Accelerator pedal dead zone

[mane2]

Hey guys, what are the steps to get rid of dead zone of accelerator pedal? Plotting potnom and pot values, initial pedal travel goes to getting potnom to positive numbers. When reaching there and car starts to move, I've already pressed the pedal about 1/3 of pedals total travel.

I have calibrated the pot. Lowest value is about 950 and highest 4095. I even tried lowering potmin to 920 or 900 but that does not feel like a right way to do it. Adding boost makes it accelerate faster once it gets to positive side, so that does not feel right either. ampmin feels to do the same as boost.

My params here: download/file.php?id=24670

Any hints?

[johu]

you can reduce regentravel further.
All else would require work on the firmware, like automatically reducing regentravel at low speed

[mane2]

Oh, so when regentravel=20 is set, means that 20% of throttle is not used at all for accelerating - becoming a dead zone. I think I have upped it from those parameters even further, like 50 .. so that sounds like the real issue here, in my case.

So when coming off from 100% throttle, first 80% nothing happens and then the last 20% it ramps up regen?

I'm just wondering how it work on commercial cars.. They don't have any deadzone and regen works even when not fully lifting off from accelerator. Like if you quickly lift from 100% fully pressed to 50%, car regens a bit and then stays on 50% of the target power. I guess software knows the speed / rpm and uses time as a factor too? Any idea how they do this? Or was my first conclusion of how regentravel works wrong?

[johu]

Say regentravel=20 and offthrotregen=30
- 0% throttle -> 30% regen
- 10% throttle -> 15% regen
- 20% throttle -> 0
- 28% throttle -> 10% acceleration
- 60% throttle -> 50% acceleration
- 100% throttle -> 100% acceleration

OEM do more or less clever things. E.g. Kia reduce regentravel when reducing offthrotregen. So when you reduce regen with the flappy pedals suddenly you accelerate a bit more.

There's not deadzone indeed in OEM cars and there's ways to eliminate it from OI. Just couldn't be bothered

[mane2]

Thanks for the clarification, that's exactly what I experienced. I really struggled to understand the regentravel parameter's purpose in the first place, now I know.

I just can't live with that dead zone. Every time I get going, I get the feeling that I haven't started the car and pressing the pedal like an idiot and then finally it starts to move Still, I'd love to have regen without fully lifting off from the pedal. I guess I need to dig into the code

[muehlpower]

A question that bothers me too. Can "regentravel" be manipulated via CAN? If so, then an external controller could set "regentravel = 0" at 0 RPM and raise it to "regentravel=60" at increasing RPM. The result would be immediate starting with little accelerator pedal and regeneration at e.g. 20% at higher speeds.

Another question would be, does regeneration via pot2 work without connecting a brake light switch?

[mane2]

While looking at the code I wonder if there is variable for target rpm (or similar) and current rpm? Calculating difference would give you good value for negative potnom.

[muehlpower]

As i understand it, the accelerator pedal specifies the desired torque, not a RPM. The RPM can be output. There is no target RPM, only an upper speed limit. The version I suggested, which can of course also be implemented in the software, has the advantage of not having a dead zone and, at higher speeds, a well-dosed regen. In Johu's example, the pedal must be depressed at least 20% in order to get any propulsion at all. On the other hand, at high speeds I only have 20% to control regeneration. If I reduce regentravel to e.g. 2%, this means an almost digital regeneration.

[mane2]

I don't like the regentravel -idea, I feel there must be better way to do it. I'd like to have something that does not affect the accelerator feel/travel.

If you increase regentravel dynamically from 0% to 60%, the same thing happens what Johannes told about Kia with flappy pedals adjusting regentravel - it affects your accelerator. Smaller and variable effective zone for accelerator pedal, might be even worse than initial dead zone.

Ok yeah, accelerator defines the desired torque, but my question was more related to finding a variable which could tell the difference between current rpm related to accelerator position. Not sure if there is any .. but something could be made up. For example just time based. Like when you lift off from accelerator, calculate the diffrerence in pedal position averaging last 0.5 second, and if it's going down and is lifted more than 5% of the previous position, apply some regen.

[muehlpower]

If you increase regentravel dynamically from 0% to 60%, the same thing happens what Johannes told about Kia with flappy pedals adjusting regentravel - it affects your accelerator. Smaller and variable effective zone for accelerator pedal, might be even worse than initial dead zone.

that is not right. With KIA, if you have e.g. 30% pedal travel and regentravel is also 30% and you drive down a slight incline, nothing happens. Not even if it gets steeper or flatter. If you then reduce regentravel to 20%, the car accelerates. That's unconfortable.
For me, if I have 30% pedal travel and I am at a speed where regentravel is 30%, nothing happens either. If the gradient becomes steeper and the car accelerates, regentravel increases and regeneration begins. Conversely, if the gradient decreases and the car slows down, it will automatically accelerate again. It creates exactly the effect you want.

[muehlpower]

It's the same with a combustion engine car. If you give it enough gas that the engine makes 3000RPM with no load, it will brake above 3000RPM and accelerate below 3000RPM.

[uhi22]

I just made some measurements on my OEM Hyundai Ioniq. Result is, that the pedal position, which leads to zero power, is dynamic, depending on the speed:
- At speed below 6km/h, the regen level does not matter at all. The car slowly accellerates also with 0% pedal.
- At low speed (20km/h), the zero-power-pedal-position is around 8%.
- At higher speed (50km/h), the zero-power-pedal-position is around 18%. (Did not test at higher speed, need to improve my setup first
This was measured with regen level set to maximum via the paddles.
For me this makes sense. The higher the speed, the bigger the pedal range for regen.

When changing the level using the paddles, the regen changes, but also the acceleration at constant pedal. This makes sense for me, because just the complete characteristic of the pedal changes. I'm NOT irritated by this behavior. I have in mind just a straight line diagram, with a certain regen power at 0% pedal, and 100% drive power at 100% pedal. The line crosses the x-axis at a certain point (which is the regentravel parameter in OI), but this point does not need to be a parameter, it is just the result of the two power values on 0% and 100% pedal. And at the Hyundai, the power level for the 0% pedal is a function of the speed, something like
0km/h -> +1kW (or zero in case of brake pressed)
6km/h -> 0kW
20km/h -> -5kW
100km/h -> -20kW
(values just guessed, and are lower if lower level is selected with the paddles)

Summary from my point of view: Discard the regentravel parameter completely, and instead use a straight line, with a dynamic calculated power for 0% pedal.

[mane2]

I think the problem here is example the case where you quickly press the pedal 100% and release it to 20% to keep the speed. You probably don't want any regen to happen, even though lifting 80% of the throttle. Johannes probably could answer to this: Is there some variable in the code which could tell you current torque difference to target torque (accelerator pedal new position)?

[muehlpower]

I think the problem here is example the case where you quickly press the pedal 100% and release it to 20% to keep the speed. You probably don't want any regen to happen, even though lifting 80% of the throttle. Johannes probably could answer to this: Is there some variable in the code which could tell you current torque difference to target torque (accelerator pedal new position)?

Can you explain exactly what kind of behavior you expect and why? With my solution, it does not depend on the previous gas position in your example, but on the speed achieved whether regeneration occurs or not.

[muehlpower]

0% pedal is a function of the speed, something like
0km/h -> +1kW (or zero in case of brake pressed)
6km/h -> 0kW
20km/h -> -5kW
100km/h -> -20kW
(values just guessed, and are lower if lower level is selected with the paddles)

Summary from my point of view: Discard the regentravel parameter completely, and instead use a straight line, with a dynamic calculated power for 0% pedal.

The difference to my suggestion is the strength of the regeneration. With me, the regeneration is always at the set value with 0% pedal, with you it is high at high speeds, but rather low in city traffic. Technically it is anyway not possible to regenerate to a standstill.

[mane2]

I don't get how would those ideas work in practice. I think the variables that can be used for adjusting regen level is time, rpm and pedal position. Software does not know if car has reached it's target speed or rpm.

[Aragorn]

I just made some measurements on my OEM Hyundai Ioniq. Result is, that the pedal position, which leads to zero power, is dynamic, depending on the speed:
- At speed below 6km/h, the regen level does not matter at all. The car slowly accellerates also with 0% pedal.
- At low speed (20km/h), the zero-power-pedal-position is around 8%.
- At higher speed (50km/h), the zero-power-pedal-position is around 18%. (Did not test at higher speed, need to improve my setup first
This was measured with regen level set to maximum via the paddles.
For me this makes sense. The higher the speed, the bigger the pedal range for regen.

When changing the level using the paddles, the regen changes, but also the acceleration at constant pedal. This makes sense for me, because just the complete characteristic of the pedal changes. I'm NOT irritated by this behavior. I have in mind just a straight line diagram, with a certain regen power at 0% pedal, and 100% drive power at 100% pedal. The line crosses the x-axis at a certain point (which is the regentravel parameter in OI), but this point does not need to be a parameter, it is just the result of the two power values on 0% and 100% pedal.

I'm not sure my Skoda ever gets to 18% travel. Might be worth doing some logging with carscanner or something now you mention it.

To me it feels like the Skoda does the first two steps.

below ~5mph there is a small forward torque with no pedal.
Above 5 that forward torque reduces towards zero. Thus if in drive with no pedal input, the car will accellerate up to around 5mph or so. If going downhill such that the speed increases beyond that, the car will ramp into regen.
Once your above about 10mph, it feels like the last 5-10% of the pedal is regen. It doesnt feel like this position changes with speed, and that doesnt seem to matter much, as at higher speeds the regen has less effect anyway.

As with your Hyundai, altering the regen level (the Skoda has 4 levels, off, 1,2,3) does change the mapping, such that holding a fixed throttle position and then changing the regen level alters torque. I tend to drive in level 2, which in the Skoda applies more or less the maximum allowed regen without illuminating the brake lights. It may well be that in Max regen mode it does alter though.

At the most basic, i think regentravel should be ignored or ramped out at low speed, every factory EV ive driven does this, and it makes the throttle pedal "feel" correct.

[uhi22]

Some measurements on the Ioniq:
Each color is a certain speed range. The samples are coming each 100ms from the CAN. The force axis shows only the range to 4000N, to see better the details. Actually the force can go higher (around 8000N).

Conclusions:
- They prevent the dead zone by making the regen force at zero pedal dependent of the speed. Between 0 and 40km/h, the force increases from 0 to ~1500N, and at higher speeds it is more stable in range ~1500N to 2500N.
- There is a smooth change between acceleration and regeneration, no knee at the zero-force.
- At higher torques and speeds, they use some non-linearity, it looks like they want to increase the comfort, so that a small change in the pedal position does not lead to a big reaction.
- When changing from the shown "eco mode" to "sport", the curves are much steeper, but the strategies look the same.

Edit: How this diagram was created? By listening on the CAN bus, using an STM32 "blue pill", and a cheap TFT driven via SPI. The software is here: https://github.com/uhi22/TFT-with-CAN

[muehlpower]

This confirms that a speed dependent pedal travel or initial recovery value is the correct approach. A fixed pedal value for transitioning from accelerating to decelerating is not it! With your car, it may be that -2500N is the maximum deceleration. I find the positive values ​​interesting. You say they go up to 8000N. If I extend the straight lines upwards, it means that I reach 8000N with a pedal force of about 50% at 10km/h. So you don't have to pedal all the way to the maximum torque

[mane2]

How to read that graph? From top to bottom, each line? For example white line, you lifted the pedal slowly from 60-70 km/h speed, or completely at once? How can the pedal percentage be about 90% when driving only 60-70? Below the pedal-line is negative force / regen? I'm at loss

[uhi22]

How to read that graph? No, there is not certain direction or time-dependency in this graph. It shows the static transfer function, torque = function_of(pedalposition, currentspeed). "Static" means, it does not care for the time.

Of course, in reality there are time-dependencies, and they are visible in the diagram, as "noise". When you press the pedal, it takes some hundred milliseconds, until the filtering is done and the torque is really on the motor, especially in the used "eco mode". That's why, starting from a certain point in the diagram, we get often 3 "strange" samples on the right side (means: pedal was pressed more, but no increase of torque for 3*100ms), and afterwards the new setpoint is found (means: next sample is up on the target line again).
The "beautiful straight lines" are just for illustration, to have a better summary. They are NOT calculated by from the measurements. Only the dots are measured.
I did some minutes of driving for the diagram, with different speeds and different accelerations and recuperations. So to say a multiple zigzag through the whole diagram. The course of the zigzag does not matter, only the pattern, which is the result. E.g. the brown dots are forming a certain line, no matter, whether they are recorded at the beginning or at the end of the trip.

How can the pedal percentage be about 90% when driving only 70? Well, just imagine you are parking and then push the pedal to 90%. Then there will be in the diagram a black dot at (lets say) 90% pedal with 8000N, and then a brown on the same place (because speed increases), and so on, until the power is not sufficient anymore to keep the torque at high speed, and so in the diagram the higher colors will be below the (brown, red, orange, yellow) dot. We see such a case in the top right corner. Will full throttle, the speed was increasing (but the color is the same for everything above 90km/h), and the torque is decreasing because we are in the power-limit range (88kW rated power).

Below the pedal-line is negative force / regen? Yes, exactly. For example at 20% pedal, we may have strong acceleration (lets say 2000N at 15km/h) or moderate regen (lets say -800N at 100km/h).

And, regarding maximum and minimum force in the diagram: The diagram cuts at +4000N, because showing the complete range would hide the interesting details. Of course, the higher forces are also interesting, but not shown here. And the negative force limit is given by the software of the car. It is the maximum regeneration which we can achive by releasing the pedal. But, if using the brake, the regen further increases (up to 88kW). I did not go so far, but a little bit it is visible on the left lower edge: The black dots (means: speed below 10km/h) at zero pedal goes down to -3500N when pushing the brakes stronger.

[uhi22]

I find the positive values ​​interesting. You say they go up to 8000N. If I extend the straight lines upwards, it means that I reach 8000N with a pedal force of about 50% at 10km/h. So you don't have to pedal all the way to the maximum torque

Yes, I find this also very surprising. And this is eco mode, the normal and sport are much more aggressive. Somehow it fits to the daily experience: Need to be careful with the pedal at lower speed, otherwise we lost a lot of rubber on the street (front driven car, accu in the back, not able to bring the torque on the street). But: Maybe the linear extrapolation is not really correct, need to make some more measurements with the full-scale torque

[muehlpower]

My suggestion looks like this. With a flat curve for acceleration at low speeds. Useful for high power configurations. Strong regeneration also possible at lower speeds => less use of the brake pedal. A new parameter needed, if it is set to 0 everything is as before. And of course no dead zone!

Regengraph.png (10.84 KiB) Viewed 9665 times

[uhi22]

And on the left side, between 0 and 20km/h?

[muehlpower]

And on the left side, between 0 and 20km/h?

The formula continues in the same way (left). It would be correct as shown in the middle. You need more parameters for that. In reality it will be like the picture on the right, since near zero the braking torque possible due to regeneration decreases. I don't think the slight difference is worth the effort.

If "offthrotregen" is already a percentage value of the currently possible braking torque, the behavior in the middle picture appears anyway.