Analog electronic soft start E-bike throttle solution.

[TommyCat]

You’ve talked me out of the simulation for now.
I think I’ll stick with the delay/soft start idea at this time for the issue of excessive start power at the beginning of throttle twist.

Do you have a cruise circuit that will try to maintain SPEED at a certain throttle voltage output? Like an automobile…

 

[amberwolf]

I don't have a specific circuit (Fechter might) but that just requires a speedometer with negative feedback to vary the throttle voltage down if the speed increases.

If the speedo is pulses, you can use an integrating opamp circuit and some caps to smooth it into a constant voltage.

You would need something (pot, etc) to create the setpoint for the speed each time you engage it, or for simpler usage (but more complex design) a button that captures the present speed and then does a form of sample-and-hold of that voltage to use to compare with the realtime speedo input.

There are various S&H chips that are more or less simple enough to use in an analog-only circuit.

sample and hold chips - Google Search

Random examples

https://www.analog.com/media/en/technical-documentation/data-sheets/AD582.pdf

https://www.ti.com/lit/ds/symlink/lf198jan.pdf?ts=1710793265538&ref_url=https%253A%252F%252Fwww.ti.com%252Famplifier-circuit%252Fspecial-function%252Fsample-hold%252Fproducts.html

 

[TommyCat]

Yes, I have a sample and hold circuit and a D flip flop for its actuation/release by pressing a momentary push button switch.

It's the first part... speed pulse integration with the components you mention that I could use help with.

.

 

[amberwolf]

There's a number of pages that explain opamp integration better than I could (it's been a long time since I worked with them), like this
https://www.electronics-tutorials.ws/opamp/opamp_6.html
or this
https://www.digikey.com/en/articles/analog-integrators-how-to-apply-them-for-sensor-interfaces
but basically the speed pulses go into the Vin, and the smoothed output (which may look like a sawtooth at this stage, depending on the RC combination chosen) at Vout, as an opposite of the input (meaning, voltage on the output should go *down* for the input rising), which if you need to you can fix with a second inverting opamp stage, which can also be used to scale the voltage for the next part of the circuit if needed.

 

[fechter]

The integrator circuit is what my throttle ramp circuit is based on. The diodes allow for a faster ramp down compared to the ramp up.

 

[TommyCat]

Will definitely be looking at this op amp circuit closer, thank you both for your inputs.

Recommendations for the signal input. I.E. speed sensor on the wheel with 1 rpm pulse. Or perhaps put a couple or more magnets for more signals? Or tap into a hall signal output?

 

[amberwolf]

More pulses per second will be easier to smooth out. More magnets on a wheel sensor gets you that easily enough, and they don't have to be perfectly evenly spaced (the result is smoother at slower speeds the more even they are, and the more of them there are).

Too many magnets too close together may blur the signal into just an "on" at too high a speed, and may also give too high a duty cycle so it doesn't vary the integrated voltage enough.

Without testing this, I *think* the easiest-to-smooth signal would be magnets on an "encoder ring" (like a PAS sensor) mounted at the hub of a wheel, where the magnets are all in the same polarity, if they don't end up too close together. Many geared hubmtoros use six magnets at around that same diameter circle aroudn the axle, just inside the cover, as a separate speed sensor from the halls.


A motor phase-hall signal will have a bunch of other noise on it that may affect the results; probably more noise the higher the phase currents are (more magnetic crap going on in the stator where the halls are mounted, and more possible induced currents into the hall wires).

 

[fechter]

If you wanted to make an analog "cruise control", one way would be to use a frequency to voltage converter like a LM2907 (see attached datasheet). You can use a hall sensor signal as the speed input and some other control (like the throttle) as the set point input. Using a hall signal will be better than a wheel sensor since it is much higher frequency, so will work at lower speeds. The LM2907 has a built in op-amp to perform the comparison of set point vs. speed input.

One issue will be when the speed is below the set point, the controller will get full throttle until it reaches the set point. You'd want to limit how fast it changes the throttle input.

 

[TommyCat]

If you wanted to make an analog "cruise control", one way would be to use a frequency to voltage converter like a LM2907 (see attached datasheet). You can use a hall sensor signal as the speed input and some other control (like the throttle) as the set point input. Using a hall signal will be better than a wheel sensor since it is much higher frequency, so will work at lower speeds. The LM2907 has a built in op-amp to perform the comparison of set point vs. speed input.

One issue will be when the speed is below the set point, the controller will get full throttle until it reaches the set point. You'd want to limit how fast it changes the throttle input.

This looks VERY interesting. Bit above my knowledge level but going to study it. Thank you.

 

[TommyCat]

... Bend it like Beckham.

A usable circuit as modified?








Delay is easily adjusted by the R7 capacitor. Working on a potentiometer model to work in LTspice.

 

[fechter]

Adding C3 will slow things down and also make the response slightly non-linear. C1 and R4 were intended to control the speed and keep the response linear.

In practice, if the throttle ramp is linear, the current draw (and torque) will stay nearly steady during acceleration.

You could try it both ways and see what "feels" better.

 

[TommyCat]

You could try it both ways and see what "feels" better.

Right now I have an exponential ramp circuit. (Typical R/C type plotj
Your linear ramp circuit.
And a logarithmic ramp by modifying yours, that was my goal to achieve in my mind.

Granted, one might not be able to tell much if any difference between the three. But will see.

Is R1 just to dump the voltage charge from the op amp Input when the throttle is released?

 

[fechter]

R1 pulls the throttle signal down. It's there to make sure you have zero throttle if the throttle signal is disconnected.

 

[hannesd]

Sorry I've been away for a while.

I've had several scooters where the low speed throttle response was an issue. It's really nice when it's dialed in properly.
If the motor controller was fancy and fully programmable, this could be fixed by proper programming, but my scooters had cheap controllers with limited or no programming options.

There are a couple of factors that make this an issue.

First, look at the response of the throttle output voltage vs. throttle positon.

View attachment 349146

If you look at the curve, you'll see that it is not a straight line linear response. It's sort of "S" shaped with curved portions at the ends. The other thing to notice is the motor start voltage is quite a bit higher than the lowest throttle output. The motor start voltage has to be higher than the lowest throttle output to make sure the motor turns off when you release the throttle. Due to large variations in construction and things like temperature drift, manufacturers make a large tolerance here.

But this makes the starup happen on the steep part of the graph. If you could lower the motor start voltage, you could get it on the curved part near the end where the slope is much flatter. Since cheap controllers don't have programmable motor start voltages, the way around this is to place a resistor in series with the ground wire going to the throttle. This will shift the whole curve upward at the bottom end as shown by the green line below.

View attachment 349147
By using a multi-turn trimmer pot, you can adjust the pot so the motor barely starts running, then back off a bit. This makes a HUGE difference in the controllability at low speeds. On mine, the right resistance was around 150 ohms.

The other big factor is the controller's "operating mode", by this I mean whether it's a speed command, torque command, or something in between. Typical cheap controllers are really neither, but "PWM" command, where the throttle input controls the PWM duty cycle independent of any feedback.

With PWM command or speed command, a small, sudden increase in throttle makes the controller put out FULL power (limited only by the controller's current limit or BEMF) for a short time as the new operating point is reached.

If the controller uses a current (torque) command, a small sudden increase in throttle results in a small sudden increase in current, which is barely noticeable.

Years ago, I built a circuit that measures the battery current by using a section of the existing controller battery wire as a shunt and an op-amp that adjusts the throttle voltage to keep the battery current at level determined by the throttle setting.

View attachment 349148

This circuit took a while to get dialed in, but once adjusted, worked great. This totally eliminates the tendency to get large power surges at low speeds. The bypass switch allows you to go back to normal PWM mode and doesn't require any mods to the controller itself.

On one of my scooters, the trigger throttle was integrated into a display unit and the hall sensor was a tiny thing on a board with no easy way to mess with the wiring. On this scooter, I was able to take a small chunk of magnet and placed it in the right spot with double sticky tape to lift the low end of the curve so it behaved like the green line in the graph above. This worked great but was tricky to find the right spot for the magnet. The stock throttle magnet is to the left of the sensor in the picture:

View attachment 349149

regarding this simple extra magnet trick, does this sort of "balance" the workings of that hall sensor?
i'm inclined to give this a try.
if it only gives me somewhat more control over the scoot, i'll be glad.at this moment i've turned the power down to about 30% to make it smoother and rideable.

 

[fechter]

The extra magnet just makes the throttle less sensitive at the low end.

 

[hannesd]

The extra magnet just makes the throttle less sensitive at the low end.

i opened the cover of the speed controller and held a steady throttle with wheels off the ground.
when i added a magnet close to the hall sensor (wich was also VERY close against the magnet of the throttle lever!) the speed dropped slightly. so i hot glued the magnet in place.
i also found the throttle lever somewhat sticky.
i tried different lubricants but it got worse, so out with all the greasyness with brake cleaner.
it feels better now but i might see if i can loosen up the throttle lever even more, maybe less spring tension and completely dry throttle pivot point and some more free play on the pivot point somehow.

so right now i feel slight improvement, but i'm not yet there.

 

[amberwolf]

FWIW, most lubricants (petroleum based, anyway) will make any styrene type plastics sticky, or even dissolve them.

If you need to make something made of styrenes (almost all plastic casings on cheap stuff) slide / etc better, and don't have a lube you *know* is plastic-safe (like "white lithium grease"), dry bar soap usually works. (it also works for window and door and drawer slides, etc, that aren't based on roller wheels)

 

[99t4]

Silicone Lubricant might work?

 

[hannesd]

Silicone Lubricant might work?

tried that, and it was even worse thought of trying wd40 but kept it dry for now
i'll go for the hard soap bar.