twitchy throttle response problem

[kZs0lt]

I have a 12 FET Infineon controller from ecrazyman, this is probably the soft start version. And an ecrazyman hall effect thumb throttle.
The throttle response is sometimes like intermittent. The throttle response is not linear, at the end the current rise is steeper. But the problem is, it seems like I don't get the same power at the same throttle position. Worst is when I am ascending a like 8-10% slope(BMC motor), I can't stay smoothly at 15km/h for example, as when turning throttle, at a point I get a burst, of around full throttle(have no ampmeter), till get to like 18km/h, then it backs off power and I slow down to like 12km/h and it kicks in again, there is not a slow ramp. It's like not having anything above mid current, but only full. I am using 66v, but it's the same on lower voltages too. I am also using a 4 step switch programmed to 30%, 50%, 70%, 100%, But I get the same effect on every setting, it seems like it is not connected to a certain speed or voltage, but more like to load. If I want to accelerate very slowly on flat, I can get to full speed smoothly, but once I want to do it a little quicker, somewhere it would kick into full current the throttle being only about 70% WOT.
I'm not sure what the throttle is supposed to alter current, voltage or duty cycle, but this is very annoying, with kick-ins all the time while I keep trying to ride with smooth handling on throttle.

Is this unusual, or others have encountered the same? What could be the cause, or how could this be remedied?
Could this be caused by the resistor that I have put in series with the throttle signal earlier(see here). I'm not sure, but I think it acted the same before that. Could it be caused by thin and long wires to throttle?
What tests should I perform, to investigate further the problem. Should I try with a resistor pot instead of the hall sensor?

 

[chet]

I have an unknown controller from ebikekit.com. The twist throttle holds power output very exact.

 

[nicobie]

I've got the same controller (35a) with a 600w BMC motor and am having the same problems. Out of the 80* available rotation, only about 10 is usable. I am going to try using a CA with the pot. mod so I can adjust the amp limit on the fly. If that doesn't work, I think I'm going to run the throttle output thru the CA. Maybe that would fix it. Clyte analog 72-36v controllers are supposed to work OK but I've never tried one.

Nick

 

[jimw1960]

Apparently, those Infineon controllers don't work with the BMC motors. There are several active threads on this topic if you want to do a search.

 

[rkosiorek]

the problem may not be with the controller at all. the problem may simply be a busted throttle.

Potentiometer based throttles have a a carbon track that can be infiltrated with dirt and cause non-linear or discontinous results. the throttle could cut in and out.

Hall effect throttles can have loose magnets. since the Hall effect measures the distance from a magnet if the magnet is loose the distance measured will change unpredictably as the magnet rattles around.

a loose or intermitant connection from the controller to the throttle could also have this effect. this could be hard to spot if the conductors are broken somewhere under the insulation or between the mating contacts of a connector.

do you have another throttle you could use to test?

rick

 

[dogman dan]

Try a different throttle to start with. I have one like that that works fine with the brushed controller it came with, but put it on a brushless one, and the last 20% doesn't work. Wierd, I can't even imagine what causes that? They are all the same WE throttle.

 

[kZs0lt]

I don't have any other throttles to check with. I don't have loose magnets, for sure, it's a fairly new throttle, and I have disassembled it once, and everything looked fine. I have measured the signal and the curve seems not very linear, but continuous. In fact I think this is a controller issue(possibly related to soft start implementation), because it is not related to a specific throttle position, and if I roll very gently, it acts as supposed, it increases speed continuously and consistently, I can get from 0 to full speed smoothly without kicks, but if I try to do this with some reasonable acceleration, somewhere it would just kick in with full power. This is only an assumption, as it feels like all 20amp controller limit, have no current measurement.
What I will try is to hook up an amp meter, and check if this is the case (hitting amp limit when not WOT). When I will have some time, I will disassemble the controller and test cruise control, will check if cruise control can keep speed or amp on a ramp continuously.

 

[ZapPat]

I find that the standard PWM-controlled throttle that most all ebike controllers use give this kind of over-sensitive response, specially with higher performance setups... this is the nature of this simple controller scheme. I'm not saying that your specific problem is due to this, but I do know that my own recent setup feels kind of like what you describe.

The factors in an ebike setup that make the throttle feel more like this (over-sensitive response) are:
- Higher battery voltages
- Low battery internal resistance
- Motors with low winding resistance (usually large motors and/or ones with low voltage windings)
- Low controller (FETs+traces) resistance and low wiring resistance

In addition, having a high current limit set will make you feel the problem more. What is happening is that the minimum throttle power point (just above the coast point any given speed) and the maximum throttle power point can get very close together in ebike setups like the one above. So if you're not a heavy person and have a fairly light and high performance ebike setup (like higher voltage Lipos/A123's and a torquy motor with a 30-40+ amp limit), you will likely get that almost all-or-nothing throttle feel.

The good news is that using a motor-current based throttle control will cure this problem on higher performance systems. My own controller design is based on this type of control (as are some higher end ebike/industrial controllers), and so I am VERY eager to try this thing out on the road! My [hopefully] final revision is being assembled, and I'll have some further programming revisions to do before getting it out for road tests though.

 

[dnmun]

didn't richard recommend adding a resistor in line with the ground lead to help with this? maybe rick remembers how that works.

 

[kZs0lt]

didn't richard recommend adding a resistor in line with the ground lead to help with this? maybe rick remembers how that works.

Yes, I have a resistor in series with the hall line (about 1.5kOhm) it is for fixing my initial problem, that it cut out above 90-95% WOT. With this mod I feel I still get full sped(there is a small gap(maybe like 2-3%), when motor is still running, but won't accelerate anymore).

About my setup:
-light and torquey BMC v1 geared motor
-light battery pack (66v 2.2Ah) so not very low resistance.
-i'm not sure about winding resistance, it's probably rather high
-4110 FETs, so low resistance, but the wiring is not overkill.
-light rider (less than 65kg)

Usually using 20a limit, sometimes 30a.
I can say that the effect is more pronounced with higher voltage, and higher current of course. Around 40v feels slightly better.

The good news is that using a motor-current based throttle control will cure this problem on higher performance systems.

Motor-current based throttle seems a good choice, as it is directly proportional with torque.

Now I'm quite confused. My throttle is supposed to alter battery current, isn't it? If I turn halfway, I should get approx. half the battery current limit, isn't it?
Mine acts as if it would try to keep a certain speed(rotation) which relates to voltage(and implicitly duty cycle). If I instantly turn halfway, I get a full acceleration(of course not instantly, I have 1 sec ramp-up because of soft start) and once I get to half speed it drops current. Sometimes (on hills) it drops more than needed to maintain this speed, slows a little, and comes back for some seconds again with considerable(full?) power. I have yet to measure if this is really full power.
My question is, does throttle really manage battery current? (if not linearly, at least according to some fairly continuous graph
What does speed limits on the infineon controllers do? Why these don't help with this issue? I think there are 2 types (one SL limits speed only, the others limit current too?)
I really wonder if cruise control could keep the speed at a given point uphill with a constant slope, without varying current.

 

[ZapPat]

Yes, I have a resistor in series with the hall line (about 1.5kOhm) it is for fixing my initial problem, that it cut out above 90-95% WOT.

I discovered that you can just set the "bar guard" option to "off" in the parameter designer software to avoid the high level throttle cut-out.

Now I'm quite confused. My throttle is supposed to alter battery current, isn't it?

No, it just blindly controls the PWM duty cycle as I described in my post just above. It also limits the battery current to a maximum value that you select. This is what is probably giving you a jumpy feel, specially since as you expect a current-based throttle response but are getting something else very different.

What does speed limits on the infineon controllers do? Why these don't help with this issue? I think there are 2 types (one SL limits speed only, the others limit current too?)
I really wonder if cruise control could keep the speed at a given point uphill with a constant slope, without varying current.

As far as I know, the cruise control will also just aim for a set duty cycle output. Same for the speed limiting settings, they'll just limit the maximum PWM duty cycle output of your controller, but the controller has no idea of what real speed you are traveling at.

 

[kZs0lt]

I discovered that you can just set the "bar guard" option to "off" in the parameter designer software to avoid the high level throttle cut-out.

Great! I just can't find it. I am using this version: http://www.endless-sphere.com/forums/viewtopic.php?f=16&t=7361#p110931 Is there a newer one?
Another note on parameter designer: I find annoying that you cannot specify LVC exactly with the software. For example there is nothing between 32.5v and 40.5v, for a 12s lipo, a very common config, would be needed something like 36v. Is there a way to tweak this. Is there someone who already tried to disassemble the program?

So, throttle controls PWM duty cycle. Usually from 0 to 100%. Meantime is probably watching for the shunt, and if the current exceeds the limit, it is lowering duty cycle till it gets below. The speed limit, probably figures out the max duty cycle for the setting, and scale down to that being 100%?

 

[ZapPat]

I discovered that you can just set the "bar guard" option to "off" in the parameter designer software to avoid the high level throttle cut-out.

Great! I just can't find it. I am using this version: http://www.endless-sphere.com/forums/viewtopic.php?f=16&t=7361#p110931 Is there a newer one?
Another note on parameter designer: I find annoying that you cannot specify LVC exactly with the software. For example there is nothing between 32.5v and 40.5v, for a 12s lipo, a very common config, would be needed something like 36v. Is there a way to tweak this. Is there someone who already tried to disassemble the program?

I guess you have a slightly older version of ecrazyman controller that is based on the infinion chip (look at the chip on the board inside to be sure). The newer controllers are based on a "116" processor and use a similar but more complete software for programming.

So, throttle controls PWM duty cycle. Usually from 0 to 100%. Meantime is probably watching for the shunt, and if the current exceeds the limit, it is lowering duty cycle till it gets below. The speed limit, probably figures out the max duty cycle for the setting, and scale down to that being 100%?

Yes; Maybe; Yes.

"Maybe" about the current limit because I believe some (most?) ebike controllers use a simple pulse skiping method to cut back the current at the set limit. The gradual lowering of the duty cycle as you describe would need to be based on some form of PID control loop, and I'm not sure if even the newer infinion/116 based controllers do this. I should check this out on the scope sometime though. The problem with the simple pulse skiping method is that it will produce more audible noise in the motor, and have more current ripple throught the motor when doing current limiting.

 

[nightowlgk]

<<<Maybe that would fix it. Clyte analog 72-36v controllers are supposed to work OK but I've never tried one.>>

Having similar trouble with my new CLYTE analog 35 amp controller ( from Ilia) and CLYTE Twist throttle from Justin. Not only that, but the controller seems to be cutting out at 15amps!!! not good. 13amps is the max limit setting on the CA that allows me to get more than 15 feet.

New Bike:

DAHON MATRIX 26" folding MTB
BMC V2 speed model from Ilia Brouk 7 sp. freewheel
CLYTE 36-72v 35 amp controller
CLYTE Twist Throttle
Cycle Analyst 2.1
HEADWAY 36v.10aH Rear Rack mounted battery 2 amp charger

 

[nicobie]

Best to talk to Ilia.

I've totally given up on my Keywin Infineon. Now I've found out from Justin at ebikes.ca that the reason that his CA doesn't work right with my controller is because Keywin always wires the CA pigtail wrong.

 

[fechter]

The twitchy throttle response you're describing does not sound normal by any means.
I have no clue at this point what to suggest however.

All the controllers I've tested linearly ramp down the duty cycle when you hit the current limit and they have a crude PID implementation.

It almost sounds like the current limiter is just not working at all. You might check the shunt wiring and try lowering the limit in parameter designer and see what happens.

 

[nicobie]

and try lowering the limit in parameter designer and see what happens.

Oh fechter, if only I could... :? :?

 

[icecube57]

Best to talk to Ilia.

I've totally given up on my Keywin Infineon. Now I've found out from Justin at ebikes.ca that the reason that his CA doesn't work right with my controller is because Keywin always wires the CA pigtail wrong.

Ha ha ha ha ha ha. Some times its right. Sometimes its wrong. Some times its right but colors are in reverse order. I have two controllers and they are no where near alike lol . Especially that damn CA connector.

 

[kZs0lt]

I have enabled cruise control and put a momentary switch. This thing is awesome, it works perfectly, and is very practical.
But the described problem persists even with cruise control. So it can't be a throttle problem, no shaky thumbs...
It is easiest to observe the phenomenon on a hill let's say about 7-8%. The figures are not exact, just an example.Let's say I give throttle about 25% to get a constant 15km/h on the slope. I get to 15km/h and enable cruise control. It uses 7-8amps and it cannot maintain 15km/h, eventually slows down to about 12-13km/h. Then it would kick in with around 15amps, or sometimes with full 20amps (measured approx.), this accelerates the bike for 4-5 secs, while it gets to 16-18km/h. Then it sits back to 7-10amps and slowly decelerates and when the speed falls below again it would kick back again. And on a nice slope I can't cruise with a constant speed @ 15km/h(with let's say 10-12amps), but instead I get accelerations and slowing down's.

The problem occurs with heavy loading, or great acceleration. When there is no great slope and I am not heavy on throttle, I can accelerate very smoothly and on flats can keep speed very constant.
I suspect it could be some software "feature" like the "nice" soft start. The soft start actually is not a bad thing, considering my nylon geared motor, but I personally hate it, could start softly myself with a good throttle response. However I must live with that, I don't find it a sufficient reason to throw money for a new controller with hard start.

Could it be something else than software problem, something workable?
If I will get a camera, will try to make a video with an ampmeter hooked up, with sound to hear the motor noise on acceleration. It may not be very soon, don't hold your breath.

Zsolt

 

[kZs0lt]

If I will get a camera, will try to make a video with an ampmeter hooked up, with sound to hear the motor noise on acceleration. It may not be very soon, don't hold your breath.

I have finally made 2 videos, with sound that could be of help.
Again this is a soft start xc846 Infineon controller set to 20 Amp, driven @ 66v nominal.
I have no Cycle Analyst, so I have attached 2 DMMs, the left one for current measurement, the right one for voltage.
First I am trying to accelerate very smooth slowly. You can see, that I can do that, the speed is gradually increasing and current is rising to approx. 7Amps, to 28km/h
When I am trying to accelerate harder, I am doing my best to engage the throttle gradually smooth, but the response is still very much like a stair. It kicks in with 12A, then it drops back, kick in again with 18Amps, drops back, kick again with 16amp while am trying hard to work smooth gradually with the throttle.
[youtube]0RrIe-cXp5w[/youtube]


The second video is probably of more interest, but unfortunately the camera is a bit out of focus.
What I was trying to show is, that while I am climbing a street of around maybe 5-8%, enable cruise control @ 17km/h and the bike after acceleration settles down to 4-5amps. The bike is slowly loosing speed by 2-3 km/h and the controller decides that it needs to supply more power to the motor to keep up with speed, so it bumps up the current to 16-18amp, falls back to 4-7amps when reached aprox18-19km/h, slows down, kick in again, fall back, and kick in again rhythmically. Listen to the sound of the motor, doing accelerations and decelerations.
[youtube]BsIVUiNAYOI[/youtube]

This odd behavior is very disturbing to me, every time I want to accelerate, or climb a hill with a steady pace, I get kicks with almost full power unexpectedly when loading is heavier. I'd wish cruise control would stick to a given current or a PWM duty cycle, but the controller is doing something else. It seams that is trying to keep speed, but it's not smooth. It also happens with different battery voltage, different current settings, different speeds.

I think this could be some odd behavior of the SLOW START algorithm. I hope I don't need to get rid of this controller.
Or could it be something else causing this behavior? Something silly that could be fixed?
Is anyone else experiencing this with slow start controllers?

Any idea or help would be greatly appreciated.
Thanks, Zsolt

 

[biohazardman]

I would think, by your description, the cruise control is werqing as designed and there is not a problem with it. Hills will always be a problem with cruise control as the parameters it operates within will change quickly and the controller must react to get things back into spec. Once there it will go back into it's regular program. Once out of spec it will add the power to get back in spec again and so on. Cars with automatics do the same thing. Alsoo cruise is meant to keep you at a steady speed once you are already there. I don't know how it reacts on acceleration if it's already in cruise mode maybe someone else can say something about this.

I have a BMC 600 and use the 35A 48V controller I got with the kit but use the throttle I got with my GM kit as it is built better. It is a bit touchy but controllable with a light touch. I have an Infineon that is posed to werq with BMC motors and am planning on using it at 60V so will be watching to see what you end up with in your search for correct throttle operation.

 

[kZs0lt]

I always thought the cruise control will stick to the current PWM duty cycle, not speed.
I don't think that is normal to increase the current 2-3x the amount needed to maintain the speed. Accelerating every 5 seconds with upto full power won't help on efficiency either. It's the same thing without cruise control, just trying to maintain speed with throttle. I was using cruise control in the video to show that the problem is not with my finger being not sensitive enough to maintain a throttle level.
The biggest problem is accelerating on the flat. I can accelerate grandma slowly with 4-6 amps, or full current 16-20amps. Nothing in between it's almost impossible, because the controller would ramp up the current every time I try to accelerate no matter if I'm just halfway on the throttle from a stop.

 

[amberwolf]

Is the throttle itself actually putting out a smoothly-changing voltage from around zero to around 5v? Or does it jump at some point from the low range to the high range? This you can test stationary, as long as the motor is disconnected or the wheel is off the ground.

EDIT: wait, nevermind. You said this happens the same with cruise or throttle. I assume that for cruise it has it's own separate electronics for outputting throttle voltage.

So that means it is likely an issue inside the controller. Can it be reprogrammed to a different response curve?

 

[kZs0lt]

So that means it is likely an issue inside the controller. Can it be reprogrammed to a different response curve?

None that I know of. I think soft start controllers cannot be reprogrammed to hard start, the chip is not rewriteable, can be only changed I think.

Do you guys think a hard start version would work the same?

 

[mwkeefer]

Sounds like a glitched controller to me... got a spare to swap in and test with?
-Mike