Another DIY GPS logger approach

Until now I have been charging through USB only, mainly though the PC since I was constantly trying changes to the software.
I bought 2 types of usb connectors on my qi receivers. Type A is the correct one, where the cable goes to the back side of the T5. The type B goes to the display an needs correction. It comes in a flexible protective cover, but you can peel it off nicely and get to the bare components. The ferrite tape is causing it either to shield it from charging or enhances it. As mentioned in the instructions of Jan you can use the ferrite tape to stick it to your case.
Since I have also a wrong type B connector, I soldered 2 wires directly on the (flexible) pcb, since it came with 2 extra pads so I can leave the USB connector in tackt for now. Measuring the voltage over the wire it was a whopping 7V and looked a bit steep for me. This was however without any load, not connected to the T5 board. Next thing is soldering the wires to the charge controler pin and gnd of the T5 board. Hopefully it is within specs and don't blow-up anything...

No need to add a non scratch coating. Can purchase scratch resistant Polycarbonate (Lexan MR10 and Makrolon AR2 I think). I was actually using this for all my prototypes in the beginning. It works well as I purchased some for another project that has been seriously abused for 8+ years now and holding up well. I ran out of offcuts so scrounged a free sheet of standard polycarbonate. Despite my current unit being heavily scratched this has come with heavy use and you don't notice on the water.

There should be piles of offcuts of the abrasion resistant stuff around as most of the Covid shields used at over the counter sales are probably made of the stuff. Will start scrounging for some.

What about buying some clearance phone protector non-glass) and cutting to size? I had one on a GW60 and it did the job.

I came to the same conclusion, I'm not worried about the height cut out, but the speed cut out is the killer.
Sorry can't remember what I settled on, "portable" does ring a bell though.

I fully understand why fixing 1 variable like height to all the unknowns would lead to better accuracy of position. I never knew however that's why they had a land and sea setting... I'm not sure if this is also true for doppler speed. If anyone can explain this to me it would be appreciated.

I looked into adding a shield beneath the gps antenna to improve the signal. When I looked into the size, the recommendedradius should be at least a quarter of the wavelength of the radio waves the antenna, to enhance the signal on the antenna side by reflection.

To block possible signals that are reflected from the water, I expect it's
just line of sight. The larger the shield, the more it is blocking signals coming from behind. I think this is the main concern for us right? So anything that extends the PCB of the BN2xx would help, preferably several times larger than the height. 1-2cm could already be significant. I bought some copper-conducting tape do do some experiments. It will probably only work/improve if the gps is horizontal to the water.

Last thing I'm struggling with are the maximum allowed acceleration criteria that remains the same. I think there is a dependency to the sample frequency. I have read in this thread that some people believe it is the chop related. But then it should be gone when traveling by car right? And it doesn't...
Having longer integration time on some measurements does increase the accuracy of the measurements, but only to a certain point. Having higher sample rates increases the amount of samples and therefore also reduces error. However the maximum acceleration is per measured point and not per distance or avg... That's why I believe we can't make use of the benefits of higher sample rates (>5Hz). I have seen individual points in GPSresult that exceeded the max of 5m/s2 allowed for a 10Hz sample. Going back to 5Hz solves it, but I'm not convinced it is less accurate. Yes, the max speed on individual sample should not be used but a 1s or 2s would still be pretty good. Perhaps some equations could prove my theory or what I see happening (1Hz =>5m/s2; 10Hz=>5?SQRT(10)m/s2). But for now the restrictions on 5m/s2 per sample does not sound realistic for higher sample rates and forces me to go back to 5Hz. Wonder how the 25Hz devices are performing on this.

Thanks Jan, so you do see the chop visible in the data between the mounts on the boom vs. helmet. That means the data is more accurate than I thought and makes the need for the maximum acceleration (per data point) need to be increased for units with higher sample rate. Alternatively it could still use an average over 1 second max acceleration equal to the current 5m/s2 and not simply filter-out the higher readings. It's not noise nor invalid. I have seen 10Hz runs being labeled as invalid by gps-speedsurfing, where it shouldn't.

This video (start at 2min) is a good example of how much a device shakes even at Luderitz:

The Motion on Bjorn's arm is a bit too loose. In my experience, if it's just done up a little firmer, it does not shake around as much as that, and does not slip around the arm. But I don't wear a slippery rashie over my wetsuit either.

I am hoping the brains trust and especially any materials engineering experts out there might be able to help me with finishing the box for my ESP32 GPS.

There hasn't been any wind lately so what do windsurfers do when there is no wind? They make GPS units to measure their speeds when there is wind.
I have reported on my adventures in building the GPS earlier.

Flex has reported on his adventures with his 3D printed boxes, which was very educational. I have had quite a rollercoaster of experiences with mine also, and will document them here so you will not make the same mistakes.

Flex 3D printed some boxes for me.
He told me that when filling his units with potting epoxy the epoxy leaked out through the very porous 3D print, so I decided to seal my boxes with some Botecote epoxy that I had. I painted both the inside and outside of the boxes. First mistake - Flex has so accurately designed his 3 D prints that there very little tolerance, and the epoxy coating prevented me being able to insert the GPS module into the recess. I had to dremel away some of the epoxy to fit the GPS module in, and also to keep the switch recess clear.
lesson 1: only seal the outside of the box, not the inside.
lesson 2. The epoxy does not like to coat the 3D print in a thin layer - it beads up. It has a minimum meniscus thickness, like molten glass, so if you want a smooth finish you have to keep topping up the dry spots until the epoxy remains uniform.
The epoxy coating did stop the potting epoxy from leaking out, so a win there.

I decided that a surround for the screen would make for a more professional looking result, and serendipitously discovered that the black surround prevents ghosting of the screen in bright sunlight. The surround I cut from a folder cover is only about a half mm thick, but that was thick enough to prevent the polycarbonate cover from sitting flush with the side of the box. Remember I said these boxes are very accurately designed. Flex has now designed one with a 3D printed surround and the extra depth to accept that, but I did not want to trouble him to print another one for me so I thought I would make do with what I had. That was mistake #2.

Flex used dichloromethane to glue the polycarbonate top onto his 3D printed box. Dichloromethane is recommended for joining polycarbonate to 3D printer plastic. It requires a close fit of the parts to be joined, but I had a 0.5mm gap, and needed a gap filling adhesive, so I thought I would use more epoxy. First I epoxied the surround to the polycarbonate to eliminate bubbles, then epoxied that to the box. That worked beautifully with the first box, with the second unit all was going well until the epoxy was almost cured and then the GPS unit failed to turn off when tested - I had to pull the sticky top of the box to try to rescue the also sticky GPS, but that is another story...

Next step with unit one, all nicely sealed and looking pretty. Fill with potting epoxy. This step went very smoothly, eliminating the gap between the polycarbonate and the display, and the unit looked very cool, if I do say so myself. Pic 1.
As there was no leakage of the potting epoxy through the sealed box there was no need to keep topping up the level of potting epoxy, although there was a tiny bit of shrinkage overnight or loss of volume from bubbles rising, that did require a few drops of topping up. The potting epoxy required a 3 day curing time. I allowed three and a half, then it was time for the last step (I thought) - dremelling the excess polycarbonate from around the edge of the box.

While dremelling the excess polycarbonate I noted some separation forming between the polycarbonate and the epoxy over the black surround. I thought this might be from deformation from the vice, so I took the box out of the vice and held it by hand while continuing to remove the excess polycarbonate with the Dremel. To my horror the separation continued around the edge of the polycarbonate in an unsightly way. Pic 2. Thinking this may reflect poor adhesion between the polycarbonate and the epoxy I tried pulling off the glued cover off another unit and it flipped off readily. Clearly polycarbonate does not adhere to epoxy. so, I decided to pull off the cover with the damaged adhesion, but whilst the polycarbonate separated readily from the botecote epoxy, it did not separate well from the potting epoxy. Not wanting to damage the screen or the visibility of the screen, I stopped at that point.

Now I had a mostly separated cover, with air in between the cover and the epoxy over the surround (which is an unknown plastic), except in a few places, which again, looked messy (pic 3) . What to do?

Flex suggested masking off the body of the box and trying to inject potting epoxy into the gap since the potting epoxy had better adhesion and was quite thin. No gap existed visibly, and there was nothing that you could get even a fine needle into, so for a couple of hours I dribbled potting epoxy onto the edge and hoped capillarity would draw the potting epoxy in and let the bubbles out. The epoxy seeped in, very slowly, but in an irregular pattern which sealed the border and trapped air within. I even tried making the epoxy thinner with acetone, but no luck. The masking tape was a total failure also and the whole box was a wet sticky mess.

Maybe some heat application would expand the air pockets and force air out? Applying heat with a hairdryer increased separation of the new potting epoxy from the polycarbonate cover and made everything look really messy, BUT it also separated the cured potting epoxy from the polycarbonate, and I was able to pull the cover off, clean up the sticky mess and I was left with a lovely shiny smooth surface all one level. Hallelujah! Pic 4.

The quandary now is how to fix the polycarbonate cover onto this unit. The front of this unit is now covered in epoxy (botecote over the surround and potting epoxy over the display). The only glue that adheres to both polycarbonate and epoxy is methylmethacrylate - which is a coloured 2 part paste with a short working time. Medical methylmethacrylate dries clear but is unavailable at present, also has a short working time and may be difficult to eliminate bubbles.

My proposal is to apply potting epoxy to the surface, carefully apply the polycarbonate cover, squeeze out bubbles or place in a vertical orientation so bubbles rise up and out, allow to cure for 3 days. Apply botecote over the sides and up onto the sides of the polycarbonate cover which is roughened by the dremel and will have improved adhesion over smooth polycarbonate. Will probably need to do this one side at a time to reduce runs. Any better ideas?

pics are not in correct order. Pic 2, 3 4, then 1.

Any ideas on how to use the built-in buttons in the GoPro case instead of magnets?

Told you, wildy inaccurate.

Thanks for clarification

Two people have now access to the source code : Freezer and myself. The reason that the sw is not completely open source is because it is now released by GP3S. There is a certain trust that the generated GPS-data is not manipulated by the sw. The same for a possible release (on sample basis) by GPSTC.
As far as I understood, they want to see GPS tracks generated by the ESP_GPS, and a other "released" GPS-logger, as the Motion.
I encourage everyone who wants to make a ESP-GPS for friends or surf mates, as I understand that most people will not take the risk to assemble and solder the parts, as this is way out of their "comfort zone".
Greetings, Jan.