Weird behaviour caused by?
Display on one of the test units (213BN 5.89 firmware) GO PRO casing, showed battery voltage lower as 2.7 volt.
I put the unit on the charger and the RED charge led was lit.
After 12 hours the red charge led was off.
So I pressed the ON/OFF button but there was no change in the display.
1 sec or 4 sec pressing the button did not change the hang up.
I removed the battery and connected a full one.
The GPS starts up automatically.
I put back the original battery and the GPS shows it at 4.1 volt, so it was charged correctly.
ON/OFF button works fine.
Why would after the charge the unit not respond on the ON/OFF switch which has a GREEN led for optical working?
Paco
Usually when "turned off", the ESP logger is in deep sleep mode, which allows it to wake up when buttons are pressed. I think the firmware is supposed to put the unit into deep sleep more if the power drops to low. That, however, may fail (see github.com/RP6conrad/ESP-GPS-Logger/issues/35 for one possible reason).
From what you describe, your unit may have gone into "hibernation" when the battery voltage dropped too low. Hibernation uses even less power than deep sleep, but seems to do so by turning off support for push buttons (see thelinuxcode.com/esp32-sleep-modes-power-consumption/). If the ESP32 is in unplanned hibernation, then only power cycling (e.g. removing and reconnecting a charged battery) would restart it (but with a logger in a GoPro housing, you can also use the sliding button on the side).
There are a couple of other things that may have happened instead. One is that the ESP32 actually went into "power off" mode when the voltage dropped too low. In that case, I would expect it to reboot after power goes up again, but I may be wrong there.
The other thing is that this may be a software (or rather firmware) freeze. I've seen this happen, but it is very rare, so I think this is least likely. A freeze would require power cycling, too.
out of interest, has anyone started looking at commercialising this or is our market just too small and there are other products out there that are "close enough".
Select to expand quoteboardsurfr said..
Usually when "turned off", the ESP logger is in deep sleep mode, which allows it to wake up when buttons are pressed. I think the firmware is supposed to put the unit into deep sleep more if the power drops to low. That, however, may fail (see github.com/RP6conrad/ESP-GPS-Logger/issues/35 for one possible reason).
From what you describe, your unit may have gone into "hibernation" when the battery voltage dropped too low. Hibernation uses even less power than deep sleep, but seems to do so by turning off support for push buttons (see thelinuxcode.com/esp32-sleep-modes-power-consumption/). If the ESP32 is in unplanned hibernation, then only power cycling (e.g. removing and reconnecting a charged battery) would restart it (but with a logger in a GoPro housing, you can also use the sliding button on the side).
There are a couple of other things that may have happened instead. One is that the ESP32 actually went into "power off" mode when the voltage dropped too low. In that case, I would expect it to reboot after power goes up again, but I may be wrong there.
The other thing is that this may be a software (or rather firmware) freeze. I've seen this happen, but it is very rare, so I think this is least likely. A freeze would require power cycling, too.
Thanks for the extensive explanation.I will test this unit by leaving it on and see if it goes again in hibernation mode.
Let you know the results.
If the battery depleats quicker at the low end cut-off voltage setting when the ESP can not write the end of session files and goes into hibernation mode instead of deep sleep mode then I can see why it COULD go wrong.
This GoPro version is charged with a wireless charge coil system not 5 volt from USB.
This time it was a GoPro version but when it is poured like other people do, than your done... :-(
Paco
Select to expand quoteRolz said..
out of interest, has anyone started looking at commercialising this or is our market just too small and there are other products out there that are "close enough".
I am slowly starting to do this in low volumes on request. Prototyping takes more time as usual.
Select to expand quotePacoRaapNL said..
Weird behaviour caused by?
Display on one of the test units (213BN 5.89 firmware) GO PRO casing, showed battery voltage lower as 2.7 volt.
I put the unit on the charger and the RED charge led was lit.
After 12 hours the red charge led was off.
So I pressed the ON/OFF button but there was no change in the display.
1 sec or 4 sec pressing the button did not change the hang up.
I removed the battery and connected a full one.
The GPS starts up automatically.
I put back the original battery and the GPS shows it at 4.1 volt, so it was charged correctly.
ON/OFF button works fine.
Why would after the charge the unit not respond on the ON/OFF switch which has a GREEN led for optical working?
Paco
I have no explanation for this behaviour. The normal behaviour should be :
If the voltage goes lower then 3.1V, the ESP should go to deepsleep, with the message "Low voltage". Normal deepsleep, every 1800s wake up and measure again voltage. I difference > 0.1V -> refresh screen.
At some point, the voltage goes below the treshold (2.7 V?) from the protection print in the lipo. Then, the power is cut off complete. Screen freezes with the last message on it.
When charging again, the ESP directly tries to boot, but as the voltage is still too low, it is send again to deepsleep. This goes on for some time, until the voltage is high enough so that the lipo protection stays on.
Maybe a issue with the protection print in the lipo ? Can you do some more tests with your setup ?
Greetings, Jan.
Select to expand quoteRolz said..
out of interest, has anyone started looking at commercialising this or is our market just too small and there are other products out there that are "close enough".
Hi Rolz, Jan made the software available open-source under MT rules at: github.com/RP6conrad/ESP-GPS-Logger?tab=MIT-1-ov-file#readme
One should be able to provide help and assistance in the build of these units as a service fee. The whole point of this community is to help and contribute. There are a few people that help and assist others. It's great to see others stepping in as well.
For myself the reliability of the devices is a concern. Some devices that could be reopened started to leak over time and I also had poured devices stopped working. Not clear what are causing the issues. Success rate <50% so quite a bit of devices "lost"over time...
Select to expand quoteFreezer said..Rolz said..
out of interest, has anyone started looking at commercialising this or is our market just too small and there are other products out there that are "close enough".
Hi Rolz, Jan made the software available open-source under MT rules at: github.com/RP6conrad/ESP-GPS-Logger?tab=MIT-1-ov-file#readme
One should be able to provide help and assistance in the build of these units as a service fee. The whole point of this community is to help and contribute. There are a few people that help and assist others. It's great to see others stepping in as well.
For myself the reliability of the devices is a concern. Some devices that could be reopened started to leak over time and I also had poured devices stopped working. Not clear what are causing the issues. Success rate <50% so quite a bit of devices "lost"over time...
Freezer,
" Some devices that could be reopened started to leak over time."
Are you talking about 3D printed cases? Or the reactanlge boxes of the first designs from Jan or the GoPro cases?
Paco
Select to expand quotePacoRaapNL said..
This GoPro version is charged with a wireless charge coil system not 5 volt from USB.
That should not make a difference. Both USB and wireless charges deliver 5 V. Some of the fancy wireless chargers used in phones etc. have feedback circuitry that checks how much charging is actually going one, but they still have to deliver a voltage that's higher than the battery voltage while charging. Charger and battery are connected in an electric loop while charging, so what the ESP board measures is something in between the battery voltage and the charging voltage. That means that the voltage read during charging is higher than the battery voltage. Then, automatic calibration will give the wrong results, and the unit may not be put to sleep in time.
I'm running a test now to see if I can reproduce the problem you saw. I'm using an older logger that's still got an old firmware version installed, so I have set the calibration value to high and will see what happens.
Select to expand quoteFreezer said..
I also had poured devices stopped working. Not clear what are causing the issues. Success rate <50%
Can you give any details about the problems you had with these? Did you use them regularly, and the just stopped working randomly, or did you not use units for a long time and then could not wake them up again?
Or did you have problems with the screen going bad?
I have 3 units that I made about 3 years ago which are all still working. They are all in GoPro housings and can be turned off completely using the slider switch. I've lost two units - one when the GoPro clip opened or broke, and one when I used a new brand of GoPro housing that turned out to be leaky. In both cases, it was really my fault. I had not used a safely line on the first unit, and did not notice when it cam off my helmet. For the second one, I had omitted to put silicon waterproofing on before testing it on the water. Water intrusion was minimal, and it probably would have survived with a thin silicon film. Interestingly, the only part that's really damaged is the screen, which now tries to emulate my old Motion display screens.
Various Openlog-based loggers I have made have held up for a few years, and a few hundred sessions, before starting to act up of fail completely. It's rather non-trivial to create a logger that holds up for many years, like the GT31 or the Garmin watches.
The high tack double sided acrylic tape patch of 70x100mm cutout for applying on the top of the main case.
Now no fiddling anymore with the locations were the previous tape needed to touch.
Enhancing the water resistance a step more.
@ Jurjen nice design. The switch is 100% waterproof?
@ Boardersurf.....understand your point of charge voltage.
Paco
Select to expand quotePacoRaapNL said..
The high tack double sided acrylic tape patch of 70x100mm cutout for applying on the top of the main case.
Now no fiddling anymore with the locations were the previous tape needed to touch.
Enhancing the water resistance a step more.
@ Jurjen nice design. The switch is 100% waterproof?
@ Boardersurf.....understand your point of charge voltage.
Paco
We will see...Ali sold them as waterproof. A momentary push button seems to be the best in terms of user friendliness, but not in terms of durability. I'll have a look and see for better alternatives
Select to expand quotePacoRaapNL said..Freezer said..Rolz said..
out of interest, has anyone started looking at commercialising this or is our market just too small and there are other products out there that are "close enough".
Hi Rolz, Jan made the software available open-source under MT rules at: github.com/RP6conrad/ESP-GPS-Logger?tab=MIT-1-ov-file#readme
One should be able to provide help and assistance in the build of these units as a service fee. The whole point of this community is to help and contribute. There are a few people that help and assist others. It's great to see others stepping in as well.
For myself the reliability of the devices is a concern. Some devices that could be reopened started to leak over time and I also had poured devices stopped working. Not clear what are causing the issues. Success rate <50% so quite a bit of devices "lost"over time...
Freezer,
" Some devices that could be reopened started to leak over time."
Are you talking about 3D printed cases? Or the reactanlge boxes of the first designs from Jan or the GoPro cases?
Paco
These are our 3D printed versions:www.printables.com/model/320425-gps-case/ and www.printables.com/model/379323-m10-winsurfinggps. Initially tried to use an o-ring, but later start using silicon. I think I have ~10 versions made, there are only 1-2 still water tight (but not a lot used). Switched to epoxy-pouring concept and save quite a bit in size (no need for inserts, plexiglass, etc). However here I also had random failures. I used extra clear, slow curing, low heat pouring epoxy. ~6-8 units failed. Later I used polymax to protect sensitive areas on the lilygo PCB that remains a bit flexible before pouring. Looked a bit better, but also here assembly issues. Charger connected got loose just before pouring, would not recharge, etc. etc... I think I bricked about 16 units, very low success rate. BigBoss had a better success rate. Units also fail over time with an improper wireless charger causing to overheat the device (likely the battery).
Select to expand quoteJurjenvdN said..
I made my own design, inspired by Paco. What do you think?
Jurjen, ensure the BE280 has a clear "view" of the sky with the antenna. This is the ceramic part. Right now you have it pointed downwards and below the PCB, this will have quite bad reception of GPS signal. Flex's design uses a 90deg angle on his boom design, while I'm trying to keep in the same plane (next or above) as the display (or slightly angled).
Select to expand quoteboardsurfr said..Freezer said..
I also had poured devices stopped working. Not clear what are causing the issues. Success rate <50%
Can you give any details about the problems you had with these? Did you use them regularly, and the just stopped working randomly, or did you not use units for a long time and then could not wake them up again?
Or did you have problems with the screen going bad?
I have 3 units that I made about 3 years ago which are all still working. They are all in GoPro housings and can be turned off completely using the slider switch. I've lost two units - one when the GoPro clip opened or broke, and one when I used a new brand of GoPro housing that turned out to be leaky. In both cases, it was really my fault. I had not used a safely line on the first unit, and did not notice when it cam off my helmet. For the second one, I had omitted to put silicon waterproofing on before testing it on the water. Water intrusion was minimal, and it probably would have survived with a thin silicon film. Interestingly, the only part that's really damaged is the screen, which now tries to emulate my old Motion display screens.
Various Openlog-based loggers I have made have held up for a few years, and a few hundred sessions, before starting to act up of fail completely. It's rather non-trivial to create a logger that holds up for many years, like the GT31 or the Garmin watches.
All related to water tightening the 3D printed housing. The go-pro housing should work great, but feels very bulky. Under estimated the time and effort to do it properly. Pouring epoxy also with limited success. Device froze over time, unclear what caused this. Some clear cases when charging heat was causing damage to case and components. Most recent example does not switch on since sunday. Not clear if this is caused by the reed-switch or the display is frozen... Difficult to debug once it is poured with epoxy.
Select to expand quoterp6conrad said..PacoRaapNL said..
Weird behaviour caused by?
Display on one of the test units (213BN 5.89 firmware) GO PRO casing, showed battery voltage lower as 2.7 volt.
I put the unit on the charger and the RED charge led was lit.
After 12 hours the red charge led was off.
So I pressed the ON/OFF button but there was no change in the display.
1 sec or 4 sec pressing the button did not change the hang up.
I removed the battery and connected a full one.
The GPS starts up automatically.
I put back the original battery and the GPS shows it at 4.1 volt, so it was charged correctly.
ON/OFF button works fine.
Why would after the charge the unit not respond on the ON/OFF switch which has a GREEN led for optical working?
Paco
I have no explanation for this behaviour. The normal behaviour should be :
If the voltage goes lower then 3.1V, the ESP should go to deepsleep, with the message "Low voltage". Normal deepsleep, every 1800s wake up and measure again voltage. I difference > 0.1V -> refresh screen.
At some point, the voltage goes below the treshold (2.7 V?) from the protection print in the lipo. Then, the power is cut off complete. Screen freezes with the last message on it.
When charging again, the ESP directly tries to boot, but as the voltage is still too low, it is send again to deepsleep. This goes on for some time, until the voltage is high enough so that the lipo protection stays on.
Maybe a issue with the protection print in the lipo ? Can you do some more tests with your setup ?
Greetings, Jan.
I can confirm this behavior, test it alot.
but as Freezer (we are on the same project) mentioned there are some strange behavior. At the moment there are 5 devices which worked perfectly fine that wont charged anymore, 2 works fine for 9 months. 2 works fine for 3 months 1 for one week. First behavior is that the percentage fluctuate a lot, then it wont charge above 40%, later its showing 0% (3.4 volt). These 3 device are stil working but wont charge above 0%. The other one wont start anymore. Strange?? it al happend in one week (last week). They are al poured so debugging is limited. First i thought it a exemplary fault of 1 battery. But after the 3 one . it's strange. My first (and oldest) poured esp-gps is still working perfect after al my sessions.
this one uses a 1300mah battery.
all the other version uses a 1500mah battery. It lokale like a battery qualityissue.
If I am talking about the ESP-GPS units I use, it is always used on the left upper arm!
@ Jan.
Firmware version 5.89The troubled unit showed the known message see picture and the voltage was below 2.6 volt.
Voltage measured on the GND-Vbat was 0.11 voltSo Li-Po battery electronic had switched off the output to the board.
Push button gives no action to wake up.
Put the unit on the wireless charger and the red charge LED comes on.
Display wakes up and goes into a loop of trying to start the GPS.When the display says the voltage is 3.1 volt the unit passes the GPS initialization.
I stop charging and put the unit in sleep manually.
So this is how it should work. Will do some more drain testing to see what happens.
I will for the first few test not fully charge the Li-Po to auto switch off by the Lilygo.
Later test will be from full charge to auto switch off by the Lilygo.
@ Freezer (Simon)
The GoPro is more bulky but when I used it on the upper arm with my GoPro holder it did not bothered me during surfing wearing a 3mm wetsuit.
The 3D design need to be minimum 6 walls or layers thick when using 0.4mm nozzle.
Water resistant is the base we are fighting. 100% waterproofing a design which 99.9999% of the time is not below the water is not needed.
Always check in a preview of the slicer that there are no locations where there are not enough layers or walls.
The first design I took from the web had this problem.
Although they looked very nice it took me a few prints to understand the problem.
After printing a main case fill it with water to see if the water seeps through. If there are spots that are not water resistant it will leak out.
Put a piece of paper kitchen towel underneath. It will quickly show leaking.
The density of water might be higher as epoxy but we are not surfing in epoxy.
So with normal water pressure the main case will not collect water easily if not put below water level for a long time.
A layer of acrylic spray in and outside will prevent very tiny holes or surface imperfections to guide water in or outside.
@ Jurjen
Like Simon says antenna must be located to the upperside (Sky) as much as possible.
When the GPS is under the Lilygo display I still reach 13 + sats easily even when I test it in the car in the center console.
If I am correct the Chrono GPS watches do not go further as 5 stats!!! So what are we talking about?During my holiday I had an ESP-GPS with me.
When I put it in the salt water swimming pool at 5cm below water level the sats went to 0 in a second.
As soon as I put it above water level the 13 sats came back instantly.
This was with the antenna flat facing upwards.
Paco
Seeing a lot of epoxy potted GPS units die.
The epoxy most guys are using is for cosmetic/timber finishes with very slow cure and very low viscocity. Easy to get. Easy to use. Not expensive. But...
This stuff usually sets hard and brittle, and shrinks a few % when it cures. This creates significant stress on the PCB components and can crack them (usually capacitors), or tear the gold plating off the pads- leaving components disconnected. So a few shocks from drops/crashes and/or thermal cycles (hot car or magnetic charging and then into cold water) will significantly accellerate failure of already mechanically stressed parts. I'm amazed the epaper displays last as long as they do like this.
Normally we use low hardness PU, or silicone formulated specifically to pot electronics. With harder potting material like epoxy (needed for some applications), you can flow coat the electronics first with a thick layer of flexible confirmal coating, or silicone. This creates a soft layer to reduce stress on the fragile components and PCB bonds.
I don't pot my GPSs. I've only had one fail but that was due to (I'm sure) a drop of sweat that happened to land on the battery charger IC when I was assembling it, and corroded off one of the pins after a year or so. There was no other sign of water in the case.
I use a magnetic contact with a USB lead from aliexpress that I made a cutout for in the case and glued in, along with an external cheap battery charger PCB with blue/red status leds that I can see clearly through the PLA case. Its been really reliable and convenient, and charges very fast. It doesnt even get warm.
There is indeed a major problem when the lipo goes in to protection. When charging starts, the ESP directly boots and try to connect to wifi. This is very power hungry, and if the (inductive) charger can't deliver this power, voltage drops and it al starts over again. The ESP comes in a bootloop. I will change the sw to prevent this :
After booting when power comes up, the ESP directly goes back to sleep. Waking up only possible by activating reed switch / button. This should prevent this kind of bootloop.
For now, my strong advice is to deep it on a minimal charge from 25% !!!
My experience with the non potted ESP is actually very good. If no water intrusion happens, the fail rate is very low. I have units from 3 year old who are still working. Sealing is crucial. The seal that comes with the electrical box is no good. I seal them now with silicone, just fill the gap in the top part, make it flat and let it dry. As the silicone does not stick to the gap, you can easily remove it and redo it if needed.
Greetings, Jan.
Thanks Jan for thinking with us,
As always highly appriciated.
Could the new firmware also have the option of automatic battery voltage calibration and a manual one?
Tick box in the EWS to change? By default automatic. For the PRO's we can change the setting and the calibration value?
Paco
As per Elmo's request heres a link to what I used. Its good enough!
www.aliexpress.com/item/4001283805395.html
I masked off the gold pads on the front and gave it a spray with clearcoat to seal it a bit better around the sides. I also put glue over the pins on the back after I soldered the wires. The pins are not waterproof- they are just pushed into the plastic.
A few heavy coats of urethane clearcoat seems to seal the PLA printed case well enough for me.
Select to expand quoterp6conrad said..
For now, my strong advice is to deep it on a minimal charge from 25% !!!
That's a problem. Some of my (non-potted) units run down the battery to 0% quickly (in a few weeks) when not used. It's pretty common to not use the units for many months (I use mine only during a 3 month period, but they can be switched off completely).
The info slowboat has about potential failure from stress is interesting. Adding a thick layer of special silicone is easy enough to do (I use Silicone Conformal Coating from MG Chemicals for some minimal protection of my units in GoPro housings). It just stinks a bit 
.
It would be interesting to know how many failures happen during a session, and how often they happen after not using units for a while. Failures in a middle of a session would likely be due to the mechanical failures slowboat outlines. Failure to wake up after not using units for extended times could also be due to battery issues. Even if the battery protection circuits turn off everything when the voltage drops too low, long storage can still put batteries into a state where they won't charge again. I got quite a few rarely-used electronic gadgets lying around where this has happened.
Select to expand quoteslowboat said..
Seeing a lot of epoxy potted GPS units die.
The epoxy most guys are using is for cosmetic/timber finishes with very slow cure and very low viscocity. Easy to get. Easy to use. Not expensive. But...
This stuff usually sets hard and brittle, and shrinks a few % when it cures. This creates significant stress on the PCB components and can crack them (usually capacitors), or tear the gold plating off the pads- leaving components disconnected. So a few shocks from drops/crashes and/or thermal cycles (hot car or magnetic charging and then into cold water) will significantly accellerate failure of already mechanically stressed parts. I'm amazed the epaper displays last as long as they do like this.
Normally we use low hardness PU, or silicone formulated specifically to pot electronics. With harder potting material like epoxy (needed for some applications), you can flow coat the electronics first with a thick layer of flexible confirmal coating, or silicone. This creates a soft layer to reduce stress on the fragile components and PCB bonds.
I don't pot my GPSs. I've only had one fail but that was due to (I'm sure) a drop of sweat that happened to land on the battery charger IC when I was assembling it, and corroded off one of the pins after a year or so. There was no other sign of water in the case.
I use a magnetic contact with a USB lead from aliexpress that I made a cutout for in the case and glued in, along with an external cheap battery charger PCB with blue/red status leds that I can see clearly through the PLA case. Its been really reliable and convenient, and charges very fast. It doesnt even get warm.
We (Simon and me) have walked the same path.
we now put a flexibel polymer at the pcb. See photos. Works fine. Now we get battery problems after a while with some devices. yes we uses fhebsame chargers (with red and blue led). Made a special case for it.
see photos. Oh and we use normal epoxy for the inside and ultraclear uv resistent epoxy for the screen and outside (in 2 steps). That makes it total water and crash proof.
The charger
The backside
There is an undervoltage protection circuit that comes inside the cell pouch to "end the life" of the battery and make it non-rechargeable if the open cell voltage drops too low (= damaged chemistry and significantly increased risk of fire when recharging).
It will keep working until the cell voltage drops to ~2.4 or 2.5V, then it shuts off, and doesn't turn on until the open cell voltage floats back to ~3.0V.
Normally the application presents enough load current (10s of mA or more peak) to cause the terminal voltage to drop to 2.5V and let the protection circuit kick in, due to the internal resistance of the battery going up significantly when it is "flat".
Then the cell voltage under no load floats back up to >3V. You notice this if you flatten an appliance and hit the power switch again it goes for a bit more... Dont keep doing this or it will kill the battery.
Our standby current is too low for this to work in the long term. Theres no button that says "hey I'm not working so I'm not going to keep trying". The standby current keeps getting sucked out until the cell voltage becomes too low for the protection chip hysteresis to save it. It never floats back high enough to turn on again.
You might get lucky warming it up...
That means you can't recharge it unless you bypass the protection circuit.
The cell is usually OK if its not too far below 3V- I've saved a few in other applications by bootstrapping the protection circuit to get it charging again. But if you leave it flat for too long the damage is done and at best the capacity will be severely reduced. Usually the cell will swell at this point- looks like a goon pillow.
Some smart batteries detect the cell has shut down from overdischarge and has a small microcontroller that detects this and writes information to EEPROM to prevent the user from bypassing the protection. DJI batteries have this- I've saved a few using an arduino tool which resets the firmware in the protection microcontroller and makes it go again... cool hack (not my work).
Ideally we need an additional undervoltage cutout- set to say, 3.5V. There is little benefit discharging the battery further than this- at 3.5V with <50mA load its already basically flat. This would turn off the GPS circuit fully (no standby current being sucked out), and allow the thing to sit there for longer periods. The charger still can charge the cell, since the cell protection hasn't activated yet.
All doable but where does it stop? Cobbling more bits together... just use a bigger cell and keep an eye on it.
