"Parallel-drive" hybrid idea...

[thelittleacura]

I've been thinking/conceptualizing this idea for a while, but I've still got a few questions for you e-kart guys about the idea of the world's first Personal Hybrid Vehicle (a low-speed two-seater based off a kart designed for commuting). This project is still HEAVILY in the concept phase, and it would be my first vehicular project (besides auto maintenance), so it would take a while before I could start on it...

Preface: I'm a hybrid-tech geek. I know how Toyota/Lexus hybrids (and others that use a similar system) work, but just have some questions about a DIY hybrid. Feel free to use the Toyota abbreviations for the parts in this dicussion.

Engine (ICE): Planning to use a 212 hooked to a torq-a-verter. The throttle would be DBW (drive-by-wire) because hybrid. The engine will be started and power could be generated for the battery by an e-scooter brushed motor (MG1).

Electric-drive/assist motor (MG2)
I've seen @Functional Artist's success with 1800W 48V brushless motors, and think a motor like that would be more than enough to assist acceleration, allow re-gen brakes, and provide a "stealth mode" at under 15 mph (if you're light on the accelerator) solely using battery.

Engine-off question: How would a torq-a-verter handle the kart moving 15mph with the engine shut OFF for fuel efficiency? I'd imagine no problem, but IDK...

This is a HYBRID, after all, not an EV, so I'm planning on using a small ebike LiPo. I'll use a voltage/current monitor hooked into the PCM to co-ordinate charging (both re-gen and the engine generator). How fast can these packs charge, current-wise?

Gearing question: I'm aiming for 25 mph max, because street-legal "low-speed vehicles" can't go faster than that. https://wisconsindot.gov/Pages/dmv/vehicles/title-plates/nev.aspx
How should the ICE's final drive be geared in order to reach that speed? If someone can provide a number of teeth (assuming a 12T sprocket), I can calculate how I would need to gear MG2 so it wouldn't overspeed.

Registration: I'll talk with my DMV office about low-speed classified hybrids.

Software: Don't ask. Hybrids are 65% software, 35% clever engineering, so given enough time, I'll eventually figure the PCM coding out...

Thanks!
- the Little Acura

 

[itsid]

Hi and welcome..
first allow me to get the easy stuff out of the way first:
road speed is a matter of overall gear ratio and that does include the driven wheel diameter I'm afraid,
so no solid number can be given at that point.

if we assume a 13" rear however and the 12T TC sprocket that'd be a 74T axle sprocket for 25.09 mph (or a tooth more for a bit less )
if that engine is goverend to 3600 rpms (a rather stock value)
Again.. wheel diameter counts max rpms can be adjusted slightly so you can still dial in the top speed a bit.

The TC belt will spin and unfortunately it will also rub against the fixed sheave.
only a bit since the clutch will not be engaged, but still rub.
(also efficiency of a TC is less than 80%... not a very fuel efficiency promising figure..
we usually don't care too much about that but you might .. given the task at hand)
Maybe you're better of using a std centrifugal clutch instead.

Tough task you picked, I must say..
having something resembling a hybrid is easy throwing enough parts and money at it..
making it actually efficient is a lot harder

Power generating...
I'm not too sure a brushed motor is a good idea to generate power.
first it's fluctuating depending on the rotational speed,
and for charging you want a constant voltage, limited current source,
electronic nitbits to achieve such with a brushed DC motor are terribly inefficient or terribly expensive(.. sometimes both)

A Brushless DC instead has a slightly better chance of generating a solid charge.
depending on the desired voltage that is.
(since they're usually three phase motors, there's less sinusoidial drop and usually a higher volt/rev ratio so you only need to step down the voltage [easier than stepping up])
[charge btw depends on the actual batpack configuration you intent to use...
some happily accept 20 Amps, some cry when delivered 8... we cannot generalize that I'm afraid]

finding an e-start 212 dictates you skipping the horror fraud stuff (predators don't have e-start OOTB and conversion while theoretically possible is not yet proven to work [noone that asked about it ever showed us any result])
Soo Ducar 212 would be your engine then I'd say.
--but read on.. it might not be the engine of choice --

And that'd make three motors (incl the starter motor) on one kart plus the engine ..
Don't get me wrong, but it'd be better if you could reduce the partscount no?

So no estart 212, and a twinshaft brushless DC in the 1.8-2kW range
drastically reduces the partscount..
two manual clutches for the motor then woul allow you to
A) start the engine with it
B) have it generate charging currents with the engine
and
C) have it power the axle without it turning over the engine's crank

I think it should even be possible to drive the axle with the engine using the motor as a jackshaft but that'd be sketchy to say the least.

it won't be cheaper though, since manual clutches are rare and expensive for motors/engines in the 3/4" shaft size.
but it should be much lighter and more efficient I'd say.
maybe replacing one with a lawn mower style gear box almost makes sense..
(cranking up the motor speed having the engine run at a fixed 2200 rpm or such (wehere it's happiest)
and since most come with a neutral as well, you may be able to skip one manual clutch.
OR do get an estart version and use it's flywheel gear to power
one of the shafts (you will likely need to adapt a retractable pinion mechanism from a car starter)

Wiring and software will be something you need to think about a lot as well.

As I said tough task, especially since ideally you still want it to be lightweight, no?

'sid

 

[thelittleacura]

Toyota uses an "integrated starter-generator unit" (MG1) and an "electric drive motor" (MG2). MG1 in this arrangement would go where the pull cord was on the original engine (the other end of the crankshaft as the main drive sprocket).
This MG1 config would be REQUIRED to work this way, because when the ICE is demanded "on" (by a sophisticated algorithm that'd take into account your driving style, inputs from the DBW throttle/brakes, battery voltage, etc.) it would need to spin to ~1000 rpm nearly instantly. Once the engine caught, the integrated starter/generator would spin up, becoming a generator. The main challenge with this approach would be as you said maintaining a constant voltage, unless MG1 is BLDC, but that would require a 2nd controller with regen capability (which would be IMO less of a hassle than the DCT approach). See below...

Interestingly, VW, Nissan, and Hyundai use a very similar hybrid system to what you're describing. One motor/generator, two clutches, and a gearbox.
The "manual clutches" you described would be easily automatable, but that'd require 2 more servos besides the drive-by-wire throttle, brake (Acura uses DBW brakes because they're easier to coordinate with regen brakes), and auto-choke servo. Switching between modes would require precision, and it'd probably be hit-or miss, jerky, and slow-to-react with this system. The two-motor CVT approach would be a better, lighter (I'm planning on using small BLDC motors from ebikes), and quicker-to-react idea IMO...

 

[itsid]

you can use electronic clutches if you so desire.. no need to buy db1s and add a servo (which frankly requires a massive servo each, and an additional servo controller to power them..)
electronic clutches use an internal solenoid triggered by a simple on of signal, no noteworthy spring tension to overcome no need for complicated electronics
(a simple 12V 0.1A signal or such)
just check on clutches for industrial applications and you'll certainly find them.

The ISG mounted directly to the crankshaft might work, if you find a modern day car alternator that might already be about the size and power you want.
while it's terribly inefficient on working like a motor,
it happily provides 13'ish volts and sufficient charging currents at low engine speeds
I think the effective ratio is close to 1:1 and it works at 900 engine rpms or so...
with the idle speed of a 212 being in the realms of 1400 rpms
it should generate an okay amount for sure;
and certainly has enough power to start the tiny engine.

but still one more magneto-electric rotator to carry around.
But No, certainly not a second controller with regen capability
(that'd be the wrong approach)
a simple charging regulator (as the car would have) and a contactor to disconnect it while powering it with a simple high amp BLDC controller (hobby store grade will do as long as the amps are high enough)
the startup signal then is a three step process:
"power up the hobby BLDC controller", "toggle contactor to disable controller" and "enable regulator" done it can even do so on it's own by detecting the current drop as soon as the engine is running.

DBW brakes is amongst the stupidest idea I ever heard about tbh..
of course you cannot skip an electronic throttle in for an electric motor,
and maybe it even makes sense to controll the engine similarly in a hybrid,
but that's it.. DBW brakes or worse steering is just a fancy way of asking for trouble.

Wait what?? "small" BLDC motors?
I smell RC motors I'm afraid..
Forgive, me but if you think outtrunner RC stuff,
then I'm actually leaning back and getting some popcorn instead.
simply no way that project would make any sense whatsover immediately..
still interesting to watch, not interesting enough to further think about it though
(not for me that is)
RC motors are far less efficient and far less powerfull than you think
and worse far less rigid than you'd need

the few crossbreds with super low Kv (90-200) might power a skateboard, but even on bikes they're almost useless (hence people throw 4kW motors at their bikes still getting outperformed by a 500-750W vehicle drive motor by Bosch or Panasonic)

'sid

 

[thelittleacura]

I'm still sticking with the 2-motor CVT approach, because I believe a dual-clutch tranny approach would be harder to coordinate, less responsive, and clunkier in operation than the (proven since the '01 Prius) dual-motor and ICE approach.

You couldn't drive the wheels with a DCT hybrid while spinning up the engine. If you accelerated sufficiently to turn on the ICE for propulsion, there would be a power cut as the engine spooled up, then the vehicle would lunge forward as everything connected. Hybrid cars with this type of arrangement have a similar problem.

Power generation: I'll be pushing 48V through the hybrid drivetrain, so I can't use an alternator. I'll have to use an existing 48V BLDC motor to work as an ISG. An existing ebike controller with regen would be the only thing that'd have the proper electronics to work with the ISG.

Motors: Nope, I'm not thinking outrunners, I'm planning on using an 1800W 48V powersports-grade (similar to the other brushless projects on this forum) motor as MG2. I refer to these as "small" because some people use industrial-grade 1-or-2-hp DC motors in karts.

Drive-by-wire: EVERY current automobile has a drive-by-wire throttle system. Alfa Romeo, the 2017-present Acura NSX, and the 8th gen Corvette (future hybrid?) have DBW brakes. DBW accelerator and brakes make sense in a hybrid application because the PCM needs to know exactly what the driver is doing in order to coordinate two power sources and regen brakes effectively. By "learning" the accelerating/braking style of the vehicle owner through DBW, the PCM can also provide smoother, more efficient operation.

 

[snowjob]

Good luck! I see you are right up the road from me in Middleton. It looks like the LSV definition may fit for a licensed driver and insured vehicle. I have had nothing but problems running a go-kart in Verona and was considering an eConversion to see if we could make it work. Turns out shipping it to Grandmas was a better choice. Sounds like a fun project, but I enjoy speed too much to put that effort into 25mph.


Sent from my iPhone using Tapatalk

 

[itsid]

you know that alternators run in 60-70 volt ranges usually right? some even upwards of 100Volts)
brakepedal position sensing is not braking by wire.
a purely digital system can fail far more easily than a mechanical connection... especially if there'S electronics in between
if the throttle fails you just won't go any faster..
if the brake fails you just won't go any slower..
now which one would you want the car heading towards your kids that are playing on the street suffer from? riiight...

bike-controller is a terrible term here..
since dedicated bike controllers do not allow for reversing for example ..
motor controllers for electric vehicles in general do..
So let's say electric vehicle instead and not care much about the dumbed down bike versions too much
regen brake on EV controllers is a binary choice, on or off
a microswitch on the brake pedal is all they take as an input.
some can be setup to also trigger if the throttle is not engaged and the vehicle is above a certain road speed.. still either on or off (and everytime terribly inefficient really)
but you're right, the moment the ICE kicks in there will be a momentarily jerk in the drivetrain.
And it'll be more pronounced if you use the vehicles inertia to start the engine up.
if you don't the only downside would be that you'd have to release the throttle for a split second in order to start the ICE up.
(decoupling the electric drivetrain to startup the engine..once it runs, there will be absolutely no difference in matching the speeds no matter how it was started)

still on a two motor (uhm two magnetoelectric rotary appliances) setup,
one should be an alternator IMHO and it should take care of the regeneration of electric storages mostly since the ev-controllers regenerative braking alone will not help much at all.
alternators have far better recuperation,
and if you use that as a starter motor for the ICE, the terrible inefficiency as a motor doesn't matter much (it runs for a second a day as a motor or so.. not enough juice wasted to care IMHO)

Any setup is only as good as it's parts..
you can copy the layout of whatever mfg you want (toyota, honda or nissan, vw whatever)
and end up with a totally different result especially if you rely on code more than mechanical connection and self alignement,
you are only better if your code is absolutely pristine the tiniest hiccup in code,
a faulty sensor a humidity issue an unchecked unsmoothed position value of any sensor and you fall far behind in terms of responsiveness, comfort and so on.
So pick waht you're msot comfortable with, but don't claim it'll be better in the end until you have tested both with your parts
(or stick with it and call it "happy with it no need to try a different approach" of course )
better? nah
some hybrid vehicles work "better" than other ones even using the same style setup.
just like a V8 is not necessarily great it could still suck where other V8s rule

there are only few fixed points in this graph that you need to keep in mind:
a motor is a terrible generator and a generator is a terrible motor
so the smartest thing to do is to keep generation to the generators
and motivation to the motors as good as you can.

'sid

 

[thelittleacura]

If you say that "a motor is a terrible generator" and vice versa, why does EVERY hybrid use brushless motor/generators? From the research that I've done, a BLDC motor for an EV is similarly efficient at both motor and generator tasks. If BLDC was bad at making power, why would automakers use it for motor/generators?

As for MG2, the PCM would handle regen, based on whether the throttle was applied. No microswitch needed, the PCM would just short the regen pins on the controller when it saw that the DBW throttle was not applied.

Some hybrids work better than others even using the same setup because hybrids are 70% software!

 

[itsid]

well point me to some datasheets of the motor you are talking about specifically,
since frankly..
you cannot downscale full sized car motors too well.. especially not maintaining their relative efficiency ratios.
And I am not aware of any car manufacturer that actually uses a BLDC tbh.
(seen one in a window lift IIRC but that's about it)
synchronous or ac induction motors are quite common.
BLDCs.. seen none in the 50+kW range for a car so far *shrugs*

Now as of today no car manufacturer
not Honda not Nissan not Toyota nor VW has published
their recuperation efficiency data,
that unfortunately rules out we could know what the three relatively
tiny motors in the Honda NSX can do in terms of regenerative braking.
(they're small enough with none being bigger than 35kW toa ctually be BLDCs.. but I suspect synchronous motors tbh)
since the overall motor setup isn't exactly made to be recuperating
but to drastically enhance the handling of the car
(which they do extremely well as far as the internet says)
I think there isn't much to expect there really..

I mean a total electric power out (hopefully out) of 89kW isn't exactly breath taking,
split into three motors with the front ones able to actively work against the other two
makes rather clear that it's not actually a car made to be
especially efficient but rather especially fun. (while being more efficient as the competitors maybe)
So let us set your Acura aside..
it's exact motor specs are not published by Honda and I don't have high hopes
it'd work in your favour here anyways.

The one exception to the unpublished numbers of recuperation I was able to dig up is BMWs i3.
if you have anything else, let me know!

BMW hasn't published any datasheet either but at least they gave some solid numbers.
Namely:

acceleration in all tests 0.16g positive as well as negative.
So its 1345kg i3

with 60km/h
acceleration uses 74Wh,
constant speed of needs 8kWh/100km
recuperation would yield 47Wh when stopping to zero again

with a 100km/h speed (~60mph)
the acceleration uses 204Wh,
constant speed 13.9kWh/100km
and decelleration yields 129Wh of recuperation
​

Unfortunately that's the only solid numbers I can find online..

It uses a reluctance drive synchronous motor with 125kW (135kW i3s available)
reluctance drive making it's 250Nm of motor torque (yikes!!)

Anyhow..
it's rather save to assume that car manufacturers to not brag with numbers
that are comparatively crappy, right?
In fact those are rather promising figures with 60++% of batpack-road-batpack values

here's why:
Our everyday BLDC motor is about 85-90% efficient if it is allowed to run in it's happy state.. a bit worse in accelerating but usually ~70% there are doable.
unfortunately the hiher the peak efficiency, usually the steeper the curve and the less efficient they are accelerating ah well you can't catch 'em all I think..

the same BLDCs (all I have looked up regenerative efficiency for so far)
have a regenerative efficiency of less than 55% (50-52% usually some even far less)
if you found a better number.. please let me know! honestly that'd be a game changer.
And I'd be happy to know about them.

Now back to the BMW numbers
old and comparatively crappy car alternators also run at an efficiency of 50-60% indeed..
since many car mfgs gave a flying fart about them being more efficient for decades.
(they're sinusoidial three phase ac machines but very well comparable to a three phase BLDC still different however)

Modern day car alternators in more efficient gasoline cars however
(with start stop automatic and low engine speeds like the 1l three cylinder VW for example)
yield upwards of 80% in efficiency, some peak in the mid nineties.
so for every mechanical watt rotated into it's shaft more than 0.8Watt drop out of their charging ports.

And that's where it get's interesting.
assuming an efficiency of ~80% each way
(electric to mechanic and mechanic back to electric)
a total of 40% would be lost when recuperating in our nice number example
(average 80% motor efficiency when accelerating and average 80% alternator efficiency when decelerating)

BMW yields just over 60% (64'ish)
so I think I know why they bragged about it a bit

UNfortunately BMW doesn't use a BLDC at all!
its an reluctance drive synchronous motor.
no brushes indeed.. but still different bred.

Doing such with any small BLDC I've seen (in the <4kW range still for EV and not toys)
the best turns out to be ~87% efficient as a motor and only ~54% efficient as a generator
and thus it yields no more than a total of ~47% battery-road-backtobattery
in it's absolute best moments! (averaging below 30% usually since acceleration efficiency is lower and the only thing we can recuperate really unles going downhill that is )

comapared to the i3s 64.2% that's a really crappy number
And I'm afraid that'd be what you would expect from any BLDC currently.

Fun fact the BMW i3 range extender uses a classic "generator" setup
on the tiny combustion engine to charge the battery..
I personally don't wonder why
(engine puts out 28kW @5000 rpm the generator makes 26.6kW @5000 rpm so it's 95% efficient!)

Now in a small vehicle setup, you cannot get a good synchronous motor setup,
no ac inductors and for sake of vehicle weight and complexity a BLDC is indeed the goto solution
(heck some PMDCs are nice as well of course but let's not add to confusion here )

So there are limits we are working with of course
and there's nothing much we can do about it
unless we spend 10grand or such to develop a dedicated hybrid kart drive system

So using premade BLDCs is fine for a motor
but if I understood you correctly, you wanted the second BLDC as a generator only..
and that still is a terribly inefficient setup!
(recuperating 50% when you can get 80% or more by just using a generator instead)

In a ONE motor setup, you are perfectly correct to just pick a good BLDC..
power out is more important than recuperation in terms of efficiency for a moving vehicle,
but if you deliberately use two electromagnetic rotators...
why not pick the more efficient ones?

Oh and "but the controller has regen braking off the shelf" is NOT an answer,
since frankly that drops recuperation by another 10% at least.. those controllers can do that yes, but never well! (losing about 20% internally isn't too promising but unfortunately very common to keep the parts costs down)
making your 50% BLDC efficiency a 40% efficiency
converting mechanical into electrical energy is just terrible really.
So yes, pick a regen braking controller for your motor (every bit is good!)

But for a dedicated generator that is about as ideal as
using your wifes hand blender as a drill.
might work, far from a good result


'sid

PS again: if you found better numbers in your internet than I found in mine,
I'd be happy to know.

 

[thelittleacura]

An off-the-shelf alternator can't be powered to start an engine instantly without an off-the-shelf EV controller (apparently an alternator is pretty similar to a permanent-magnet BLDC at heart). I'd HAVE to use the existing controller's recuperation (because I'd need to spin MG1 up to work as a starter). That'd mean I'd lose 15-20% in the process of conversion, but it'd still be more efficient than using a BLDC motor as MG1.

Inefficiency as a starter due to MG1's design wouldn't matter much (MG1 only works as a starter for <1 second). But then again, you might be demanding the engine "on" every time you accelerate and "off" every time you coast, but that would assume you drove like an idiot, and the PCM hadn't "learned" how you drove and known to keep the ICE on. The alternator idea would still be more efficient for it's primary purpose, though...

 

[itsid]

pretty much what I said.. yes your ordinary alternator is about close to an ordinary BLDC
and yes it can (with minor modification) shortly be powered by a BLDC controller.
I'd say an EV controller wouldn't be necessary to start an engine
all you need is a short time of decent power to start the engine
and no actual 'control' about it nor sensing much less signaling or such;
so a cheap RC controller in the correct power realms should do already

it itself can be master-controlled by whatever controller you like and a small microprocessor for the finicky 'translation' bits... an attiny10 or such should suffice.

And correct I doubt the few seconds a day it has to run as a motor
matter much in terms of overall system effieciency.

I'd say you need less than 250W to start a 212 engine
Let's say it runs for 1seconds to start the engine
it then needs 0.07 Wh and even if we assume the efficiency of the alternator
to be only 50% when powered as a motor,
that means you'd lose 0.07Wh per start on average..
means to lose the equivalent of a quality triple-A Alkaline you need to start about 26 times ... I think that's not too much to worry about

Talking about "learning" machine learning can be a double edged sword
and is harder to implement than one might think for a dynamic system like that.
I'd rather drop to some old school fuzzy logic at first with dumbed down parameters
(say if electric motor is drawing more than XAmps start the engine to help, if current demand then drops to (X-Y)Amps cut off the helping engine again)
then maybe dial in X and Y at runtime when necessary and store values in EEPROM,
with two microswitches far easier to implement (less code, less preset parameters).
And maybe a fixed capacitance to start teh engine up in range extender mode.

if done as an 'override' switch so that
if you think the engine should start you press button A (stores and switches immediately)
and if you think it should've cut out already button B
storing previous values then gives you a chance of reading EEprom back into your computer
to see what some specific values were and get a better idea how to create the learning algorithm.

personally I'd say once you compiled a hundred values or so
you get a pretty good average that doesn't need to be updated programmatically too often.. if at all
but that remains to be seen.. good code is always an iterative process
In any case it'd give you a solidly working road test that can be tampered with at runtime
(to adjust) and you end up with real world values for your specific demands.
and you have some 'save mode' values you can revert to in case the learing algorithm has a bad hiccup somewhere in the middle of the woods (say some sensor drops dead)

'sid

 

[thelittleacura]

As for software, it's not going to be based on the current draw of MG2, because the vehicle's primary power source is the ICE. The accelerator pedal will be a "percent-of-torque-demand" pedal (like every DBW car). The system will choose the most fitting combination of MG2 power, ICE power, and MG1 generation based on how much torque MG2 can supply (compared to the estimated total system output), battery state-of-charge, estimated vehicle speed, etc.

The learning aspect would look at driver input over a long period and create an "aggression" value to tune the engine on/off battery state of charge (higher minimum SoC if driver was more aggressive) and throttle response ramps. Before the PCM had "learned" enough to start its tuning, generic values which would be reasonably efficient and responsive would be used as fillers. As the PCM learned, it would become more suited to the way you drove.

A Hall sensor on each front wheel could even allow vehicle stability control (if wheelspin or oversteer was detected, the PCM would ignore the DBW pedal's percent torque demand to slow the driving wheels down).