Interesting.

 

I saw a set of spare idler gears for 95RX on AliExpress yesterday and they were 18/40 teeth, same as mine are.

For an older version 95 perhaps? Mine was supplied a year ago.

 

Yose quoted 4.4, consistent with what you have found. I counted the gear teeth, which suggested 8.15*. That gave P1 = 88 (Yose) and 163 (teeth count). When I set P1 to 88 and max speed to 25kph, the motor cut out at about 9 mph. When I changed P1 to 163 the motor started cutting out at 15.5 mph. So I assumed the reduction ratio of 8.15 was about right. (But you know what they say about assumption...)

 

Yose said the motor is wound for 190 - 200 rpm wheel speed. That works out at about 25kph (200 x 2100/1000 x 60), but I know that the motor goes well over that speed when max speed is set high (60kph). It was pulling well at 17mph when I went past 3 cops pretending to pedal hard!

200 rpm @ 4.4:1 = 880 rpm motor speed.

200 rpm @ 8.15:1 = 1,630 rpm motor speed.

I have no idea whether either or both figures are plausible!

 

There must be a logical answer but atm it's confusing.

 

 

* Helical gears - motor 18 teeth, idler 40 teeth. Straight cut gears - casing 66 teeth, idler 18 teeth. I counted them all several times to be sure. So, 40/18 x 66/18 comes out at 8.15:1. What I wasn't sure about was the effect of the idler gear assembly rotating in the opposite direction to the idler gears, but since P1 = 163 worked well I stopped worrying about it. Perhaps idler assembly rotation makes a big difference and Yose was right with 4.4, but then what about P1 = 88 being so far adrift? I might be missing something?

I think you're right. The reduction must be 8.15:1. To get the max motor RPM, lift the wheel off the ground and spin it up to maximum unrestricted, then compensate for the battery voltage by multiplying that by 36 divide by actual battery voltage. 201 rpm is 15 mph with a 26" wheel. Do the maths to convert your speed to RPM.

Max. no-load speed, level 5 = 18.9mph = 30.4kph. ( I tried C2 = 0 and C2 = 4 with exactly the same results.)

 

Battery voltage 39.8V (Digital voltmeter. Assumed no significant voltage drop due to negligible current.)

 

Max speed normalised to 36V: 30.4 x 36/39.8 = 27.5 kph or 27,500 m/hr (=17.1mph).

 

Tyre circumference 2165mm or 2.165m. (Measured on garage floor, 1 revolution. +/- 5mm? Marathon Greenguard 700-32 has 3mm latex under tread, hence slightly higher value than 2152mm obtained from tables seems plausible).

 

Wheel rpm = 27,500 / (2.165 x 60) = 212 rpm. (Slightly higher than Yose Power figure 190 - 200 rpm.)

C15 is the walk assist speed (6kph is the maximum legal permitted I think). I think it's a new feature in the KT software. I bought a replacement display earlier this year and it was on that, but wasnt' on the previous one.

Edited October 21, 20232 yr by RogerA

Thanks Roger. So that's not an issue in the context above. Good to know.

Max. no-load speed, level 5 = 18.9mph = 30.4kph. ( I tried C2 = 0 and C2 = 4 with exactly the same results.)

 

Battery voltage 39.8V (Digital voltmeter. Assumed no significant voltage drop due to negligible current.)

 

Max speed normalised to 36V: 30.4 x 36/39.8 = 27.5 kph or 27,500 m/hr (=17.1mph).

 

Tyre circumference 2165mm or 2.165m. (Measured on garage floor, 1 revolution. +/- 5mm? Marathon Greenguard 700-32 has 3mm latex under tread, hence slightly higher value than 2152mm obtained from tables seems plausible).

 

Wheel rpm = 27,500 / (2.165 x 60) = 212 rpm. (Slightly higher than Yose Power figure 190 - 200 rpm.)

Great work. Thanks.

PS With max speed set to 25kph (P1 =163, i.e. 8.15 x 20), the no-load speed is 15.8mph.

Update:

I'm coming to the conclusion that either:

 

That's just the way it is.

The motor seems significantly stronger than it did with the JYT speed controller, which may have had a lower current limit - no way of knowing. (Label on JYT says 15A max, 8A rated, but programmed max current could be lower?) So it would be noisier/harsher due to higher torque, and may get quieter as the gears etc. bed in. I'll try altering C14 =2 (default) to C14=1 and see what happens. (C14=1/2/3 = "Weak/General/Stronger assist strength of motor". Not sure what that means... Perhaps it alters the spread of power levels so that level 1 can be more gentle or stronger ?)

 

Edit:

C14=1 felt immediately quieter, however the Watts in the lower power levels was reduced substantially (around 60% reduction in level 1 and 40% in level 2). In higher levels it still sounded a bit rough. At least I know now what C14 does.

 

 

I may have damaged* a Hall sensor wire?

Would that result in the controller defaulting to square wave? (L3=0) If so, presumably it would be a bit harsher/coarser sounding. If it's just running square wave I'll live with it. But how to tell?

*With my arthiticky thumbs playing up I struggled to grip the motor lead plug hard enough to push it into the socket. So frustrating. To get it right home I had to wriggle it side to side. Then I found I'd forgotten the rubber end cap and had to unplug it again which was even harder. D'oh! I don't think I would have done any damage as I wasn't holding the cable, but it's always possible.

 

Any other thoughts?

 

I've learnt some useful things from this exercise and feel a lot more confident about configuring the controller. Thanks Saneagle.

Edited October 24, 20232 yr by Slightlypedantic

Update:

I'm coming to the conclusion that either:

 

That's just the way it is.

The motor seems significantly stronger than it did with the JYT speed controller, which may have had a lower current limit - no way of knowing. (Label on JYT says 15A max, 8A rated, but programmed max current could be lower?) So it would be noisier/harsher due to higher torque, and may get quieter as the gears etc. bed in. I'll try altering C14 =2 (default) to C14=1 and see what happens. (C14=1/2/3 = "Weak/General/Stronger assist strength of motor". Not sure what that means... Perhaps it alters the spread of power levels so that level 1 can be more gentle or stronger ?)

 

I may have damaged* a Hall sensor wire?

Would that result in the controller defaulting to square wave? (L3=0) If so, presumably it would be a bit harsher/coarser sounding. If it's just running square wave I'll live with it. But how to tell?

*With my arthiticky thumbs playing up I struggled to grip the motor lead plug hard enough to push it into the socket. So frustrating. To get it right home I had to wriggle it side to side. Then I found I'd forgotten the rubber end cap and had to unplug it again which was even harder. D'oh! I don't think I would have done any damage as I wasn't holding the cable, but it's always possible.

 

Any other thoughts?

 

I've learnt some useful things from this exercise and feel a lot more confident about configuring the controller. Thanks Saneagle.

KT controllers are not dual mode, so they won't switch between sensored mode and sensorless like some other controllers do.

 

AFAIK C14 adds a ramp to the PAS startup. The different settings are the steepness of the ramp.

I've ridden the bike a bit more and, from watching the wattage on the display, it does seem that C14 adjusts the spread, and possibly spacing, of the power levels. For example: In level 1 C14=1 gave around 25-30W and C14=2 gave around 60-80W. Likewise, level 2 gave around 90W and 130W. The readings fluctuate rapidly so it's hard to give definite figures. I haven't yet been able to assess the higher power levels since consistent hills would be needed to keep below the 25kph limit, and there's yet to be any need for levels 4 and 5 anyway.

 

It's possible that all levels change together such that level 5 power varies, or that the spacing of the levels changes such that level 5 always gives full power. I haven't tried C14 = 3, but expect that to be excessively lively. C14 = 1 is may be better for battery life.

 

I couldn't tell if C14 affected ramping as well, as I felt the different amounts of power could be masking this.

 

I obtained an up to date KT manual from the AliExpress supplier and noticed that three previously blank settings in manuals downloaded from the internet (C5 = 0, 1, 2) have now been populated with alternative ramping settings. This must be a recent software change and probably explains what I found with C14, whose function may have been changed at the same time.

 

On the motor noise, I decided to fit wider tyres so, while I had the wheel out, I swapped the motor internals with my spare motor. The noise was similar but more subdued and less objectionable. The motor performance was identical. So I'm thinking that the noise might actually be mechanical noise from the reduction gears, in which case it should reduce as they bed in.

 

 

EDIT:

 

C14 "Assist strength of motor"

I found a long and fairly steep hill this afternoon. With C14 = 1 ("Weak assist strength of motor"), I found that levels 1 to 4 were all lower in power than with the default (C14 = 2), with a big step up to level 5, which gave full power (bike accelerated uphill until it hit the 15.5mph limit).

 

With C14 =2 ("Normal assist strength of motor"), levels 1-4 are stronger and the difference between levels 4 and 5 is less marked. I haven't yet tried C14 = 3 ("Strong assist strength of motor") but it seems reasonable to suppose that levels 1 to 4 would all be stronger still, with a smaller step up to level 5.

 

My feeling from this ride is that KT have actually programmed 7 power levels, mapped as follows:

C14 = 1 - ("Weak assist strength of motor") - selected levels 1-4 are mapped to actual levels 1-4.

C14 = 2 - ("Normal assist strength of motor") - selected levels 1-4 are mapped to actual levels 2-5.

C14 = 3 - ("Strong assist strength of motor") - selected levels 1-4 are mapped to actual levels 3-6.

 

In all 3 cases selected level 5 is mapped to actual level 7 (= full power).

 

This would be relatively easy to implement from a programming point of view and fits the C14 parameter description well.

 

C5 Current limit/Soft start

C5 is currently set to 10 (default), which allows the full 17A for which the controller is rated. I'll try playing with C5 = 0,1,2 sometime to explore these (new) soft start settings. C5 = 10 seems quite satisfactory except if pulling away in a higher power setting - perhaps that's where the soft start settings come into their own.

Edited November 5, 20232 yr by Slightlypedantic

I've ridden the bike a bit more and, from watching the wattage on the display, it does seem that C14 adjusts the spread, and possibly spacing, of the power levels. For example: In level 1 C14=1 gave around 25-30W and C14=2 gave around 60-80W. Likewise, level 2 gave around 90W and 130W. The readings fluctuate rapidly so it's hard to give definite figures. I haven't yet been able to assess the higher power levels since consistent hills would be needed to keep below the 25kph limit, and there's yet to be any need for levels 4 and 5 anyway.

 

It's possible that all levels change together such that level 5 power varies, or that the spacing of the levels changes such that level 5 always gives full power. I haven't tried C14 = 3, but expect that to be excessively lively. C14 = 1 is may be better for battery life.

 

I couldn't tell if C14 affected ramping as well, as I felt the different amounts of power could be masking this.

 

I obtained an up to date KT manual from the AliExpress supplier and noticed that three previously blank settings in manuals downloaded from the internet (C5 = 0, 1, 2) have now been populated with alternative ramping settings. This must be a recent software change and probably explains what I found with C14, whose function may have been changed at the same time.

 

On the motor noise, I decided to fit wider tyres so, while I had the wheel out, I swapped the motor internals with my spare motor. The noise was similar but more subdued and less objectionable. The motor performance was identical. So I'm thinking that the noise might actually be mechanical noise from the reduction gears, in which case it should reduce as they bed in.

You can't test by riding the bike because the speed of the motor is changing all the time, and as the speed goes up, the current goes down. The best way to test is to lift the motor wheel off the ground, power up with the throttle and apply the brake to get the speed down to about 4mph under full power. Read the current or power from your LCD wile the wheel is slowed right down. Obviously, don't hold the wheel with the brake for too long.

Afaik the C14 parameter is a PAS pick up strength signal setting that is weak - strong.

Afaik the C14 parameter is a PAS pick up strength signal setting that is weak - strong.

It sets a ramp on the power from the PAS. The ramp is actually on the current. The idea is that you don't get such a fierce startup when using level 4 or 5. Instead of giving the max 15A as soon as you start pedalling, it ramps up the current from zero to 15A over a few seconds. The number of seconds is determined by the C14 value. The lower power on the lower levels isn't a problem for startup, so it's not necessary to use the ramp unless you leave your power on level 4 or 5 all the time, or maybe if you had a 22A or more controller.

 

Below is a picture of how it works with say a 15A controller,pedalling from startup to maximum speed:

Thanks for the graph. Visual information is always good. I understand the concept of ramping and it's depiction in the graph.

 

When I start off in level 1 or 2, I think can feel the motor ramp up for a couple of seconds (with C14 = 2). It's hard to say if this effect changes when C14 is changed since the change in motor assist strength dominates. To explain this, I've added to the graph to show what I am experiencing (subjectively):

 

 

Important: The levels shown represent C14 = 2 (default). The differences between levels 1 to 4 are meant to be equal, whereas the differences between level 0 (off) and level 1, and between level 4 and level 5, are meant to be larger.

 

If I select C14 = 1, then levels 1 to 4 move downwards, i.e. they become weaker. Level 5 remains unchanged. The motor feels distinctly weaker than C14 = 2, except in level 5. There is a bigger and very noticeable step up in power between levels 4 and 5.

 

If I select C14 = 3, then levels 1 to 4 move upwards, i.e. they become stronger. Level 5 remains unchanged. The motor feels distinctly stronger than C14 = 2, except in level 5. There is a smaller step up in power between levels 4 and 5.

 

These changes in power are not minor and are experienced throughout a ride, not just when starting off.

 

In level 1, ghost pedalling, the maximum speeds achieved on the same stretch of level road and same wind conditions (i.e. little or none) vary significantly depending on C14 setting.

 

Level 1 is useful in a headwind when C14 = 2, but too feeble when C14 = 1. In the recent strong headwinds along the coast road, which I ride frequently, I've found myself using Levels 1 and 2 (C14 = 2), or levels 2 and 3 (C14 = 1).

 

I've attached a copy of the latest LCD3 manual that I can find (V4.0, 08/10/2022). The description of the C14 settings is: "Assist strength of intelligent pedal motor" (Values: 1 = Weak, 2 = Normal, 3 = Strong). Ramping settings appear in C5 (= 0, 1, 2). I have yet to try these out and have left C5 = 10 throughout.

 

I hope this makes my previous posts a bit clearer.

2022-10-08 KT-LCD3 V4.0 with C15.pdf

Edited November 9, 20232 yr by Slightlypedantic

Does the amended graph below represent the lower (blue) power levels better with current eventually being limited by back emf only, due to the controller being unable to supply any more current?

 

I'm not sure why the first red segment (level 5) is sloping. If, in the highest power level 5 , the controller provides 15A (controller maximum), with pulse widths increasing progressively to maintain that value as speed and back emf increase until current is being delivered 100% of the time, should the first red segment not be level rather sloping? I've probably missed something and would appreciate being put right.

 

Once cumulative pulse widths reach 100% of time, rising back emf and falling net voltage cause the steeper red segment because the controller can no longer maintain the current.

 

Since the lower power levels (blue lines) merely represent lower current limits, the controller can maintain them at higher speeds before pulses occupy 100% of the time and the target current can no longer be maintained. Although I have now drawn the blue lines straight rather than kinked, I have kept them parallel to the first red line (level 5), but should they be level or sloping? And why?

 

I appreciate that soft start profiles are superimposed, however it's the ongoing behaviour that interests me rather than the first few seconds.

 

Thanks in advance for any answers.

 

Edited November 15, 20232 yr by Slightlypedantic

Does this amended graph better represent current limited by back emf only, due to the controller being unable to supply any more current? I.e. Blue lines represent 36V pulses for up to 100% of time?

 

[ATTACH=full]55119[/ATTACH]

The blue lines would be horizontal. I've never checked level 4 - maybe it dips off a bit as the speed goes up. You should watch your LCD to see what happens. On the lower levels, the current is exactly flat, regardless of speed until the back emf limits it. The chart below is an actual record of what happens with a KT controller. The orange line is watts and the green is current, which ar exactly sychronised because the voltage was constant during that record. You can see that I used three levels during the record. The bits where it goes up and down is because I paused pedalling or hit the speed limit. I had it set with C14=1 or whatever the default minimum is, so no ramp:

Interesting graph...

 

Since the current is constant for each power level, regardless of speed/back emf variations, the controller must be compensating by increasing the pulse widths to maintain a constant current. Once speed/back emf have increased to the point where current is being supplied 100% of the time, further compensation is not possible and current declines according to any further increase in back emf.

Interesting graph...

 

Since the current is constant for each power level, regardless of speed/back emf variations, the controller must be compensating by increasing the pulse widths to maintain a constant current. Once speed/back emf have increased to the point where current is being supplied 100% of the time, further compensation is not possible and current declines according to any further increase in back emf.

The controller is limiting the speed to 15mph, so you can't see any effect of back emf.

 

I don't believe that the controller has the intelligence to adjust current other than the global limit. I'm pretty sure that at each power level it gives pulses at a fixed sine-wave pattern and duty cycle, with a different duty cycle at different PAS levels. Any effect of back emf wouldn't be noticed at those low power levels and speeds.

Since the current is constant for each power level, regardless of speed/back emf variations, the controller must be compensating by increasing the pulse widths to maintain a constant current. Once speed/back emf have increased to the point where current is being supplied 100% of the time, further compensation is not possible and current declines according to any further increase in back emf.

I think you are correct. The software has current control loops as shown even in basic controller below and the duty cycle is constantly been adjusted. The shunt is used to generate a current signal which is connected to one of the analogue inputs of the micro-controller.

 

By the way, I like to set C14 =3 as it scales up the P1 to P4 settings only and reduces the 'Gap' between P4 and P5 by increasing the P4 value in my case from about 55% to about 70%.

 

I think you are correct. The software has current control loops as shown even in basic controller below and the duty cycle is constantly been adjusted. The shunt is used to generate a current signal which is connected to one of the analogue inputs of the micro-controller.

 

By the way, I like to set C14 =3 as it scales up the P1 to P4 settings only and reduces the 'Gap' between P4 and P5 by increasing the P4 value in my case from about 55% to about 70%.

 

[ATTACH=full]55314[/ATTACH]

That loop is for throttle control, not PAS. Is it from the open source firmware project?

I've been looking online for explanations of PWM current control. Although I haven't found anything e-bike specific, I keep coming across graphs like Fig 2 of the Wikipedia article "Pulse-width Modulation":

 

Figure 2

 

"The simplest way to generate a PWM signal is the inter-sective method, which requires only a sawtooth or a triangle waveform (easily generated using a simple oscillator) and a comparator. When the value of the reference signal (the red sine wave in figure 2) is more than the modulation waveform (blue), the PWM signal (magenta) is in the high state, otherwise it is in the low state."

 

The saw-tooth or triangular waveform determines the frequency of the PWM pulses, say 20kHz.

 

The sine-wave frequency and timing are determined and synchronised using the motor Hall sensor outputs.

 

Sine-wave frequency is likely to be less than 1kHz (c. 550Hz max in my motor) depending upon motor design, reduction gearing and road speed. This implies at least 20 PWM pulses per sine-wave cycle: 10 pulses in the upward part and 10 pulses in the downward part, with the actual number of pulses per cycle increasing as rpm falls.

 

The sine-wave amplitude (height) could be determined by the current required for a particular PAS setting and controlled by a feedback loop. The red line on the graph shows full power. For lower PAS levels the red sine-wave curve would be correspondingly flatter, with the maximum pulses widths being reduced accordingly.

 

At PAS level 1 the current required might be, for example, 20% of the maximum current allowed by the controller. This could be interpreted as PWM pulses being 20% on and 80% off. This means there is plenty of scope to increase the pulse widths so as to maintain current and torque as back emf increases. This could be achieved simply by varying the sine-wave amplitude (height) via a feedback loop.

 

It's probably more complicated than that, but hopefully these basic ideas are somewhere near the mark.

Edited December 16, 20232 yr by Slightlypedantic