smoothing out the throttle

GMUseless said:
The analog solution is pretty simple, it's like 4 or 5 components...use a dual op amp like a TL072 as a voltage follower with a pot in between and a capacitor to ground. It's a standard slew or lag circuit. You can also add a diode to bypass the pot to give a single sided ramp (i.e. ramp up...but instantly drop off....like what you'de want in a throttle.) When I get back to my work computer, I can draw up a schematic is anyone is interested.

Yes, please. As an electronics caveman, I have to look at the easiest one to build.
 
gwhy! said:
here is a basic diagram to use a picaxe08M for a throttle interface.


it could be made to be hard wired into the throttle lead and cover it with some heatshrink.

GWhy,

Thanks for that. Where do S/out and Sig/out go? I don't understand the 2 outputs.
 
John in CR said:
gwhy! said:
here is a basic diagram to use a picaxe08M for a throttle interface.


it could be made to be hard wired into the throttle lead and cover it with some heatshrink.

GWhy,

Thanks for that. Where do S/out and Sig/out go? I don't understand the 2 outputs.

The 3 connections s/in, s/out and gnd are for programming the chip and do not need to be normally connected.
the sig in connects to the throttle signal wire and the sig out connects to the throttle input on the controller.

I am waiting for some chips to arrive then I will make one and post up some pictures, and also post up a program for it.
 
The analog approach could look something like this:

throttle-ramp.jpg

This circuit would plug in-line with the normal three wire throttle interface. As I understand it, the standard throttle signal is 1-4v...so powering the op-amp with a single sided 5v supply should be fine.

It's technically a little different from what I described earlier...the voltage follower setup would give log response. For linear, it's just two op amps buffering the signal. What's going on is your using pot/trimmer R4 to vary the charge time of capacitor C2. This is the same type of circuit you use for portamento or glide in the synth world.

The input impedance of 1M keeps you from loading the controller throttle circuit. From there, R1 is really there just to provide a current limit to protect the op-amp from a no power condition if C2 still has a charge. At full adjustment, the combination of R4 charging C2 should give about a 5s ramp from 0 to wide open. To decrease this total range, you can decrease the value of C2, down to say 470nF for more like 2-3 seconds range. Of course, the actual ramp time is controlled by the pot...the cap value is for the total range. The final output resistor gives low output impedance with a little protection for the op-amp. C1 is just a decoupling capacitor for the op-amp...probably not needed.

The diode D1, allows R4 to be bypassed in one direction, such that C2 can discharge directly through R1. Since the R1 value is low, discharge time is really fast...like 100uS or so. So, you get a linear ramp up as you hit the throttle...but get an immediate cut off when you let off.

I've built this type of circuit several times for other purposes, but never as a throttle interface...so some testing / adjustments might be required.
 
dam there are some clever people on this forum. Sounds like the perfect solution. I don't possess the skills to make one of these so I've ordered one from Lyen, I'll get a snap of it when it arrives to see how close it is in design to what you just planned out. Something like this should save a few fuses from popping and probably extend the life of those plastic gears in the mac.
 
Great subject "nonlineartom"

This has been an issue with me also, and i use a 40amp controler, will reprograme the controler to 30amps next. Should be fine because it wasn't too bad to begin-with anyways.

But I'm curious about how these circuits work for you guys, maybe one of you teck dudes can manufacture them for us for a fair price.
 
nonlineartom said:
Lyen PM'd me and claims to sell a magical device which fixes all my woes. Sits between the throttle and the controller, he claims it's an analogue circuit that slows the ramp up of the throttle so it takes a second or two to reach full throttle no matter how quick you slam the throttle, seems perfect to me, too good to be true?

After trying EVERYTHING, this is what was suggested to me. But i realized it was not a proper fix when i noted that partial throttle caused the motor to cut in and out trying to maintain the speed. If you experience the same problem ( usually at voltages higher than 36v and speeds faster than 20mph ), another controller is your only fix.

[youtube]AFyi34LfqfU[/youtube]

Long video, but i attempted to demonstrate the MAC/BMC & infineon problem here. People did not believe me and thought it was all sorts of other things.
 
I cancelled my order from Lyen, his module did ramp up and ramp down only, in my personal opinion having the throttle on after you let go off the throttle, even if it's for a second is just too dangerous, when I ask the motor to stop, i want it to stop NOW. I showed that schematic posted earlier in the thread to a friend who's very nifty with a soldering iron and he said he can build one with parts he has knocking around, using a pair of op amps. It might be a week or two until I get one to test but I'm sure he wouldn't mind selling them on to ES members at a fair price if they work well. On the design he's doing there will be a potentiometer to adjust the ramp up time so you can tweak it during the ride to really soften the blow the controller on takeoff, it will have a male JST one one end and a female JST on the other and sit inside a little weatherproof box to sit between the throttle and controller.
 
neptronix said:
nonlineartom said:
Lyen PM'd me and claims to sell a magical device which fixes all my woes. Sits between the throttle and the controller, he claims it's an analogue circuit that slows the ramp up of the throttle so it takes a second or two to reach full throttle no matter how quick you slam the throttle, seems perfect to me, too good to be true?

After trying EVERYTHING, this is what was suggested to me. But i realized it was not a proper fix when i noted that partial throttle caused the motor to cut in and out trying to maintain the speed. If you experience the same problem ( usually at voltages higher than 36v and speeds faster than 20mph ), another controller is your only fix.

[youtube]AFyi34LfqfU[/youtube]

Long video, but i attempted to demonstrate the MAC/BMC & infineon problem here. People did not believe me and thought it was all sorts of other things.


David, my controller doesn't seem to have the same problem as yours with the cutting out, if the wheel is freewheeling faster than the motor is spinning, the motor will still work, so I can be going down a big hill faster than the motors unloaded speed and the CA will show 0mph with no throttle, if I give it full throttle the CA will almost instantly go to full unloaded speed, still not fast enough to propel the bike any faster but the motor is still spinning inside slower than the wheel is spinning, I've never had a problem with the motor cutting out like you have, It's a cell_man infineon 12fet controller and so far it's spot on very little dead zone and I can make it really crawl along, but because it can deliver such high amp spikes, and the motor can take such a high load it does make it very flip happy with a hard poke of the throttle hence the need for this smoothing module.
 
recumbent said:
Great subject "nonlineartom"

This has been an issue with me also, and i use a 40amp controler, will reprograme the controler to 30amps next. Should be fine because it wasn't too bad to begin-with anyways.

But I'm curious about how these circuits work for you guys, maybe one of you teck dudes can manufacture them for us for a fair price.

I have been using my throttle interface for about 8 weeks now, It took awhile to find the best log ramp programming for the throttle but I now have it dialled in for my needs. There is no delay up or down the throttle range. The best way to look at it is like have a 3 speed switch wired in without the need to switch the speed: the first part of the throttle range is set for a 40% speed setting but the speed percentage increases as the throttle is twisted through its range upto 100%.
 
our controllers are nearly all speed based, meaning even an innocent little 1mm of throttle is enough to demand say 6kph (sounds fine, till you realise the controller will give you absolute full tilt until this speed is reached, then nothing once you are there) thus we come off the back from this 100% torque responce to our request for say 3-20% that we asked for/expected.

we need torque based throttle here, not speed, as its torque/or force that flips you. torque is what we feel and naturally need to be in control of and this is allready the case for car/bike gas motors that we are used to, and comfortable with.

seems straight forward enough, but unfortunatly in around 99% of cases, the human brain is not capable of accepting any of this, even once explained (in writing here or in person after they got flipped by a geared bike), so little progress can be made till people get there heads around it, more and more folks are asking about it now though, so give it few more years.. :wink:

-ps thankfully justin is onto this now with the new v3 cycle analist in beta testing with torque throttle conversion and programmable throttle curves, this could well be the answer cos even a 20kw+ beast will be nicely tamed for a slow wander up some stairs, but with full torque on tap (but only when YOU ask for it)
 
toolman2 said:
our controllers are nearly all speed based, meaning even an innocent little 1mm of throttle is enough to demand say 6kph (sounds fine, till you realise the controller will give you absolute full tilt until this speed is reached, then nothing once you are there) thus we come off the back from this 100% torque responce to our request for say 3-20% that we asked for/expected.

we need torque based throttle here, not speed, as its torque/or force that flips you. torque is what we feel and naturally need to be in control of and this is allready the case for car/bike gas motors that we are used to, and comfortable with.

seems straight forward enough, but unfortunatly in around 99% of cases, the human brain is not capable of accepting any of this, even once explained (in writing here or in person after they got flipped by a geared bike), so little progress can be made till people get there heads around it, more and more folks are asking about it now though, so give it few more years.. :wink:

-ps thankfully justin is onto this now with the new v3 cycle analist in beta testing with torque throttle conversion and programmable throttle curves, this could well be the answer cos even a 20kw+ beast will be nicely tamed for a slow wander up some stairs, but with full torque on tap (but only when YOU ask for it)

I have just found that a current limit based throttle is what is needed to take away all the twitchyness, less likely to wheel spin and makes the bike much more controllable for my project http://endless-sphere.com/forums/viewtopic.php?f=12&t=32042&start=105 :mrgreen: , my current limit is adjusted on the fly via a clutch lever but it could be straight forward to incorporate some sort of ramp up the current limit up in connection with the throttle.
 
gwhy! said:
toolman2 said:
our controllers are nearly all speed based, meaning even an innocent little 1mm of throttle is enough to demand say 6kph (sounds fine, till you realise the controller will give you absolute full tilt until this speed is reached, then nothing once you are there) thus we come off the back from this 100% torque responce to our request for say 3-20% that we asked for/expected.

we need torque based throttle here, not speed, as its torque/or force that flips you. torque is what we feel and naturally need to be in control of and this is allready the case for car/bike gas motors that we are used to, and comfortable with.

seems straight forward enough, but unfortunatly in around 99% of cases, the human brain is not capable of accepting any of this, even once explained (in writing here or in person after they got flipped by a geared bike), so little progress can be made till people get there heads around it, more and more folks are asking about it now though, so give it few more years.. :wink:

-ps thankfully justin is onto this now with the new v3 cycle analist in beta testing with torque throttle conversion and programmable throttle curves, this could well be the answer cos even a 20kw+ beast will be nicely tamed for a slow wander up some stairs, but with full torque on tap (but only when YOU ask for it)

I have just found that a current limit based throttle is what is needed to take away all the twitchyness, less likely to wheel spin and makes the bike much more controllable for my project http://endless-sphere.com/forums/viewtopic.php?f=12&t=32042&start=105 :mrgreen: , my current limit is adjusted on the fly via a clutch lever but it could be straight forward to incorporate some sort of ramp up the current limit up in connection with the throttle.


That seems like a good solution. At the moment I use the 3 speed switch just like you use your clutch level system. By switching to lowest speed I've effectively tuned the throttle resolution, so 10 degrees of twist equals 5mph not 15mph it allows for greater control when it built up spaces / pedestrians about, when on the open road and up to speed I click into the higher speed, which can cause a jolt if you aren't careful. Your solution seems really nice.
 
changing the throttle resolutions works ok, but having a current limit based throttle is a even better and a 100% fix :mrgreen: . I will have a play and see if I can incorporate a way of just linking it to use just a throttle as I am using the clutch arrangement at the moment which will not be very good on a pushbike setup.

edit: I think if the 3 speed switch was still used but selected between current limits this would be a very good compromise.
 
Will you have the same problem that you have when current limiting through the CA? that is, the system doesn't respond fast enough to override the throttle command. I can set the CA current limit down to 10amps, quickly twist the throttle and it will lurch forward with the full 50A for a fraction of a second before the CA realises what's going on and inhibits the signal to limit to 10A, because it doesn't sample the current value often enough to realise it's gone over. With a high torque geared motor this can lead to a real bum clencher.
 
nonlineartom said:
Will you have the same problem that you have when current limiting through the CA? that is, the system doesn't respond fast enough to override the throttle command. I can set the CA current limit down to 10amps, quickly twist the throttle and it will lurch forward with the full 50A for a fraction of a second before the CA realises what's going on and inhibits the signal to limit to 10A, because it doesn't sample the current value often enough to realise it's gone over. With a high torque geared motor this can lead to a real bum clencher.

Im still working on finalising the electronics on the bench at the moment but it looks very responsive so far with a fixed load of 50A but limiting to 20A, it over shoots between 4A-8A when you crack open the throttle but returns to 20A in around 100ms I dont know how this is going to feel when its on the bike and capable of 150A , I will have to wait and see. When I prototyped tested the electronics on the bike and winding the current limit up very slowly the watt meter did not see any peaks above 60A when I was testing initially so I think it will be fine. ( but the watt meter is pretty slow at capturing peak currents)
 
could you combine your method with a ramp up of the throttle response to stop the lurch? inhibit the throttle signal so that it can't spike to full throttle in less than 100ms. Are you using the wires that plug into the DP-CA or using your own shunt?

It was what I suggested originally in this thread, something that ramps the throttle input on a logarithmic curve, the only solution offered had a ramp down as well, which I deemed not safe as i want to power to cut out INSTANTLY
 
nonlineartom said:
could you combine your method with a ramp up of the throttle response to stop the lurch? inhibit the throttle signal so that it can't spike to full throttle in less than 100ms. Are you using the wires that plug into the DP-CA or using your own shunt?

It was what I suggested originally in this thread, something that ramps the throttle input on a logarithmic curve, the only solution offered had a ramp down as well, which I deemed not safe as i want to power to cut out INSTANTLY

I have tried so many ways now to try and get this under control, but Im sure this is the answer, I am using a 150A IC current sensor that gives a output voltage in relation to current http://www.allegromicro.com/Products/Current-Sensor-ICs/Fifty-To-Two-Hundred-Amp-Integrated-Conductor-Sensor-ICs.aspx , a additional ramp on the throttle could help as it will need only to be a very short , like you say needs to cover the 100ms, Once I have tested it out on the bike with the on-the-fly current limiting I will see what it will be like as a pure current base throttle i.e the throttle turns the current limit up this should help with the initial pulling away but will still have the potential for someone to go to mad with the throttle and it all going tits up.
 
I've done a couple of current mode throttle setups. It does give much better control and feels more like an ICE. The Allegro sensors are great. Quite accurate as well.
One problem you can run into is oscillation around the set point, which results in a pulsating or surging in the motor. Reducing loop gain is probably the best way to avoid this without slowing the response too much. If you're lucky, you never experience this.

I've also looked at some circuits that will slow the ramp up but keep instant ramp down. These are easier to make and don't have any posibility of oscillation but not as nice as a true current mode.
 
fechter said:
I've done a couple of current mode throttle setups. It does give much better control and feels more like an ICE. The Allegro sensors are great. Quite accurate as well.
One problem you can run into is oscillation around the set point, which results in a pulsating or surging in the motor. Reducing loop gain is probably the best way to avoid this without slowing the response too much. If you're lucky, you never experience this.

I've also looked at some circuits that will slow the ramp up but keep instant ramp down. These are easier to make and don't have any posibility of oscillation but not as nice as a true current mode.


:D It is one of your circuits that I used a starting point ( I hope you dont mind fechter ) , yes I did run into the problem with oscillation at the set point, I managed to minimize this by reducing the loop gain , like you say . I am waiting for a different opamp to arrive hopefully this will improve things even more.
 
I don't mind at all. The only way these things get perfected is by messing with them.
I think the loop gain can be really low and still have the desired effect. If it gets too low, it behaves like a normal speed based throttle. Somewhere in between, you get a 'blended' effect, which is not bad.

I tried all sorts of loop filters, but anything that integrates results in slow response. I don't like any delay in throttle closing. It may be possible to make a loop filter that integrates on increasing throttle but not on decreasing throttle. This would limit how fast you can increase the throttle, but not slow the decrease.
 
fechter said:
I don't mind at all. The only way these things get perfected is by messing with them.
I think the loop gain can be really low and still have the desired effect. If it gets too low, it behaves like a normal speed based throttle. Somewhere in between, you get a 'blended' effect, which is not bad.

I tried all sorts of loop filters, but anything that integrates results in slow response. I don't like any delay in throttle closing. It may be possible to make a loop filter that integrates on increasing throttle but not on decreasing throttle. This would limit how fast you can increase the throttle, but not slow the decrease.

The replacement opamp (TLC272CP) ( fechter, It may be worth making a note on your original circuit diagram ) as this replacement have basically cured all the crossover oscillations for the feedback values I am using and the circuit is working very well (yet to try it on a bike ) when its set to limit the current I have the feedback values set to give a 1 sec ramp from around the mid point current set point i.e if the current is set to 100A max then there is no ramp up to 50A then the next 50A is ramped over 1sec there is no ramp down. This I hope will work very well for my purpose because I have a clutch lever i can adjust the current limit on-the-fly whether I want that rampup or not. I will also have a play with it on a normal bike at sensible current levels to see how well this small ramp up works out even if you are not adjusting the limit on the fly.
 
Figured i'd update this old thread..

Going to an EB3 controller from cell_man cured all the issues i was having with stalling, the on/off throttle control, high speed wavering etc.

But the throttle was still too punchy from a stall to about 10mph.

So what i did was set my battery:phase amp ratio rather low 1:2.25 ratio.
This causes it to have lower torque at low speeds, but pick up torque at higher speeds. It flattens out the torque band.

This is good because it encourages me to pedal to get started. It also stresses the clutch/gears out far less. I have more control at lower speeds, and unintentional wheelies aren't a problem anymore.
 
Back
Top