PilotFish Mk.III - If at first (and second) you don't succeed...

by Web FishMar 24, 2015 @ 08:49pm

If at first you don't succeed, try, try, try again.

Another year - another PilotFish design. Since this is the second major reboot of the project, we should be getting very good at learning from our failures :)

What did we learn from Mk.II? Here is the short list:

  • Structural integrity is both paramount and difficult to achieve. Actually, this statement is only partially true. We can make things reasonably strong, but in the process they get quite heavy. Reliable mounting points on thin-walled structures are a challenge;
  • Weight is important. Really important. Mk.II floats got extended vertically twice to compensate for the ever growing total weight;
  • Free board is overrated. Well, not really. But in our case it doesn't seem to make a big difference - it only increases windage with little effect on everything else;
  • Ease of build. Spending 4 man-years building something that has a limited chance of survival doesn't sound fun. And now that the MicroTransat challenge is open to powered vessels, this is becoming important.

 

So what's next for Project PilotFish?

 

 A simpler, lighter, stronger and hopefully better Mk.III design:

 

 

As we all know, third time's the charm. Apart from that, this is what Mk.III has going for it:

  • Simpler tubular floats made from multi-layer plastic stock - light and strong with reasonable drag;
  • No need for risers and cross-members in the deck structure - light and simple;
  • Super-strong mounting points while still allowing full disassembly for maintenance / repairs;
  • Very low vertical profile driving down windage factor;
  • Natural compartmentalization of float volume;

And a few challenges which will need to be addressed during the build process:

  • Limited displacement / floatation. Adhering to the design weight budget is paramount;
  • Switching to a smaller size / lower power rating solar panels will put extra stress on the power budget;
  • Watertight mounting for rear and aft float sections;
  • Mounting / orientation of the GPS and satcom antennas with floats having limited freeboard;

 

Let the journey begin...

 

Mk.II Floats - Week 3

by Web FishMay 10, 2014 @ 12:40pm

Final finish. An external coat of epoxy applied and finished to a polish. We are now ready for mounting on the vacuum base plate:

 

With combined weight of float model and ballast of 19.5 lb. we still have 2" of free board. In a catamaran configuration this gives us a target total weight of 40 lb:

Mk.II Floats - Week 2

by Web FishMay 6, 2014 @ 11:26am

More hand plane, sanding, filler, sanding. The float plug takes shape:





 

Mk.II Floats - Back to Square 1

by Web FishMay 2, 2014 @ 01:22pm

As we mentioned in our previous post, the new design called for increased displacement and lower weight. Going back to the early days of the project, the original plan was to have the float shells thermoformed / extruded from polycarbonate plastic. This plan proved to be too ambitious and was abandoned early in favor of fiberglass shells (and later on composite wood/foam/fiberglass construction). With the new weight target in place we decided it was worth revisiting the plastic shells idea. Instead of polycarbonate (which, even though very strong, turns out to be less than ideal for thermoforming) it was decided to stick with more traditional polystyrene or ABS floats. The two materials have similar properties with the ABS route adding extra flexibility to the structure (which might or might not be a desire-able feature). Both materials lend themselves very nicely to vacuum forming.

There are two major hurdles when going the vacuum forming route: the cost of the process makes it impractical to run really short series of pulls (we really only need 4 floats for now) and you need a solid (and strong!) plug (positive model). There isn't really much one can do about the former (other than run a larger batch and store the extra shells for future use ;) so we focused on the latter.

Week one results:





 

PilotFish Mk.II - Le roi est mort, vive le roi!

by Web FishApr 17, 2014 @ 12:33pm


The king is dead, long live the king! Our last post was more than 6 months ago and some of our followers have been asking what is happening at PilotFish headquarters. The short answer is: A LOT! After we assembled the basic frame of the boat (4 floats and a deck structure) it became evident that building a craft which is strong enough to withstand the challenges of an open ocean voyage, yet light enough to stay afloat and move under 50 watts of power or less, while fitting in a 1m3 cube has its challenges. In retrospect, hoping to get it right the first time was a bit optimistic. Here is a summary of the issues identified during the build and initial testing:

  • Windage factor. This should have been obvious from the get go, but somehow it wasn't. The symmetrical design approach called for two-sided hulls stacked on top of each other. The result was a massive "superstructure" towering above water, combined with relatively shallow draft. Increasing the draft is not really an option, as every inch of draft adds an inch of height above the waterline as well. Once on the water it became clear that in fresh breeze the boat would drift significantly and require constant course correction (wasting energy in the process). This was one of the items discovered early enough in the process to addressed as we went along (by splitting the original hulls into separate top and bottom floats) but the required changes ultimately caused some of the other issues listed below;

  • Build method. The original build plan called for two solid vertically-symmetrical hulls built out of fiberglass and mounted on a solid deck structure. As the build progressed this approach was morphed into a honeycomb composite float with glue-on fiberglass skin. While this produced relatively strong floats, it further complicated mounting of the components;

  • Overall strength. The hull mounting points were originally designed as an integral part of the hull frame in order to ensure structural integrity and torsional rigidity. As the build transitioned to the four-float design, the mounting points ended up being attached to the float deck (and through that to the float honeycomb structure). The deck structure itself was based on three 0.75" aluminium tubes with limited cross-bracing. In addition, extra reinforcement was needed on the mounting point assembly to ensure the design allowed disassembly while still withstanding the forces in the joints. Although there is no clear evidence that it would have failed under normal conditions, the design just didn't "feel" robust enough;
     
  • Limited adjustability. The horizontal mounting points of the floats predetermined the deck position, the clearance between the floats and the overall free board. This further complicated some of the other issues as it was difficult to make adjustments to address them;

  • Battery chemistry. The initial design assumed the use of Ni-MH batteries (same type used in RC car/boat models before the advent of Li-Po packs). At the end the Ni-MH charging model (and the Li-Po one for that matter) combined with solar panels as a power source and the requirement for on-line charging turned out a bigger challenge than expected. Adding to that the incurred power losses, as well as the limited longevity of these battery packs in deep-cycle applications pushed the implementation to a different chemistry - sealed lead-acid (SLA). These battery packs check all the boxes with one added nuisance: relative weight. More on this below;

  • Navionics and battery bays. The original plan was to store main battery banks in the hull structures and the navionics bay and antennas in the deck structure. With the changes of the design and the build process the payload had to be relocated in the inter-float space. This partially blocked the wind slot between the floats, thus negating its design purpose, while leaving components exposed to the elements;

  • Propulsion pod design / mounting. The original design design called for traditional motor-shaft-prop arrangement. With the change to the float build the approach became impractical. This item warrants its own series of posts, but in short: building a miniature, geared, salt-water rated propulsion pod is not as easy as it sounds;

Each of these problems is bad enough in itself. Yet probably none of them is insurmountable. The one issue that resulted from all of them and tipped the scale was overall WEIGHT. Our original weight target for the solid hull design was < 45 lb. Switching to the four float model looked like a great way to save weight. Unfortunately, with the added weight from the new battery packs and all the reinforcements needed to make the new design sea-worthy the overall weight was creeping up towards the 35lb. mark even before the final deck structure was in place. With the reduced buoyancy of the new floats we were looking at scaringly low freeboard. 

After countless hours of calculations, numerous back-of-the-envelope sketches and proper amounts of coffee and pizza, in one swift act of bravery the most important decision of the project was reached:

 

Give Up and Move On with Our Lives!

 

And so we did. The final result: PilotFish Mk.II. The next series of posts will detail how The New Bigger and Better Vessel (OK, maybe not bigger - 1m3 rule still applies) came to be and why sometimes going full circle is the best thing that can happen to a project.

Power Supply Segmentation

by Web FishJul 11, 2013 @ 10:31am

For the last couple of months we've been trying to make our mind on power system segmentation. On one hand having separate circuits for the navionics package and the main propulsion seems like a no-brainer - less interference from the powerful brushed motors and speed controllers, separate management of consumption leaving the central microprocessor powered even during deep drain of the propulsion battery bank, being able to store and run at different voltages, etc. On the other hand, the main propulsion bank offers 160+ Watt-hour of capacity which comes at a cost - total SLA battery weight (without the water-tight enclosures) is 8.5 lb. Adding extra capacity as SLA cells adds more weight, adding extra solar panels adds either more weight (glass crystalline) or cost (thin film plastic) to the project. In addition, we have built excellent redundancy with the 4 SLA cells and 4 solar panels which will now need to be replicated on the navionics side (at lower power level / capacity). On top of that the Iridium sat-com modem can draw upwards of 2A on its 12V bus for nearly 60 seconds, so we still need plenty of juice on the navionics segment for normal operation.

For now we will be powering the main electronics from the 4.4V SLA bank through a separate voltage regulator and see if we can work out a good capacity meter to be able to throttle down / cut-off propulsion at a predictable residual level and potentially further adjust comm-link session frequency to save power. This will allow the vessel to drift in case extended periods of low solar harvest rates (weather, panel damage) without completely loosing telemetry and control. Testing will tell if this is the right way of handling it - we still have space on the deck frame for 2x5W auxiliary panels on each side. But we can definitely live without the extra 4 lb of weight...

Vertical Orientation Sensors

by Web FishJun 18, 2013 @ 11:03am

In order to operate normally (or at least efficiently) within the current design concept, the central micro-controller will need to be able to determine which side of the boat is currently facing down. This allows the power and control to be routed to the proper set of motor pods and rudders. This might sound like a trivial task, but under open sea conditions it gets complicated by the constant forces acting on the vessel. Assuming the surface of the water was (near) perfectly still, you can rely on the accelerometer reading to determine the current position relative to the earth center of gravity (or "down"). This is how most cellphone / tablet firmware operates. Ideally one would want to deploy a 3-axis gyroscope to get an accurate reading on orientation. We are trying to get away without one, so currently testing these mechanical tilt-switch sensors for overall vessel orientation. They add an inertial component to the accelerometer readings - hoping to develop an algorithm that relies on a combination of both sensor types to determine with sufficient accuracy the actual vertical orientation of the boat on the water. We don't need precise vertical orientation - just Side A vs. Side B currently down. We will still rely on compass readings for fine readings in the horizontal plane.

Local Communication Link

by Web FishJun 12, 2013 @ 08:38am


Once the basic navionics package was assembled it became clear that we'll need some means of real-time monitoring of the system for tuning the command-and-control algorithms of the boat. The primary satellite link has a fairly steep cost factor associated with it and the SBD channel has an inherent delay which makes it impractical for real-time feedback and control. In order to maintain two-way real-time communication with the boat during development and testing (short of chasing the prototype in the pool with a laptop and USB cable :) we decided on a low-power radio link. This piece will evolve once boat moves on to large-scale testing as Bluetooth has limited range but for now it provides an easy way for communication with any BT enabled host device. 

  • Current selection: JY-MCU Bluetooth Serial Port
  • Status: Temporary / Stage solution

  • Criteria: Convenience, compatibility, power consumption, cost 
  • Finalists: JY-MCU Bluetooth, SainSmart Bluetooth XBee Shield, Pair of generic NRF24L01+ 2.4GHz transceivers
  • Main decision factors: Availability, cost, ease of integration

 (click image for larger view) 

Spec highlights:

  • 1.73 in x 0.63 in x 0.28 in (4.4 cm x 1.6 cm x 0.7 cm)
  • 0.25 oz (7 g)
  • Emulates standard COM port

 

Caveats:

  • Limited range suitable for close-range monitoring and control;
  • As the project evolves, we will be switching to the Nordic 2.4GHz transceivers. Those will require a separate host micro-controller on the PC/Tablet side to interact with, which makes the setup a bit heavier;

 

The Solar Panels

by Web FishApr 25, 2013 @ 04:49pm


The solar panels provide the main (and for now - only) source of power for the vessel. Electric power generated by the panels is used for driving propulsion pods, rudders, as well as all navionics (processor / controllers / comm). The harvested power is split between current consumption and battery bank charging. Battery banks kick in when solar power is below pre-defined threshold. 

  • Current selection: HighFlex Solar HF35W
  • Status: Finalized

  • Criteria: Weight, power output, efficiency (watt/sq.ft.), mounting, cost 
  • Finalists: Solbian SL40Q, HQRP 30W Flex panels, SunFlex 50W panels 
  • Main decision factors: Weight, availability, cost

 (click image for larger view) 

Spec highlights:

  • 27" x 13" x 1/8" panel size
  • < 2 lb panel weight
  • 4.7 V no-load output voltage
  • 4.5 V output voltage under nominal load
  • up to 7 A @ 4.4 V short current

 

Caveats:

  • Low output voltage requires adjustments to the rest of the power package;
  • These panels are MAGNIFICENT! At less than 4 ponds per side, we are harvesting up to 60-65 Watt power at high noon. Very, very exciting!

 

The Rudder Servos

by Web FishApr 16, 2013 @ 06:31pm

Rudder control servomechanisms. Control the position of each of the four rudders of the vessel.

  • Current selection: 4 x Hobby People HP-A38N
  • Status: Undecided (under testing)

  • Criteria: Cost, size, reliability, ease of control by Netduino (PWM)
  • Finalists: Hobby People HP-A38N, Hitec HS-322, Airtronics 94102 
  • Main decision factors: Cost

 (click image for larger view) 

Spec highlights:

  • Weight: 38g (1.34oz)
  • Speed @ 4.8V: 0.20 sec
  • Torque @ 4.8V: 3.4 Kg-cm (47.2 oz-in)
  • Dimensions (mm): 40 x 20 x 36

 

Caveats:

  • Water-proofing the enclosures. Going with servo-arm seal boots glued on to control arms;