LoRa Gateway overall Power Consumption

I am part of a research team working on energy models for scalable LoRa Networks at Politecnico di Bari, Italy. I was going through the data sheets for various LoRa Gateway manufacturer’s (i.e. Microchip, multitech, linklabs,cisco) to understand their total power consumption (in Watts) but I could not find power related information on their data sheets if someone can provide a precise estimate.

Moreover, I would be grateful if you can also guide me about the average power consumption of different LoRa Gateways/ concentrators belonging to different vendors. I have gone through the datasheets for a gateway from symphony link, and they believe it to consume around 10 W but I am sure not all the gateways consume almost similar value of power (for example Cisco’s LoRa gateway, contrarily consumes around 40 W according to them). Can we safely estimate an average power consumption value for a LoRaWAN gateway or a lower and upper bound of consumption ? Your cooperation would highly be appreciated.

Looking forward to hear from you at the earliest.
Many Thanks in advance.

I’m looking to roll out some gateways at solar powered sites. I’ve started looking into this.

I have just built a Raspberry Pi based gateway that uses the new Raspberry Pi 3 B+ and an MTAC-LORA-915 card.

At idle (receiving approx 50 messages per day), the gateway (RPi & MTAC card) averaged 5.5 watts (1.11A at 4.95v) over a 24 hour period.

5.5 W/h?

Pi 3 B+ consumes much more than a normal 3B in idle state… strange choice for a solar powered application

I am currently using the LorixOne whose consumption sholud be below 3 W/h

Alex

I note that with each generation the RPi’s are burning more power and will often be main ‘consumer’…new RPi 3B+ is perhaps ~2x over old Pi2 or PiZero/W. For some new builds I have bought 3/3B+ where power less of an issue but have also just bought a bunch of Pi2B’s for more sensitive apps and where tight packing means internal heat a potential issue longer term…just my 2penneth :wink:

BTW think I saw a recent post on the forum comparing pkt fwdr performance with quad core/multi thread RP3B with single thread on Pi Zero but quick search didn’t turn up to post here - may be worth you digging…

Also inline with last night’s posting wrt Laird RG191/186’s I’ve found that if I run the Laird off lower voltage then current consumption is generally lower c/w my RPi based platforms which may help your overall use…

I replaced the RPi 3B+ of my ‘lab’ setup by a Zero, and the processor is not used a lot.

I haven’t measured the power consumption, but another factor is heat dissipation. 3B (and 3B+) needs to dissipate all this power – no secrets – and it can be problematic in an enclosure…

Only downside of the Zero is that you either need to go wireless or add an USB network adapter…

Note that there are 2 versions of the RPi 2, the first batch has an ARM V7 processor while the last one has the same V8 chipset as the RPi 3 and is more power hungry…

Power Consumption Benchmarks RPI’s

I understand the later Pi 2’s using newer chip also then have potential SPI timing issue flagged (with work around) elsewhere on Forum.

Here is graph from Iss 68 (April) MagPi Mag showing relative power draw of the generations…think this reflects earlier RPi 2’s

image

PiZero WiFi only is also a pain when trying to use external IP rated box (typically metal such as the E-Zy housings and others), hence my choice of 2B’s

Yes, I would think about making a 3D printed “skeleton” to hold the Zero and the concentrator, that could be fixed in an IPxx box. But I am not in favor of Wireless for “production” use, so I do agree 2B first gen is a better choice.
Do you have a reliable source for 2Bs? I am currently swapping devices at home and trading with friends who are usually happy to get an upgrade :wink:

“Do you have a reliable source for 2Bs?” sadly not guaranteed to be old ones

“currently swapping devices at home and trading with friends”…good tactic! Pity I had a bunch of Gen 1’s before getting into this and sadly few Gen2’s so have to rely on what the postman brings… :slight_smile:

Thanks very much all for your input. We recently published an article based on the consumption of several lora vendors, if it might be of your interests! I am sharing the link below:

1 Like

They must be using a Linear regulator. I’ve just ordered a Laird RG191 and will have a bit of a play.

No, 5.5 watts (132 W/h consumed over a 24 hour period).

The Pi 3 B+ is what the stores here in NZ are now selling, and to be honest, I’ve never considered using the Pi zero for any of my various projects.

Most of my sites already have a RaspberryPi based ADSB receiver - This is a CPU intensive process while the LoRa gateway seems to need minimal resources. Maybe the best option for me is to upgrade the Raspberry Pi and do both ADSB and LoRa gateway from the same RPi.

Uhmm… I think you’re making a bit of a mess with the units of measurement

Alex

How? What’s wrong with watts and watt hours?

If your application consume 132W/h it means that it needed 3168W in a day :scream:.

I think you mean that it consume 5.5W/h that is 132W in a day.

Alex

Its 5.5W and 132Wh per day (thats Watts x hours, not Watts / hours)

My input on this is that the SX1301 concentrator is the main power hog. I say that because the MPU is a variable that can be scaled by processing load, clock speed and silicon process, but the SX1301 is on all the time in receiving mode.

Microchip LGx271 module datasheet (as used in TTN gateway) puts it at about 3W and this feels about right. It certainly runs hot.

So where you’ve seen 40W total, maybe that was a 64ch gateway, with 8x SX1301’s ???

A standard 8ch gateway should be around 5W constant.

Assuming solar power would need ~24hrs autonomy, that would need a pretty large lithium cell, like automotive sized, or a standard lead-acid battery

Right, I see the issue. As @JDP posted, I incorrectly used W/h (Watts per hour) instead of Wh. But your understanding of power units is also incorrect.

Watts are the used to measure a rate/speed of consumption (1 joule per second).
Watt hours, Kilowatt hours, etc are a volume/measurement of use over time.

White W/h could be used to measure a rate of change in wattage over time, it’s probably not a very common unit to use.

I’ve done some tests on the RG191 and found that this is not the case.

Voltage (v)Current (A)Power (W)
130.1752.275
120.192.28
110.2052.255
100.2252.25
90.252.25
80.282.24
70.322.24
60.3752.25

From my tests, it draws ~2.25 Watts regardless of input voltage

I took my RG191 apart. It uses a TI TPS54525 switching buck regulator which supports up to 18V input. The output is set to ~4.55 volts.

Based on this, the RG191 will run on an input voltage range of 5-18 volts. In other words, it will happily run directly off an AGM battery and the fully charged voltage of almost 15V.