Final Testing Session For 2016

The beginning of November saw us return to the Solent again for an informal weekends sailing with London Corinthian Sailing Clubs offshore group.

The Not Race and Hot Toddy evening have become a traditional end of season outing. Its a simple format, we get together at Hamble Point on a Friday night in November to eat and drink. Saturday morning we have the Not Race before sailing somewhere for dinner and to compare Hot Toddy recipes. Sunday is recovery time before we sail home.

The event brings together an interesting mixture of boats. This year we had everything from a 1900’s gentleman’s racing yacht to a selection of production charter boats. The Not Race is so called because its not a race in the traditional sense. Boats start at staggered intervals and aim to complete a set course around the Solent in the shortest possible time. The staggered starts mean that if the conditions are kind and the race officers maths correct, big if, the yachts should reach the destination at the same time.

Jengu being the fastest boat in the fleet, by some distance, had to give the other boats a considerable head start. Aeolus, the oldest boat, started first and over the next 45 minutes the rest of the fleet set off in pursuit. The course consisted of a series of broad and close reachs with one beat across the Solent; nothing too taxing.

With a brisk and building Northerly force 3-5 breeze it took about an hour and a half for Jengu to catch Aeolus. We overhauled the other boats along the way before we got to Aeolus somewhere between Norris and Browndown on a screaming close reach.

As a days sailing it provided an excellent opportunity to run all of the collection software and systems that we have been developing this year. We had setup the route in the data collector software when the sailing instructions were released; in the warm on dry land. In the run up to our start slot we turned off the engine and started the collection software. Picking the correct, boat, route and configuration option then creating a new recording session. On Jengu the laptop connects to an Actisense USB-NMEA gateway to collect the data from the instruments. This was all connected up and tested before we left the dock. We could see that data was flowing correctly into the software.

We left the collector running and sailed the course. Once we finished we stopped the collector, crucially, before we started the engine to motor up the Medina past Cowes.

Reviewed the data we collected a couple of things immediately became apparent.

Firstly; we had only recorded about an hours worth of data which seemed strange as the collector was still running collecting data when we can came to stop it. Some more detailed examination of the power settings on the laptop we were using showed that when running on battery it was setup to go to sleep after 60 minutes. Windows will put itself to sleep if there are no keyboard or mouse inputs; even if the computer is doing things like collecting data in the background. When we touched the mouse to stop the collector the laptop woke up and the collector started running again. From a testing stand point this was a little frustrating. We had an hours test data most of it the start sequence !

Secondly we had a number of corrupt values. Our original test boat two years ago would generate messages where the sentences were mangled, there is no other way to describe them, with very high values. We thought it was just a glitch with the wiring on that boat but it looks like Jengu has the same problem. So we had a boat speed profile that looks like the one in the image below.

We can clean the data up using the data collectors NMEA Data tab; we are able to filter the displayed data down to show only the VHW Water Speed and Heading sentences. We are also able to identify the invalid records that had been collected. On the next image the invalid data is in column C8. You can filter for just the invalid elements on the NMEA Data tab using one of the drop lists at the top.

Its worth noting at this point that the Data Collector will use a checksum to make sure the messages it receives are valid. The source system generates a check sum value based on the sentences content and a standard shared algorithm. The receiving system can then regenerate that checksum to make sure that no corruption has occurred in transit.

We have an option built into the Data Collector to either accept or discard sentences that fail the checksum. Generally we tend to accept the invalid messages and weed them out later. Early testing showed that some systems were not correctly generating the checksum value so we were getting a lot of false fails leading to us reject more records than we needed to. Its also possible to manually rebuild a sentence where, for instance, an individual element has been merged with its neighbour.

Having removed the invalid records we needed to recalculate the summary data. Summary information and Polar values are calculated as the information is collected. Invalid data like that identified above will distort the summary and polar values. We use the Recreate Session Summary Data function on the recording session to perform the necessary recalculation.

This deletes the summary data for the session and runs all the collected messages into the collectors sentence processor again. As it works its way through the collected source NMEA 0183 sentences the summary information is rebuilt. With a long collection session this can take some time. Our hour long session had collected 30000 NMEA 0183 sentences; multiply that up for a days collection and you are looking at a lot of data to churn back through. It takes longer to process the sentences that feed into multiple calculations; Our 30000 NMEA sentences could generate 100000 calculations. The recalculation time is also very dependent on the performance of the computer performing the calculations; generally though it should be significantly less than the original collection time. We left the laptop churning and went to lunch…

A short while later we had the data cleaned up and the graphs displaying nicely.

The tabular data is complete too.

We can view and export the full summary data for analysis in Excel.

The track can be exported and uploaded to Google Earth; though the track we generated was mostly us circling prior to the start you can see us sailing to the first mark.

Sadly we seem to have missed the part of the sail we spent hammering up and down at 10 plus knots.

So that is our final on the water test for 2016. Now that we have a working test boat and have the software that talks to it we will be focusing on releasing an updated version of the data collector in early 2017. There will be more details of that in the next post.

Bill December 2016

Originally published on December 16 2016