H5000 Hercules System

[Dockhead]

Anyone using this for cruising?

During our race last year I became acutely aware of the limitations of our sailing data, particularly apparent wind data which is apparently distorted by heeling and upwash.

I am very keen to find a solution to this which will help not only our racing, but also cruising under sail -- having better target speeds, more accurate polars, etc.

I guess an end run around the problem would be to simply develop own polars based on the raw data from my particular wind instrument (an LJ Capteurs CV7) uncorrected for errors.

But I do know that the Hercules system allows for making all these corrections. It is massively expensive however when you add the wind sensor, which alone costs $3000 and up.

I know there are some software systems which allow making these corrections; I was written to privately recently by someone concerning this system: SeaMODE Speed Lab

I don't know whether any of these systems other than the Hercules will output corrected wind values to the ship's network, so maybe there is no substitute.

How are others approaching this? I'm especially interested in the views of keen performance-oriented cruisers like WingSail and JPCates (paging you!).

I felt so much despair from my wind instrument readings last summer that I reverted to my old prehistoric, pre-instrument practice of watching the windex. I really want to do better this year.

[mglonnro]

My +1 half-solution to this:

- I use a Raspberry to process the data from the NMEA network and make a few additional corrections to calculate TWS+TWA.
- I have 10 Hz pitch, roll data on the network, so I can make the adjustments for that.
- I don't have (what the Hercules includes) acceleration data to use in the adjustments. I could get that from the Raspberry HAT, but haven't done it yet.

At the moment I don't write anything back to the network for real-time usage (but I guess I could). I use the data just to analyze things after the fact.

Wind shear is interesting as well. I saw some device that estimated/accounted for that using some mathematic models, but don't know exactly what they are doing.

[Dockhead]

Originally Posted by mglonnro

My +1 half-solution to this:

- I use a Raspberry to process the data from the NMEA network and make a few additional corrections to calculate TWS+TWA.
- I have 10 Hz pitch, roll data on the network, so I can make the adjustments for that.
- I don't have (what the Hercules includes) acceleration data to use in the adjustments. I could get that from the Raspberry HAT, but haven't done it yet.

At the moment I don't write anything back to the network for real-time usage (but I guess I could). I use the data just to analyze things after the fact.

Wind shear is interesting as well. I saw some device that estimated/accounted for that using some mathematic models, but don't know exactly what they are doing.

That's very interesting. If you could get your system set up so that it could OUTPUT corrected wind data into the network, that would be a killer app. I'm not sure how you would do that -- would need to identify as a device on the network, and you can't do that without certification, can you?


HOWEVER, why couldn't you take the data in over N2K then do the OUTPUT over 0183? Then put the corrected wind data back into the network via the plotter or whatever other device you have which converts 0183 to N2K. That would for sure work on my boat, because my Zeus plotters take in 0183 and will convert nearly all possible 0183 sentences to N2K PGN's.


Surely this is realistic to program? This would be something fantastic. Wonder if the OpenCPN wizards could do something with this?

[Dockhead]

Or maybe I should just bite the bullet and buy an H5000?


Big question whether the H5000 needs analogue wind and STW data in order to provide corrected data over the N2K network, or whether it can take N2K data and output the corrected datea with the H5000 as the source device. Anyone know?


Because I will not be installing analogue sensors. I just ordered the Airmar UST800 ultrasonic speed log, which is native N2K, and I will continue using my LJ Capteurs CV7 ultrasonic wind transducer, which is native 0183 but converted to N2K at the masthead using the LJC device.

[mglonnro]

Originally Posted by Dockhead

That's very interesting. If you could get your system set up so that it could OUTPUT corrected wind data into the network, that would be a killer app. I'm not sure how you would do that -- would need to identify as a device on the network, and you can't do that without certification, can you?

Yes, I can! The Yacht Devices USB-N2K gateway is the certified device and when I write to that (from the Raspberry), it is the one responsible for behaving well on the N2K side. I _guess_ it is visible and selectable from the Zeus as well, but don't know. Will have to look into it..

As to your next post, I think "biting the bullet" and going for the true and tested system would be a good choice. For our own boat, I'm stuck in a "change as little as possible" philosophy, so I'm trying to make the most out of the Raspberry

[ChrisseH]

The challenge here is that either you go consistently with one of the major boat electronic brands or then you do as mglonnro and I and build a hybrid system. Either way there is no shortcut to happiness, as all systems require large emphases on calibration before they can deliver anything that is accurate. I know several yachts that have very nice B&G systems, but haven’t properly calibrated them. And when I talk about calibration, I am not only referring to the calibration of the speedlog and wind sensor.
So, if you want to get wind data that is good enough for picking TBS values from your polar table, you need at least the following calibrations and compensations:
1) Gain and preferably also heel compensation of the speedlog value
2) Calculation (or measurement) of the boat’s Leeway angle and if you calculate, you need to determine the boat’s Leeway coefficient.
3) Offset calibration of the wind sensor’s direction measurement
4) Calculation of Upwash (in case the wind sensor is located within the upwash region)
5) Adjusting the wind speed and wind angle values for heel (easy if you use a mechanical wind sensor)
8) Normalize the TWS value to 10meters standard height in order to get values that can be used with ordinary ORCi or ORC Club polar tables.
Adjusting TWA for wind veer is not easy because veer is only present in stable or neutral atmospheric conditions and even then, there might be micro meteorological phenomena distorting the veer.
And as I said, no matter if you choose the brand or the hybrid approach the big challenge is not to get the calculations done, but to get the system properly calibrated. And some of the coefficients can only be determined empirically…

[Dockhead]

Originally Posted by ChrisseH

The challenge here is that either you go consistently with one of the major boat electronic brands or then you do as mglonnro and I and build a hybrid system. Either way there is no shortcut to happiness, as all systems require large emphases on calibration before they can deliver anything that is accurate. I know several yachts that have very nice B&G systems, but haven’t properly calibrated them. And when I talk about calibration, I am not only referring to the calibration of the speedlog and wind sensor.
So, if you want to get wind data that is good enough for picking TBS values from your polar table, you need at least the following calibrations and compensations:
1) Gain and preferably also heel compensation of the speedlog value
2) Calculation (or measurement) of the boat’s Leeway angle and if you calculate, you need to determine the boat’s Leeway coefficient.
3) Offset calibration of the wind sensor’s direction measurement
4) Calculation of Upwash (in case the wind sensor is located within the upwash region)
5) Adjusting the wind speed and wind angle values for heel (easy if you use a mechanical wind sensor)
8) Normalize the TWS value to 10meters standard height in order to get values that can be used with ordinary ORCi or ORC Club polar tables.
Adjusting TWA for wind veer is not easy because veer is only present in stable or neutral atmospheric conditions and even then, there might be micro meteorological phenomena distorting the veer.
And as I said, no matter if you choose the brand or the hybrid approach the big challenge is not to get the calculations done, but to get the system properly calibrated. And some of the coefficients can only be determined empirically…

That sounds reasonable, but how do you get calibrated data to your instruments without a device like the H5000?

What I learned last summer racing was how pathetically calibrated my instruments are and what a problem it's been all these years. I really want to break through this somehow.

[ChrisseH]

If I have understood your situation correctly you need some kind of “black box” that is converting the raw data from the heading, water speed and wind sensors on your N2K network to calibrated and motion compensated information for your Zeus plotters and maybe also for a PC for the route optimization. This could, for instance be done by a Raspberry PI with a suitable software. I do not know exactly what is included in mglonnro’s RPI software, but if something is missing perhaps that could be added to the current version. I think it is good if the software would be able to simultaneously log both the raw data and the computed parameters, so that you easily could follow how the coefficients and compensations effect the output. In addition, the UI should of course also enable the change of coefficients, which means that you need at least a touch display on the RPI.


The PC based hybrid system I use has all the required algorithms, calibration and motion compensation features, but the system has currently no capabilities to forward the refined information to external instruments in the NMEA formats. Instead I have a Panasonic Toughpad at the steering wheel that is mirroring and controlling the laptop on the chart table. The VNC software I use for the mirroring is RealVNC, which is truly bi-directional and provides also for cross-platform remote access between Windows, macOS, Raspberry Pi and Linux computers. This means that I also can connect to the laptop from my Android phone.


There is a multitude of references with algorithms for the wind calculations. Here a simple example:
Sailboat Instruments: True wind, VMG and current calculations


Note that there are different opinions in what order and how some of the compensations should be done. B&G is in many cases using compensation tables instead of formulas. Either way you need empirically to find the suitable values for your boat. The main thing is however to choose the calibration coefficients so that they suit your algorithms and that you can verify that they give you the accuracy you need.

[mglonnro]

Originally Posted by ChrisseH

If I have understood your situation correctly you need some kind of “black box” that is converting the raw data from the heading, water speed and wind sensors on your N2K network to calibrated and motion compensated information for your Zeus plotters and maybe also for a PC for the route optimization. This could, for instance be done by a Raspberry PI with a suitable software. I do not know exactly what is included in mglonnro’s RPI software, but if something is missing perhaps that could be added to the current version. I think it is good if the software would be able to simultaneously log both the raw data and the computed parameters, so that you easily could follow how the coefficients and compensations effect the output. In addition, the UI should of course also enable the change of coefficients, which means that you need at least a touch display on the RPI.


The PC based hybrid system I use has all the required algorithms, calibration and motion compensation features, but the system has currently no capabilities to forward the refined information to external instruments in the NMEA formats. Instead I have a Panasonic Toughpad at the steering wheel that is mirroring and controlling the laptop on the chart table. The VNC software I use for the mirroring is RealVNC, which is truly bi-directional and provides also for cross-platform remote access between Windows, macOS, Raspberry Pi and Linux computers. This means that I also can connect to the laptop from my Android phone.


There is a multitude of references with algorithms for the wind calculations. Here a simple example:
Sailboat Instruments: True wind, VMG and current calculations


Note that there are different opinions in what order and how some of the compensations should be done. B&G is in many cases using compensation tables instead of formulas. Either way you need empirically to find the suitable values for your boat. The main thing is however to choose the calibration coefficients so that they suit your algorithms and that you can verify that they give you the accuracy you need.

That's cool!

1) I think a good question is: is the H5000 system enough to get reasonably good end results? Does it have all the required components as far as the hardware and software go? (I would assume so.) There's a lot of moving parts to it all, and equally important are the processes a) to get the initial setup/parameters good enough and b) to adjust/verify over time.

2) It seems like an exciting option to put more time into the Raspberry stuff (that could be used in any SignalK etc application as well, I guess), so I'll have a go at that.

Here are the true wind calculations I'm using: https://github.com/mglonnro/truewind

I'll start by checking them against the link ChrisseH sent.

They do account for pitch and roll but not for some of the other stuff. And most importantly (and embarrassingly), I've been running on SOG only since (I think) it has been good enough over here in the Baltic where we don't have a lot of permanent currents (but wind currents instead, obviously).

[Dockhead]

An update:


I did acquire a used H5000 from Fleabay, and installed a new racing version of the CV7 wind transducer.



I have not had much time to work on this so last year I was using the CV7 connected to the main network via the WindyPlug, and with all the default corrections in the H5000. Last year I didn't have good STW (my DX990+ arrived dead) so didn't attempt to get good True Wind, but I was very happy with the apparent wind.


Right now I'm reconnecting the CV7 directly to the H5000 via 0183 which will eliminate the latency through the N2K network. And I grit my teeth and bought the H5000 motion sensor, so I'll connect that as well.



I have reinstalled my old UST850 ultrasonic speed log so I'll have at least some kind of STW data. I've got a Signet blue top on order, and I've sent the DX990+ back to Airmar, so I hope that later in the summer I'll have some better options.


Maybe with the much better apparent wind, and with whatever corrections I can make in the H5000 to the STW data, some kind of roughly usable true wind will emerge.


I'll report as I gain experience.