You are here

Delayed Startup with ADC Throttle

12 posts / 0 new
Last post
Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12
Delayed Startup with ADC Throttle

Hi,

I'm using an FSESC4.20 on a mid-drive ebike.

The throttle is in current control mode and mapped between 0.88v and 2.55v. When the throttle goes over 0.88v the green light comes on and I can hear electrical activity but no motor movement. As I increase the throttle the motor current goes negative. This continues until the throttle voltage reaches ~1.25v by which time the motor current is under -2A. Now the motor kicks in sharply and the motor current goes up past 3A. At this point I am able to backtrack on the throttle and find the low power control that's missing before startup.

Being on a mid-drive, this 'popping' startup gives a very aggressive whack as the motor takes the slack out of the freewheel. I can avoid this from standing start by applying pressure to the pedals before I throttle. But this is rarely possible when I'm moving. So if I slow to 5mph to turn I have to accept that no matter how softly I apply the throttle I'm just going to get a huge kick in the chain when the motor starts back up.

Any ideas?

district9prawn
Offline
Last seen: 1 year 7 months ago
Joined: 2018-04-26 12:18
Posts: 123

Are you on sensorless or hall sensors? This sounds like what my mid drive bike does in sensorless.

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

No, I have hall sensors. All seem to be configured ok... but then, how would I know.

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

I increased the minimum current to 2A and this seems to give a more consistent, controlled startup. The motor now starts pretty much when the electrical activity (bright light) starts.

However, increasing the minimum current has also increased the effective deadband of the throttle. With deadband set at 1% and minimum voltage set at 0.88v, electrical activity doesn't start until the throttle hits ~0.98v. This leaves quite a big dead space at the start of the throttle.

So now how can I get this 2A minimum current to start at 0.89v throttle?

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

OK, I have an interesting work around.

Turn off 'safe start'
Set min adc to 0.80v (throttle off actually reads 0.88v)
Turn the min current back down (0.5A atm)

Now as soon as I power on or release the brake, the throttle is active (bright light). But nothing happens because of the low power startup issue. However, I give the thumb throttle a slight flick (on/off) and this nudges the motor into starting up but at minimum power. This takes up the slack in the freewheel so now I can give as much throttle as I like without the aggressive kick. And the power doesn't stop completely until I brake. It's a bit like using a clutch.

Of course I'd rather it just started up properly in the first place but I'll take it out for a proper ride and see how practicable this technique really is.

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

Well, that technique works ok at higher speeds but at slower (walking) speeds it takes the slack nicely but the motor stops which makes it useless.

So now I've gone for a full on clutch simulation mode where on releasing the brake the motor will startup consistently at low power. This means that unless I'm braking, the motor is powered/spinning all the time. So when I do hit the throttle there's never any kick... just a bit of g-force. ;) And when I do brake and release it, the slack is immediately taken up with a very gentle kick.

This seems to work really well but I'll have to wait for a break in the weather to have a good ride on it to find any unforeseen problems and see how it impacts range. Of course, I also need to fit a toggle switch to the brake signal for a parking brake... A hairband over the brake lever isn't very elegant.

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

So the clutch simulation mode is great. It's intuitive, smooth and doesn't seem to impact range much.

Well, I say it's smooth but that's only until the motor starts juddering (repeatedly stop/starting). I don't know what's causing this but it can happen while starting, accelerating or decelerating (not braking). Usually it's quite gentle and overcome by a slight push on the throttle. However, occasionally pushing the throttle makes it much more aggressive.

On a full battery this was happening maybe every 2-3 minutes. On a low battery this was pretty much happening constantly. I'm sure there's a clue there for someone who speaks fluent vesc.

So as it stands now, I haven't got a clue where to start with this and it doesn't look like I'm going to get any help here (but thanks to @district9prawn for validating me). So I guess I'm stuck hoping I randomly stumble on a fix. It'd help if I could find some comprehensive doc's explaining all the settings. Anyone?

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

Well I've tried messing about with most of the settings including all the 'may help at low speeds'. I wouldn't know if I've tried the most appropriate values but nothing made any noticeable difference to my juddering issue. I also tried upping the minimum current to 5A (far more than I want for this) and even at this level the motor wouldn't consistently push against the chain to keep hold of the slack at walking speed. Yet at 0.8A with the chain off, it spins up every time and keeps going until I brake.

This is still better than it was with the initial standard setup though. As even with all the juddering, it's still relatively gentle on the gears. It's so frustrating though because it's so very close to perfect but so annoying that I'm not sure if I'll actually use it like this.

velolac
Offline
Last seen: 1 week 3 days ago
VESC Original
Joined: 2019-11-03 09:57
Posts: 41

Hi,

I am riding an ebike too, but with completely different setup, front drive with big gearing. So I dont need high torque and low speed from motor, as it can spin a lot. As a result I don't have a hall sensor, and rely only on sensorless mode. I think your problem is with the transition from sensored to sensorless. I have no experience with it but you should be able to find some info on forum to get it setup correctly. Also you may try to run it full sensorless to see how it copes.

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

Cheers velolac. It never occurred to me that it would be transitioning at such a low speed. But you're right. Looking at the erpm, the transition point is right around walking pace.

I did a quick test walking it up and down the kitchen. Changing the transition point both above and below the default seems promising. I'll get out a bit later and hopefully find a happy place for this value. And if I can't get a good transition, I'll give sensorless a go.

Cheers.

Marctwo
Offline
Last seen: 3 years 12 months ago
Joined: 2020-06-18 20:04
Posts: 12

I've done some testing and changing the sensorless erpm does seem to help at slow/walking speed but the juddering issue is still there, it's just better now at low speed. The lowest I can go with this setting and maintain consistent startup is 800. As soon as I'm going fast enough to keep balance I'm already well passed this level. This gives good performance but I still get the random juddering. At full speed the motor does about 20000 erpm so I tried various values upto 25000 (to avoid transition) but found no improvement. So it seems the sensorless erpm setting may well have been exacerbating the juddering issue but was not causing it.

pwd
Offline
Last seen: 1 month 2 weeks ago
VESC Free
Joined: 2023-01-05 16:20
Posts: 1

I'm also struggling with a similar issue as the original poster and am also using current control on an electric bicycle. The drivetrain makes a "clunk" as the motor spins up to engage the rear hub's freewheel; either from a complete stop or when in motion. One observation I have is that duty cycle mode starts up much smoother and at much lower RPM; nearly eliminating the "clunk".

When I switch to current control mode; the motor shaft begins spinning at a noticeably higher RPM vs "duty cycle" mode even with a load on it. Is there any way to improve "current" control type so that it starts spinning the motor at a lower rpm? Adjusting throttle curves doesn't seem to help.