm still working full time and I haven't had time to work on my own projects for a while, so I think I'll just keep whittling away for now whenever I have a spec of time. I'm focused mainly on some ESP8266 related projects for the foreseeable future. And I have this CNC machine I really want to put to use soon

Thanks,

-Nico

Hello ntewinkel,

They are headquartered out of British Columbia and they do have pretty helpful staff (if you can get the phone). If you get a MangOH Red ($99) it comes with 100 MB of data with no expiration.

I was hoping you might be more in the middle of this huge country, but it sounds like you are more on the west coast. If you would like to collaborate (help with a nubee) I would be more than happy to work on a project with either a MangOH Green or Red anlong with you.

Kas

Thanks Kas kas.lewis !

That's very useful to know. I also just now (finally!) had a closer look at Sierra Wireless - sounds like an interesting company to keep an eye on. I've never tried MangOH yet, I'll have to add that to my list. Plus they're headquartered somewhat close to here - I think Sierra Systems (related, I think?) was one of the companies in the early 90s that was hiring us co-op students from UVic

The no-end-date part of the data plan is especially nice for those little side projects. The 90 day trial is also very nice - it allows a no-cost way to check off plenty of the unknowns before diving in deep.

Cheers,

-Nico

ntewinkel,

I had a call with Sierra Wireless today and they were pretty helpful. What was interesting is if you plan on using them they can provide upto 5 SIM cards for free that have I think 100MB of data on them for you to do system testing with. This would allow testing for 90 days at which point they would talk about more commercial style contracts. I should mention that on a regular contract (such as the 100 MB that comes with the MangOH Red) the data does not expire but is more like a pay as you go with no expiration.

I was looking for only a few SIM cards so they said they would try put me in touch with a reseller as they deal with more volume style contracts (more than just 5 SIM cards).

With regards to the endless comments here about difficulty getting cell modems working and dealing with AT commands, I have not had this issue with the MangOH boards. They run a Linux framework called Legato that is open source and has a decent community. They are more expensive than the chinese brands but maybe they will be easier to use. I can't say as I've only used the MangOH boards and also I'm newish to using linux in this way. If you want more info on these boards I have posted a few blogs here about them while I have been slowly working on a side project to use one in a weather balloon application.

Kas

I used MultiTech cell modems and SMS protocol.

As I recall, there was virtually no documentation in the MultTech manual describing the sequence of AT commands necessary to discover and connect to the cell network.

Each AT command was described in the manual, sometimes poorly, or incorrectly, but nothing tied it all together.

As I mentioned, I figured how to discover and connect to the cell network through application of logic and a lot of experimenting.

During development I wondered if all the misconfigured requests I sent to the network would draw attention.  Didn't seen to be an issue, and within a week or so, I had it figured out.

rscasnyased on some of the comments here perhaps a tutorial on how to discover and connect to cell networks using AT commands, or whatever is currently in vogue, might be helpful for those getting into cellular IoT.

At this point I'm wondering which module did you guys use. I'm currently using the SIMCOM 5320

I'm in a similar situation.

I'm trying to port over a project that uses a ESP8266 to upload data to a cloud platform using MQTT, to a cellular modem.

Getting the ESP8266 to work was pretty easy - give it the credentials and use a couple of commands to format the data, and then upload it.

The cellular module on the other hand is complicated - you need to send multiple AT commands (which aren't documented very well) to initialize the module, make it connect to the network, open up a connection to the server etc. The connection tends to work for a while, and then disconnects.

Using HTTP/TCP seems to be a little simpler, but it seems that there is a lot that can be done to simplify all of this.

I'm in the extremely interested minority, but I've had quite a few years experience with cellular comms, from early analogue through the various flavours of GSM, and am soon hoping to deploy some 4G modules around the Midlands in the UK.

The necessary AT commands are usually an interesting challenge, as is dealing with the various errors you get when the connection is dropped, or can't be established.

Although trying, it's also quite satisfying when it does work and just survives out in the field and can be contacted - it's not quite so rewarding as deploying and communicating with a Martian rover, the journey time is usually less, but it sometimes feels almost as hard to get to the locations!

As for Luis' comments about video over cellular, I guess because I've been at this so long I try and be very minimalist - do most of the processing locally, discard as much as you can that's boring and can be summarised somehow, and just save that expensive bandwidth for something you know to be interesting / urgent. It's not always easy (or even possible!) to work that out before you deploy the device, but if you don't it can soon get pricey, as pretty much everyone above seems to have noticed too!

A

I completed a project several years ago that used GSM cell modems.  The modems were in a package that included environmental sensors.  The client wanted alerts sent when temperature or humidex thresholds were exceeded.

Each device had a SIM card and a $100/year voice and text plan. I built a total of 6 of these devices.

My biggest headaches were similar to neuromodulator 's.  The AT command documentation was incomplete and sometimes wrong.  Plus, the actual process for discovering and connecting to a network was not explained in the modem documentation, so I figured it out using trial and error.  Probably not the best method, but it worked.

I also discovered that text messages have a variable latency travelling through the network, in some rare cases arriving hours after they were sent.

Not a great feature when sending temperature trip alarms.

I'm currently using a 3g module on an IOT project (SIMCOM 5320). The AT command documentation is 540 pages long and is written in bad chinglish, even worse, the document specification of the module responses is not even right for a few commands. I'm opening a TCP socket to a server to send and receive data, to do that it takes around 8 AT commands, which can return errors that should be properly delt. I'm still not done programming the drivers, I'm using around 12 AT commands, and still need to implement a few more edges of the state machine (loss of TCP connection, loss of mobile connection, etc) and properly parse asynch messages. The 2G modules are far less expensive than the 3G, the traffic cost as others points is not cheap, here the lowest bracket is 1.5 USD for 1 mb/month, but is followed by 3 USD for 100 mb/month. As I want to stay below 1 mb/month I'm using the least number of bits to represent data, which I'm accomplishing through the designing of a binary communication protocol, mapping of floating point numbers into fixed point numbers and differential compression. The strength of these modules is geographical coverage, which may be surpassed only by satellite communication.