It depends on the module, some modules have battery backup, some do not.
The presense of a supply to the backup pin on a GPS, which does not need to be a seperate battery, does mean the the emphersis and other stuff is stored and subsequent power ups can produce a fix in a few seconds.
In general if you power up a GPS from cold and as soon as there is the first fix you turn it off, it will struggle a bit on subsequent hot fixes. But eventually the GPS settles down after some hours and average hot fix times can be in the 5 second region, depending on the GPS. The L76 averages around 5 seconds in my tests.
As for the theory that letting the GPS initially run for 15 minutes or so before powering it down will then subsequently (significantly ?) reduce hot fix times and in turn power consumption, I cannot say.
Is there any pratical evidence that modern GPS benefit from this initial power on period ?
I need to revisit my combo of documentation & testing as it has got quite old and we now have the much better L2C signal to receive. My first Garmin used to take many minutes to do a warm fix and when I used it in the States in 2006 it took well over an hour to get enough satellites before it became vaguely useable.
Testing a NEO-6M based module recently in the office, downloading assist data via the u-Blox software certainly kick started each one but I wasn’t paying much attention to the details - finding a position in the rafters to put an antenna so I didn’t have to string multiple SMA to SMA cables to get one outside was taking up my attention. But it was clear from a brief test between rain that the most beneficial thing was an active antenna outside with a 45° cone of visible sky, second best was the same active antenna in the rafters but away from other cables / flourescent tubes etc.
Most of the personal trackers are pendant based with a ceramic block antenna - I don’t know how that particularly differs from one of the mag-mount active units but I bet being sideways to the sky doesn’t help. But I do know a NEO-M8N with ceramic block wasn’t picking up anything when put in the same place as the active antenna.
I’ve a box of NEO-6M’s which I think would represent a typical entry level GNSS module that have never been used, maybe I can figure out a test setup and try things out - included the all important current consumption.
I’ve looked into schematic diagram and I’m confused a bit.
PIN of ublox is Backup battery and according my module (MAX-7Q) accepts voltage between 1.4 - 3.6 V. This pin is in schematic marked “BAT++/2.4C” but haven’t found opossite side of the signal.
Datasheet says that MAX-7Q supports GPS and Glonass but if trying to set AT+NMEA353=1 then got “Parameter error” as well.
Have to check directly on board and measure if backup PIN is powered or not. And if isn’t then try to connect 3.3 V via cca 4k7 resistor since backup current is 14 uA.
Unfortunatly the ‘myth’ of the 1 hour before a ‘factory fresh’ GPS gets a fix persists to this day, even though the most basic of tests will show that a modern ‘factory fresh’ GPS will get a first fix in the 40 second mark.
Older GPS had fewer channels so the almanac data, when available, could allow the GPS to use the limited number of channels to seach for GPSs that should be in view.
In general the Ublox GPSs dont perform so well with poor antennas, whn compared to the Quectels for instance, see here for some tests;
I had spotted some time ago that leaving a GPS powered up for a couple of minutes after its first fix from cold did reduce the subsequent hot fix times in the short term. But still hot fix times could be quite long.
With an L76 (with backup) setup to get powered up, get a new fix and transmit it every hour, these were the average hot fix times over the next 24 hour period;
No startup delay, average hot fix, 23.9 seconds, total powered uptime 631 seconds
5minute powered startup delay, average hot fix, 19.9 seconds, total powered uptime 792 seconds
20minute powered startup delay, average hot fix, 19.6 seconds, total powered uptime 1691 seconds
So beyond a short startup delay, the GPS does not seem to make good use of the potential almanac data it has downloaded to quickly identify which satellites are in view, at least in the short term.
There are many many conflicting “official” pronouncements on how long the ephemeris & the almanac are valid for. About the only thing that is consistent is the time to transmit.
I ran a few tests just to get my bearings again - once it’s heard something in the last few days (so it has the almanac on board) a M8N with ceramic top stuck on top of a rafter under a flat roof (felt + some puddles & leaf matter) gets a fix after a few minutes. A 6M with active antenna next to the M8N is totally deaf unless I take the antenna outside and once it’s got a fix, it will then work happily inside. Not hugely scientific but it shows how the almanac can help get a fix under edge conditions but the newer chip is overall much more superior. The 6M clearly needs to know exactly what to listen for to have a fighting chance of getting going when reception is spotty.
The reason I go for poor sky view is that seems to be the sticking point for most use cases (and support issues). TTFF when you are outside with at least a 45° cone of view never seems to be a problem unless it’s a small PCB antenna with the PCB & 18650 battery between it & the sky.
I’d have to clear the almanac for that and I’ve only just screwed it’s case back together! I’m sure u-Center has a button but v1 seems to have been modelled on the flight deck of Concorde and I haven’t got to registering for v2 yet.
But you’ve peaked my interest so I’ll drape the damn thing over the side of my monitor as a reminder to do it today.
Remove it from the power and\or the backup supply and leave it a few minutes, unless the almanac has been programmed into flash from an on-line source, dont know much about dealing with that.
The Ublox M8s do perform a good deal better than the older 6M series, and with a lot of fake 6Ms out there, comparisions are sometimes not good.
I’ll have to open the only recently closed case to pull the battery.
But I did use the online assist from v1 recently on that device but as it’s the v1 rocket surgery GUI it was hard to tell if it ‘took’ - seemed to. Whereas some other random series 6 module couldn’t cope with a download so that may be a fake.
I don’t discount testing fakes because it is all too easy for someone to buy some in good faith and then start asking questions - but my backstop is that it has to come from a known distributor for the acid test however achingly expensive. But also learning to cope with the fakes isn’t so bad as they do a basic job OK, so can be put to use for non-critical applications - tracking JCBs for instance - only need a 10m fix for them, the rest I can figure out by looking for the big yellow metal thing.
I’ve did measuring on ublox PIN6 (backup voltage) and can confirm that voltage is presented when tracker is switched on only. Once you power off the tracker, voltage disappear. It’s OK but I hoped that backup voltage is not used and log hot fix is caused by that.
I have several MB of logs from an M8N & a 6M to process - both had a radical battery-ectomy (removal with pliers). Hopefully it will show more detail but I need some downtime to get those logs in to order.
The 6M of questionable provenance could get a fix from power up inside 30s but had an active antenna outside with a good view of the sky.
The M8N drone puck with ceramic top stuck in the rafters with a felt roof above it took between 3.5 and 4.5 minutes to get a fix.
I’ve started downloading various documentation to try to see if there is a consensus on GNSS. There is an interesting ST doc that shows various modes for their chipset that minimises power vs keeping enough almanac or ephemeris data available to perform the quickest fix in line with the power policy - so sometimes allowing the ephemeris to expire but then using hard math to extrapolate it to get a better than warm fix.
None of the vendors are very explicit about how they may combine GPS (USA), GLONASS (Russia) and the built-in-the-UK satellite constellation called Galileo that the Europeans have stolen sequestered. Apparently do-able but not simple.
Looking at the official GPS docs it seems that the ephemeris runs to about 4 hours now, but GLONASS is only 30 minutes - which would explain why a hot fix is possible way beyond 30 minutes as I thought.
The GNSS world is like a LoRaWAN cliche of special people who mainline TLA’s and move the goal posts on the transmissions with each new series of satellite with more & more refinements to increase accuracy and keep the rest of us on our toes. The situation is further complicated with a new constellation that is just coming on-line now that there are enough of them so all our modern chipsets should start doing even better.
I’m going to get an M8N module from Uputronics, a very trustworthy source / manufacturer and setup a spare Pi to run tests as I think of them. But the creation of a summary of how GNSS works I think I’m going to have to delegate to someone with more time & a less full brain.
To solve the question as to whether the GPS hot fix time reduces if the GPS has been powered for a period, its easiest, in my view to use simple point to point LoRa setups.
I currently have a ‘tracker’ running in my garden, sending the fix time via LoRa which is picked up by another LoRa device with an SD openlog on the serial port.
Around Thursday this week it will have been running a full week, so the logs on the SD should reveal what the long term average of hot fix is. The last ‘hot fix’ was 16.9 seconds.
So next week, I will let the GPS stay powered for a long time, an hour maybe, and see in practice how much difference it makes.
