Making – Caving – Diving

15 August 202315 August 2023

Static Laser Scanner

Whilst my cave surveying efforts to date have been mostly focused around grovelling around in small passages trying to use a Disto and/ or Gopro and lights to capture data to create accurate models via photogrammetry my mind was slowly turning on how to survey bigger chambers, laser scanning is the obvious answer.

Cut through of GB’s Main Chamber created from a compilation of scans

In 2016 I used a Geoslam Zebrevo to survey a few local show caves and other sites, this worked very well however the cost and chance of damage to the device in real caving situations is not something I would like to risk. I had heard that some people had attempted to create a DIY version as the SLAM code was apparently available open source, this triggered lots on internet searches on the subject. I wrote off the Cave-a-tron type system as it wouldn’t easily fit into a dry tube to be dived through sumps to survey passages beyond which is something I wanted to be able to do.

Sadly I didn’t find anything I felt I could build within my skills or cost means that could recreate a SLAM type scanner, however I did find some people had successfully built tripod scanners using fairly simple and off the shelf components. What was even better was that second hand laser units they had used were readily available on Ebay at quite a cheap price relative to the cost of a new unit.

The base laser unit itself was a Velodyne VLP16, a small compact unit that I had used professionally on a few occasions, a short time ago these retailed at around £5000 new but are now in the region of £2500. I snapped up two used ones on Ebay for: $400 and $250 (I’ve since damaged the $400 one).

At this stage I should give massive credit to the originators of this idea who’s designs and softwares have allowed me to realise my own version of their scanners. The first is Jason Bula’s:

https://github.com/jason-bula/velodyne_tls

His scanner and Matlab based code is quite rudimentary but paved the way for another person to further refine and inspire my device, Donny Mott’s:

https://github.com/Rotoslider/TLS_Pie

So to reiterate, I have simple re-arranged components they have used into a form factor which suits my needs and have used their software /code to process the data, nothing massively clever on my part. The result is a compact tripod based scanner that is far cheaper than any commercial offering, I can carry all components myself readily into dry passages or through sumps without external help as the scanner fits in a small pelicase or a small dry tube. I had recently signed up to CREG journal and partway through this process saw an article showing how a Cave-a-tron had been converted to be used on a tripod like the device I was building at the time, however the laser is very primitive compared to a Velodyne unit.

It took quite a few months of trial and error with various components to get data of an accuracy that I was happy with, this was mainly due to the use of a stepper motor with 50:1 ration planetary gearbox. It didn’t quite rotate at a consistent speed meaning that when overlaying two scans from the same location features were not in the same place as they should have been. I replaced it with a different stepper motor with a 30: 1 harmonic drive and finally was happy with the results. Many other things were tried or swapped in this time before the gearbox was identified as the cause such as power supplies, wiring looms and Arduinos so it wasn’t as straight forward as it sounds.

The main components are:

Velodyne VLP16: Laser scanner mounter vertically on a slowly rotating frame

Raspery Pi 4: Logging of Ethernet data from VLP16 using code written by Donny

Arduino: 3 x push button controls of the stepper motor and logging commands to the Pi4

Stepper motor and Harmonic drive: Provides the smooth slow rotation

Lipo battery: Power for all components

Various other DC-DC converters, switches, wires etc necessary to interface the components.

The scanner is mounted to a tribrach bolted onto a tripod and leveled, either button can be pressed to initiated a scan. The scanner then starts to slowly rotate and the Raspberry Pi4 logs the data to a file, no results are visible in realtime. Once the scan has finished the tripod is moved to the next location, leveled again and another scan initiated. In this manner progress through cave passage can be made, usually 7-10 m at a time between scans but this depends on the nature of the passage. Once back at home the raw scans are turned into point cloud files by Donny’s excellent piece of software and can be aligned together using cloud compare’s manual tools initially then its fine align tools. Below is a video taken by Duncan Price of the scanner in use in Wookey Hole.

Time lapse of use in Wookey Hole, video by D Price

Scans have so far been undertake in:

Badger Hole

Wookey Hole Chamber 20/ passages beyond

Wookey Hole Chamber 22

GB cavern (all of the large main passages and Great Chamber)

Numerous tests in my garden

Interested in how accurate my scanner and methods were I surveyed a close loop around a house, the loop was 76 m in length and contained 9 scanner locations, the misclosure was less than 20cm in XYZ between the same point visible in both the first and last scans.

This I believe to be far more accurate than what can be achieved with a Disto in a cave though it lacks alignment to either magnetic or true north so its accuracy is only in a relative sense. Alignment to a disto based centre survey could be used matching up common reference points in Cloudcompare to align the laser scanner data.

A commercial laser scanner would achieve accuracy far greater than this but the cost would be 15-100x more along with a large sense of paranoia with regard to damaging the unit in the cave environment. The Velodyne laser doesn’t log point colour but is does log intensity so this can be interesting to colour the resulting point clouds by.

The below video shows the data data I collected and aligned from GB cavern, Donny made the fly through after I sent him the data to show him what I had been doing with his ideas and softwares, it starts at the mud run in at the top of the major passage and goes down to the choke just below ladder dig.

At this point I feel I have succeeded in my goals, I have a relatively cheap laser scanner than I can take nearly anywhere (whether I want to is another matter, eg transporting through Daren Entrance crawl..) that produces results accurate to a few centimeters and can scan large chambers or passages with ease, now it is a case of working through suitable sites and producing laser scans of places that otherwise might never be scanned unless someone invested a lot of time or man power or money into carrying in commercially available scanner.

2 May 2023

Ravens Well Update

Well its been a little longer than I anticipated updating this page with what I have been working on but rest assured this isn’t due to a lack of activity more a lack of internet at home.

Numerous projects have been going on both on the surface and underground but I thought I would start with some updated data from Ravens Well. For Christmas I treated myself to a small drone (the DJI Mini2) mainly so I could add some surface features to the underground surveys I have been working on though flying it around aimlessly is equally fun !

The learning curve I found to be very easy and by lunch on Christmas day I had produced a model of my house. After a bit more practice and experimentation I was ready to capture the images required to create a surface model of the area above Ravens Well around the three lamps junction of Bristol, ever curious each time I walk/ cycle/ drive over the top as to where the tunnels are exactly beneath this would provide the answer more accurately than just overlaying the underground model/ survey roughly in Google Earth as the Axbridge Caving Group have done.

Ravens well and surface model of the three lamps junction

The flight took about 20 minutes well within the drones ability and Metashape processed the images very efficiently. I used RTK GPS to measure the positions of some prominent features on the ground mainly road marking which were clear and well defined in the point cloud. These were then matched up in CloudCompare resulting in a reasonably accurate surface model of the area. I once again scrambled down to the entrance to the tunnels and to my surprise found that I had fixed RTK status next to the entrance so I marked a temporary point and quickly surveyed from the GPS point to a point marked inside so that the drone data and underground data could be tied to the same reference and overlaid.

Surface point being measured with RTK GPS

Same point in model with tunnels visible underneath

Overall a successful addition to projects, easy to do and gives a better understanding of the relationship between above and underground features.

Three lamps Junction area with tunnel underneath (long shadows from surface features caused by low winter sun)

8 November 2022

Laser FAIL !

Just a quick post to say that the underwater Disto didn’t work quite as intended, I made a number of trips to Sump 9 in Porth Yr Ogof this summer with the aim of surveying it accurately. The water is usually crystal clear on entry to this sump however the laser was not able to read the short distances I had hoped it would. For shots that should have been 1 to 2 m in length it was just returning valves of around 0.15 m every time, the bearings will be useful but I need to return and measure the lengths traditionally.

11 August 2022

Underwater DistoX2 Housing

I will start by saying that this idea has been copied from other cavers/ divers, this isn’t the the first time someone has wanted to use a Distox2 underwater. My reason for wanting a waterproof Distox2 is allow me to use the same methodology for underwater photogrammetry geo-referencing that I use for above water photogrammetry geo-referencing to preserve accuracy through sumps. Lasers do have limitations underwater, light is attenuated by water quite readily especially red light or red lasers. so the range is going to be very limited even in very clear waters.

The main body of the housing is made from Acetal, a small cut out sealed with an o ring allows the laser to emit and receive through the front of the box and is covered by 5 mm thick perspex. The lid is 12 mm thick polycarbonate and houses two brass buttons sealed with o rings to actuate the on/ off buttons. The springs have been taken from a Gopro housing and are the most ferrous part, making the bearing swing by 0.2° when they are swept close to the Distox2, all screws are brass. A standoff is fitted to the rear to extend the rear reference point to make it easier to align to the survey station. The gland sticking out the side doesn’t have a function other than to seal the hole present in the side ( the boxes previous life was a waterproof box for an Arduino Mega).

Due to the difference in refractive index of water and air the distance readings taken underwater are no longer correct when compared to the same measurements taken in air. The refractive index of air is around 1.00 and approximately 1.33 in the water I would be using the device in.

To prove this I made a test in my bath, I put two pencil marks at either ends of my bath, with the Distox2 in the housing I made a measurement with the marks and device underwater then again with the water drained, I also measured with a tape measure for a sanity check.

The tape read 1.415 m, the dry Distox2 in the box read 1.31 m and the wet shot read 1.70 m.

The distance from the rear of the disto to the reference extension stick needs to be added which is an extra distance of 0.109 m, this added to 1.31 m gives 1.419 m which is very close to the taped measurement.

Converting the underwater shots distance requires the offset from the front of the device to the rear to be subtracted first as this is a fixed offset added by the Disto and doesn’t need to be scaled from water to air. There is also around 0.02 m of air in front of the disto before the laser passes through the housing lense and into the water and this doesn’t need to be scaled either, i’m purposefully ignoring the 5 mm of perspex as this is very small compared to the distance of the shots being taken and probably only accounts for a few millimeters of difference.

So starting with the wet shot distance of 1.70 m we need to subtract 0.134 m which is the length of the disto plus the air gap in front. This gives 1.566 m, which we need to scale by the refractive index of water 1.33, which gives 1.177 m. To this we need to add the length of the disto plus air gap and the extension distance so 1.177 + 0.134 + 0.109 + 1.420 m which is very close to the taped distance and the dry measured distance – Result !

Cat testing

The distoX2 manual available here:

Click to access DistoX2_UserManual.pdf

It states that the wavelength of the laser is 635 nm, using this calculator here we can confirm the refractive index of water is around 1.33xx.

https://www.staff.tugraz.at/manfred.kriechbaum/xitami/java/H2Orindex.html

The next phase is some pressure/ wet testing of the housing without the Disto installed in case it leaks then I can begin to put it into use, sump 9 in Porth Yr Ogof is the first place I have in mind for it so that data collected in Parker Series can be accurately aligned to the rest of the cave.

17 June 2022

Broken Aquazepp :(

After a successful two hour dive following Marcus around Vobster whilst he photographed the walls around the 22 m area the afternoons dive didn’t go so smoothly. Feeling a bit scootered out after the mornings dive we jumped back in anyway and whizzed around this time unencumbered by camera equipment . Nearing the end of the dive my scooter started to make strange noises, the normal Aquazepp racket had taken a turn for the worse and didn’t sound healthy at all so I swam it the short distance back to the entrance point, pondering what might be causing it. I initially suspected that the motor might have come loose on its mounts and the gears might not have been meshing optimally.

Once back at home I had a look in the tail cone but found the motor still firmly bolted into place, instead I could see that the large drive gears teeth had partially stripped. No prop jams or other issues had been encountered on the dives so why this happened is a bit of a mystery, old age perhaps ? The drive gear attached to the motor shaft was fine, this is only a few years old.

This happened in February 2022 and writing this is June I still haven’t bought a replacement part, I have been busy with other things and the scooter doesn’t get much use over the summer when the caves are a better proposition for diving than Vobster. I think I might take this as an excuse to rid the scooter of the noisy drive train and go with a direct drive motor, at the moment replacement motors for the CUDA are on sale and are suitable for direct driving a scooter.

https://dive-xtras.com/products/cuda-motor

It seems recently that someone has taken over ownership of what was left of the Aquazepp brand, more information can be found in this video below:

8 December 20219 December 2021

Drill Bag

The humble battery powered SDS drill is largely responsible for the modern exploration of Mendip caves, either through drilling holes for various rock splitting/ removal methods or for faster bolt/ aid climbing where the easy ‘just the follow the open passage’ phase of exploration has long passed. The ability to transport a drill through a sump opens up more opportunities where climbing or passage enlargement may not have been possible before.

A dry tube would be an obvious choice for some but to fit a powerful drill it would have to be big, and thus require a lot of weight to sink and this would complicate the transport element in the dry passage to the actual work site, dry tubes also tend to be quite expensive but on the plus side are usually capable of passing very deep sumps.

A previously well known/ used method is simply to get a large section of inner tube, put the drill inside and methodically roll and fold the ends over, this works in shallow sumps but for me posed too much risk of flooding and requires lots of faffing to seal properly.

Drill, drybag and inner tube with clamps

I chose to create my own version of this, still using a section of inner tube but instead of relying on folding I made two sets of metal clamps which are used to seal each end. This has proved very reliable and is often bone dry inside even after passing multiple sumps up to 20 m depth. the metal clamps are 8mm thick stainless steel closed by M8 bolts and wingnuts, one end is clamped very tightly and never removed and the other end is used to open and close the inner tube, a spanner or bolting hammer is useful to help tighten and loosen the bolts. The drill is packaged in a sturdy drybag in case it does flood.

Drill in drybag ready to go into sleeve, as much air removed as possible

The packaged drill doesn’t require any additional lead to sink, if care is taken and all the air is removed from both the drybag and inner tube sleeve before sealing its about 2 kg negative near the surface but this increases slightly with depth due to further compression. I use this as part of my weighting system, combined with some climbing gear its more than enough to sink me in a wetsuit.

Drybag sealed inside inner tube, as much air is removed as possible

I use a very well worn tackle bag that has half the bottom missing, this makes for easy drainage when standing up to get out of the water fully kitted up, other gear goes in another bag without a massive hole in. The bag is side mounted over the top of diving gear, the drill bag on one side and climbing gear on the other balances out quite nicely underwater.

Inner tube bag inside an old tackle sack, sidemounted for easy transport

Using this method this drill was carried to Chamber 24 in Wookey Hole on numerous trips and was used to forge the dry link from Chamber 20 so that dry cavers can now visit without the need for diving. It occasionally gets a small pin hole sized leak, because of the inner dry bag this doesn’t really matter but its easy enough to patch them with bike repair kit bits. I’m not sure how deep this would work but it has been great in the UK for shallow sumps, it would probably be fine to 30 m, if I intended to take it deeper I would trial it with a block of wood inside to represent the shape of the drill as I did when originally testing it.

10 May 2021

DIY Helmet Mounted Caving/ Diving Light

There are multiple options for cavers who like diving through sumps to further their caving trips to choose from in the lighting department, the likes of Scurion, Phaeton, Rude Nora and other manufacturers have been making suitable lights for years.

This being the making category of this website you can probably see where this is going !

Sometime in 2018 I discovered the LED driver board (the heart of any modern lamp) for the Phaeton was available for purchase from its designer/ manufacturer in the States which is:

https://www.taskled.com/

I duly ordered a couple of boards, some CREE leds cobs and a suitable switch (tricky to track down and a few month wait for stock). Once these had arrived I soldered the bits together then proceeded to procrastinate for a while whilst trying to figure out how to solder and assemble the 20 pence piece sized board inside a small water proof housing.

This went on to the point where I became more interested in other things and forgot about the parts I had, occasionally coming back to the problem but finding no solution.

My whole cave/ sump diving career has involved the use of a hotch-potch of different hand held lights attached to my helmet, great for redundancy but heavy out the water. For caving trips only I would remove these lights and attach a Petzl MYO which is nice and light in comparison.

I’m not one for caving with super bright lights, its nice to have the option to occasionally use full beam to light up distant parts of passages but for the most part i’m happy to cave on dim settings. During the late summer of 2020 the diggers of Wookey 20 (website in links) had broken through into what they have named ‘The Land of Hope and Glory’ in which was an enticing aven was found and to be climbed by myself and the bolt climbing veteran Tom Chapman.

We duly arrived with equipment and the other proceeded to use their highly powered caving lights to light up the roof of the aven some 30 m above. I set my Petzl Myo to full and could barely see anything !

This lack of power is what reignited my want for a powerful light that was waterproof enough for any diving I had planned and brighter than what I had already. Its main use would be for caving the other side of sumps, centrally head mounted lights give lots of backscatter so are of limited use as a primary light underwater in normal British cave diving conditions.

The parts I had obtained in 2018 were dusted off and I had a fresh look at the problem, mainly the soldering and assembly in such a confined space to keep the overall size and weight of the light down. No futher progress was possible with the idea in my head or in CAD modelling. so I decided to just start making a housing and and to try and assemble it and make it up as I went along.

Strangely, once I had roughed out the housing the ideas started flowing and after a few eureka moments I found a way to assemble it so construction started in earnest.

The body is made of Acetal, with a 5mm Polycarbonate front and an Aluminium heat sink for a rear both sealed by o rings. The battery box houses two 18650 batteries and is made of Acetal.

After having assembled a working version in Autumn 2020 I took it for a few trips, although the light remained dry it worked very nicely in Raven’s well, it didn’t overheat which was one of my concerns using a plastic body and metal backing plate and the combined spot and flood lenses gave a nice pool of light to cave with. A more testing trip a week later in Swildon’s Hole involving free diving to sump 6 however proved it to be less than waterproof. The light worked really well on the trip and owing to the conformal coating I had put on the exposed electrical connections it continued to work even when wet inside but on inspection on the surface it had a fair amount of water inside given the shallow and short nature of the sumps.

The possible source of the leak perplexed me for a while, I kept taking it apart checking things then taking it for a dive only to have it consistently leaking. Whilst descending from the surface with it in my hand so that `I could visually see where the water was coming from showed that it was leaking by the switch which has an o ring seal on.

Detail of switch showing threads not cut up to sealing o ring

Taking it apart and drying it again and inspecting the switch showed that the threads didn’t go all the way to the sealing surface on the switch lip, meaning that when it was screwed down onto the body it couldn’t go down enough to actually engage the o ring and the source of my troubles. These switches are meant to be mounted in a panel using a clearnce hole for the threads and a nut the other side instead of being installed into a threaded body like I had done. What I needed to do was cut a small relief diameter to the top of the threads on the light body so that the switch could be screwed all the way down to compress the O ring and make a seal. I tried this on the light I had made already (v2) but messed it up so version 3 was ‘born’ with a funkier cut away shape and the extra clearance for the threads.

Version 3 from the front with lenses removed

Pressure testing the light on dives and fixing it in between had become a rather exhausting process so I decided to fashion a small pressure pot out of some clear pipe, some end caps I had used on a small dry tube previously, an old bike inner tube valve and a bike pump. Using this setup I was able to cautiously test it to 80 m depth in my back garden.

I am happy to report some months later that after a few diving trips it has been working very well, I have since changed the LED’s to those with a warmer colour temperature, I find this easier on the eyes. Caving with the light on the second dimmest setting gives ample light and even with the occasional bursts to full I am getting many hours before I am having to recharge it.

I have it mounted to my original Petzl Spelios helmet (with the Duo removed) and found that due to the foam in the helmet when inadvertently left in a sump pool the whole lot floats which is a nice benefit, no more worries about dropping and loosing a helmet into the murk.

13 September 20203 November 2020

Underwater Survey Device Assessment

One of the previous posts on this website details the device I have assembled in the hope to speed up underwater cave surveying and at the same time make it more accurate than using the traditional divers compass, depth gauge and slate.

Using the Adafruit BN0055 ‘9 DoF IMU’ inside a waterproof housing as the tilt compensated compass should give a reasonable degree of accuracy but just how accurate is it going to be ?

To find out I ran some tests using a DistoX2 for comparison.

A small wooden jig was constructed that allowed easy foresight and backsight alignment of the home built device and the DistoX2 so that comparable shots could be easily collected.

Disto X2 in wooden Jig, Plastic pegs used for alignment

Survey box in wooden jig, axis of BN0055 co-incident with alignment of plastic pegs

The sizing of the recess in the wood is such that when the box is rotated for the foresight/ backsights and pushed up against the right and left hand edges the sensor of the BN0055 is in approximate alignment with the plastic pegs used to align the DistoX2, the BN0055 is mounted around 90° out from the long axis of the box so its raw reported bearings is around 90° different.

Forty comparable foresight and backsight shots were taken with both devices and the data entered into a spreadsheet. The first task was determining the average difference between the DistoX2 data and the box data (I should come up with a decent name for this device…) The average difference between the two was 89.56°.

The Raw data from the box was then corrected by 89.56° and re-compared to the DistoX2 data. Average difference to Distox2 and Standard deviation values were calculated.

The foresight and backsight differences were also calculated to give a quality check for the shots as the jig wasn’t moved until foresight/ backsights were taken with both devices. The DistoX2 foresight/ backsight differences were far smaller than those calculated for my home made device.

These tests were conducted on a flat surface so further tilted tests will be done to assess this aspect, overall I am happy with the results so far, I have been able to buy an off the shelf sensor and without any complicated calibrations or maths have a sensor that is able to report magnetic bearing to within a few degrees of a DistoX2.

Assuming the tilted performance isn’t much worse then any large errors underwater will come from the ferrous metal equipment carried by the diver (or in the sump) and the ability of the diver to align the device with the dive line which is another challenge itself which needs thinking about.

2 September 20204 September 2020

Underwater Cave Survey Device

A commercially available device for underwater cave surveying is available to purchase called the Mnemo, in keeping with traditional cave survey methods it logs distance, depth and bearing of the line used in caves to guide cave divers.

Inspired by this concept I set about designing and making my own version, it is a work in progress and in its current form can log depth (via a pressure sensor), bearing, temperature, pitch and roll of the device (useful for assessing how still the device was during logging, inclination (pitch) combined with depth change can also be used to estimate distance between belays using basic trigonometry).

The line measurement aspect of the Mnemo might be more difficult to implement in British caves as the line diameter varies greatly from cave to cave and sometimes even within the same sump so I have ignored that bit for now until the rest of the measurements are proven to be of reasonable accuracy.

Housed in a waterpoof box I have:

Adafruit Feather M0 SD (control and data logging)

Adafruit DS32231 RTC (timestamping)

Adafruit BN0055 (9 DOF IMU)

Blueorobotics Bar30 (pressure sensor)

Small screen (data display)

IP68 Momentary Piezo switch

18650 Battery

Assorted resistors, capacitors and a power switch

The components are mounted on a custom made isolation routed single sided PCB and hand soldered onto header pins.

The BN0055 IMU was chosen as it does the complicated sensor fusion on the board and outputs a heading, pitch and roll solution (it can also output raw data if required but the maths and programming is beyond me). This is much easier and hopefully more accurate than having to read and compute data from the separate IMU components.

The device is powered on by activating the latching on/off switch accessed by removing a 3/8″ UNF regulator blanking plug from the side (must be done out of water). When the program starts the battery voltage is displayed before showing the calibration status of the three sensors which make up the IMU which are ; a gyro, an accelerometer and a magnetometer. It is important each sensor is calibrated before use but this doesn’t take very long and once calibrated this status is held until the device is powered off. In between survey shots the status of each sensor and the overall system status is displayed on the screen

Once ready the device can be aligned with the dive line next to a belay, the button can be pressed then after a short delay the device writes 10 values at 10Hz to the SD card, it then waits for the next button push. The screen does display the shot data momentarily but as the screen is small this is more for reassurance.

In this manner it could be used to replaced the compass and depth gauge readings taken by a diver, line distance still needs to be measured traditionally and noted.

By automating the bearing and depth measurement and recording aspect of underwater cave surveying I hope to speed up the process and increase the accuracy of the data collected, this should prove useful in resurvey projects of caves which are thought to be close by to other caves.

The device could also be reprogrammed and repurposed as a DPV navigation console, or mounted to a camera and used to provide accurate camera orientation and depth data to improve under water photogrammetry image alignment (inspiration for this idea was taken from https://youtu.be/YKw3lBXX6vM ).

Building this device was the first goal, testing and appraising its accuracy is the second goal (currently ongoing) then if suitable putting it to use in some projects is the third and main goal.

Links to the various parts used are shown below:

https://learn.adafruit.com/adafruit-feather-m0-adalogger

https://www.adafruit.com/product/3028

https://www.adafruit.com/product/2472

26 August 202014 September 2020

Aquazepp LT30

I bought an Aquazepp LT30 (short body, single speed, 12v with headlight) in the summer of 2017. Once a new battery was installed it was fully functional despite being well used and quite old. I was getting around one hours use out of the battery and measured my underwater speed at 33 m per minute, faster than swimming but not that fast for a scooter.

Though this was ‘slow’ around the same time Marcus Blatchford was scootering around Vobster Quay with a camera attached to the nose of his vehicle at similar speeds taking photos to be processed into a model (via the process of photogrammetry) so it was ideally suited to the task of accompanying him for which many a pleasant dive was had cruising through the wintry atmospheric waters of Vobster Quay.

I was already planning on upgrading my batteries to lithium based chemistry and had been searching Hobbyking for suitable packs, some 16 ah packs came up at a good price so I purchased six of theses giving me 96 ah Vs the standard 33 ah plus the benefit of a higher voltage as these were 4 s packs which give 16.8 v when full and 12 v when empty.

I was unsure if the standard 12v motor would be able to handle the extra power but installed the batteries anyway and took the scooter for a dive. A very enjoyable 90 minutes was spent whizzing back and forth to the end of Brixham Breakwater and back, and back and back…until the scooter seemed to loose power, I had assumed the packs were just flat so ended my dive very happy that my modification had improved both the speed and the duration of the scooter. Brixham breakwater is about 900 m in length and it had taken me just under 20 minutes to reach the end so a big improvement in speed over the lead acid batteries had been achieved.

The scooter was taken home and the batteries charged, and I took it for another dive at Vobster Quay expecting to cruise around faster than before, unfortunately once the scooter was in the water the propeller was making a feeble effort to turn much like it had at the end of my Brixham dive so a bit miffed I left it at the surface and went for a swim without it. Once I got home I took it apart and began to check the batteries had actually charged which they had, on opening the rear motor compartment a burnt smell was apparent and the insides were covered with a thin layer of black dust, clearly the motor had burnt out given the extra power that it was being asked to convert from electricity to rotation of the propeller.

A source of motors from the larger Aquazepp models which run on 24 v was provided by a friend and an order quickly placed. The motor was a direct replacement for the 12 v motor, it just needed the drive gear fixing to the shaft which was achieved using a small screw fixed through the motor shaft. The 24 v motor is heavier than the 12 v motor so I had to remove some of the lead ballast that was required when swapping from lead acid to lipo batteries.

Drive gear fitted to 24v motor (metal gears are why the Aquazepp has such a distinctive sound)

Confident that a motor rated to work at 24 v could handle 4s lipo packs some more test dives were performed using a power meter to assess the average current draw. I found that I was now able to travel at around 44 m per minute with the motor drawing 16 amps so I now had a maximum range of nearly 16 km (6 hrs) ! Not bad for an old Aquazepp !

Power meter used to asses battery usage post dive

Next on the improvement list was a rear handle like most other scooters have, the scooter needs two hands to drive comfortably in its standard configuration. A delrin handle and bracket was made and fixed to the shroud, a small PVC trigger box was welded onto the main body which contains a magnet and spring which are actuated by a lever on the trigger. This dramatically improved the comfort when scootering, its now easy to control with one hand and only needs light steering input to instigate a turn.

The scooter is now capable of running at different speeds, a 50 amp Syren brushed DC motor controller has been installed which is controller by an Arduino UNO. The sketch loaded onto the Arduino reads the output from a hall effect sensor mounted to the inside of the hull. On the outside of the hull in the same location is a small plastic rod with magnets embedded which can slide back and forth to increase or decrease the speed. Whilst noise is always going to be loud when using an Aquazepp due to the metal gears the motor seems to run better now it is being supplied through the controller. The speed as its currently set can be run anywhere between around 65 m/ m and 40 m/ m, this can be adjusted further on the surface by editing the Arduino sketch if required. The controller outputs a PWM signal and even running at my current top speed I’m not at 100% duty cycle yet but 65 m/ m seems fast enough for now if dive kit is worn, if snorkelling it is even faster !

The downside to using a variable speed over fixed speeds is monitoring battery consumption, with fixed speeds battery usage can easily assessed post dive and then used as reference in the future, with variable speed unless it is run at fastest or slowest then its hard to gauge where you have been on the scale.

https://www.everything-ev.com/24V-750W-DC-PM-Motor

https://www.dimensionengineering.com/products/syren50