Location Network

[lbjstetson]

I'm looking to build a system that will report back GPS coordinates of some of the equipment & machines that my company owns. We're in construction and have many machines(200+) spread out between many different projects. My thought is to put main units on a few of the company trucks that travel between the jobs a lot. The units on the equipment would just be a simple moteino with a radio and would simply respond with a unit number. I'd want to put them in some sort of a magnetic puck or something that could be easily moved. I'd assign the units to machines in my online system. The main units would have a celluar GSM & moteino w/radio that would send out a request to the units on the machines.

My questions:
     1) how long could I expect the battery life to be and what should I use for batteries?
     2) I'm located in Wisconsin. What would be the best of frequency radio to use?
     3) Any ideas for a housing for the simple "puck" units?
     4) What would be the best radio option for range?

Thanks!

[ChemE]

1) Decades on 2xAA Energizer Ultimate Lithium batteries if you remove the LDO from the Motes, sleep them often and just have them wake periodically (once an hour?) to broadcast out a "here I am" pulse

2) 915MHz in the US

I think the way to set up your network is to have the GPS/cell units on things with big batteries that would never notice a constant 40mA load and then put bare Motes on smaller things like compressors and blast pots for instance.  The smaller units could easily run 10+ years on a pair of AAs if you slept them appropriately.

[lbjstetson]

I was thinking that they could be on listen mode constantly. The main units would initiate by periodically sending out a signal every few minutes and the machine units would respond with a unit #. The main units are on trucks that come and go in a short period of time from our projects . I could have the machine units sleep for a while after they answer to prevent them from constantly responding. How would that impact battery life?

[ChemE]

I think some folks here have had long term problems with listen mode.  Does the equipment set change so rapidly that a TX every 5 minutes wouldn't be recent enough?

[Felix]

lbjstetson,
May I add some thoughts based on the description/requirements.
For best battery life, I would recommend having the units transmit their location at a constant rate, rather than add the complexity of having to ping each node, which would need to be awake all the time to receive the ping.

I think even with default settings the RM69HCW (with ATC) can give you plenty range while keeping cost down (LoRa radios are more expensive). The 915mhz is for the US as ChemE suggested.
For antennas you might require external mounted antennas - uFL to coax to case SMA to antenna. Dipoles might be the best - and see captcha's thread about PCB dipoles.

It sounds like each location would have a central unit (with the GSM module/gateway?) which would handle the data gathering at that location?
Are units being monitored while in transit from one location to another, or do you only care to see them when on a site that has the GSM gateway?

[lbjstetson]

Felix:
All that the nodes will need to report is a unique ID, 5 digits would probably be plenty. I'll assign them to machines and equipment in my equipment location app. I'm planning on having the location come from the central unit since they'll have a cellular gsm with GPS anyway. How long could the battery last if they sent out their pings every 5mins or so? I could have it sleep for a few hours when they receive a response from the central unit. We have a fuel truck which goes to our jobs every day but it's only on the jobs for 20-30mins so I'd want to have the nodes reporting fairly frequently. If it works well, I could put central units on other trucks (quads,semis,etc) in order to get more frequent location updates.

I'd probably want to go with the LoRa radios for a few bucks more per unit if the range would be better.

Would pcb dipoles like that be able to be placed on the top of a truck?

Correct, the central units would collect the data and transmit via gsm every 15 minutes or whatever seems to work best.

Are there any guides for making a good weatherproof housing and securing the mote inside of it?

ChemE:

Every few minutes would probably be fine. That'd give the nodes a few chances to make contact per vehicle visit. What would that do to battery life? Thanks for the advice, btw.

-----

This is unrelated to everything above but I have a mote based sensor network in a field right now with about 35 nodes and we're having issues with units needing to be manually reset when we move them. Right now the motes are loose in a simple case, see photo below.

[ChemE]

Probably easiest to consider the various functions of an equipment node separately.  I'll assume we wake every 32 seconds to broadcast w/o asking for an ACK.  If there is a collision, they will rebroadcast again in 32 seconds.

Sleep current if you remove the LDO is 4,400nA
Active time while the uC is speaking SPI to the radio is 10nA
Broadcasting (assumes 300kbps) is 326nA
Total average current across one 32 second cycle is 4,736nA

2xAA L91s at such a teeny tiny current have about 3,400mAh or 3,400,000,000nAh of capacity.

3,400,000,000nAh / 4,736nA = 717,905h = 82 years

In truth the self discharge rate of the batteries will trim this to 10-20 years but you get the idea!

[lbjstetson]

Ok, excellent. So basically, I don't have to worry about battery life. Thanks!

[lbjstetson]

I think the way to set up your network is to have the GPS/cell units on things with big batteries that would never notice a constant 40mA load and then put bare Motes on smaller things like compressors and blast pots for instance.  The smaller units could easily run 10+ years on a pair of AAs if you slept them appropriately.

The problem is that if I were to connect to the main battery, I'd likely get in trouble whenever there was a battery issue regardless of whether or not my unit could possibly be the culprit. I really want to avoid that. We recently had a OBDII plugin GPS tracking unit that caused a truck to not shift above a certain gear.... don't ask me how because those things aren't supposed to impact anything like that.

[perky]

You might need to look at FCC compliance here, for commercial things that is probably the elephant in the room.

Mark.

[ChemE]

He won't be close to 1% duty cycle but you think that frequency hopping might also be required?

[lbjstetson]

To be clear, these would not be sold but used by my company.

[Felix]

lbjstetson,

For LoRa you will have to use a 3rd party library - RadioHead is the popular one, see this page for getting started with that. I am sure there are plenty of examples on the web on how to better utilize that library. I would personally start with a pilot using RFM69HCWs and see how that does.

The antenna is important especially in mobile applications like yours. RF does not like enclosures, especially metal. Any antenna will love to be elevated as much as possible, with no obstacles in the line of sight, keep as much to that principle as possible to maximize the chance to receive packets.

I would not go to 300kbps in real practice except when range requirement is minimal or for short range wireless programming. I think the default 55.5K is good, and perhaps if the question comes, you can lower it to 19.2k or even lower.

To maximize battery life you need to sleep everything inside the mote that might use power, when the mote is not in use. There are various ways to monitor battery life. The simplest is a voltage divider to bring the battery voltage under the 3.3v operation of the Moteino, and read that on an analog pin. Several examples in the RFM69 library (motion mote is one).

I would not focus too much on removing the LDO at the moment, that will save you 2uA and I think your effort is better spent elsewhere.
Make sure to sleep the FLASH MEM chips as well since I see you have that in the photo.

I do NOT recommend powering a Moteino or any electronics straight from a car battery, it can be damaged by unregulated spikes and transients during cranking.

For enclosures, look around in the projects forum for inspiration, there have been lots of great examples how people got creative with cases in various projects. I think it's easier to develop your motes around an existing case rather that the reverse. Choose a case that makes sense, there are lots of possibilities. Perhaps choose one with a rubber ring top that mounts with screws. You can mount your 3xAAA or similar power pack on the top with velcro or hot glue, for the RF use coax cable pigtail of 1/4 wavelength to route the signal from the mote to a panel mount SMA (you can find some on ebay precut to length). Then use external mounted antenna with matching SMA mount.

[Felix]

This is a pretty cool project which uses a piezo transducer to detect vibration of a vehicle and send GPS location via GPRS/SMS using the SIM5320e module.
It's hooked to an arduino that sleeps most of the time, and only wakes on vibration, and also controls the power to the SIM module via a mosfet. Nice implementation.

https://www.youtube.com/watch?v=CeStrH-5Llo

[lbjstetson]

This is a pretty cool project which uses a piezo transducer to detect vibration of a vehicle and send GPS location via GPRS/SMS using the SIM5320e module.
It's hooked to an arduino that sleeps most of the time, and only wakes on vibration, and also controls the power to the SIM module via a mosfet. Nice implementation.

https://www.youtube.com/watch?v=CeStrH-5Llo

Interesting. I recall a company that pitched me on devices that tracked operating hours for machines based on vibration. It reported in a similar way that I'm hoping to make my system work so there was no monthly charge per unit. There are tons of companies out there offering equipment tracking but they all seem to be cellular on the $15/mo or more schedule plus you have to hook them into the machine's electrical system somehow. When you're talking 200 or more units, that's a lot of money.