Severe Motor Chattering Problem?

[Ken McNeill]

I have a few bikes I built that I use for e-bike tours. When I built them in 2019, they worked great. But lately, one by one they have begun to severely chatter under load. The stuttering start has really been hard on clutches, even shearing an axle key. 100-200 miles if that. I have changed motors & controllers with no effect. Once it’s rolling, it’s silent and perfect.

I’ve attached a video of what I’m experiencing. I just replaced the clutch/gear assy in this motor. Could it be noisy electrical power from the controller? Hall sensor issue? Is it possible the CA3 is causing this? Could it be a noisy throttle signal?

CA3
40 amp Infineon Motor Controller -
EM3ev 48v 16ah Battery
MacMotor - Ebike Hub Motor For Sale | MAC

Video link -

 

[eee291]

Hall sensors maybe broken.

 

[Ken McNeill]

Hall sensors maybe broken.

I went ahead and performed a Hall Sensor & Motor Coil Test and it checked good. Video Below.

Then I went ahead and performed a motor controller output test. This did not look normal. Possible issue. With the throttle applied steady… out stops.

Here’s a video I did a couple years ago on how it should look.

 

[eee291]

I've never seen this behavior, I'm guessing the Hall combination is correct or does not need to be with this controller in order to work.
The throttle can be tested with a multi meter, while powered measure between ground and signal,
should be around 1v idle and roughly 4.5v at wide open throttle.

 

[Ken McNeill]

I've never seen this behavior, I'm guessing the Hall combination is correct or does not need to be with this controller in order to work.
The throttle can be tested with a multi meter, while powered measure between ground and signal,
should be around 1v idle and roughly 4.5v at wide open throttle.

I also get a throttle input signal on the CA3 indicating correct voltage between 1v - 4v, so I’m leaning towards a bad controller.

 

[neptronix]

What happens when you take the cav3 out of the equation?

 

[Braddudya]

I also get a throttle input signal on the CA3 indicating correct voltage between 1v - 4v, so I’m leaning towards a bad controller.

I have had EXACTLY this symptom using a my1020 motor on a Pitbike. The chinesium controller that came with the motor turned out to be the culprit. It said it was a sine wave controller but I assumed it was trapezoidal. I still don't know if the controller was bad or if it just was a POS.

have you changed controllers? It was a problem that solved using a Flipsky 75200 controller and it's a world of difference.

 

[Ken McNeill]

What happens when you take the cav3 out of the equation?

I haven’t tried that. I can swap the CAV3 easily, so maybe I’ll give that a shot first. Might rule out a CA issue?

 

[Ken McNeill]

I have had EXACTLY this symptom using a my1020 motor on a Pitbike. The chinesium controller that came with the motor turned out to be the culprit. It said it was a sine wave controller but I assumed it was trapezoidal. I still don't know if the controller was bad or if it just was a POS.

have you changed controllers? It was a problem that solved using a Flipsky 75200 controller and it's a world of difference.

I haven’t tried other controllers, but when they were new they worked great. Slowly, over time the motors have been getting noisier. I have seven bikes and 2-3 are starting to have similar issues. I figured the MC’s would either work.. or not work. Not some where degraded in between?

 

[neptronix]

I haven’t tried that. I can swap the CAV3 easily, so maybe I’ll give that a shot first. Might rule out a CA issue?

There's a number of settings on the CA related to soft start/throttle ramping that require special attention to get right on MAC/BMC motors, i think grin has some articles on them.

Just removing it temporarily and using a regular throttle should rule that out.

Is your 40A infineon an original one from em3ev?


Soundwise this sounds like a gear or clutch issue. However, controller issues/controller tune can produce these same kinds of sounds.
Hard to diagnose; swap parts between bikes if possible to help rule out what the problem component is.

 

[Ken McNeill]

There's a number of settings on the CA related to soft start/throttle ramping that require special attention to get right on MAC/BMC motors, i think grin has some articles on them.

Just removing it temporarily and using a regular throttle should rule that out.

Is your 40A infineon an original one from em3ev?


Soundwise this sounds like a gear or clutch issue. However, controller issues/controller tune can produce these same kinds of sounds.
Hard to diagnose; swap parts between bikes if possible to help rule out what the problem component is.

I’ve used the ramping quite extensively 1vps-3vps, based on rider skills. I hope that didn’t cause the issues. The bikes were pretty ‘hot’ when I first built them, so I toned them down a bit through ramping. I’ll have to look to those articles? They are all set at 3vps currently.

These are the 40A Infineon’s fro EM3ev. I agree it sounds mechanical, but I’ve laced in brand new motors and they did the same thing. After that controller test, I think I’m narrowing the possibilities. Hopefully.

 

[neptronix]

TBH i built a MAC bike with similar power and reaaaaaaaallllyyy had to tune the CA3 to get it to not stutter/jolt right out of the box.
I think it comes pre tuned more for DD hubs.
Honestly the MAC has lots of poles and is difficult to drive in the first place.

Ebikes.ca may have information on a special tune for the gmac with their cav3... check their page.

With the infineon controllers i would program in a 2.25:1 phase amp to battery amp ratio and this would lower the clutch beating initial torque on those controllers.
Example: 36A x 2.25 = 81A batt, 36A

I haven't bought an infineon for 6 years so i don't know how em3ev tunes them today, but i believe they were tuned more like a 2.75:1 which gave a little too strong initial torque and i knew this would eventually beat up the clutch/gears.

If you haven't reprogrammed your infineons then the cav3 may be fighting this strong initial torque.

 

[amberwolf]

I haven’t tried other controllers, but when they were new they worked great. Slowly, over time the motors have been getting noisier. I have seven bikes and 2-3 are starting to have similar issues. I figured the MC’s would either work.. or not work. Not some where degraded in between?

My bet would be capacitors in the controller. They degrade over time, especially with heat. The big electrolytic cans have a limited lifespan even when they're good ones run within their specs.

Sometimes the problem is visible, such as in the images here:

Badcaps Electronics Repair Forum & Schematic Search - Badcaps

Badcaps Electronics Repair Forum & Schematic Search

badcaps.net

My next bet is just the batteries aging and sagging more over time. So under higher loads they sag enough for the controller (or the CA) to begin limiting power and causing stuttering/chattering.

If it's the CA doing any limiting, you'll see a capitalized letter on the diag screen, which is one left arrow press from the main screen you get when turning the CA on.

I haven’t tried that. I can swap the CAV3 easily, so maybe I’ll give that a shot first. Might rule out a CA issue?

Has anything been changed in the settings, or firmware? If not, it's unlikely to be the CA itself.

If settings or FW have been changed, first just set the CA throttle mode to BYPASS, which feeds your throttle output straight to the ocntroller unmodified. So if the CA *settings* are causing a problem, this will prevent that.

Alternately you can physically bypass the CA by connecting your throttle directly to the controller's throttle input.

 

[Ken McNeill]

My bet would be capacitors in the controller. They degrade over time, especially with heat. The big electrolytic cans have a limited lifespan even when they're good ones run within their specs.

Sometimes the problem is visible, such as in the images here:

Badcaps Electronics Repair Forum & Schematic Search - Badcaps

Badcaps Electronics Repair Forum & Schematic Search

badcaps.net

My next bet is just the batteries aging and sagging more over time. So under higher loads they sag enough for the controller (or the CA) to begin limiting power and causing stuttering/chattering.

If it's the CA doing any limiting, you'll see a capitalized letter on the diag screen, which is one left arrow press from the main screen you get when turning the CA on.


Has anything been changed in the settings, or firmware? If not, it's unlikely to be the CA itself.

If settings or FW have been changed, first just set the CA throttle mode to BYPASS, which feeds your throttle output straight to the ocntroller unmodified. So if the CA *settings* are causing a problem, this will prevent that.

Alternately you can physically bypass the CA by connecting your throttle directly to the controller's throttle input.

Thanks! I concur with everything you said. Nothing has changed with CAV3 FW or programming. Just a slow degradation resulting in noise and then chattering, resulting in quick clutch failure from the hammering effect. Batteries are in great shape.

The MC is mounted in an unventilated housing which may have added to potential heat degradation. They're mounted in the custom fiberglass tank assy.

My gut tells me I need to order a bunch of MC’s, but I’ll start with one to see where I’m at.

 

[Ken McNeill]

TBH i built a MAC bike with similar power and reaaaaaaaallllyyy had to tune the CA3 to get it to not stutter/jolt right out of the box.
I think it comes pre tuned more for DD hubs.
Honestly the MAC has lots of poles and is difficult to drive in the first place.

Ebikes.ca may have information on a special tune for the gmac with their cav3... check their page.

With the infineon controllers i would program in a 2.25:1 phase amp to battery amp ratio and this would lower the clutch beating initial torque on those controllers.
Example: 36A x 2.25 = 81A batt, 36A

I haven't bought an infineon for 6 years so i don't know how em3ev tunes them today, but i believe they were tuned more like a 2.75:1 which gave a little too strong initial torque and i knew this would eventually beat up the clutch/gears.

If you haven't reprogrammed your infineons then the cav3 may be fighting this strong initial torque.

I’ll see what I can find on that tune and compare to my configuration. Thanks a bunch.

 

[amberwolf]

Thanks! I concur with everything you said. Nothing has changed with CAV3 FW or programming. Just a slow degradation resulting in noise and then chattering, resulting in quick clutch failure from the hammering effect. Batteries are in great shape.

The MC is mounted in an unventilated housing which may have added to potential heat degradation. They're mounted in the custom fiberglass tank assy.

My guy tells me I need to order a bunch of MC’s, but I’ll start with one to see where I’m at.

As long as the controllers have not actually blown up, then if it's failed or degraded capacitors, you can just replace those easily enough in the average controller. It's usually just the electrolytics on the main power / FET bus, the largest ones. Sometimes it's the ones on the LVPS / input as well, and rarely it's also the ones on the lower voltage rails.


I recommend Rubycon, Panasonic, etc brand name caps, with 105C or higher temperature rating (the higher the rating, the longer they'll last even at lower temperatures), at at least the voltage rating they are already at, or the next one up, and same capacitance. If you get the caps from Mouser or Digikey, Farnell, etc., then you are much less likely to get counterfeits so they'll actually do the intended job for as long as they should.

 

[larsb]

Would the behaviour seen on the tester be caused by a weak cap? It looks like there is a phase/FET combination missing in the cycle.

Weak FET or FET driver mayhaps? Or a poor connection, corrosion? These controllers aren’t waterproof.

 

[Ken McNeill]

As long as the controllers have not actually blown up, then if it's failed or degraded capacitors, you can just replace those easily enough in the average controller. It's usually just the electrolytics on the main power / FET bus, the largest ones. Sometimes it's the ones on the LVPS / input as well, and rarely it's also the ones on the lower voltage rails.


I recommend Rubycon, Panasonic, etc brand name caps, with 105C or higher temperature rating (the higher the rating, the longer they'll last even at lower temperatures), at at least the voltage rating they are already at, or the next one up, and same capacitance. If you get the caps from Mouser or Digikey, Farnell, etc., then you are much less likely to get counterfeits so they'll actually do the intended job for as long as they should.

I may try that as MAC Motor wants $199 each + expensive shipping from China. I paid $90 each from EM3ev. Does anyone know of a more local USA source, or a better replacement? I really don’t need 40A as 25A would do.

 

[Ken McNeill]

Would the behaviour seen on the tester be caused by a weak cap? It looks like there is a phase/FET combination missing in the cycle.

Weak FET or FET driver mayhaps? Or a poor connection, corrosion? These controllers aren’t waterproof.

They are never exposed to water and are in an enclosed housing.

 

[Ken McNeill]

I did perform some resistance tests between batt-in +/- and FET output phase wires. No shorts. Motors run fine once spinning, but chatter under load. Found many Controller Tester videos online indicating if the light doesn’t continually rotate, controller is bad. Don’t have a ton of knowledge on the internal circuitry of these motor controllers, especially without a schematic. Do you think Caps could cause this issue?

Since I operate a tour business, and having working bikes critical, I may have to just order them @ $199 each - 8 pcs minimum order + $200 shipping. Ouch!

 

[amberwolf]

The caps are there to smooth out current flow, so fi they aren't working right the controller may not operate as expected when there is high current demand by the motor, but work fine when the current drops enough.

If you're willing to open one of the problematic controllers, caps are easy enough to check visually for the worst-case problem; the site linked previously has a lot of examples of good vs bad. You might not even have to take the board out, just take an end cap off and look down in there in good lighting (direct sunlight is best), to see if any caps are swollen or otherwise visibly failed.


It's also possible the caps have actually broken off the board if they werent' glued down, like this one:

Can't believe it still works: ancient Crystalyte controller edition

So... the other day the funky old four-wire throttle that came with my old Crystalyte 5305 and 35A controller (bought direct from Crystalyte in China in 2003 or 2004) soiled its trousers, and the bike stopped working suddenly, without warning. I cut the controller loose from my cargo bike so I...

endless-sphere.com

In either case, it's a fairly quick matter to replace the caps if they are a problem. If you have a local electronics store they may carry usable replacements (just take the controller and caps with you to make sure the new ones will fit).

If you need to test the caps, the easiest way is with a meter with that specific capability; most of the cheap ones only go up to a couple hundred uF though, while most of the big caps in controllers that fail are 680-1000uF or more.

 

[Ken McNeill]

The caps are there to smooth out current flow, so fi they aren't working right the controller may not operate as expected when there is high current demand by the motor, but work fine when the current drops enough.

If you're willing to open one of the problematic controllers, caps are easy enough to check visually for the worst-case problem; the site linked previously has a lot of examples of good vs bad. You might not even have to take the board out, just take an end cap off and look down in there in good lighting (direct sunlight is best), to see if any caps are swollen or otherwise visibly failed.


It's also possible the caps have actually broken off the board if they werent' glued down, like this one:

Can't believe it still works: ancient Crystalyte controller edition

So... the other day the funky old four-wire throttle that came with my old Crystalyte 5305 and 35A controller (bought direct from Crystalyte in China in 2003 or 2004) soiled its trousers, and the bike stopped working suddenly, without warning. I cut the controller loose from my cargo bike so I...

endless-sphere.com

In either case, it's a fairly quick matter to replace the caps if they are a problem. If you have a local electronics store they may carry usable replacements (just take the controller and caps with you to make sure the new ones will fit).

If you need to test the caps, the easiest way is with a meter with that specific capability; most of the cheap ones only go up to a couple hundred uF though, while most of the big caps in controllers that fail are 680-1000uF or more.

Great information! I will pull it apart tomorrow and visually inspect tomorrow. My VOM is on the higher end so I believe it will test caps in that range.

I do think I found a source for replacement controllers, but I like better understanding what is happening, and why.

Thank you!

 

[amberwolf]

Don't forget to discharge the caps before putting the cap meter on them--some meters have no protection against the high voltages that can be present on these.

If you find caps have failed (especially if swollen or popped), then the most likely cause is heat, especially if they are low-temperature-rated caps, or no-name "brands". The fix for that is to provide some way to get the heat out of them, and the "best" is to connect the controller's heatsink to the outside world, either directly by putting it into the airflow, or indirectly by bolting that heatsink to one that does stick into the airflow (depends on the environmental conditions which one is possible, and exactly how you can do it).

If it's not caps, I'd be surprised, but the next part that "ages" badly (other than connections, wires, solder joints, crimps) is the big fat resistor from teh batery positive to the LVPS input; sometimes it changes value from the heat, and this changes the votlage present at the LVPS input--if it makes it lower instead of higher, it *might* cause the LVPS to be unable to deliver proper voltage to the FET drivers or even the MCU during high enough current draw. If this is the case it would be worse the lower the battery voltage is (the more discharged the battery is)--if that's not the case, it's unlikley to be the resistor.

 

[Ken McNeill]

I popped off one end to take a peek. Caps looked normal and there was absolutely zero electrical smell. Like new. My e-bikes are run at 8-12 mph on level ground. Motors barely get warm, let alone the motor controller. I don’t think high heat is the root cause. I think it’s some sort of failure in the timing/sync circuit.

The good news is EM3ev had 9 units in their old inventory. They made me a deal on the lot of them.

 

[amberwolf]

It doesnt' have to be high heat, and the caps don't have to have visible failure indications. It just has to be enough heat over a long enough time, plus normal aging, to degrade the capacitors enough to no longer filter correctly, so that the previously described problems occur.

Might not be the caps, but they're really the only part (besides that resistor) that ages and fails, in these controllers. There's nothing else I can think of (besides connections and wires that could corrode, or solder joints crack) to fail over time the way you see happening, and all of those other things would cause random types of failures, not the same one in every controller. The caps *would* cause the same type of failure (assuming the same type of cap failure, which is likely given the same conditions and same controllers, etc), so they are the most likely cause, and the thing I'd still check if I had this issue.

I suppose the MCU or gate drivers or even FETs could fail, but the MCU is unlikely to fail the same way in every case, so it's probably not the cause, and the gate drivers or FETs are more likely to just plain fail than to gradually do so--usually on whichever channel is more stressed than the others for whatever reason (including part stress that happened before they were installed in the controller at the factory, like handling damage or ESD).

The only other things that should cause the issue you see is controller or CA settings changes, but again, these aren't something that just changes over time (especially not the same way on every controller), they'd have to be something changed on purpose gradually over time, whcih seems unlikely (and it would be something youd' expect to notice the association between the increasing problem and the settings changes being made). (if it was an accidental change, it wouldn't happen the same on every system, and wouldn't accidentally change the same setting a little more each time, either, so no gradual change, just a sudden "whoops" and then there's a problem).

If the system shows the same failure without the CA between throttle and the controller, then that completely eliminates the CA as a cause.


If the batteries were aging and dropping severely in voltage under load, the controller could shutdown as it dropped below it's own LVC, but as noted in a previous post about the LVC resistor, that would not be the same at all states of charge of the battery--it would be worse the emptier the battery gets, so if this problem does not change from full to empty, it's not that either. (if the battery were dropping below *its* LVC, it would power off the whole system, and you'd notice that, so it's unlikely to be the cause).