par. 4.4 and 4.5

4.4	Fixing the runner on the shaft and aligning it	previous
		index page

A problem with using second hand alternators of different types is that they all have slightly different shaft diameters, pulley shapes and pulley dimensions. So a way to fix the runner on one type of alternator might not work at all for another type.

What all alternator types have in common is the general shape of the pulley: It has a flat rim with a diameter of some 50 to 70 mm at the front and the nut with which the pulley is fastened on the shaft, is more or less sunken into the pulley itself. Now a simple and universally applicable solution for fixing the runner on the shaft: An M6 bolt, 20 mm long that pulls the runner against this rim on the pulley, see both drawings in fig. 4.12. This bolt itself can not fix the runner properly. It's the friction forces between the runner and the rim that hold the runner in place. Consequently, the bolt itself will conduct a constant pulling force only and will not break due to metal fatigue. For easy assembly, a bolt with hexagonal socket head is best because it sticks on top of an Allen key. Normal bolts with hexagonal head are just as good, but more difficult to stick through the washer and the hole in the side disk.

In practice, the procedure for fixing the runner is:

First, something one should not do: Do not file off material from any surface that influences the alignment of the rim of the pulley. Normally there is a series of parts on the alternator shaft: A distance bush, the alternator fan, maybe a kind of support disk and then the pulley that might consist of two halves. As long as you don't change anything, the surfaces where those parts touch one another, are well aligned. But suppose you wanted to make the distance ring a few mm shorter and on one side, it ends up 0.2 mm shorter than at the opposite side. Then the pulley rim will be misaligned by some 0.8 mm and every time this distance ring has been taken loose and fitted again, it will be misaligned in a different direction.
Drill a 5.0 mm hole at least 20 mm deep axially into the shaft. Starting in the middle is not difficult because most shafts have a cone-like centre point but the difficult thing is to keep drilling well-aligned. Things are easier on a drill press. If you have to do it with a hand-held drill, you can only see whether you keep the machine well-aligned from above. So have someone else look from the side whether the drill is well-aligned in that direction. Especially with a hand-held drill, a new, sharp drill bit is a great help since the material of the shaft could be quite tough. Both with a hand drill and on a drill press, it is worthwhile to drill a few mm, then rotate the alternator shaft 1/4 turn, drill another few mm etc. In this way, a misalignment will be corrected or at least you will notice it in time.
Tap an M6 screw thread in the hole. Make sure to screw in the M6 tap well-aligned with the shaft. Some taps will continue straight-on if you started at an angle with the hole itself, eating themselves into the side of the hole as they go. The sets of 3 taps that make a complete thread in 3 goes, are self-centering and will align themselves automatically into the hole.
For unexperienced people: A tap must be oiled before every hole. To start, press it down quite strongly, hold it as well aligned as possible and screw it in the first few turns. This is a bit tricky, better try it out on a piece of scrap metal first. Since the hole in the shaft is a `blind' hole, eventually the tap will get stuck at the bottom of it and require more and more force to screw in. If you continue just a bit too long, the tap will break, quite likely you will never get out the remains from the hole and you will have to think out another way of fixing the runner on the shaft. So don't try to get the thread as deep as possible, better drill a bit deeper.
For a proper fixing, the bolt should pull the runner against the rim on the pulley and not against the end of the shaft or the large nut on it. The tightened bolt will make the side disk bend outwards by some 0.5 mm and therefor the end of the shaft should be sunken 1.0 mm within the rim, see fig. 4.12. Probably the end of the shaft has to be filed off a bit and maybe there is no room any more for the spring washer because then the nut might stick out.
With the runner mounted, the M6 bolt already holds the pulley onto the shaft. However, the large nut on the shaft itself prevents the pulley from tilting and it should not be left out or weakened too much by filing it thinner. If there is no way to get the nut and end of shaft sunken within the pulley rim by 1 mm without weakening the nut too much, make a ring that fits between the runner and the rim of the pulley. To keep the ring at its place, it can be glued onto the runner with two component glue.
By itself, the 3 mm side disk is too weak to stand the pulling force of a well-tightened M6 bolt: It would bend outwards much more than the 0.5 mm stated above. That is why there must be a 6 mm thick, 20 mm outer diameter washer below the bolt head, see fig. 4.12
Probably the alternator side disk has an irregular surface with a layer of brass of varying thickness. So the area that touches agains the rim of the pulley should be filed flat in order to have a good fit between the two. But better postpone this until the alignment is checked. Then for the moment, the bolt should not be fixed tightly because this would make the whole runner bend.

Just one bolt of only 6 mm thick might seem insufficient to keep the runner in place. But in fact, even with an M4 bolt (of quality 8.8 and well-tightened), the charger could be used up to the maximum head of 15.6 m. But the 6 mm thick washer, the sunken end of the shaft and a proper tightening of the nut that holds the pulley are essential. If the end of the shaft is so thin that it would be weakened unacceptably by the hole for a 6 mm bolt, you could safely use a 5 mm one, as long as they are of 8.8 quality and well-tightened. When the outer diameter of the thread on the shaft is 12 mm or above, the hole for a 6 mm bolt will not weaken it significantly.

Fig. 4.12: The seal and the way to fix the runner on the shaft.

When the runner is fixed on the shaft, it becomes clear how straight or misaligned it is after soldering and what should be done about it in case it wobbles too muchreference to judge alignment. . A problem is what to use as a This depends on which parts were made most accurately:

Choose the blades if all blades had nearly the same width.
Choose the outer radius of the side disks if these were better.

Fix the alternator in a vice or place it firmly on blocks on a table. Rotate the runner and hold a carpenters square upright from the table against the runner circumference. Or just look along its circumference to a fixed point in the background.

First check the alternator end of the runner. If this end is off-centered, there are 3 possible causes:

The hole in the side disk was drilled off-centre.
The hole in the shaft was drilled off-centre.
The M6 bolt has play in either the side disk or the shaft, or both.

With respect to the top 2 causes: Probably both errors are present and this means that by rotating the runner with respect to the pulley, a position can be found where the error is minimal. Rotate the runner until you find the side that sticks out most towards the square and estimate how much the runner should be moved to become well-centered. Then loosen the M6 bolt, hold the runner steady and rotate the pulley until the runner has moved as close as possible to the desired place and tighten the bolt again.

The third cause is tricky to deal with. It means that every time the runner has been taken off and put on again, it will fit differently. The easiest way to get it well-centered is by tightening the bolt very loosely and then carefully hammering the runner towards its proper position. Don't use force on this because if the holes are poorly aligned as well, it might make the bolt bend. When this would happen, the runner could become misaligned again once you try to fasten the bolt. Or when the runner stays as it is, the bolt will have to bend back and forth as you fasten or loosen it and it will break soon.

Once you have found the best position for the runner, make corresponding marks on the pulley and the runner indicating how the runner should be mounted from now on. Also make a mark on the shaft the next time the runner is removed.

If it is not possible to get the pulley end of the runner less than 0.5 mm off-centered (0.5 mm off-centre means that you see the runner move 1 mm while rotating it, from 0.5 mm towards one side to 0.5 mm towards the opposite side) by changing the position of the runner itself, the hole in the side disk should be filed out in the desired direction. Mark the direction in which the runner should be moved and estimate how much it should move. Then file out the hole in the side disk in that direction.

Now the bolt has play inside its hole in the side disk, so use the procedure described above: Tighten the bolt very lightly and hammer it to the right direction.

With the alternator en well-centered, check the free end of the runner. This end is likely to be even less well-centered than the pulley end because it is difficult to solder the runner straight. If it is more than 0.5 mm off-centre, it makes sence to have the runner tilt a little with respect to the pulley. Mark to what side the runner should tilt and estimate how much. The easy way is to file off material from the rim of the pulley because this is just a few mm wide. But this means deliberately mis-aligning the rim of the pulley to correct an error of the runner. Then when another runner has to be fitted, problems are even bigger. So to keep runners interchangeable, it is best to change the runner itself and keep the pulley's well-aligned.

In case there is a lot of brass on the alternator side disk, maybe filing or grinding away the brass at one end will make the runner tilt so much that it will be well-aligned. Do not file off more than say 0.5 mm from the side disk itself. If the problem cannot be solved this way, file the side disk flat and glue (with 2-components glue) a ring onto it where it touches the rim of the pulley. Now this ring could be filed off until the runner becomes well-aligned.

Only when the alternator side disk and the rim of the pulley make good contact all around, the M6 bolt can be fastened. So check whether the alternator side disk is flat and if not, correct this.

Then the outer circumference of the runner itself can be filed to shape. First check whether the runner is well-fixed and all marks are unambiguous. Have a vernier callipers ready and check the diameter regularly. To find the places where material must be removed, use the square standing upright from the table. First work on the ends of the blades, filing off a bit of the side disks as well if these stick out. Once both ends are well-centered and have the right diameter of 75 mm, it is easy to see how much has to be filed off from the middle of the blades. Take care not to file off too much and keep using the square and the vernier callipers.

Filing off blades makes them weaker. If the runner was poorly centered, filing off blades until the runner becomes well-centered would mean that some blades are weakened unacceptably and the same happens if some slots were not cut deep enough. Especially when it is intended for use at a rather high head, it is adviseable to stop at 75 mm diameter, even when not all blades are touched yet.

Box 4.11: Machining runners in series.

The way to align runners described above takes quite some time. To get runners to fit perfectly well on all alternators that have the central hole bored in properly, both the outer circumference and the contact surface with the rim of the pulley must be machined. The easiest way is to use a lathe:

Get a piece of 20 mm shaft of about 100 mm long. Fix it in the lathe, sticking out about 65 mm and bore a 5.0 mm hole in it. Tap an M6 thread Machine the end of the shaft until it is flat, then machine the innermost part a bit deeper so that a rim of some 2 mm wide is left at the outer circumference.
Now slide the free end of a runner over the shaft and fix the alternator end with an M6 bolt + a 6 mm thick washer. Now it is fitted in the opposite way as it would be on an alternator: The shaft sticks through the runner and the bolt is at the outside.
Like with a runner fitted to an alternator, the runner must be well-centered before starting to machine it. First check the alternator end and expand the 6 mm hole in the desired direction if it was poorly centered. If the free end is poorly centered, the 20 mm hole in that side disk must have been off-center: Expand it in the desired direction.
Now both the contact area with the rim and the outer circumference can be machined. Take care that:
The alternator side disk won't end up too thin, glue a ring onto it if a lot has to be machined off.
With the outer circumference, take off little by little since the blades aren't too rigid. With an ordinary runner on an ordinary lathe, the runner will rotate in such a way that the blades will bend inwards when they are hit by the chisel of the lathe. With a runner designed for the opposite direction of rotation, or with a lathe that works differently, the opposite could happen and there is a chance that the runner gets jammed when one tries work too fast.

I haven't tried out this yet. If it reveals that machining is difficult because the runner isn't fixed rigid enough on the shaft, try fixing the free end of it better by fitting wedges of thin metal sheet between the shaft and the free side disk.

Another way of machining runners is by fitting them on a M12 x 150 mm bolt. Fit the bolt in an electric hand drill that is fixed rigidly on a table with the shaft in horizontal direction. Make sure the central hole with M6 thread in the bolt is as well centered as possible mark how it was fitted in the bore head. To fix the free end of the runner, use an M12 nut with the inside thread bored out and the outer circumference filed round to 20 mm and slightly conically. Shove this over the bolt until it gets stuck in the 20 mm hole in the side disk.

For the machining itself, have the electrical drill run at its maximum speed and lightly hold an angle grinder against the runner. If the hand drill bearings have no play and it is fixed so rigidly that it can not vibrate, the angle grinder will only touch the protruding parts of the runner and make it become well centered. Regularly stop for checking the diameter since one easily grinds off too much. Like with a lathe, both the contact area with the pulley and the outer circumference can be machined. For a nice finish, hold sand paper against the rotating runner.

4.5	Seal around the shaft

The seal is about the most trickiest part of the charger:

It is difficult to predict whether water will come through a certain type of seal.
If some water comes through, likely the brushes will be the first parts that are affected: When wet, the resistance over brushes and conductor rings increases, field current decreases and the charger will not produce a decent charging current. This might be misunderstood as an electrical problem.
Other negative effects might only show half a year later, when bearings wear out or the inside is badly corroded. Then operators might try to solve these secundary problems rather than dealing with the root cause and could become disappointed when the charger breaks down again.
There is very little room to build a proper seal. It must fit in the narrow space between the rim that supports the runner and the fan of the alternator itself. So in effect, the seal can only be made at the place of the pulley.
Like with the problem of fixing the runner on the shaft, the shape and dimensions of the pulley affect the possibilites for building a seal. Only now there is no easy solution that will do for a great variety of alternators.

Of course a solution to all this would be to fit the runner on a separate shaft with a standard seal and have it drive the alternator by means of a V-belt transmission. This means going back to a more conventional charger design. Having the runner mounted directly on the alternator shaft has some major advantages: It is cheaper to build, has less moving parts that can cause trouble, the frame is simpler, it is lighter and more compact so it is easier to transport, and it could achieve a slightly higher efficiency because there are no transmission losses. Therefor I think it doesn't make sense to have the runner on a separate shaft just to make it easier to build a proper seal. Only when a transmission is needed because runner speed becomes too low for the alternator (so: When a large capacity is needed at a low head site). Then one must take for granted that in the firefly, the seal is a tricky part to build and that it deserves special care.

The type of seal used is a labyrinth seal. The rotating and stationary parts do not touch one another so there is a path along which air with water droplets could pass through the seal. But if it is designed properly, the amount of air and wate coming through is so low that it can do no harm any more (when working, the alternator produces considerable heat and the alternator compartment is ventilated, so a small amount of water will immediately evaporate and is harmless). What make a labyrinth seal effective is:

Narrow clearances between rotating and stationary parts: 0.6 or 0.5 mm.
Many bends in the path the air has to follow: Because there is so little room, only a few bends can be made in the firefly.

Since the parts do not touch, they will not wear out because of use. Still it is best to use quite thick steel sheet (1.0 or 1.25 mm) because then parts hold their shape better and it is easier to make things fit accurately. To avoid oxidation, choose galvanised sheet when that is available.

Since one might have to improvise when coming across a type of pulley for which the standard solutions of fig. 4.12 are not appropriate, it is good to understand the effects that play a role in this:

The alternator fan causes the trouble. It is designed to suck air through the alternator itself, but its blades also suck in some air from the other end, so through the seal from the aturbine compartment. If some water droplets come through, the fan also disperses them into very small droplets. It is this fine mist that causes damage because it will be be sucked into the air inlets of the alternator. Some large droplets that would settle out on the inside of the covers, would do little harm. So:
- A seal that worked fine at a low speed, might leak at a high speed because then the fan sucks harder. So chargers that will be used at high head need a more carefully designed and built seal.
- In case a roughly made seal turns out not to be effective enough, consider having the alternator run without a fan. Of course this severely reduces the flow of cooling air through the alternator so this is only adviseable if the charger will be used at low capacity only. Producing 100 W without a fan should be no problem and probably 165 W is possible also.
- Have the fan as far away from the separation sheet as possible. Then the fan will merely make the air circulate inside the alternator compartment rather than sucking in air through the seal.
The gap between a rotating disk and a stationary sheet also acts as a pump. The rotating disk will make the air rotate as well and then the centrifugal force will make it flow outwards. When the gap is minimal, the amount of air flowing through the gap will diminish accordingly. But the pressure this pump can deliver, does not: For a small gap it will be even higher. However, the main variables that govern the pressure this pump can deliver, are the diameter of the disk and the speed of the alternator. So:
- The ideal situation would be if this pump action creates a pressure that is large enough to overcome the suction created by the fan. Then dry air from the alternator compartment will flow to the turbine part rather than the other way around. Such a seal will function at any speed, since both the suction of the fan and the pressure created in this gap will go up with the speed.
  To make sure that this pump works in the right direction, the rotating disk should be at the turbine end of the stationary sheet.
- Then a rotating disk at the alternator end of a stationary sheet creates pump action in the wrong direction. To minimise this effect, either give this disk a relatively small diameter, or make a hole somewhere so that this pump action will pump around wet air from the turbine compartment rather than pump this air to the alternator compartment.
Once wet air is rotating somewhere in the seal, water droplets experience higher centrifugal forces than air since they are heavier. This makes that droplets will settle at the outer circumference of such a cavity. Once too much water gathers there, some of it will pass through the seal. So either water droplets have to be thrown out directly into the turbine compartment, meaning that there is no such cavity. Or there should be a drainage hole at the lowest point of such a cavity to the turbine compartment.
Then on where the water comes from that might pass through the seal. The main flow through the runner poses no problem: It leaves the runner still with considerable speed as a broad jet and will not come near the seal. Problems arise with:
- Obstacles in the course of this main jet that make some water to splash backwards towards the seal. This could be especially problematic if it could enter the seal cavity through the drainage hole.
  The connection of the rear legs to the square of the frame could form such an obstacle. To minimise this effect, the rear legs are welded onto a triangular plate bolted to the square, so that there is no large surface that could deflect water backwards. The bars that make up the protection shield are too thin to make any water splash back.
- Water leaking away through the gap between runner and nozzle. Only the water leaking away at the alternator end poses a problem: It hits the separation sheet or first sheet at high speed, making it disperse into fine droplets.
  The fact that water leaks through this gap means that this water was still under pressure. According to crossflow design theory, this should not be: Once water leaves the nozzle, all its energy should be converted into speed. Clearly this was not the case with the first 4 chargers I built. The new nozzle design presented in this book should largely solve this: It spreads the water over a 25 % larger admission angle to the runner so that the runner can absorb this flow without causing back pressure.

There are two basic forms for the seal:

For pulley's that consist of two halves. Here the rotating disk can be clamped between the halves. It rotates within the seal cavity created by the separation sheet, two distance rings that are open at one end to form the drainage hole, and the first sheet, see the large drawing in fig. 4.12.
For pulley's that are made in one piece. Here the pulley itself will act as the rotating disk so it must have a small clearance to the separation disk. To make it more like a labyrinth seal, there is a another rotating disk at the alternator side of the separation disk. This disk should have a small diameter since otherwise it would create pump action in the wrong direction. There is a distance ring between the pulley and this small rotating disk. To reduce air flow even further, the outer diameter of this distance ring should have a small clearance with the corresponding hole in the separation sheet. See the inset drawing in fig. 4.12.

Both versions have been tested and worked fine up to free running speed at respectively 12 and 15 m head.

Addition to internet version:
Anti-splashing ring for seal for pulley in one piece.

With the seal type for a pulley in one piece, there was still a leakage problem: Some water comes through and gathers at the bottom of the alternator compartment. It does not pose a real danger as it comes through as a liquid and not as a fine mist of water droplets that can be sucked into the alternator: One could even drill a hole through the separation sheet at its lowest point to get rid of it.

Probably the water can come through because there is a jet of water leaking away between the runner and the nozzle. This jet hits the separation sheet just outside of the pulley and apparently, some water is forced inwards through the narrow gap between the pulley and this sheet. A solution would be to fit an anti-splashing ring of 0.5 mm galvanised iron sheet, inner diameter = 6 mm, outer diameter = 100 mm between the runner and the pulley (see fig. 4.12). This ring should prevent the water jet from splashing against the separation sheet at high speed. I haven’t tested this solution yet, please let me know whether it works.

With a pulley that consists of two halves, there is a problem if the contact surface between the two halves isn't flat, see the large drawing in fig. 4.12. Then a rotating disk fitted between the two will bend if the nut on the shaft is tightened, but it might not bent completely into the shape of the two halves. As a result, the rim of the pulley to which the runner is fixed, might be poorly aligned. And every time the pulley has been taken apart and assembled again, the alignment might have changed.

The easiest answer to this is to have the two halves fit directly onto one another and fix the rotating disk in another way. This could be done by pressing it against the flat rim on one of the halves by means of a rubber ring. Or it could be screwed to one half with tiny M3 screws with sunken heads, meaning that holes must be bored and screwthread tapped into one half of the runner.

Whatever design of seal is used, it must be made quite accurately to make sure that clearances between rotating and stationary parts are minimal (about 0.5 mm) while still these parts won't touch one another. In fact it doesn't matter much if they would touch one another a little as long as the runner still rotates quite easily: Then these parts will wear out within 5 minutes once the charger is runnning, or when it is driven by an electrical drill. When they touch so bad that the runner can only rotate using excessive force, the parts might be grinded away completely by the time they are worn out enough not to touch one another any more. To make it easier to build the seal accurately, the design has the following characteristics:

Rather thick steel sheet is used (1 mm or 1.25 mm). This holds its form better during the building.
All parts are flat, so no bending is required. This is easy to check and, if necessary: To correct.
Very important is the fixing of the alternator: It should be at the right distance from the separation sheet, with its shaft going exactly through the middle and at right angles. A procedure for this is described in par. 4.7.

Fig. 4.13: The separation sheet.

Fig. 4.14: The other parts of the seal.

index page