There are many good print and online references for lacing bicycle wheels. Many will defer to Sheldon Brown, Brandt's Bicycle Wheel Book, and others. However, a lot of the points and details with a conventional bicycle hubs don't necessarily apply to hub motors with their much larger flange diameter.
Number of Crosses
Spokes are crossed so that the tensioned spokes have a tangental force component that can transmit torque between the hub flange and the rim. For regular bicycle hubs, the spokes will usually cross over each other 3 times in order to achieve sufficient angle. On a hub motor, the flange is so large that a single cross will usually result in an even larger angle than triple cross in a regular hub. We've seen many times where people have suggested that hub motors should also be laced 2 or 3 cross to be sufficiently "strong", and this is rubbish. A double cross lace on hub motor looks very cool, but it is an extra challenge to lace and results in spoke nipples coming through the rim at steep angles.
The radial or 0-cross lacing pattern is in vogue for stylish looking front wheels. It is by far the easiest pattern to lace, but in order for a torque to be transmitted between the hub and the rim, a certain amount of torsion has to occur between the hub and the rim in order for the spoke tension to have a tangential component.
With conventional rear wheels, this torque comes from the pulsating pedal forces from the rider, and resulting flex can cause fatigue wear of the spokes and loosening of the nipples in pretty short order. Conventional wisdom is that you should never ever use radial lacing on rear wheels or in any application where there is a torque transmission between hub and rim (such as a front wheel with disk or drum brakes). However, with a hub motor, the flange diameter is so much larger that the relative amount of twisting to cause a torque transmission is far less:
It used to be that all hub motors, even the 16" and 20" ones, would use a cross lacing pattern, often nearly butchering the spokes in the process. Then in early 2008 we started seeing most manufacturers switch to a direct radial lace with their 20" wheels. At the moment all the manufacturers we deal with (eZee, Crystalyte, and Nine Continent) do this. In spite of the popular wisdom not to use radial lacing on drive wheels, empirical experience has been that this isn't really an issue with the large hubs in small rims.
Hub motor wheels are typically supplied by the manufacturer with 13 gauge (2.3mm) spokes. Meanwhile, the spokes available from bicycle stores are almost always the thinner 14 gauge (2.0mm). Some manufacturers such as Crystalyte had so many issues with even their 13g spokes breaking that they went to very thick 12 gauge (2.5mm) spokes.
|Crystalyte:||2.5mm (12 gauge)|
|Most other brands:||2.3 mm (13 gauge)|
|Standard bicycle wheel:||2.0mm (14 gauge)|
In principle, a heavier spoke should be stronger and less likely to fail, but when the failure mode is fatigue from bad fitting, then even that won't help much. There are many downsides to fatter spokes. With 12 gauge, that means that any rim which you use will need to have the eyelets drilled out to accommodate the large 12 gauge nipple, and you'll have a hard time sourcing replacement spokes, nipples, and spoke wrenches anywhere. Thirteen gauge spokes are used on some tandem and touring bicycles and the 13g nipples can fit through the hole in a regular rim, so there are fewer issues. But even then 13 gauge equipment is still pretty rare, and although many bike shops can make a custom length 14 gauge spoke, few will be setup to do this in 13 gauge.
In practice, a well built wheel using 14 gauge spokes is more than strong enough to withstand the forces in even powerful hub motor setups. But most hub motor flanges are drilled out to accomodate 13 gauge or thicker spokes, and the small head from a 14g spoke won't seat well and can sometimes even can pop through.
One of the best solutions to this is to use butted spokes that are 13 gauge at the bend, where the fit through the hub flange, and taper to 14 gauge at the threaded end. Then they can be cut and threaded to any length by better equipped bike stores, they use standard 14 gauge nipples, and they still seat properly in the holes in the hub motor flange.
Seating at Bend
A common reason for spokes to fail on hub motors isn't because the motor puts extra strain on the spokes, or because the spokes aren't a thick enough gauge, it's because of fatigue failure from spokes that aren't held snug against the flange. If the spoke bend radius is too large or too far from the head, then it can flex up and down at the bend with each wheel rotation, eventually causing it to crack and fail.
This problem has been legendary with overseas built hub motors, and we had some Crystalyte shipments where about half the customers would experience spoke breakage on a recurring basis. Ideally the distance between the head and the bend in your spoke will match the thickness of the hub flange, and you won't have problems. But if not, there are two ways to address the situation. One is to insert a washer under the spoke head. The second way is to lace the wheel in an over/under pattern, such that the spoke tension compresses the bend part of the spoke into the flange.
It is standard with rear bicycle hubs that the spokes ont he right side are shorter and more vertical than the spokes on the left, because of the space that the freewheel takes up. On a hub motor, this is no different, except that the flanges are often closer together, and this results in the freewheel side spoke becoming nearly vertical to center the rim. This can be addressed to some degree by lacing the wheel such that on both the left and right set of spokes, all of the spoke heads come out on the right side of the flange. This will offset the base of all spokes to the right by the flange width.
On symmetric hubs like Nine Continent or pre-2008 Crystalyte, the dishing issue can be pretty bad and this step is required. On asymetric hubs like eZee and BMC, the amount of dishing required is a fair bit less. In 2008 Crystalyte changed the design of their rear motors so that the flanges were so asymetrically offset that their rear motor requires no dishing at all.