Main Winch
Hitec drum servo
Waverly Models integrated winch
Protoyype 1 Servoless winch
Protoyype 2
Screw drum concept
R/C control PCB
Screw drum prototype A
Screw drum prototype B
Self Tailing Winch Unit
Self Tailing Winch — Hitec servo
Self Tailing Winch — MFA motor
Brass pulley block

Main Sail Winch Specification

The starting point was to calculate the Main Sheet Load and Travel.
A PHP Calculator program was written to calculate these using information scaled up from the 1/20 model. Data in cm, cm2,

Main Sail
10560cm2 area
52 C of E to mast (Centre of Effort)
87 C of E to waterline
Top Sail
2680cm2 area
22 C of E to mast
168 C of E to waterline
Combined Sails
104 height of wind at C of E
5 Beaufort wind speed
13240cm2 total sail area
46 Mast to C of E
60 Mast to Boom sheeting point
60 Mast to Deck sheeting point

Calculator results
5 Beaufort wind speed
100 cm C of E above WL
7.32 m/s Height corrected wind speed
0.33 g/cm2 Wind pressure on sail
13240 cm2 Sail area
4.34 Kg Load on sail through C of E
46 cm Distance from sail’s Centre of Effort to mast
60 cm Distance from sheet attachment point to mast
1.30 Mechanical Advantage (Sheet/C of E distances from mast)
3.33 Kg Sheet Load (sail load / MA)
4.70 Kg Sheet Load with boom at 90 degs
60 cm Distance from sheet attachment point to mast
60 cm Distance of deck attachment point from mast
84.9 cm Sheet Travel with boom at 90 degs
Main Sheet requirements
Sheet working load 5kg
Mechanism safety load 10kg
Sheet travel 65cm (60+ 5, model measured, boom hits rigging)


Hitec sail drum servo

Hitec HS-785HB sail winch. @ 6V. 4 turns
As this winch only rotates 4 turns it would require a 5.1cm diameter drum to wind in 65cm Sail Sheet. Torque required would be which is the maximum the winch could handle.
The Action-Electronics Servo Morph P97 extends the winch travel to 7 turns. This would reduce the drum diameter to 3cm reducing the torque to which is nearly half the winch’s specification.
The winch was tested with a 4cm drum and various loads. The maximum load it would lift was 4kg ( Over 3kg, the sheet load was starting to overcome the servo power.
Hitec 785 video


Waverley Models intrgrated servo winch

Complete system from Waverly Models.

The standard configuration for a sail winch is a double drum on a sail winch and a Cord Loop running around a pulley fixed to the boat’s hull. The Sheet is attached to the Cord Loop.
The Waverly winch puts everything together in one stand alone unit so it can be installed and removed easily. It uses the Hitec sail winch on a plate, a 3.8cm drum and made to have a 60cm Sheet travel. The end of the Cord Loop ran through a semi-circular tube and connected to the winch by a sprung loaded square tube.
The system was tested using the Servo Morph. Again the maximum load it would lift was 2kg. Over this point, the load overcame the spring tension shortening the distance to the winch allowing the Cord Loop to drop off the winding in drum. Great idea for small loads.

Uploaded 7.4mb .wmv in 3min 480x360. Displayed 480x385.



Servoless winch prototype 1

This winch uses the idea of a stand alone unit but this time strengthening its construction and using a fixed Cord Loop Pulley running on bearings to take the high loads and reduce friction.
The idea of using a Servo to control a sail Sheet has always seemed odd. It requires the joystick to be held continually in one position to fix the sail’s position. Jolie Brise has three sails and a rudder to control with only two hands!
My sailing experience is in 10m boats, we would set the course and trim the sails to the course and tie off the sheets. The rudder changed the boats heading to make the best use of the varying wind strength. If the wind was steady, we would lock the wheel. 14 hours from the Needles to Cherbourg.
Sometime ago I came across the idea on the Model Boat Forum of using micro-switches to control the end travels of the Cord Loop and using the motor in a Servo for the power. This is controlled by a Speed Controller not the Servo. It also used the neat trick of putting diodes across the switches to allow the current to be reversed when the switches were “open” at the end travel point.
Miniature Sail Winch by Dave Petts
Mfa motor with gearbox and 4cm drum
A 919D 35mm motor from Mfa was used as the power source, controlled by a 10A Viper Speed Controller. A 100:1 gearbox was chosen as it gives an output of and 79rpm @6V; 7 turns would take 5sec.
The 4cm double drum was sufficient to allow a 3mm ply strip to separate the Cord Loop into two halves and support the micro-switches.
The End Stop mechanism was made from wood and was a bit sloppy but it proved the system worked.
The Sheet was attached to the Cord Loop by wrapping them with thread and then gluing. It was enlarged to create the stop which moved the bar to trigger the fully in micro-switch. Another knot was positioned for the fully out micro-switch The Cord Loop Pulley was too narrow as the Stop Knots kept on catching when there was no load on the Sheet.
servoless sail winch
The motor was powerful enough to lift 5kg but over 3kg the load would overcome the inherent gearbox friction and unwind which defeats the object of this winch.
The next perennial problem is what to do with the excess Sheet Cord when the loop is moved to the fully out position and there is no load on the Sheet. It has to go somewhere. I could just box in the whole structure.
Another major consideration was the mechanism footprint was much too large. The motor was at right angles to the Cord Loop.



Servoless winch prototype 2

R/C Servoless sail winch on Vimeo.
To solve the large footprint I put the motor in-line with the Cord Loop. A gearbox was made to:-
House a worm and spur gear to change axle directions,
100:1 speed reduction
Support for two separate drums
Use the worm gear properties of not being able to be rotated by the spur gear.
Finding suitable gears is a problem, in the end I bought a MOD 1: 4 start worm and a 25t spurs gear in Delrin for £35 from Ondrives
An expensive mistake. I was told this would give me 100:1 reduction, in fact it is only 6.25. Also they have helical cut teeth so work in both directions.
My original intention was to use a simple motor so I had to revert to the 100:1 gearbox. The combined ratio of 625:1 solved one problem as the gear friction was greater than the load so held the sheet in any position.

Splitting the 2 drums apart resulted in having to use a large diameter Cord Loop Pulley. It was made wider and worked well. As it is at an angle, it needs to be enclosed to stop the Sheet from getting trapped when slack.
The micro-switch mechanisms worked well. The binding and gluing the Sheet to the Cord Loop eventually broke. I stitched them together with fine brass wire off a wine bottle and then wound on and glued some thread to create the Stop. The problem of the motor running on after the current is switched off is still a problem when the sheet is pulled fully in. It can be minimised by pulling the Sheet in slowly just before it’s fully in.
Tests were run for different loads to check the current draw with 6 and 12V and the speed to wind in 60cm. Tests were stopped at 6kg as the mechanism was beginning to bend

LoadKg 0 0.51 2 3 45 6
Current6 & 12V 0.120.51 0.56 0.65 0.7 0.8 0.91.0
Time 6V  24 2425 33 34 40 43
sec 12 12 1212 12 12 13 13

A very odd thing happened when tests with 12V were started. When the micro-switches were operated the current should have turned off and stopped the motor but it carried on at a low speed. Much scratching of head and rechecking.
The fault lay with the diode. When it was taken out, the motor stopped when the micro switch broke the circuit, when it was in, the motor ran on slowly. Turning the diode around made no difference. The diodes had come from my odds box and I had not bothered to check their spec. They had a max reverse voltage of 6V so broke down at 12V and passed some current. Diodes are used here like non return valves, they stop 6V but leaked at 12V.
The video shows the Cord Loop stretching as the load increases.
The footprint is narrower but still as long. It could be even narrower if reverting back to the double drum configuration and a vertical Cord Loop Pulley.
The problem of the Sheet Cord flapping about when there was load on it, has not been resolved.



Servoless Screw Drum Winch Concept

Servoless winch
The Electronic system using micro switches mimics the real method of setting the sail and tying off the sheet
The next problem was to reduce the footprint by removing the Cord Loop completely and replacing it with a single sheet drum.
To solve the problem of what to do with the slack sheet caused many sleepless nights until I hit upon the idea of rotating the drum on a threaded shaft.
When the motor rotates to wind in the sheet, it rotates the threaded shaft which in turn moves the drum along the shaft to one end to hit a stop. Now the shaft rotates the drum winding in the sheet.
When the motor rotates to let out the sheet with tension on the sheet, the drum rotates. However, if there is no load on the sheet, the drum does not rotate but moves sideways along the shaft.(slight friction on the drum is needed to stop rotation)
When the sheet is pulled out it rotates the drum until it hits the end stop.

A common practice to lock a nut on a threaded shaft is to use another nut, tightening them together. The drum runs on a 5mm threaded tube and rotates until it hits the end stop. The 5kg force could lock them together. To avoid this, Driving Dogs are used to the drive the drum.
Servoless winch gearbox
The gearbox gives a 3:1 ratio giving a 300:1 final ratio. 2 pairs of gears are used to bring the drum shaft in line with the long Ali motor shaft reducing the overall height. (they were bigger than expected)
The 5mm brass studding is glued in the short threaded Ali tube. (two views are shown).
The driving dog is made by brazing two 1mm brass strips to a steel washer. This is then brazed to a nut (2 different temp rods). The nut is glued to the shaft.
A similar dog is made and glued to the drum. Its location found by winding the drum on the shaft until the dogs fully engaged. It is held in this position until the glue sets.
Servoless winch dog drive
Again 2 pairs of plastic gears are used to drive the cams at 8:1 ratio to the drum inline with the drum axel. The cams turn less than one rotation when the drum turns 7 times.
The cam discs were 6mm ply cut and turned on a mandrel held in a pillar drill. The cam was made with Plastic Padding shaped to suit.
Servoless winch micro switches
The system works and the gearing holds a 7kg load without unwinding. 7kg was the max load needed.
There was very little sheet overpull when the motor was turned off.




LoadKg 0 0.51 2 3 45 6
Current6 & 12V 0.50.6 0.6 0.8 1.0 1.2 1.41.5
Time 6V  15 1617 18 19 20 22
sec 12  7 8 8 9 9 10 10

I had hoped that by halving the gear ratio would halve the 600mm sheet travel time, it was more like a quarter for 12v. This could be due to friction caused by more gears and mis-alignment, the current increased from 1 to 1.5A.
Adding the 3:1 gearbox took up a lot of room and having the drum and cams in the same alignment made the foot print even bigger.
Using a combined motor/gearbox adds to the length and reverting back to a simple motor and worm/spur gears would allow a gear reduction allowing the sheeting speed to increase using 6v supply. The drum could be put at right angles to the motor. Using bevel gears may allow the cams to be put on top of the motor. All reducing the footprint.
Changing to 6mm studding will allow for slightly thicker dogs.
Mechanical components cost £61.24. Electronic components cost £27.56. Total £88.80

Mechanical Components
1. Motor 919D 38mm 100:1 Gearbox19.00
1. Motor bracketinc
1. Universal joint 6/6mm9.00
2. 2 x 30t brass pinions7.58
2. 2 x 17t brass pinions8.58
2. 2 x 58t plastic spur gear3.22
2. 2 x 20t plastic pinion2.60
2. 6 x 8mm bore flanged ball bearings3.80
3. 8mm thick wall aluminium tube 1.00
3. 6mm aluminium rod0.50
4. 5mm brass studding x 300mm2.10
4. 5mm tapered tap2.65
4. 2mm tapping drill1.21


  1. Mfa
  2. Technobots
  3. brilliant2buy
  4. Model Making Supplies
  5. Rapid Electronics
  6. Model Dockyard
  7. Stock

  Servoless winch electronics components
PCB Components
1. 65 x 35 PCB2.38
2. PCB Fuse 0.16
3. 3A Fuse 0.14
4. 4 x PCB terminal blocks 1.36
5. 2 x Schottky Zener Diodes 3A 30V 1N5821 0.36
6. 1 x 10A Viper speed controller21.98
7. 2 x Micro switches1.18

  Circuit diagram Servoless winch
  Servoless winch PCB


Servoless Screw Drum Winch Prototype A

Servoless winch prototype A build
This prototype is getting close to the final design. It raised a few problems, notably, it could only life 1kg not the required 7kg.
A straight cut worm and pinion were added to stop the sheet drum winding out when no load is applied resulting in the sheet falling in the hull.
One of the fist designs used helical cut worm and pinion which I descovered worked in both directions. (you learn something new every minute).
Many of the parts of the previous design were used again, flanged & thrust bearings, sheet drum, cams & micro switches and speed controller PCB.
Servoless winch prototype build
The bevel gears are the take off for the timing cams. The panel over the motor is for the speed controller PCb.
The threaded sheet drum shaft is now 6M to give a slightly larger pitch resulting in a larger contact for the drive dogs on the shaft and drum.
The worm/pinion ratio is 1:30 and the spur gears 20:58 total reduction of 87:1.
Servoless winch prototype A timing cams
The 10T small pinion is driven by bevel gears on the threaded drum shaft. The gear layout achieves a ratio of 8:1 to give sufficient room for the cam and micro switches (10:40, 30:30:60). The top cam is adjustible to alter the sheet travel.
Servoless winch prototype A motor
The aluminium tube in the sheet drum was replaced and threaded 6M. A thin metal strip was added touching the drum to stop the drum from turning when there is no sheet load. The friction can be altered by twisting the dowel.
This combination of previous designs worked. It was designed as I went along so the accuracy of the holes caused the axles to slop about a bit. Pieces got added so I cannot take it apart without distroying it. The main problem is that it will only left 1kg. So a redesign with a motor with five times its power.
On stripping down the model to re–use the parts it was dicovered the worm and pinion were nearly worn out.
To calculate the torque of the motor needed to pull 7kg, a bar was placed in the universal coupling and a lead weight hung on the end. 30g at 17cm lifted 7kg slowly.
Servoless winch prototype A calculating torque
  Main sheet drum
5kg Sheet load
65cm Sheet travel
3cm drum
9.43 circumference (22/7) x 3
6.9 turns (65/9.43) torque (5 x 1.5)
Sheet Travel Time
7 sec @ 6V 3cm drum 7 turns
1 rps = 60 rpm
Sheet Drum Shaft
M6 studding 1mm travel per turn
Design for 7mm sideways travel
Mfa 35mm RE540/1 motor
6200 rpm @6v output.
Ratio 87:1 = 6200:71rpm (1:30 Worm:spur 20:58 Spur:spur)
End Stop Cams
Gear ratio 8:1 (7 + 1 turns)
Cams rotate 87% of one turn.


Servoless Screw Drum Winch Prototype B

The motor power is increased, the cams are moved to get rid of the bevel gears and the drum only rotates 6 turns to reduce the number of gears.
The casing for the winch is made from 5mm 3 core ply (multi core would be better). The holes for the bearings and cam drive shaft were drilled together on a pillar drill before assembling. I found afterwards that drill was not vertical (read next section). The two large holes are for the motor mount.
Servoless winch prototype b

The shaft on the left shows the top gear which is driven by the worm. The centre shaft is the Drum drive shaft. You will see two brass strips which were used to accurately align the two shafts (they are shown on next but one pic). A new piece of ply was used for the top bearing. The shaft on the right is the cams drive shaft.
Servoless winch prototype b

The centre drum drive shaft has a 6mm threaded top section screwed and glued into an 8mm aluminium tube. A 4mm brass shaft is glued into the bottom.
The worm gear is fitted into a seperate housing to get the best engagement with its pinion. I found it difficult to accurately align the worm and pinion in the previous prototype.
The brass tube fits over the aluminium tube to act as a spacer to stop the shaft from moving upwarde.
Servoless winch prototype b

The flexible joint is used for shaft alignment, ease of removing motor and to give sufficient space for the bottom cam gear. The 15A Speed Controller justs fits under the motor (more by luck than design). Its switch can be seen on the right in the picture above.
The connector block is used for the mass of wiring and the two dioeds. The external fuse is at the top right. (the previous prototype had inside!)
Only one dog is used, now 2mm thick, to ease alignment.
Servoless winch prototype b


Self Tailing Winch unit


No part of the design of this Self Tailing Winch may be reproduced
by any means without prior permission in writing from Anthony Bell
Intellectual rights are the sole property of Anthony Bell.
© anthony bell 2011


From the previous winch, the next step was to simplify the design of the self tailing drum and make it into a self contained unit which can be driven by different sources.
Self Tailing winch.
To apply friction to the drum, a strip of thin tin plate is glued to one of the aluminium support tubes. The tension is adjusted by the end screws.
The ends are made from 6mm MDF. Drill the holes for the bearings before cutting the sides accurately and drilling the supports holes.
This top was made from 6mm MDF and screwed to the suport tubes. The brass tube through the two mdf blocks locates the pulley block which guides the sail sheet cord.
MDF has a problem as it resembles layers of compressed paper. It can be easily split with a knife. Gluing these two blocks to the top is not a good idea as they can break off under the strong side force of the sheet load. Make all of the top pieces from ply.
The two pine blocks are glued and screwed.
The following video shows the self tailing winch in operation driven by a Hitec servo.



Self Tailing Winch with Hitec sail servo

Self Tailing Hitec winch.

I found it stalled @ 6kg with a torque of (the box says The working load is 4 kg giving a 280mm sheet time of 11 seconds.
The video shows the servo moving in its mountings, this was caused by the driving shaft not being concentric on this first unit.
Hitec HS-785HB Winch Servo. Torque — @ 6v
Turns 3.5 (7 with Action-Electronics P96 servo morph)
Servo/winch flexible drive by servo arm driving two bars glued to steel plate attached to the drum threaded shaft.
Self tailing drum 25mm dia. Sheet travel 280mm (560mm with P96 servo morph)
Winch 140 x 90 x 70 high - weight 340g
Battery life for 6v — 2300mAh is 3.4hr @4kg continuous working
Low load   0.5 kg — 6.4V — 0.18A — — 7 sec sheeting time
Working load 4kg — 6.0V — 0.68A — — 11s
Stall load        6kg — 5.7V — 1.00A — — 25s




Self Tailing Winch with MFA motor

Self Tailing MFA winch.
The following picture shows how I aligned the winch unit to the motor as a straight coupling is used to reduce length.
The winch was positioned to align with the 6mm bar by putting a 1mm shim under one end. In futire I am going to drill the bearing holes before cutting out the square ends for accuracy.
Self Tailing MFA winch.
The Jolie Brise will have the Main, Fore and Jib sails controlled separately. It will be impossible to hold the sheet positions static and control the rudder at the same time using standard transmitter and servos.
My experience of sailing large boats is that you set up the sails to the wind conditions and steered course, quite often locking the steering wheel and letting the boat get on with itself. (14 hours from Southampton to Cherbourge). Using a R/C sail servo does not meet my design specification as it needs constant control.
This winch will be operated by a microprocessor which will count the number of sail drum turns to set the sheet travel. It works as a speed controller with Forwards/Off/Reverse so each sail can be set then left alone. The gib can be backed to get a fast tack through the headwind.
The Mfa motor has limitations but has a small footprint. The gib will need two winches working in contra directions controlled by one microprocessor.
MFA 919D with 100:1 gearbox Motor RE540/1
Shaft Torque — 6kg @ 6v — 2.1A — 79rpm
Rx output
3 — 5v peak to peak square wave pulse
Hitec    0.9ms to 2.1ms — 1.50ms central
Futuba 1.1ms to 1.9ms — 1.52 central
Peak refresh rate 50Hz (20ms)
Deadband 20 — 40 micros
Motor speed controlled by microprocessor
(speed controller not required)
Direct drive + shaft rotation counter sensor disc
Self tailing drum 25mm dia.
Sheet travel controlled by microprocessor
Max sheet load is 4kg.
Over this, the gearbox friction is not enough to stop the sheet load unwinding the sheet, so control is lost.
Winch 220 x 60 x 70 high (110mm with pulley) - Weight 540g
Sheeting time 7 to 10 sec
Low load     0.5kg — 6.0V — 0.60A — — 7 sec
Working load 4kg — 5.7V — 1.50A — — 10s



Brass pulley block

3mm thick pulley has two 1mm side checks that fit inside a 1/4" brass square tube. A 3.5mm wide strip is curved to fit round the pulley to stop the cord falling off. The square tube just fits inside a brass tube so it can be rotated to any direction.
pulley block production.
pulley block production.



Self Tailing Winch Microprocessor

This Radio Controlled winch needs to be operated by two separate processors; one for the motor’s speed and direction and another for the sheet drum rotation as this controls the length of sheet travel.
The most important feature of the sheet travel is its fully in position as if this is not accurately controlled, the motor can wind in the sail sheet too far and break the sheet, deck fittings ... The actual length of travel is not critical.
Unlike R/C servo sail winches which continually operate with the joystick having to be held in position. This winch simulates real conditions of letting out and pulling in the sail sheet to the desired position then tying off the sheet. It will work similarly to a speed controller, forward/off/reverse.
This boat will have 3 sails to be controlled separately and the rudder. I cannot see how this can be done with existing R/C transmitters/sail servos.

     Motor Controller Processor

Manufactures of R/C equipment use different PWM characteristics.
3 to 5v peak to peak square wave pulse.
Peak refresh rate 50 Hz — 20ms
Hitec    0.9 to 2.1ms — 1.50ms mid point
Futaba 1.1 to 1.9ms — 1.52ms
R/C transmitter’s joysticks give an incremental increase of pulses over their full range of travel. This is not necessary in this situation. For speed control we need:
Off                           1.45 to 1.55ms — 0% full power
Slow       Forwards  1.55 to 1.68ms — 20%
               Reverse    1.32 to 1.45ms
Medium   Forwards  1.68 to 1.80ms — 50%
               Reverse    1.20 to 1.32ms
Fast         Forwards 1.80 and above — 100%
               Reverse    1.20 and below
MFA RE540/1 Motor will run off 7.2v NiMH pack allowing for a 1.2v drop across L298N H–Bridge Driver.
Current drain 1.5A @ 4kg max working load. Not working continuously.
Processors will run off a common 4.8v NiMH pack to avoid transmission of motor’s noise.
motor control flowchart.
Motor control flowchart.PDF

     Rotation Counter Processor

The rotation counter has to take into account the fact that the motor continues to rotate under its own momentum when the power is turned off. This means the counter must act on its own and continue to count the number of rotations and sense the shafts direction of rotation so it knows to add or subtract the number of turns.
1. The processor is turned off/on with the sail sheet fully in, giving the processor’s “Counter” a fixed nil reference point each time it is switched on.
2. At start up, no reverse power can be applied so the motor processor software has to be overwritten.
3. For general use the counter processor does not affect the motor controller. It just counts up and down.
4. When “Counter” reaches it maximum number of forward pulses (max sheet travel) the counter processor switches off the motor power. Any motor overrun is not important but it must be counted. No forward power can be applied after max number of turns.
5. Reverse power applied winds in the sheet until it is nearly fully in.
6. To avoid motor overrun, at count = 6, the motor power is turned off.
7. Pause to allow the motor to stop.
8. Short bursts of reverse power to the motor to wind in the sheet until counter reaches “Nil”
9. No reverse power can be applied with counter at zero.
Sheet length = 900mm - 1 drum turn = 100mm – 9 turns
Sensor disc has a ring of 6 holes with two sensors A and B offset by 30 degrees to sense direction. A then B = forward. B then A = reverse.
6 counts/turn = 54 counts max (possibly alter counts from 30 to 90 via external preset resistor)
Motor 80rpm = 1.34rps = 4.5 counts/sec - 1 count = 0.22s
Width between direction sensing pulses = 0.11s
Processor will run off the 4.8v NiMH pack.
PRINT Microprocessor Spec.doc
motor control flowchart.
Motor control flowchart.PDF

PICAXE details