Deviation table

[Twister_Ken]

Last weekend there was next to nothing in terms of wind, so we decided to swing the compass. Attached, the resulting table. Errrr - it's not quite the nice serpentine shape, curling either side of the central axis, that the RYA YM books suggest. Quite happy to accept that one or two of my HBC readings might have been a smidge out, but are there dev tables that actually look a bit like this?

 

[Kurrawong_Kid]

I saw a photo of a professionally produced deviation card that had all the deviation to east of North-South line. Most surprised!

 

[johnalison]

If you are going to correct the compass you only need N-S and E-W readings, or, since they will be complimentary, one of each pair.

 

[Twister_Ken]

If you are going to correct the compass you only need N-S and E-W readings, or, since they will be complimentary, one of each pair.

Not going to correct it, just want to know how much it's off. And should anyone ask, I know that it's not often possible to steer a yacht that small, but still worth knowing, even if it helps me decide which way to aim 'off'.

 

[peterb]

Ken

The program I usually use is set up for 30 degree intervals. It's not that difficult to modify it for 22.5 degrees, but I can't do it until tomorrow.

First glance at your readings suggests that there is a bias by about 0.25 degrees. Probably due to the compass not being perfectly aligned with the boat's centre line. That's the usual reason for readings being biased one way or the other. It's also why you need more than just two readings.

 

[oldbilbo]

Ken


First glance at your readings suggests that there is a bias by about 0.25 degrees. Probably due to the compass not being perfectly aligned with the boat's centre line. That's the usual reason for readings being biased one way or the other. It's also why you need more than just two readings.

Yup.

That error - the algebraic sum of all 16 of the recorded deviation data points divided by the number of samples - is usually called 'A Error' or 'Systematic Error'.
On that set of readings it is of the order of +0.31° ( 5/16 ) and inconsequential at less than 0.5°. It is unlikely one could realign the compass to improve on that.

As for the other readings, they seem quite fair for a sailboat's compass. Are you quite certain no portable deviating influences are close by....? Wot about your iPad stowage...?

If you have corrector magnets, you could attempt to reduce 'B' and 'C' errors a little, but in all likelihood you'd only increase the deviations.

 

[Malabar]

I saw a photo of a professionally produced deviation card that had all the deviation to east of North-South line. Most surprised!

Called "A error". Means the compass is not aligned for and aft.

 

[Twister_Ken]

Yup.

That error - the algebraic sum of all 16 of the recorded deviation data points divided by the number of samples - is usually called 'A Error' or 'Systematic Error'.
On that set of readings it is of the order of +0.31° ( 5/16 ) and inconsequential at less than 0.5°. It is unlikely one could realign the compass to improve on that.

As for the other readings, they seem quite fair for a sailboat's compass. Are you quite certain no portable deviating influences are close by....? Wot about your iPad stowage...?

If you have corrector magnets, you could attempt to reduce 'B' and 'C' errors a little, but in all likelihood you'd only increase the deviations.

Nothing nearby except a couple of TackTicks alongside, and a solar panel just ahead.

 

[Mudisox]

I am not sure of what method you used to make the swing but I am surprised that you can differentiate 1 or 3 degrees with the compass shown, and display. I wouldn't worry too much about it in any case. I presume that the readings were taken with the engine running, I often found a difference with engine off and on, when swinging compasses.

 

[Twister_Ken]

Engine running, flat sea, autopilot ±1 controls used to get the steering compass as near as dammit on the exact reading.

 

[prv]

Engine running, flat sea, autopilot ±1 controls used to get the steering compass as near as dammit on the exact reading.

I've never really swung a compass properly - I've compared varying headings with charted bearings to check it's not grossly out - but I'd like to. Am I right in thinking you motored on exactly the cardinal headings on the right hand side of the diagram, then measured the degree figures on the left? How did you measure those? Hand compass bearing to whatever was directly over your bow? Have to say I've never found the conventional practice of checking a steering compass against a hand compass very convincing - why on earth should I assume the hand compass is correct?

I guess there's no point me doing a swing now, as part of my plan for this winter involves rebuilding the binnacle. I'll give it a go in the spring. With the luxury of land all the way round (in Lake Solent or Southampton Water) I think I'll do it by charted bearings from my known position to various distant objects, rather than a second compass.

Pete

 

[Tempus]

I know very little about swinging compassed, but I do know quite a lot about data analysis and error rates.


Here's a best fit of your curve to a corrected sine wave.



Things to note - I'd be very surprised if your accuracy is better really than +/- 2, this curve is mostly within that

I'd recheck the 22deg reading - it looks like you may have the sign the wrong way around, particularly as it's twin (the 202deg) is the same sign.

Curve is offset by 0.3 as has been noted earlier.

 

[oldbilbo]

Nothing nearby except a couple of TackTicks alongside, and a solar panel just ahead.

That looks very smart. And no sticky fingermarks on the dark glazing! You must be well proud....

 

[Twister_Ken]

That looks very smart. And no sticky fingermarks on the dark glazing! You must be well proud....

Second pic taken when boat was new, not even any control lines or halyards rigged.

For reality...

BTW, it wasn't in that state when I swung the compass!

 

[Twister_Ken]

I know very little about swinging compassed, but I do know quite a lot about data analysis and error rates.


Here's a best fit of your curve to a corrected sine wave.

Things to note - I'd be very surprised if your accuracy is better really than +/- 2, this curve is mostly within that

I'd recheck the 22deg reading - it looks like you may have the sign the wrong way around, particularly as it's twin (the 202deg) is the same sign.

Curve is offset by 0.3 as has been noted earlier.

Tempus, thanks v much for that. Looks quite sensible, will check the 22º hdg next time we're out.

 

[Twister_Ken]

I've never really swung a compass properly - I've compared varying headings with charted bearings to check it's not grossly out - but I'd like to. Am I right in thinking you motored on exactly the cardinal headings on the right hand side of the diagram, then measured the degree figures on the left? How did you measure those? Hand compass bearing to whatever was directly over your bow? Have to say I've never found the conventional practice of checking a steering compass against a hand compass very convincing - why on earth should I assume the hand compass is correct?


Pete

Pete,

Almost. Motored exactly on the numerical bearings on the left - the cardinal on the right are just there to remind me how to arrive at the numbers on the left. The difference between the steering compass and the HBC is plotted either side of the central axis, depending on whether it was greater or smaller than the steering compass heading.

As to the accuracy of the HBC, I checked that on the berth, against a plotted bearing. Slap bang on.

As to where to stand, I was centred at the back of the boat, sighting over the steering compass to the slot on the back of the mast. Quite easy for a degree or two of error to get in here if my eye strayed off the centreline by a smidge.

Before starting I checked that position didn't have deviation for the HBC. I sighted on a distant object (chimney at Fawley) and drove the boat round in a tight circle to see if the bearing changed. It didn't, so I reckon it was deviation free there.

 

[oldbilbo]

For reality...

Still looks OK to my practiced eye.... apart from the faded grey ST winches.

 

[davidfox]

Nothing nearby except a couple of TackTicks alongside, and a solar panel just ahead.

Looking at the depth it wouldn't be on the East Coast then!

 

[Hydrozoan]

This - if I've managed to attach it! - is mine. Blue line is Excel polynomial fit to blue points, all readings taken on one occasion. (The thin green line is a sine curve with half amplitude 4 and zero offset; the red points are readings from the same occasion, rejected as outliers; the yellow points are readings taken on a different occasion and year.) I agree with Tempus that getting to +/-2 would be the very best one could reasonably hope to achieve with a simple HBC

 

[peterb]

Generally speaking the best fit is given by a Fourier analysis, taking just the fixed error (misalignment), the fundamental (hard steel) and second harmonic (soft iron) effects. Theory suggests that the only real errors must fall within these effects; any other variations are measurement artifacts. Here's my go at it.



As far as I can see, the errors are mainly hard steel effect (2.6⁰), with soft iron giving less than 1⁰. I'd agree that the two suspect readings are at NNE and SE.