Select to expand quote

PacoRaapNL said..
Thanks for all the testing Jan and Chris and all others...
A magnetic latching switch.
You cant fix one thing where an other problem pops up.... Life is a compromise from start to finish.
Paco

When you complete cut the power from the T5 board, the charge controller is also cut from the lipo. So, charging the battery is not possible anymore. Another issue is closing and saving files, so you need a signal before the power down.
Another possibility is a "High side switch", this is like a old fashioned "start" and "stop" circuit with a latching relay. You start the switch with a momentary switch, the T5 is powered and a GPIO set on high takes over. The same switch can then start the power down cycle. But there is still the charging problem.....
I am now looking to reduce the power @ boot, and return to sleep when voltage is low.
@boardsurfer : The ESP boots always @ 240 MHz. We could directly lower the CPU freq @ boot, and only switch to the normal freq when bat voltage is on a minimum level. For Wifi, we need at least 120 MHz. The normal current@240 MHz (no Wifi) is indeed 40 mA. With Wifi on, peaks over 200 mA are possible.
More reading stuff about SD cards : thecavepearlproject.org/2017/05/21/switching-off-sd-cards-for-low-power-data-logging/
Greetings, Jan.

Select to expand quote

PacoRaapNL said..
Thanks for all the testing Jan and Chris and all others...

As this is a non profit DIY solution we cant expect professional solutions although with thinkering together we can come a long way.

@ BigBoss 213BN with USB Micro

Now for all the Gopro casing users the battery issue is no problem as you can easily switch off the system from the battery by opening the case. Repairing is also a no brainer.

For systems poured or closed it is a different thing.
During my 45 minutes drive from work to home I thought how can we use a switch to switch off the battery from the T5 system but waterproof.
Now when I started to build the first unit I used the reed switch as everyone did to make the system wake up or go to sleep.
That is a momentary switch action to pulse.
I now use a simple press button under the PVC cover plate and I do not need a magnet.
But what if the case still has a reedswitch inside, but this switch is placed inbetween the battery positive wire.
A magnetic latching switch.
Make at the outside of the case a part that can hold a small clickable magnet while the unit is in use or need to be charged.
I do not know if the reed contact can handle the shaking and vibrations the system is under going during use.
Also the saturation of the reed switch from the magnet when removed and not open up needs to be tested.
You cant fix one thing where an other problem pops up.... Life is a compromise from start to finish.
Will try next week if this is usable in my closed designs.

Lets keep thinkering.

Paco

Nice thinking Paco.
omdenken. so not use a magnet to Connect the positive wire, but use a magnet to disconnect the positive wire. Thats the other variant of a reedswitch

so in normal situation the battery is connected. With a magnet you can disconnect the battery.

Select to expand quote

BigBoss said..

PacoRaapNL said..
Thanks for all the testing Jan and Chris and all others...

As this is a non profit DIY solution we cant expect professional solutions although with thinkering together we can come a long way.

@ BigBoss 213BN with USB Micro

Now for all the Gopro casing users the battery issue is no problem as you can easily switch off the system from the battery by opening the case. Repairing is also a no brainer.

For systems poured or closed it is a different thing.
During my 45 minutes drive from work to home I thought how can we use a switch to switch off the battery from the T5 system but waterproof.
Now when I started to build the first unit I used the reed switch as everyone did to make the system wake up or go to sleep.
That is a momentary switch action to pulse.
I now use a simple press button under the PVC cover plate and I do not need a magnet.
But what if the case still has a reedswitch inside, but this switch is placed inbetween the battery positive wire.
A magnetic latching switch.
Make at the outside of the case a part that can hold a small clickable magnet while the unit is in use or need to be charged.
I do not know if the reed contact can handle the shaking and vibrations the system is under going during use.
Also the saturation of the reed switch from the magnet when removed and not open up needs to be tested.
You cant fix one thing where an other problem pops up.... Life is a compromise from start to finish.
Will try next week if this is usable in my closed designs.

Lets keep thinkering.

Paco

Nice thinking Paco.
omdenken. so not use a magnet to Connect the positive wire, but use a magnet to disconnect the positive wire. Thats the other variant of a reedswitch

so in normal situation the battery is connected. With a magnet you can disconnect the battery.

Hmm wont work either. If you want to turn the esp on, both reedswitches wil react hmm.

Select to expand quote

BigBoss said..
Hmm wont work either. If you want to turn the esp on, both reedswitches wil react hmm.

Place the reed switches at opposite sides of the housing - far enough apart so a magnet affects only one at a time. With a cube relation between distance and power, this should work. The question about reliability when shaken remains, but maybe putting a capacitor somewhere in line would help to smooth out brief disconnections. I'm not an EE so can't help you with the details, but think this should be doable.

Select to expand quote

rp6conrad said..
When you complete cut the power from the T5 board, the charge controller is also cut from the lipo. So, charging the battery is not possible anymore.

This does not seem to be totally correct. When I switch the power off using the slide switch on the board, this does indeed seem to be a hard off switch - I cannot measure any more current flowing from the battery. However, charging the battery when the power is switched off this way works perfectly fine. I do this all the time, and always use a USB meter to see how much current is flowing.

I got curious and measured the deep sleep current for 4 of my loggers. For 3 of them, the deep sleep current ranged from 1 - 2 mA (0.99, 1.13, and 2.04). Removing the SD card while the units were in deep sleep reduced current by about 0.1 mA, so the SD card is not the major power consumer in deep sleep (using 16-32 GB cards from 2 different brands).

On one unit, the deep sleep current was a lot higher: 16 mA. This unit (more than 2 years old and with probably 100+ sessions) also had the oldest firmware version (5.74). I don't think that the problem, though, since I had noticed that this unit always ran the battery down faster when I forgot to use the sliding off switch. It also had some visible corrosion on the battery terminal. I'll have to clean this up a bit and see if this changes anything. All tests were done with the same battery (at 3.8 V).

There's definitely quite a bit of variation in the sleep current. Even excluding the last unit, the difference is about 3-fold between what Jan reported and what I measured.

While playing around, I also managed to get one of the units in a boot loop. I had forgotten to switch the power meter back to a higher setting, so it was still at 20 mA. The logger then complained about the voltage being too low, went too sleep, and woke right back up, only to repeat everything. Could be that the voltage on the wake-up win went too low?

Select to expand quote

boardsurfr said..
Great explanations, Chris. The batteries we use are different sizes and from different manufacturers. So what you say is that the very low deep sleep current basically prevents the protection circuit from working as intended, which allows the batteries to discharge too deeply, to a point where they cannot be recharged? Did I get that right?

yep. Due to the low sleep current, the open cell voltage is allowed to discharge below the reset threshold voltage of the protection circuit reset (~3V). Normally with higher load current, when the protection circuit activates there is >0.5V drop through cell resistance when its flat, and the protection circuit trips/recovers/trips/recovers etc... until you kill the battery completely or stop trying to use it.

20mA seems way too high for any SD card sitting just idly... are you sure its not in a bootloop?

Select to expand quote

rp6conrad said..

PacoRaapNL said..
Thanks for all the testing Jan and Chris and all others...
A magnetic latching switch.
You cant fix one thing where an other problem pops up.... Life is a compromise from start to finish.
Paco

When you complete cut the power from the T5 board, the charge controller is also cut from the lipo. So, charging the battery is not possible anymore. Another issue is closing and saving files, so you need a signal before the power down.
Another possibility is a "High side switch", this is like a old fashioned "start" and "stop" circuit with a latching relay. You start the switch with a momentary switch, the T5 is powered and a GPIO set on high takes over. The same switch can then start the power down cycle. But there is still the charging problem.....
I am now looking to reduce the power @ boot, and return to sleep when voltage is low.
@boardsurfer : The ESP boots always @ 240 MHz. We could directly lower the CPU freq @ boot, and only switch to the normal freq when bat voltage is on a minimum level. For Wifi, we need at least 120 MHz. The normal current@240 MHz (no Wifi) is indeed 40 mA. With Wifi on, peaks over 200 mA are possible.
More reading stuff about SD cards : thecavepearlproject.org/2017/05/21/switching-off-sd-cards-for-low-power-data-logging/
Greetings, Jan.

Sorry Jan you not fully understand my thinking.

But anyway.
I do not know if NORMALLT CLOSED reed switches exits in small formats like Bigboss mentiod.
If they do that would do the trick too.
In that case when the reed switch is not activated, the unit will work as it does now with the normally closed reedswitch.
Capacitor needed, who knows but a 0.47F - 5/5volt is easy to obtain.

You can choose the reed switch for inbetween the battery or you solder the wires on the T5 at the solderpoints that are next to the ON/OFF switch and put the switch to OFF.

When you finished using the unit you put a magnet on the NORMALLY CLOSED ON/OFF reedswitch.
This will prevent the battery to be drained as the battery is disconnected or the T5 is switched OFF.
So the second reed switch is used as storage option if needed to stop the draining of the battery when long time not in use.
More then 3 months....!

Does this make sense Jan?

BTW I now do not use a reed switch in my designs but a normal press button.So for me that is different to those who already use a reeds switch.

Paco

Select to expand quote

slowboat said..
20mA seems way too high for any SD card sitting just idly... are you sure its not in a bootloop?

I did not think it was, but may be the case. When I did see a device in a bootloop, I also saw larger currents (> 50 mA) and current fluctuations, and then the screen displayed "voltage too low, going to sleep" before starting over. But a shorter bootloop that never gets to the part where the screen updates could be the reason for the high current.

There are a few things that speak against the problem being a boot loop. On is that the "sleep" current is very constant, without any fluctuations (but that's with a meter, I don't have a scope). 16 mA is also a bit low for a running ESP. Also, the logger still reacts to the wake up button, and then boots normally.

One peculiarity about this logger is that it goes into a boot loop if I use my amp meter set to 200 mA - none of the 4 other units do that, they all boot up fine. This is also the only logger with visible corrosion on the battery terminal.

So these reed switches exist in NC Normally Closed versions.

Specs say 0.5 A switching current.Will order some for Proof of concept.

Paco

There are a few threads where people complain about the deep-sleep current draw. They also indicate how they solved the issues:
hackaday.com/2021/03/02/wireless-low-power-e-ink-weather-gadget/#comment-6327714
www.reddit.com/r/esp32/comments/idinjr/36ma_deep_sleep_in_an_eink_ttgo_t5_v23/?rdt=40643
There are quite some versions of the 213 boards around, BN, B74 with/without usb controler and/or leds. While some things have to be handled within the firmware, some boards might have additional problems.
I cannot recall that we had these issues 2-3 years ago, but the b74 without leds are less easy to purchase. The newer ones might have the issue and need the leds to be removed... Other HW modifications like pull-up resitors could help as well. It might not be a single recipe solution with all the board versions around.

Select to expand quote

Freezer said..
I cannot recall that we had these issues 2-3 years ago

You would not notice any issues for quite a while. With a 2000 mAh battery and 1 mA deep sleep current, the battery would remain charged for more than 2 months. There seems to be some variation between units, so there's probably a range of 1 month to 4 months. So the earliest you'd possibly see any issues would be the second season you use a unit. Realistically, though, you probably need multiple breaks where the units are not used for many months. The batteries typically have over discharge protection, so just letting them run the battery down once is unlikely to cause damage (although it is possible that some batteries don't allow charging anymore after triggering the protection, or that the protection does not kick in, as slowboat has explained). Typically, you'll probably need a few complete discharges, and extra storage time after that, before problems become noticeable. It's a bit like the problems Julien encountered with the original Motions, where the screens were great for a season or two.

The simplest solution would be an on-off switch that completely disconnects power. The normal-on reed switches are an interesting idea, but even if they work well, it's an open question if they hold up over years of use. That's also true for other kinds of waterproof on-off switches, though.

One possible solution for batteries dying from deep discharge when loggers are not used for months would be to put the logger on a charger that's connected to a wifi-controlled power outlet, and have the outlet turn on for half an hour or an hour once a week. That's a cheap and easy solution, but you'd need a charger for every outlet. Personally, I'd be a bit skeptical about unsupervised wireless charging, and would prefer a cable connector like the one slowboat mentioned (www.aliexpress.com/item/4001283805395.html), but maybe I'm old fashioned.

Select to expand quote

PacoRaapNL said..

rp6conrad said..

PacoRaapNL said..
Thanks for all the testing Jan and Chris and all others...
A magnetic latching switch.
You cant fix one thing where an other problem pops up.... Life is a compromise from start to finish.
Paco

When you complete cut the power from the T5 board, the charge controller is also cut from the lipo. So, charging the battery is not possible anymore. Another issue is closing and saving files, so you need a signal before the power down.
Another possibility is a "High side switch", this is like a old fashioned "start" and "stop" circuit with a latching relay. You start the switch with a momentary switch, the T5 is powered and a GPIO set on high takes over. The same switch can then start the power down cycle. But there is still the charging problem.....
I am now looking to reduce the power @ boot, and return to sleep when voltage is low.
@boardsurfer : The ESP boots always @ 240 MHz. We could directly lower the CPU freq @ boot, and only switch to the normal freq when bat voltage is on a minimum level. For Wifi, we need at least 120 MHz. The normal current@240 MHz (no Wifi) is indeed 40 mA. With Wifi on, peaks over 200 mA are possible.
More reading stuff about SD cards : thecavepearlproject.org/2017/05/21/switching-off-sd-cards-for-low-power-data-logging/
Greetings, Jan.

Sorry Jan you not fully understand my thinking.

But anyway.
I do not know if NORMALLT CLOSED reed switches exits in small formats like Bigboss mentiod.
If they do that would do the trick too.
In that case when the reed switch is not activated, the unit will work as it does now with the normally closed reedswitch.
Capacitor needed, who knows but a 0.47F - 5/5volt is easy to obtain.

You can choose the reed switch for inbetween the battery or you solder the wires on the T5 at the solderpoints that are next to the ON/OFF switch and put the switch to OFF.

When you finished using the unit you put a magnet on the NORMALLY CLOSED ON/OFF reedswitch.
This will prevent the battery to be drained as the battery is disconnected or the T5 is switched OFF.
So the second reed switch is used as storage option if needed to stop the draining of the battery when long time not in use.
More then 3 months....!

Does this make sense Jan?

BTW I now do not use a reed switch in my designs but a normal press button.So for me that is different to those who already use a reeds switch.

Paco

Yes that was exectly what i meant.

And indeed the reedswitch you found too. I use now the other version (normaly open) to switch on the esp

Select to expand quote

veton said..
About the top values they are indeed small, but I was more pointing to the fact that the stats screen is a bit poor, would be great to have a second one with 100m/250m/500m/top . I mean while sailing having the speed in big is enough, but once you stop at the shore or after a run is very useful to check the scores to adjust gear or just see how its going.
In Spain/France we use a lot those fixed distances more than speed over time (10 sec and so on). Just a suggestion for the future, to toggle to other screens to see some scores other than 2sec and alfa.
Thanks for the help

New SW update SW5.9 is on Github (github.com/RP6conrad/T_display), with next changes :
Added more stat screens
After boot from Power on, direct deepsleep. This to prevent bootloop when lipo goes in to protection.
Can be override when reed switch is on @ boot !!!
If sw-reset, normal boot
Adapt FIR filter bat reading 0.1 -> 0.02 (more stable bat reading)
Added GPSTC posting in webserver

Greetings, Jan.

Wonderfull Jan,Just a question about the RUN screens.
Is there a (RUN) screen setup that shows the actual current speed (LARGE) and current max speed (over 2 sec) for that tack and max speed (over 2 sec) for the current session in the topbar?
The average speed is a nice value but more for after the session and looking at home for the overall stats.

Paco

Speed_screens are now like the T5 e-paper : Upper row is configurable. I noticed there is a bug in the SW5.90 : memory overflow. The webserver configuration does not work anymore. SW5.91 will fix this.

Fantastic job Jan.
Thinking on reliability of the reed switches I've looking around and find these magnetic sensors with no moving parts.
?si=lvv3pwDOMRXC2izW

They are called hall effect sensors, no clue if they are suitable for our projects or they are sensible to voltage or any other drawback.
Regards

Select to expand quote

veton said..
Fantastic job Jan.
Thinking on reliability of the reed switches I've looking around and find these magnetic sensors with no moving parts.

?si=lvv3pwDOMRXC2izW

They are called hall effect sensors, no clue if they are suitable for our projects or they are sensible to voltage or any other drawback.
Regards

Hall sensors are oa used in car engines, reed switches are used in alarm systems, so both are reliable, but this site tell it all:

www.thomasnet.com/insights/how-to-decide-between-a-reed-switch-or-a-hall-switch/

Now there are reed switches in a plastic sealed housing that have not the breaking disadvantage.

This is my design, yet to resin..
Still on trials. I wasn't able to connect the led. Is It on 15+vcc, but red on vcc and black to 15?
Thanks !

Select to expand quote

veton said..





This is my design, yet to resin..
Still on trials. I wasn't able to connect the led. Is It on 15+vcc, but red on vcc and black to 15?
Thanks !

That looks nice.
In the back of the board at the corner there is a blue led that lights up when the unit is charging.
Remove the smd led and solder the wires from the external led to that position.

Before you resin the unit you might wait a while or you must be in a hurry... :-)

In the background lots of stuff is going on and tinkered to prevent unwanted battery draining and enhance battery life when the unit is not in use or stored for a long time.
There are 3 routes to go where the base is to disconnect completely or almost completely the battery from the 0.8 mA drainage when the unit is in sleepmode.1] extra reed contact (normally closed but opens with magnet)
2] highside switch (full electronic)
3] regular waterproof on/off switch

Paco

Select to expand quote

veton said..
Is It on 15+vcc, but red on vcc and black to 15?
Thanks !

Yes, that is correct, red on 3V3 and black on GPIO15 . LED will go on if reed switch is active.

I spotted a diy gps on the last video of the luderitz speed challenge youtube channel.
It seems a bit different from the epaper one. Any info about this one ?

Around 1:28
youtube-com/watch?v=TxPrHk3pS4M&t=89

Select to expand quote

yekokata said..
I spotted a diy gps on the last video of the luderitz speed challenge youtube channel.
It seems a bit different from the epaper one. Any info about this one ?

Around 1:28
youtube-com/watch?v=TxPrHk3pS4M&t=89

Thats a custom 20Hz uBlox based logger build by Manfred Fuchs (www.gps-speed.com/impressum.html ) ,all Luderitz riders use the during the event

Select to expand quote

yekokata said..
I spotted a diy gps on the last video of the luderitz speed challenge youtube channel.
It seems a bit different from the epaper one. Any info about this one ?

Around 1:28
youtube-com/watch?v=TxPrHk3pS4M&t=89

It seems like Brendan Loho has one ESP-GPS by the looks of this screenshot:

Who has built that left one?

Select to expand quote

PacoRaapNL said..
Who has built that left one?

That is lisawindsurfing.com as indicated on the picture.

Anybody tried using magnetic pogo-pins to charge the ESP? I'm just experimenting. Although I could connect all 4 pins to USB to make it connect fully to USB, the connections are too tiny to connect on the PCB. For now I only use 2 for 5V and GND. The magnets on the side are reversed (N/S) such it connects only 1 way. Curious to see how these pins hold against corosion. I'm not sailing in salty water so that helps. Reason to try this is to prevent heat during wireless charging when the airbag gets too big.
I also added a watertight button instead of the reed contact for simplicity and tactile feedback.

Hi all,

I only provived Brendan with a water proof case for Luderitz ;) . This is indeed a ESP GPS here, so not a LISA dev (not yet )

Select to expand quote

LISAWindsurfing said..
Hi all,

I only provived Brendan with a water proof case for Luderitz ;) . This is indeed a ESP GPS here, so not a LISA dev (not yet )

If you start to sell them, you could have a few takers!!!!

Select to expand quote

PacoRaapNL said..
Who has built that left one?

It's and ESP with his sponsors logo superimposed on it.

Very cool, I have a bicycle helmet eith the same connector.