Make: Fechters' Braking Regen Add-on for BLDC Motors

[steveo]

ITS THAT SIMPLE....



i will try it out on the weekend and let you know if it works..

-steveo 8)

 

[steveo]

Fechter... ggoodrun .. i've finally added the resistive braking .. working good so far .. very impressed with it to be honest ..!

added on resistive braking .. about 18feet of 22awg .. with a 20.5 amp switch to active the rear brake.

 

[fechter]

Cool. See how hot the resistor gets after a long hill. I was thinking you could wind the wire around an existing part of the bike frame, but your frame doesn't have any good spots.

The way the Crystalyte motor is wound, the braking will heat two of the three windings. Since the windings overlap, the heat transfer between windings will be quite good, so I don't think you'd ever have an issue with uneven heating.

I'm still working on a true regen version.

 

[GGoodrum]

Very cool. 8) The version on my wife's bike is still going strong. It provides a whole lot better braking than the rear hub brake in the Nexus 8 hub, especially at higher speeds. The faster you are going, the more braking power you get, which I think is perfect.

-- Gary

 

[steveo]

Very cool. 8) The version on my wife's bike is still going strong. It provides a whole lot better braking than the rear hub brake in the Nexus 8 hub, especially at higher speeds. The faster you are going, the more braking power you get, which I think is perfect.

-- Gary

Exactly!.. thats why i think its so great! ... and it all at the convinece of pushing a button!

 

[Toorbough ULL-Zeveigh]

People have no idea what they're missing out on & no amount of explaining it is as convincing as that first grab-ass sensation.
It's like when the first time you hear stereo, then it's like 'whoa baby, where have you been all my life'.

 

[diver]

OK, here is the simplest possible configuration:


By only tapping two motor wires and no bridge, the braking current will be only passing through 2 of the 3 motor windings.

If you were doing really long downhills, it could possibly cause the motor windings to overheat, but with an X5 motor, overheating would be pretty hard.

The two coil configuration can evenly spread the load in the motor windings, but requires an additional switch contact. This may be unnecessary.

The length of wire in the resistor coil depends on what kind of motor you're running and your top speed. A 5303 will need fewer turns, a 5304 or 05 will need more turns.

If i understand this right, once i release the throttle and apply brake, the motor makes a current that goes to the resistor coil and thats it.

 

[fechter]

Yep, that's it. When braking, the controller is not doing anything. If you make the coil of wire too short, you could skid the wheel. The coil can wrap around the bike frame if you have room.

The thing I would worry about is the torque arms and dropouts. The braking force could be stronger than the acceleration.

 

[diver]

wow thanks so much for the direction. with that much of a resistor even if the brake and throttle were engaged at the same time accidental, it would not harm the controller, am i right thinking this ?

 

[fechter]

with that much of a resistor even if the brake and throttle were engaged at the same time accidental, it would not harm the controller, am i right thinking this ?

Yes, the resistor should protect the controller.

 

[GGoodrum]

I second the comment on torque arms. When you brake, the force is pulling the opposite direction, so you really need two torque arms, one oriented in each direction. This is especially true for front-mounted motors, with the standard "J"-shaped torque arms. Here's what mine looks like:

This setup is pretty rock solid. I'm using a 3-phase rectifier on mine, and originally had two coils, with two buttons:

One of the buttons was wired to just the larger coil, and the second one through both, but I found out the single coil action provided way too much braking force. With both coils active, the braking force is perfect, and proportional to the speed. The faster you are going, the more braking force you get. My wife now only uses the rear hub brake to stop completely.

-- Gary

 

[Dr. Shock]

Would this work equally well on a brushed motor? Simply switch the motor wires into a big resistance coil?

 

[fechter]

Would this work equally well on a brushed motor? Simply switch the motor wires into a big resistance coil?

Yep. The first version of this I used on my brushed Zappy motor.

 

[EMF]

What about a 408?

 

[fechter]

It will work with a 408 too, but the length of wire needed might be different.

If the wire is too long, the braking effect will be weak. If the wire is too short, you risk getting thrown off or breaking something.

 

[lazarus2405]

Update! Fechter, seemingly at my prodding, has been working on the regenerative braking system some more. Here's a summary:

The "new and improved" regen setup avoids using the bridge by using a brushed controller that has 6 FETs (it's what I had lying around). Something with 3 FETs would work too.

The FETs are normally all in parallel in a brushed controller, so I separated the drains and kept the sources connected to ground. Other than that, it is the same as the bridge rectifier setup. The advantage is there is no bridge, which would not perform well under switching conditions, so the amount of power actually recovered to the batteries would be more. A bridge rectifier would need a large heat sink, which would be eliminated by going direclty to the FETs.

So, let's see if I understand this. The brushed controller has a bunch of parallel mosfets used to PWM the battery current into various DC voltages dependent upon the throttle signal and then the PWM gate driver. In normal operation, all the gates are connected in parallel to the driver, all the sources are connected to the battery, and all the drains are connected to the motor. The throttle governs how the MOSFETS modulate the voltage to the motor, thus controlling the motor's speed.

Erm, wait... these controllers do the controlling on the negative side of the circuit... I may have mixed up the sources and drains? Please correct as needed.

In our modification, we're running things backwards. We're disconnecting the drains of three of the FETs from the controller and connecting our motor phase wires to that. The source is connected to ground, which is the controller's battery negative. Then those three FETs are used to rectify the 3-phase AC into high-voltage DC, drawing more or less current based on the PWM signal based on the throttle.

I think you got it pretty much right. Inside the brushed controller, the M+ and B+ wires are connected. The FET drains go to the motor, and the sorces are grounded.

The general configuration looks like this: The B+ and B- wires go to the battery. Very little current will go through these, so the wires don't need to be big. The motor (-) buss is chopped into 3 parts and phase wires attached to each part.

When the FETs turn on, they all turn on at the same time because the gates are still tied together. This shorts out the motor windings. When the FETs open, the collapsing magnetic field will circulate current back through the body diodes in the main controller. The freewheel diodes in the braking controller won't be doing anything, so they can be removed. The PWM duty cycle of the braking controller will determine the amount of braking.

When the braking controller is off, all the FETs are open, so there is no drain on the main controller.

And fechter's test unit, a Vego controller, being modified for the phase wires:

Here's my test unit before

I then cut the motor- buss into 3 segments with a dremel tool:

Then I added wires to each of the 3 segments to bring out to the phases:

Then when where does that current go?

When the regen controller shorts out the windings, current flows through it. When the PWM opens the switches, there's nowhere for the voltage to go execpt through the body diodes in the main controller, back into the batteries. If the CA is in series with the batteries, it will read the regen current. Even at very low speeds, you will get regen back into the batteries. The regen current will be passing from the motor back through the main controller (even though it's off).

So, on with the show!

 

[lazarus2405]

After fechter's suggestion that I look at the TNC Scooters controllers, I hit upon this page, http://www.theworkshop.ca/energy/dirt_e/2/2.htm, from Nov 2007, where they disassembled a YK42 controller and beefed it up with components on a daughterboard for use in a dirtbike. Apparently, this was with fechter's help. :wink:

The inside of a YK42-x controller:

For this mod, the regen controller needs to withstand 84v in order to charge the batteries. That means replacing the FETs with 100v ones, and hopefully nothing else. (The caps won't need to be replaced, will they?)

I'm looking at the YK42-3 and -4 models now. The -3 is rated for 36v 40a, while the -4 is rated for 48v 40a-100a (whatever that means). The price difference is $10. Either one's voltage rating is meaningless, since the FETs will be replaced. The current rating, however, is interesting. These controllers use a shunt to limit the current just like clyte controllers, but the -4 may be built heavier for the extra current. The higher the max current, the more powerful the regenerative brake can be, stopping the bike faster.

The higher current, though, means a higher charging current into the batteries. Nice for those running a123s, but my e-molis can only be charged at around 3C. With my current pack, that's only 18a, and 27a with my planned pack in the future. Looks like I'll have to lower the current limit. As far as braking, that's still >1500w flowing from the motor. If that isn't enough power to stop an ebike, something is very wrong.

So, your thoughts, fechter? How is your testing coming? Promising? Should I pull the trigger, get a $28 YK42-3, and order some 100v FETs? And for the FETs, what do you suggest? I can get 4110s for $2.50, but they seem like such overkill for low current and very intermittent use.

 

[fechter]

That one looks like it should work great. Only the FETs need to take the full voltage. They might take quite a bit of current though, so 4110's would not be overkill. The FET current might be several times the regen current to the batteries. The freewheel diodes get removed, and the main capacitor and logic circuits can run at whatever voltage the thing was originally made for. The trick is to isolate the voltage regulator and make sure the input of that stays at the right voltage. On that model, there's a 78xx regulator fed through a resistor. If you pick the right resistor value, you can have just about any input voltage. The main cap does not need to be (probably shouldn't be) across the full pack voltage. It might be best to just remove it, or replace it with a much smaller capacity one that has a high enough voltage rating. The power wires on the braking controller just supply enough for the PWM and FET gate driver circuits, around 100ma most.

No test yet...

 

[lazarus2405]

Of relays and ebrakes:

I want to set up both the regenerative and resistor coil setups, so I can have them both as the situation warrants. I want both sets of brakes to override and kill the main throttle when engaged, but I need a little help with diagramming it. So, the basic things I need to know:

How is the throttle override done? Is the Thr line normally grounded and the throttle is killed when opened, or the other way around?

What do I need to drive a relay so that I can mount a NO ebrake with small wires on the handlebars to close the circuit for the resistor coil brake?

How can I use a Hall sensor throttle on the regen controller to kill the main throttle? Maybe a 2v Zener diode?

Edit: I now see that the clyte controller have a connector for ebrakes. Odd, though, that I see it is 3 pins, while the ebrakes I see only have two wires. Which wires are for which? Also, I see that they are designed to replace the stock brake levers, to kill the throttle when normal friction brakes are applied. If I was using the ebrakes to activate a relay to engage the resistor brake, I'm not sure whether that would be desirable. I mean, friction brakes would have to be connected anyway just to be the spring that holds the switch open, right?

Also, could the resistor brake be engaged as a parking brake? If the motor was at rest with the windings shorted through the coil, would it resist turning? Would it lock the wheel? Or rather, what would be an effective way to electrically lock the motor while parked with the controller off, so that someone attempting to roll the bike would find it very difficult?

 

[fechter]

I'm still head scratching on that one a bit.

You can kill the main controller by using the brake switch (not sure which pins activate it on the Xlyte) OR you can pull down the throttle signal.

Another idea I had was to use a mechanical switch on the brake lever that switched the +5v between the main throttle and the braking throttle. This way there would be a little dead time where both were off preventing any chance of both being on at the same time (smoke). A relay could be used here too, if you wanted to use an existing switch or magnet activated reed switch.

I think it would be really cool to stick a magnet on the brake lever in such a way that it could throttle the braking action. A hall sensor from a throttle could be mounted on the stationary part of the lever housing.
I'm not sure how easy this would be.

Another approach is to simply use a fixed amount of regen throttle set by a pot. Relay or switch activates regen.

The resistor brake could be activated by a relay or with a sufficiently large push button switch. A lever actuated microswitch might be good here.

When parked with the resistor brake on, you would still be able to roll the bike slowly. The braking force increases with speed.

 

[lazarus2405]

So, I've been working on it, and I got inspired by some throttle discussion in one of the recent Kelly controller threads, and here's what I've come up with:

Two ebrakes from TNC. Connect both to the ebrake pins on the main controller. Then, connect the mechanical brake lever part of the ebrakes to lines connected to boxes, in the manner of the waterproof throttle thread, here: http://endless-sphere.com/forums/viewtopic.php?t=2522

For the ebrake controlling the regen, use a linear hall sensor in the place of that slide pot. Thus, a hall-effect throttle for braking that, well, looks like a brake.

For the coil brake, I'd use another ebrake connected to another box in such a way. The thing I haven't figured out yet is how to wire that. I'd like to make the coil extra long with taps and switches at various intervals, and then switch them based on the pull of the cable from the ebrake. That way, a little movement on that ebrake would activate a switch at 30' of wire, more movement would switch to the tap at 25' of wire, etc.

So, my first diagram with ExpressSCH:

 

[GGoodrum]

With the wire coil-type ebrake, you get proportional braking, based on speed. The faster you are going, the greater the braking power. My point is that you don't need a bunch of different taps/switches.

-- Gary

 

[fechter]

The "pot box" setup with a switch would work.

The way I'm thinking of wiring it on my scooter is to use a SPDT switch to toggle the 5v throttle power between the main throttle and the braking throttle. This ensures that both can't be on at the same time.

It would look something like this:

 

[lazarus2405]

With the wire coil-type ebrake, you get proportional braking, based on speed. The faster you are going, the greater the braking power. My point is that you don't need a bunch of different taps/switches.

-- Gary

D'oh! Forgot that one!

 

[Nimbuzz]

Wonderful work men! I'm lurking here as my dream of regen is invented before my eyes by the heroic underground engineers of the future. It's a bit over my head so a quiz if I may;

-- Does the most recent circuit have any provision to adjust the charge rate so the pack will receive the optimal current for charging and longevity?

-- If not, might the coil/regen dual system have an adjustment allowing this optimal amount to go to charge and any excess to go to the coil?

Thanks,