|"Thinking Outside the Four-Wheeled Box"
If there was some way of extracting all of the great tech support, diagnostic, and troubleshooting info from our 68,000 email messages we've sent our customers to date, well it could fill up an enclopedia or two. In the meantime, we have been producing some stand alone PDF documents to outline the more common issues we help people with, and have listed them here for your reference:
|Instructions on how to test for a correct throttle signal in your ebike.|
|Guide to using and crimping both Anderson and JST-SM type connectors.|
|Guide showing the common connector hall and phase wiring pinouts between the various eZee, Crystalyte, Nine Continent, and Infineon controllers and motors that we carry.|
|How to test the hall sensor signal voltages in your hub motor system.|
|Instructions on how to open up both geared and direct drive hub motors for inspection and repair.|
|How to replace the hall sensor(s) inside a geared or direct drive hub motor.|
|Testing a controller for blown mosfets without opening it up.|
This is possibly the single most frequent problem that we've had to deal with selling motors here on the West Coast. Whether the motor is ridden regularly or even occasionally in wet weather, water manages to get inside the hub one way or another. At first this may have little or no effect, but over time the water starts to corrode the iron and aluminum inside the motor, and this corrosion makes the water electrically conductive. Even though the hall sensors are potted in epoxy, the wetness eventually undercuts whatever sealant is in place and effectively bridges between the hall signal line and the ground return. The hall signal is pulled up to 13V by a 30K resistor on the Crystalyte controllers, even a fairly weak bridge to ground can pull this signal down so that instead of toggling from 0V to 13V, the signal instead only goes from 0V to say 6V, and this is not enough to trigger the controller hall effect decoding logic. The result of a single hall effect signal failing is that the motor spins with a very heavy jerky motion, draws a lot of current, and is generally quite unpleasant to ride.
People often assume incorrectly that this is the consequence of a damaged hall sensor but this is not usually the case. The sensor itself is just fine but the wetness has interfered with the signal that it outputs. Once the motor has been opened and thoroughly dried out again it will work fine, though after this has happened once it is generally more succeptible to developing the problem than it was originally. An apparent solution would seem to be sealing the motor against water entry, but at least from our experience this is not the case. A disproportionately high number of these failures occured on motors that had had the wire entry to the hub thoroughly sealed with silicone or other goop. Rather than keep the motor dry, water inevitably leaks in and is then trapped inside. Unless you can be sure that you have sealed 100% against water entry, it would be better to leave the wire port as is and to always store the bike indoors, preferably in a warm place, after riding in the rain so that the hub innards get a chance to dry out.
We have seen a few cases where there were hall effect problems that were not apparently related in any way to water damage. For instance, the motor works just fine but after 10-20 minutes of use one of the hall sensors doesn't toggle fully, perhaps related to the motor warming up. Or more commonly the motor leads got shorted to one of the hall signal lines this will fry the sensor. In these cases, replacement of the hall chips inside the motor is required.
The entire torque of the hub motor is transferred to the bicycle frame via a fairly small 12 or 14mm axle. The axle flat slotted in the bicycle dropout is not in and of itself strong enough to restrain against this torque. If the axle nut is loose or not seated properly the torque will be sufficient to cause the axle to spin, spreading open the dropouts and wrapping the motor wires around it, sometimes shearing/shorting the motor leads in the process and sometimes spinning the wheel right out of the bike frame.
A common cause of this failure is when people install front hub motors on forks designed for quick release wheels. These forks have 'lawyer lips' to prevent regular front wheels from falling out if the quick release mechanism comes loose, but these lips interfere with the much larger axle washer and nut so that they don't sit flat, and as a consequence don't hold the motor in properly. For these forks, either the lawyer lips need to be filed or ground off, or a small spacer washer installed under the nut so that the large diameter nut clears the lips.
We've seen a few instances where front suspension forks have cracked at the dropout, but this happened during or due to improper installation and had nothing to do with the torque or extra weight of the hub motor. In the case below, there was a small fillet on the axle shoulder that caused a spreading action of sorts, and the brittle cast aluminum dropout on this fork cracked in half as the axle nut was fully tightened. In general, forks that have a welded metal plate for the wheel to slide into are much stronger in this regard.
Spokes sometimes break on regular bicycles. They break a lot on touring bikes, stunt bikes, cargo hauling bikes, and of course electric bikes, but they don't need to. In the case of Crystalyte hub motors, what we usually see for regular riders is that the wheel holds up just fine for the first few months, and then the spokes start to break right at the elbow. Perhaps one or two a year, or for some people it can be several spokes a month. The reason for this fatigue failure as has been documented elsewhere in the bicycle industry is that the bend radius on the spokes is too large, so they don't sit tight against the hub flange, and at every wheel rotation the bend flexes a little, and that flexing wears it out. A simple fix to this problem is to install a small washer under each spoke head so that it is held quite snug against the hub, and on wheels we've built this way still using the 13 gauge Crystalyte spokes, the spokes have never broken.
The newer rear Crystalyte motors with the offset hub have a thicker flange, and so we're expecting that washers won't be necessary in this case. Also, the larger 500 series motors use quite heavy 12 gauge spokes which is a less elegant way to solve the problem for most people (though these spokes still manage to break for some). In general, the standard 14 gauge spokes used for bicycles are more than adequately strong even for 500 series motors, so long as they are a quality brand.
While it is good to keep the spokes nicely tensioned, if they are too tight it can pull in and crack the rim on either side of the eyelets. We've seen this on several occasions.
Another consequence of overly tight spokes can be cracking at the hub flange. This did not used to be a problem, but in the last year Crystalyte went from steel to aluminum for the flange, which is naturally not as strong. The rear hub was redesigned somewhat in late 2007 with a thicker flange that should hopefully handle the forces better. This problem is more likely to occure in radial-laced wheels than those with a cross pattern.
The aluminum side covers of the Crystalyte hub motors aren't engineered by NASA. Sometimes there are voids or weak spots in the castings and sometimes this can have consequences. We have seen a few instances where riders have managed to shear off the threaded shoulder that the freewheel screws onto. This usually happens when they are standing pedalling in the easiest gear, putting a maximum amount of torque on the freewheel. Because the threaded portion has a sharp inner corner rather than a proper fillet, it experiences a high stress concentration and is prone to fatigue here as a result. The 500 series rear motors use steel threads and we've never seen one of those fail.
The Crystalyte motors use sealed ball bearings which should provide a good many years of maintenance free service. But because both the Crystalyte axles and bearing housings appear to be hand machined, the tolerances aren't always tight enough and the ball bearing gets crushed a bit when pressed into place. A damaged ball bearing has a noticable grinding feeling as the wheel is rotated, and it can cause enough additional drag that the no-load current of the motor is abnormally high, upwards of 2-3 amps. It is often necessary to replace not just the ball bearing, but the side cover as well, since that is frequently the origin of the problem.
Though quite uncommon (we've seen it once) it is possible for the axle to actually get loose within the stator. The axle is a separate piece of splined steel that is press fit into the cast aluminum stator frame. Normally quite a large hydraulic press is need to push it in or out, but if the tolerances were off the stator can work itself loose with the consequence of rubbing and vibrations and shorted/twisted wires inside the hub.
We always tell people to really do the axle nuts up as tight as possible to prevent axle spinout. However, Crystalyte turns the threads on a lathe rather than with a die and sometimes the tolerances are off. The nuts can be hard to spin on, or they can be so loose that the threads strip when the nut is tightened.
This is hardly ever seen with new motors, but if the motor is ever taken apart and put back together again it is not uncommon for there to be rubbing between the stator and the hub. With a bit of trial and error rotating the side cover plates in different orientations you can find a position that doesn't rub, but the best solution is to put aligment marks on the side cover plates before you take them off so that they can go back on in exactly the same way.
Having supplied hundreds and hundreds of Crystalyte motors, we've been quite surprised at how well the windings have held up. In fact, even in situations where the motors were used well outside of their rated currents and voltages the motors windings have never been trouble. That said, there have been a few of the Crystalyte DC motors that have had or developed shorts between the windings. You can tell when this happens because there is considerably more drag when rotating the wheel than normal, and the unloaded current draw is higher as well.
When controllers fail and you see a puff of smoke it is almost guaranteed to be a blown mosfet. There are two things which usually cause a mosfet to blow, #1 is too much voltage, and #2 is too much current. Unless you connect say 72V battery to a 36V-rated controller (usually has 55V rated mosfets), then it is almost certainly too much current that caused the damage. This will happen very predictably when smaller controllers are used with motors that have low winding resistance, say using a 36V 20A controller with a 5303 hub motor. What happens is that even though the controller limits the battery current to 20A, there is nothing that limits the output current to the motor, and because of current multiplication through PWM this can be many times higher than the current you see being pulled from the battery. You can get a first order estimate of this current by looking at the power input to the motor controller at a stall, about 720 watts on a 36V 20A unit. All of this is getting dumped as heat in the motor windings (P = I^2 R), so the motor current is given by I = SQRT(P/R). With a 5303 the winding resistance is just 0.11 ohms and so in this setup you would end up with some 80 amps through the motor, spelling disaster for the single mosfets in the 20A controller. With a 408 motor, the winding resistance is 0.57 ohms, so the stalled motor current (SQRT(720/0.57)) is a mere 35 amps, which is no problem for the transistors.
The current that a mosfet can handle before frying depends on how much ON resistance the mosfet has and how well the mosfet casing is heatsunk. In general, higher voltage mosfets have higher ON resistance, and so are more likely to blow from too much current. This was certainly the case with the pre-2008 Crystalyte controllers. Their 36V controllers had mosfets with 8mOhm of resistance, while the 72 controllers had nearly twice this at 14mOhm, and their 48V controllers were nearly as bad at 13mOhm. The result was that these higher voltage controllers were much more prone to failure since they would heat up more for the same current. People often mistakingly thought that the bigger voltage rating meant it would be more robust, which was not the case.
When the insides of a Crystalyte motor controller get wet all kinds of spurious things can happen. Usually the effect is a chugging motor similar to when the hall signals are compromised, because they are, only now it's due to the inside of the controller. But it is also possible for the controller to engage in a full throttle ON position, and in other cases it simply won't work, with no output at all. Generally speaking, once the controller has been dried out it will work fine again, but because it was wet when voltages were present, there will usually be traces of corrosion all around the circuitboard which will make it more prone to weird behaviour in the future. It is pretty difficult to properly seal the Crystalyte controller boxes, with the buttons and key switches and many wires and all. However, it is not that difficult to open them up and apply waterproof conformal coating directly on the PCB. This makes the controller much more resilient to wet weather.
The ON/OFF push buttons in the Crystalyte controllers aren't the best. Sometimes they stay on, sometimes they stay off, and sometimes they need a couple pushes to work right.
The flat plastic connectors used for the throttle line, the CycleAnalyst line, and in older versions the hall sensor and ebrake lines; those connectors are inexpensive and surprisingly reliable given how cheap they look. But they can be the source of problems if not attached properly. Most instances of a broken wire occure when one wire in the multi-strand cable is shorter than all the others, so if the connector gets tugged all of the force falls on this single wire until it breaks. Properly done connectors of this type have glue holding all the wires together as a single cable, so that no one of them holds the load.
Another possible problem with these flat multi-pin connectors occures when one of the pins hasn't properly clicked in place and gets pushed back into a retracted position. This is usually seen on the male side of the connector, and it can cause an intermittent or entirely open connection. On the throttle connector, if either the 5V pin or signal pin aren't making contact then the bike simply won't take off. On the other hand, if the break occures in the ground line then the throttle will engage as though it was fully on, causing the ebike to take off the moment it is powered up.
It also happens that the anderson connector pins are not properly clicked in place as well. This usually occures because the pin is not crimped properly and has a wide mushroomed section that prevents it from sliding all the way in the housing, as depicted below. An anderson pin that pulls back like that will cause intermittent contact and/or heating, arcing and melting of the plastic housing. If the interuption is on one of the motor leads then the wheel will only have a fraction of the power and it won't feel as smooth. If it happens in the battery leads then naturally this causes loss of power to the controller.
Basic troubleshooting of motor controller
So your ebike stopped working? Most of the time this is from a loose connection on either the hall effect or throttle cable, an ignition key toggle switch that's become flaky, or something similar. Another possibility is that a pin on the inline throttle or hall effect connector got pressed into the housing and does not make contact with the mating side, so even though it looks like things are plugged in, they are not. If this is the case, you can simply pull the pin back out with a needlenose plyer. However, sometimes there are failures in the electrical hardware itself, and it is usually easy to find the source of the problem.
Check the power
Make sure that the motor controller is actually getting a voltage from the battery. With a multimeter, you should see the full battery voltage (typically ~39V) on the end of the battery cable that plugs into the controller. When you plug this into the controller cable, there should be a spark on the contacts as the inrush current charge up the capacitors.
The next thing to check is that the throttle is behaving. The inline wire connector does not provide access for probing unless you cut the heatshrink away, so it is easier to remove the controller cover and measure straight on the circuitboard header. To do this test, make sure that the controller is powered up, the throttle is plugged in, but that the motor cable is disconnected.
Just a note of caution here: Keep all metal tools and objects away from the controller when it is uncovered, since the exposed leads are live and an accidental short could easily fry chips and components, and if the controller wasn't working before, it certainly ain't gonna work after being zapped in the wrong place with the battery voltage.
Measure the voltage between ground and the throttle signal line as depicted below. It should be approximately 1V at no-throttle, and an increase to 4V when the throttle all the way on. If you don't see a voltage here, verify that there is 5 volts actually going out to the throttle.
If there is 5V but still nothing on the throttle line, then either the inline connector is not making contact on one of the pins as described above, or the throttle is faulty and needs replacing.
Hall Effect Signals
When there is a fault on one or two of the hall signals, it will manifest itself by very jerky motion of the motor. If all three hall signals have a fault, as for example would happen if either the 12V or Gnd line to the sensors were interupted, then the controller would simply not work.
Check the hall signals with the controller powered up, the motor HALL Sensor cable plugged in, and with the actual motor leads still disconnected. Measure the voltage between the Ground pins and each of the coloured hall signals. While doing so, rotate the wheel slowly by hand. Each signal should alternate between 0V and 12V.
If none of them are working, then chances are that there is a break along the 12V power to the hall sensors. If only one or two are not working, then it is possible that there is a break along that signal wire, or the hall sensor was damaged from the heat in the motor or corrosion. This latter situation is uncommon but not unheard of, and the usual remedy is simply to use a pedal first controller as they do not use the sensor cable. In the past, the somewhat delicate hall wires were prone to breaking where then entered axle of the hub motor. But with newer designs there is better sheathing of the cables and a plastic shield which protect the cables against abrasion and wear from the sharp axle shoulder.