500 Watt Mid-Drive

[Ron Z]

I have always been a fan of thru the gears drives. I have both hub motor and thru the gears bikes. I do a lot riding on crowded bike/walk paths (I live on the sea shore). A geared drive allows me to select the appropriate gear for conditions (slow/fast) and not burn battery energy heating up the hub because of inefficiency. With the motor spinning at an efficient rpm, acceleration is brisk and instant. To have more mounting options, I wanted to have a narrow drive that would fit between the arc of the pedals.

Shown in the first two pictures is my mid-drive with and without cover. The motor, a Neumotor 1907 (Kv 675 rpm/volt, Rm 0.03 ohm, Io 0.7 amp) is 1.5 inches long. The motor is run at 12.8 volt (due to high rpm/volt) at up to 40 amps continuous (500 watt). I use a standard Castle Creations 80 amp rc controller with a motor phase current sensor added to one phase wire (next picture). The peak motor phase current is limited to 60 amps. The gearbox, a Neugart PLE 40 two stage precision planetary, is 25-1 for a 20 inch wheel, and 32-1 for a 26 inch wheel. This motor can strip the gearbox if motor current is uncontrolled. Cooling is by conduction to the cover.





Shown in the next picture is a folding bike using this drive. I replaced the SA 8 speed (unsuitable for ebike) with a Nexus 3-speed internaly geared hub. The 3 speed mates perfectly to the drive. First gear is for hills, second gear is for cruising, third gear is for speed. Top speed is 19MPH. A comfortable cruising speed is 15-16 MPH.

A technical note: The Neugart PLE 40 precision planetary gearbox is too small for general ebike use. A better choice is the Neugart PLE 60, which I have used up to 1000 watts (later posts).

[recumpence]

So, how are you doing phase limiting? Please elaborate.

Matt

[Miles]

Welcome, Ron.

Good to see an example of an original mid-drive, here.

I'm curious about the phase current limiting, too.......

[liveforphysics]

Ron, you did it!

You're controlling current at the only point that matters for the motor and controller!

That is an allegro hall effect based linear current sensor, and the bridge you're looking at isn't a shunt, it's just a conductor to run current past the sensor. This enables an accurate yet FLOATING current reading on a phase leg, as the hall sensor is only magnetically coupled to the phase leg. This solves the phase leg current measurement troubles in a neat compact package!

Then you run this input into a uController (and it could be done with analog easily enough as well), which intercepts the throttle position, looks at the phase current, and decides what to output. It also easily enables true torque control as well for an RC bike! Yes, true torque control! The control solution we've been needing forever that makes the bikes so damn twitchy at lower speeds.

Ron you are awesome!

[recumpence]

So, how do we copy it?

Matt

[StudEbiker]

I look at it and it excites me, I read Luke's post and feel his exitement, but unfortunately I am not smart enough to know what he said. All in all though this seems like a VERY good thing the OP has created.

[liveforphysics]

So, how do we copy it?

Matt

Maybe Ron would be so kind as to share his circuit and code as part of the public domain?

Or if not, just seeing this clever compact solution for bi-directional phase current sensing is enough for it to be built from scratch in a day or two by someone with Burties skills or Method's skills, or RickNZ could likely build it in his sleep.

Every ebike, hubmotor, RC motor, etc, who wants torque control throttle will want this to interface with their throttle.
It's actually exactly what Fetchter's box was dreamed to do, only he did it on the battery side which did more harm than good for RC controllers. This does it where it counts, on the phase leg.


Awesome awesome work Ron!

[John in CR]

Torque control throttle, yes I need that! Put me on the "want to buy some" list. I only wish we could recruit Ron into the power craving crowd.

[Ron Z]

Liveforphysics is exactly right in his description of the motor current control loop. The problem is I did not exactly do it that way.

I use an Allegro Microsystems active current sensor in one phase wire. Unfortunately, it is a unipolar sensor whose output goes to a single rectifying diode and is filtered. I made the rash assumption that the motor will always be spinning. Two guest riders have stripped gearboxes by trying to start out in third gear. The motor is essentially stalled at that point, and if the phase current is in the wrong direction (not being measured) the current control does not work. I never changed the circuit, because it is easy to always start in first gear. A Nexus or SRAM 3 speed internal hub can be shifted at any time.

I do not have a computer ready schematic, but will provide the circuit in a day or two, with caveats.

[liveforphysics]

Ok, not quite perfect, but still damn good Ron. And better than anything we have for BLDC torque control by a mile.

So, two of those sensors in series (one flip-flopped). Or 4 of those sensors if you wanted to have 2 phase legs (for real perfect torque control), a comparator and/or bi-input op-amp to combine the signals and send this signal to a uController and you've solved the deficiencies of your original design.


This is huge.


Ron, you Rock!

But we are going to have to get you out of the 500watt thing. 500watts is like an acceptable no-load current.

[Kepler]

Can someone explain to me again why phase current at low RPM and high load isn't proportional to battery current.

[John in CR]

All it takes is one test ride to get him out of lazy biking mode into how do we make these fun machines maximum fun. Powerful ebikes work just fine when you want to go slow too, and the difference in efficiency is minimal at the same speed. Rons little box solves being twitchy at low speeds, and I suspect it will pay even greater dividends up hills, because it will be easier to know when we're asking too much and give a means of controlling it. Even with the 24fet controller I still cringe when traffic slows me down on steep hills, especially since the retards in cars lack the ability to maintain a constant speed.

[John in CR]

Can someone explain to me again why phase current at low RPM and high load isn't proportional to battery current.

It is, but it gets chopped up into such short pieces that the phase current pulses can be shockingly high. I blew one controller on a 200 yard 0 grade road at less than 5mph because of it. The road was very bumpy forcing me to go very slow, and many of the bumps caused slight throttle pulses. Pulsing throttle at 3-5% avg duty made the battery current limits of the controller irrelevant, and PWM caused the phase currents to spike probably over 200A+ in a controller set for a 20A limit at the factory that I modded to 50A battery side limit.

As I understand it, Ron's black box gives the controller real time knowledge of the phase currents, so they can be limited more directly. It still limits via PWM, but knowing instead of guessing the phase currents is the key.

[liveforphysics]

Can someone explain to me again why phase current at low RPM and high load isn't proportional to battery current.

Because the inductance of the motor and the PWM from the controller make a bucking DC/DC converter.

I was just doing some low speed tests on the motor dyno at work, 5.5amps of battery current (at 50v) was generating about 65amps of phase current (at ~4v phase voltage.)

Bucking the current over 10x what was being drawn at the battery.

[Kepler]

OK that makes sense. Thanks guys.

I have been wondering if I needed to add a second current sensor to RC ESC interface I am involved with. The present design reads battery current (and wheel speed) but sets a max current against a wheel speed map effectively stopping you loading up the motor too much at low RPM. We also use information from the current sensor for battery usage information so its not just a simple matter of moving the current sensor to a phase wire. Hence I am thinking that having current sensors on both the battery lead and the phase lead will give the best results. We have a a few inputs left on our processor so its not a problem to add the second sensor. Just adds a bit of complexity to the installation. The current sensor we use straps to outside of the cable so theres no need to break into the cable.

[katou]

How do these planetary gears last? How long have you run one of these setups?

I understand just enough about motor control to recognize that you have created one of the missing pieces in the big jigsaw.

Very clean! It looks like you sell them commercially? If so, where?

Katou

[Miles]

Ron's website: http://ronzebike.com/Home_Page.html

[spinningmagnets]

I've read this thread a few times, and though I'm certain I've only grasped the tip of this particular iceberg of awesomeness, I feel drawn to it.

I'll have a garage to work in soon, and one of the projects I'd hoped to do was to make up a dozen or so hall-sensor plug-and-play kits for the 63mm RC motors. Although,...when I approach anything electronic, I feel like someone who's learning eye surgery through a mail-order course, and "would you mind helping me with my homework?...all you have to do is lay there still"

A $100 Lyen/Xie-Chang 6-FET with an RC-hall kit would still find some customers, but a tiny RC-ESC with one of these current-sensing/limiting boards 'sounds like' it has the potential to be just as good at "low-RPM sensorless ops", while also being smaller and lighter. I realize halls have proven to be reliable, but occasional hall failures still occur from wet weather shorting and overheating.

Its seems that systems which allow a smaller motor to use two or three gears (like this one) have the potential to greatly extend the range of a small battery pack, compared to the watt-draw of a larger motor with a single reduction to the axle (though I would still aim for a minimum voltage of 22V-24V to increase battery options)

[Ron Z]

Shown is the circuit that provides the motor current control function, with interfaces to the current sensor and the throttle. I tried to draw this on a drawing program, but it takes too long, with learning curve.



This circuit requires an unipolar sensor.
The current sensor is an Allegro Microsystems ACS 755-130, avalable from Digikey. The sensor is out of production, but Digikey reports an inventory of over 500.
The dual op amp is an LM 324 or equivalent, available from You-do-it Electronics.
The rest of the parts are available from Radio Shack, or other hobby electronics distributors.
All passive pats are +- 20%.

This function can be implemented in a microcontroller (beyond my skill set). The sample rate should be at least:
(number permanent magnet poles) x (Max rpm / 60) x 10 This may be be a problem if the micro controller is doing other functions.

I've selected the components to work with most systems. Ra might have to be selected for other motor, voltage, controller drive systems, or other current sensors. If the current control function is too slow, reduce Ra (by factors 0f 2 -5). If the motor current hunts or surges, increase Ra by the same factors. (hunting or surging on a high power system could get exciting.) Iterate into optimum response for your application.

[Ron Z]

The precision planetary gear box is rated at 20000 hours at full load. My daughter has used my first prototype nearly every day, to "walk her dog", for the last three years. If you ride 5000 miles at 10 MPH, thats 500 hours.

These gearboxes use staight cut gears, but because they are precision, they are very quiet. Just a very faint whine.

[katou]

I totally just cut and pasted that. Thanks a ton for the longterm durability info, that stuff is often very hard to come by.

Hopefully some of the more electronical minds will ring in on this!

Katou

[Starson]

Ok, so now what??? It's been a week and what are you going to do with this missing piece of the puzzle? Who is interested in this and what are your intentions with the use of this information? It sounded like something great had happened and then ,... nothing! What's up???
Gary

[spinningmagnets]

I'm sure a new thread will be started on this soon. There are several electronics enthusiasts here who have programs that can quickly show how to design a circuit board to perform these functions, and also how to populate the board with appropriate components so that there are no physical conflicts. Ron has stated that the circuit is working well now, and since he published the schematic, it can now be vetted, and perhaps even imporoved in some small way.

But it really is an important advance, and also really deserves its own thread. I would call it a 'feedback loop' to enable smart phase-current management.

[flez1966]

diptrace is good and not too hard to learn

[MattyCiii]

WOW that is great workmanship. I look at the detail of the whole bike power system and am floored. THANKS RonZ for posting the pix! Thanks Miles for posting the link! I'm in Boston, so now it's time to head on up to Marblehead. One of these electric bikes would be perfect for the girlfriend.