E-bike conversion economics.

Buk___

10 kW
Joined
Jul 28, 2017
Messages
750
The basic question is can a competent DIYer put together reasonably efficient, reasonably light, reasonably powerful(legal, or legal+ a little bit) from parts, significantly more cheaply than buying a low-end but reputable kit.

(And yes, I'm aware that its a big, fluid and ill-specified question.)

Eg. I'm fairly sure that a competent electronics hobbyist who already has the knowledge and tools to make and populate his own circuit boards would be able to manufacture a controller for less than an equivalent item could be bought; but as someone with minimal electronics and no skills/tools/experience of putting circuit board together, it is unlikely that I could do so.

Eg2. A 36v 15Ah battery kit with mounting kit and charger seems to run about £300+-. I see posts suggesting buying powertool batteries and I see 12v3Ah packs for £22, 15*22=£330 is no saving; and that before you consider charging balancing, modifying to reduce combined caseweight etc.

My thinking is that unless there is some area or component of an e-bike setup that I can make a substantial saving by doing some DIY, a pre-assembled kit is probably the best option for me?

Anyone care to dissuade me of that conclusion?
 
Buk___ said:
The basic question is can a competent DIYer put together reasonably efficient, reasonably light, reasonably powerful(legal, or legal+ a little bit) from parts, significantly more cheaply than buying a low-end but reputable kit.

I suppose that would be a qualified "yes", if by "parts" you mean finding separate battery, controller, motor, and other assorted bits and bobs, vs buying all those in a box supposedly already wired up to plug right in and go. ;)

With a part of the qualification being that shipping costs for individual items can easily be much more than for all those same things in one box, depending on where they come from and what customs/etc they have to go thru, etc.

And another part of the qualification being that you already know what you are looking for, and which things will work together and how to wire them up together (changing connectors/etc as needed).



Eg. I'm fairly sure that a competent electronics hobbyist who already has the knowledge and tools to make and populate his own circuit boards would be able to manufacture a controller for less than an equivalent item could be bought; but as someone with minimal electronics and no skills/tools/experience of putting circuit board together, it is unlikely that I could do so.
For a basic controller, probably not--they are so cheap that it would be cheaper to buy one, especially since powerstage layouts are not all that easy to get "right" without experimentation (costly in parts blown up and time wasted).

For something more advanced like the Lebowksi units, it's another story, but even those, unless you use a powerstage layout already perfected by someone else, would need developing that just like the other.






Eg2. A 36v 15Ah battery kit with mounting kit and charger seems to run about £300+-. I see posts suggesting buying powertool batteries and I see 12v3Ah packs for £22, 15*22=£330 is no saving; and that before you consider charging balancing, modifying to reduce combined caseweight etc.
Just keep in mind that the volts and capacity of a battery are not the only characteristics you have to look at. You ahve to check it's C-rate, or find out what cells it's actually made from, to know if it can actually power your kit at the current draw the kit will need in the way you use it, without damaging the battery over time.

Also, many cheap batteries are badly made, or made from (literally) garbage cells. Not all of them; some are great bargains. But you have to be careful what you get, and pick from what others have successfully used for a while, so you know it's good and can handle what you want out of it. Anything that hasn't been tested by others is a crapshoot.


My thinking is that unless there is some area or component of an e-bike setup that I can make a substantial saving by doing some DIY, a pre-assembled kit is probably the best option for me?
It's possible to save quite a bit in some areas, but it's often a compromise between performance (or weight or size or other characteristics) and cost.

For instance you can build a battery out of cheap or even free recycled laptop or power tool cells (there's a few threads about this). But because these cells vary a lot, there's a lot of time invested in testing and sorting them. Also, they will generally be lower performance than new ones, so you end up needing a bigger battery to make up for that. Sometimes a lot bigger, so it's heavier and takes up more space and is less practical to carry on the bike. Depends on the power level and range you'd need from it.



There's even DIY options that can cost nothing but time, for motors and batteries and everything else, if you collect junk others don't want, off curbside collection or using things like your local version of Freecycle or Craigslist or whatever. Or even keeping an eye out here on the Online Market section, where occasionally people give away old stuff to those willing to come pick it up. Take a look at my DayGlo Avenger thread (and http://electricle.blogspot.com posts) for various DIY motor options. And CrazyBike2's early powerchair drivetrains. But they took a lot of time and iterations to get right...even if they cost almost nothing to make.


Hubmotor kits are so cheap now that the only really expensive part is the battery (because many of the cheap ones are garbage).

Middrive kits...not quite as cheap.
 
Ready-made batteries are more or less the same cost as a self-made one. I doubt that many people could make a controller for the same price as what you can buy one.

To answer your question, I'd say that the answer is yes, but the most important benefit of converting a bike to electric is that you can get it exactly how you want by choosing exactly the components that you need to get the perfect match. OEM electric bikes always seem to be lacking in some area, like speed, range, weight, torque or control system. Then, there's the cycle parts. On the whole, that's not normally such an issue, except some of the cheaper OEM e-bikes don't have suitable gearing for their speed.
 
d8veh said:
Ready-made batteries are more or less the same cost as a self-made one. I doubt that many people could make a controller for the same price as what you can buy one.

Whilst looking around at phone/tablet batteries -- the softpack 'non-replaceable' type -- on the basis that they are:

a) the most ubiquitous;
b) need to be light;
c) must withstand daily recharging for at least 1 year if not 2;
d) fitted to the fastest evolving and most competitive product in history; and will therefore have been developed further and faster whilst needing to be 'safe'
(Note 7 and a few Iphones excepted, which was mostly down to trying to fit a quart in a pint pot; which doesn't affect e-bikes as space isn't at a premium.)

I found some that are 4400mah @ 4.35v nom. for ~£1.95.

Assuming I'm doing the math right (no guarantee), then 3 parallel banks of 9-in-series gives 39V nom. and 13.2Ah for ~£54. The packaging wouldn't be a problem -- unless someone tells me otherwise -- just don't try to pack them into too small a space.

The charging regime is a little more complex. Looking at RC forums, they seem to have various rules of thumb like: never above this voltage when charging, never below that voltage when discharging. The technical references talk about Delta-V detection for constant-A fast charging followed by a v.low trickle charge to compensate for self-discharge; and then complicate that by specifying hard-limits for temp rise rates; which vary from brand to brand.

But then the C-rate stuff cropped up and I do not 'get' that yet.

d8veh said:
To answer your question, I'd say that the answer is yes, but the most important benefit of converting a bike to electric is that you can get it exactly how you want by choosing exactly the components that you need to get the perfect match. OEM electric bikes always seem to be lacking in some area, like speed, range, weight, torque or control system. Then, there's the cycle parts. On the whole, that's not normally such an issue, except some of the cheaper OEM e-bikes don't have suitable gearing for their speed.

My current thinking is a XiongDa 2-speed -- front or rear I haven't decided -- to fit my £150(new) BSO. As you can tell from the above, I'm not into spending large sums of money on stuff -- retired early on a fixed and modest income -- but I do have time, some pretty good mech. eng and computing skills.

If memory serves me -- his pages don't seem to be responding right now -- you are the ES member that Bruce Teakle refers to on some of his pages with regard to this hubmoter?



I've yet to see any hard and fast pricing information for this motor shipped to the UK; but reading between the lines, I'm guessing that if I got a hub rather than a full wheel to save weight, I might be looking at around £250 including shipping. (I'm not sure if that is would be the original 36V or the newer 48V version.)

As yet I do not understand the difference/significance of the sinusoidal vs non-sinusoidal controllers? Nor have any notion of what if any difference there is in cost.

I am aware of the plastic sun gear problem, and that there is/was a potential fix in the works; I do not know if this has come to fruition.

I'm not looking to: a) set any land speed records; 10mph on the flat is fine; b) climb any alps; the steepest hills around are mostly much less than 10% and short; c) go more than 50 miles on a charge and I'm quite happy to peddle most of the way, on the out-bound leg anyway.

I'm looking to extend my excursion range, but I live (nearly) at the top of one of the highest hills in my county, which means I tend to be going down hill outbound whatever direction, and thus hit the wall when returning. Not steep, but persistent. My intent is to save (most) of the energy in the batteries for the return journey. I'm never in a great hurry, but I am a 60y/o life-long smoker, so ...

In short -- having been so verbose -- I'm looking for the cheapest, reliable, way to provide me with comfortable 50 mile round, assisted-return day trips.

I'm not sure there is a question there -- but cost info would be helpful; and if any of that stirs you to comment, good or bad, I'm all ears.

Thanks for your reply, Buk
 
amberwolf said:

Amberwolf. You're crazy! But you know that :) You're also talented, dogged, persistent, and original. I envy you all of those. (But not the dayglow :))

I'm moderately resourceful when it comes to recycling/upcycling stuff that I would otherwise discard; and have a tendency (severely curbed by my house proud wife) to hang onto old stuff "just in case" -- and I'm not beyond a little dumpster diving -- but in the last 30 years I've seen exactly one discarded (probably stolen, broken and dumped) bike and when I went back past there a few days later -- I was out walking with the wife when I saw it -- it had gone.

And whilst I still have a couple of old laptops -- IBM thinkpads circa 199x -- and a old PS/2 m80, I wouldn't know what to do with the parts.

Essentially what I am saying is that I simply do not see much in the way of discarded stuff where I live. What is dumped tends to go into wheely-bins, thus out of sight, or directly to the municipal dump with resident ex-diddycoys who have keen eyes and a very shrewd sense of second-hand worth. Even the junk yards around me are laid out more like supermarkets than junkyards. Everything is disassembled a priori, labelled (not always correctly), categorised(NAC), and placed on racks; pre-priced and non-negotiable. "Where there's muck, there's brass" has long ceased to be a cottage industry and is now big business. Long gone are the days of my youth when I would spend happy Saturday afternoons crawling over heaps of junk finding and dismantling parts and chucking the guy on the gate a note or two for the bits he saw me carry out :)

Having read a few of your threads (here and on your own site), my take from it is that the only really successful build was when you bought (albeit at a knock-down price) and designed-for-purpose hub motor. (I cannot see anyone gifting me a similar bargain.)

Sure, you learnt a lot and had some fun from building the wiper motor/fan motor chain drive/friction drive early attempts; and we your readers learnt a lot from reading of your experiences; but what we learnt, mostly, was that a design-for-purpose hub motor was vastly superior to a hacked together re-purposing of auto-components.

It is a valuable lesson, but one that doesn't need to be repeated.
 
You need to show us a link to a 4400maH cell for £1.95. If it were possible to make an e-bike battery for £54, we'd all be doing it.
 
Well that 54 BP is like $75usd
You could have a battery, but the Ah would be so low!
-------------------------------------------------------------------
Used laptop batteries can be bought by the pound (weight) if you find the right supplier.

Used tool pack batteries with the strips still on them for easy solder.

Cheap Chinese Shit of Unknown Quality, little specs, off brand cans

RC Lipo

Department Store Yard Equipment Batteries are $60CAD/Ah in the 24/36/48V flavors.

Quality Built Batteries from a reputable company. ebikes.ca, em3ev and the like. Still $60/Ah same as above.

Buying your own cans and soldering or tab welding them yourself is about half that price above if you solder and have the tools. Otherwise thats more money, $20 soldering iron, solder, alcohol, brush, or DIY tab welder $75-$150.

Buying a used battery from another ebiker in the used section. Or finding one on online classifieds. Of unknown discharges, charges, and to what levels.

Buying a complete package from EM3EV.

Add in the price of the charger(s). I still use my anemic rc 8S charger from HobbyKing. But my primary one is Meanwell. I have 5 of them now, all in $225cad.
 
Most who are going through the troubles of finding and mod of various parts to put together and build a bike, are not doing it to save money. They are doing all that work to succeed building better than what is offered in kits. There are some exceptions, builders who did an amazing work to make a bike out of free junk gathered around. Still, if you want cheap, you are probably better with a cheap kit, unless you don't value your time and have the will and skill to build from junk recycling. Building from junk is a state of mind. Some people are looking at a pile of scrap and see instantly something they could build with it. Others have a plan in their mind, and the will to search everywhere around to realize it at minimal cost.
 
d8veh said:
You need to show us a link to a 4400maH cell for £1.95. If it were possible to make an e-bike battery for £54, we'd all be doing it.

I don't think this is the exact link -- I scribbled the numbers on my jotter, but didn't keep the link -- but it is close enough.

(Update: And I've just noticed that the detail info for that link says 1700mAh despite the headline saying 4400mAh. I'll try ploughing through my browser history to find the original link; but I looked at a lot of pages :( )

(Update2: I think this was the one.)

I realise that getting them for £1.99 (or $1.99) would probably require some massive order size, and even then is not in any way guaranteed given the source of the information; and quality could be ... well, totally missing.

I also realise that given an expert eye, there may be other details that make this particular battery, or even all batteries of this type, a non-starter for e-bike use?

But if it could be made to work, and sufficient people were interested to come up with a bulk order...
 
MadRhino said:
Most who are going through the troubles of finding and mod of various parts to put together and build a bike, are not doing it to save money. They are doing all that work to succeed building better than what is offered in kits. There are some exceptions, builders who did an amazing work to make a bike out of free junk gathered around. Still, if you want cheap, you are probably better with a cheap kit, unless you don't value your time and have the will and skill to build from junk recycling. Building from junk is a state of mind. Some people are looking at a pile of scrap and see instantly something they could build with it. Others have a plan in their mind, and the will to search everywhere around to realize it at minimal cost.

Understood. Though I'm not criticising anyone. My requirements are lowly, my finances fixed, but my time is my own.
 
Buk___ said:
But then the C-rate stuff cropped up and I do not 'get' that yet.
C-rate is a way to describe how much current you can draw from a cell. 1C means 1x the Ah capacity = how many Amps you can draw, so a 4.4Ah cell with a continous 2C "c-rate" can handle 8.8A without major voltage sag, internal heating, or cell damage. A burst c-rate of 3C for a 2Ah cell would be 6A for a few seconds or less, usually, while the same cell might handle 1-2C (2A to 4A) continuously, for example.

And sellers too-frequently lie about both the capacity *and* the C-rate of their cells and packs. You'll see stuff like Ultrafire (and other *fire) cells that give capacities of 10Ah or even 20Ah in a single 18650, which is currently far from possible, and might list C-rates of 5C to 10C that could probably set the cell on fire from the heat generated...assuming the cells even worked at all when you got them ;) since those *fire cells are typically recycled garbage rewrapped as if they were new.

The pouch cells you link to could actually be genuine; the page says MOQ is only 1, but they probably charge some exorbitant handling or shipping fees; you'd have to check them out to see. It's also possible that their C-rate is very low, which just means you ahve to parallel a number of them to draw much current from them. You'd have to determine how much current your controller needs to pull from your pack, under the loads (hills/weight) you'll be putting it under, for the continuous length of time you have to do that for.


Regarding assembly, one problem with pouch cells is that they usually need to be constrained equally across their large flat surfaces, such that they dont' swell up with gas formation inside, when used at higher (for their specs) currents. They may have to be constrained no matter what their usage is.

My present packs are pouch cells, but they're also mounted in the hardware the manufacturer designed for that purpose, which makes it easier for me to package and handle them without worry. Iv'e still killed one cell with overdischarge accidentally, puffed it up right good. :/ (but I think it took actually reversing it to do that to this one--hobby RC cells I've puffed up just sitting there. :/ ) The problem with the ones I use is they're expensive and have to either be gotten used from someone else (like mine) or in large quantities from the manufacturer.

If you know anywhere in your area that scraps out electric or hybrid cars, the packs from those are often still usable at least in part, and may have cells you can use for your own pack.

But the DIY and other costs may still make it cheaper to buy prebuilt from a trusted seller.

(trusted, because I think we see more "crappy battery" problems here on ES than anything else. )




As yet I do not understand the difference/significance of the sinusoidal vs non-sinusoidal controllers? Nor have any notion of what if any difference there is in cost.
Probably not much cost differnce. Sine controllers are generally going to be a lot quieter with typical motors vs trap controllers. Might also be more efficient (cooler) depending on the motor/etc.


I'm looking to extend my excursion range, but I live (nearly) at the top of one of the highest hills in my county, which means I tend to be going down hill outbound whatever direction, and thus hit the wall when returning.
.
Just keep in mind that if you setup a system that uses regen for braking that you won't want to (or possibly be able to) use it going down that hill outbound, if you start out fully charged. Won't matter if you use any of the geared hub motors or middrives, just DD (direct drive) hubmotors.

Buk___ said:
Amberwolf. You're crazy! But you know that :)
Yes, yes I am. Muhahahahaaaaaha..heha..:chokes:

You're also talented, dogged, persistent, and original. I envy you all of those. (But not the dayglow :))
I dunno about original, but I'll grant the others (even if my talents are rough and strange). ;)

At the time, dayglo paint was the best way to be seen on a "regular" bike in daytime, and even at night under most lighting conditions (especially with white LED or fluorescent/UV downlighting).

Nowadays I don't bother with that due to the size of my contraptions being large enough in daytime to ensure most drivers go around me rather than over or thru me, or attempt to run me off the road. (I have a feeling most of the ones that consider it decide not to as it might actually damage their vehicle to hit me). At night I'm so well lit that the only way they wouldn't see me is if they were driving with their eyes completely closed for the mile before they got to me till after they'd passed me some distance. ;)






but in the last 30 years I've seen exactly one discarded (probably stolen, broken and dumped) bike and when I went back past there a few days later -- I was out walking with the wife when I saw it -- it had gone.

This past decade has seen the recyclable junk curbside diminish severely here, too--but until so many people became desperate enough for money to do pretty much anything for easy money like that, there used to be a lot of it. I still find them at yardsales and such, but not much curbside.


Essentially what I am saying is that I simply do not see much in the way of discarded stuff where I live.
Yeah, it makes it hard to do what I like to do; much of what I use these days comes from work or friends/acquaintances that remember I might be able to use something.... :) but I used to do a lot of dumpster diving until they made it illegal (too many idiots making messes instead of cleaning up after they were done).


Having read a few of your threads (here and on your own site), my take from it is that the only really successful build was when you bought (albeit at a knock-down price) and designed-for-purpose hub motor. (I cannot see anyone gifting me a similar bargain.)
It was certainly longer lasting than my friction drive made from the skate wheels and radiator fan motors--but I think if I'd had better batteries and more clue to what I was doing, I could've made that one work out better and last just fine. :)

CrazyBike2 was successful enough as a powerchair motor chain drive...but I had little idea what I was doing and was learning as I went--I hadnt even found this site yet, IIRC. I think if I were to go back and retry that (I still want to) it'd be a lot better built--partly cuz the frame wouldn't be so bendy (the cause of most of the destruction of parts was derailing due to misalignments under torque).


Sure, you learnt a lot and had some fun from building the wiper motor/fan motor chain drive/friction drive early attempts; and we your readers learnt a lot from reading of your experiences; but what we learnt, mostly, was that a design-for-purpose hub motor was vastly superior to a hacked together re-purposing of auto-components.

It is a valuable lesson, but one that doesn't need to be repeated.
Probably true. ;)

It depends on the resources of those reading, I suppose. You should look around at the threads for Numberonebikeslover (Naeem) here on ES--now THAT man can really build useful stuff from junk! (his resources were probably a lot less than mine)


For most people, going with a kit is "cheaper" because of time...and lack of DIY skills. Basic DIY skills can make individual parts (controller, motor, prebuilt battery pack, etc, all from different suppliers or even the same one but not sold as a kit) cheaper.
 
amberwolf said:
Buk___ said:
But then the C-rate stuff cropped up and I do not 'get' that yet.
C-rate is a way to describe how much current you can draw from a cell. 1C means 1x the Ah capacity = how many Amps you can draw, so a 4.4Ah cell with a continous 2C "c-rate" can handle 8.8A without major voltage sag, internal heating, or cell damage. A burst c-rate of 3C for a 2Ah cell would be 6A for a few seconds or less, usually, while the same cell might handle 1-2C (2A to 4A) continuously, for example.

Thanks. I read the wikipage about that yesterday; but it'll take me a while and working through a few examples before it'll click into place.

amberwolf said:
And sellers too-frequently lie about both the capacity *and* the C-rate of their cells and packs. You'll see stuff like Ultrafire (and other *fire) cells that give capacities of 10Ah or even 20Ah in a single 18650, which is currently far from possible, and might list C-rates of 5C to 10C that could probably set the cell on fire from the heat generated...assuming the cells even worked at all when you got them ;) since those *fire cells are typically recycled garbage rewrapped as if they were new.

The pouch cells you link to could actually be genuine; the page says MOQ is only 1, but they probably charge some exorbitant handling or shipping fees; you'd have to check them out to see. It's also possible that their C-rate is very low, which just means you ahve to parallel a number of them to draw much current from them. You'd have to determine how much current your controller needs to pull from your pack, under the loads (hills/weight) you'll be putting it under, for the continuous length of time you have to do that for.

Indeed. That is the biggest headache with dealing with overseas suppliers. Establishing bonafides and trust. I dislike reading people who condemn anything and everything Chinese as crap; but I'm not naive enough to take website and sales pitches on face value. To that end I've sent them an email (direct, bypassing alibaba'a mechanisms) and I'll see where I go next from their response.

amberwolf said:
Regarding assembly, one problem with pouch cells is that they usually need to be constrained equally across their large flat surfaces, such that they dont' swell up with gas formation inside, when used at higher (for their specs) currents. They may have to be constrained no matter what their usage is.

My present packs are pouch cells, but they're also mounted in the hardware the manufacturer designed for that purpose, which makes it easier for me to package and handle them without worry. Iv'e still killed one cell with overdischarge accidentally, puffed it up right good. :/ (but I think it took actually reversing it to do that to this one--hobby RC cells I've puffed up just sitting there. :/ ) The problem with the ones I use is they're expensive and have to either be gotten used from someone else (like mine) or in large quantities from the manufacturer.

I'm currently working my way through the tutorials at The Battery University. Beyond the physical packaging requirements, the other part to get right is voltage sensing to deal with bad cells. The simplest way seems to be to use one chip per battery@$0.10perand (perhaps) a RaspberryPie to monitor, log and control.

amberwolf said:
If you know anywhere in your area that scraps out electric or hybrid cars, the packs from those are often still usable at least in part, and may have cells you can use for your own pack.

I looked into this, but under UK law e-car/hybrid manufacturers are responsible for recycling the battery packs, and breakers are legally obliged to turn the packs over to them for some nominal removal fee; and there are strict paper trails involved.

amberwolf said:
But the DIY and other costs may still make it cheaper to buy prebuilt from a trusted seller.
(trusted, because I think we see more "crappy battery" problems here on ES than anything else. )

Understood. I picked out the particular supplier because I read they have (as of y/e 2015] 2.5% of the (massive) Chinese home e-bike battery market. Needs confirming, but its a good starting place. They also seem to be well certified and have an outward (from China) looking perspective; key personnel with English language skills and listing on Linked-in etc. Time will tell.



amberwolf said:
As yet I do not understand the difference/significance of the sinusoidal vs non-sinusoidal controllers? Nor have any notion of what if any difference there is in cost.
Probably not much cost differnce. Sine controllers are generally going to be a lot quieter with typical motors vs trap controllers. Might also be more efficient (cooler) depending on the motor/etc.

Hopefully D8ave will have some insights on this as he has/had the type of hub motor I'm interested in.


amberwolf said:
I'm looking to extend my excursion range, but I live (nearly) at the top of one of the highest hills in my county, which means I tend to be going down hill outbound whatever direction, and thus hit the wall when returning.
.
Just keep in mind that if you setup a system that uses regen for braking that you won't want to (or possibly be able to) use it going down that hill outbound, if you start out fully charged. Won't matter if you use any of the geared hub motors or middrives, just DD (direct drive) hubmotors.

The Xiongda double-speed motor I'm interested in uses two sets of planetary gears and two roller clutches operating in opposite directions to give two speeds by the expedient of reversing the direction of the motor, so regen isn't an option. The upside is that the motor runs at a more efficient rpm, more of the time.

I've been looking into this for some time and from a variety of different angles; from the thread here where the guy built his own motor using nothing more complicated than a stand drill; to golf buggy motors and all points between.

If money was no object, I'd be looking to wrap one of these 1.5kg 500w motors around a harmonic reducer drive into a custom bottom bracket driving a NuVinci CVT. It would make for a fine, powerful infinitely flexible and theoretically quite efficient system. Maybe when I win the lottery -- which as I've never yet bought a ticket :?

But even then, I still need to buy batteries to run it.
 
A few variables to consider
1) If money was no object
2) If a lot of time is on your hands
3) Knowledge and skills

For myself, I'd have someone make me a full suspension bicycle, or have a brand name bike modified. The look wouldnt stray too far away from the look of a bicycle.
That idea is either for mid-drive placement and/or custom battery placement inside "something" perhaps tubing. I am thinking of someones recumbant (thinking its amberwolfs) where the cans were placed inside square tubing.

Motor selection, I have no clue mate. Heinzmann is talked about, dunno much about it, only know its expensive.
 
Having spent a lot of money on ebikes over the years. If I were to build a cheap ebike from scraps I would buy electric scooters as you can get them for dirt cheap once the battery dies 8) . But, the kits you can buy now are very efficient and work. I would rather ride ebikes than build them..
 
Buk___ said:
Beyond the physical packaging requirements, the other part to get right is voltage sensing to deal with bad cells. The simplest way seems to be to use one chip per battery@$0.10perand (perhaps) a RaspberryPie to monitor, log and control.
The "simplest" way is more likely to be a pre-made BMS board (there are thousands of models out there, a number discussed in varying levels of detail in various battery threads here on ES, mostly in the Battery Technology subforum). All you have to do is wire each channel to it's corresponding parallel group of cells in the battery, then wire the most positive contact of the battery to the BMS's main + and thence to a discharge connector and a charge connector, then wire ground from the most negative contact of the battery to the BMS's main - , and then the BMS's charge - wire to the charge connector and the discharge - wire to the discharge connector. Some of them don't have separate charge and discharge wires.

Or just add a balancing connector that has a sense wire to each cell and the main pack + and -, and run that out to RC hobby type alarms / meters / loggers (depending on the level of monitoring/complication you're after), and then you just listen for the alarms and stop using the battery. (or stop charging it).

The BMS is the least user-tending to deal with, but you have to make sure and get a reliable one with a good design, or you can end up with overdischarged (or overcharged) cells depending on how you design and use your pack.


Individual cell monitors and an MPU programmed to monitor them might be simple to use, but probably not to design and setup. ;) Similar things have been done here on ES over the years by various members, though there's no index of them. :/



If money was no object, I'd be looking to wrap one of these 1.5kg 500w motors around a harmonic reducer drive into a custom bottom bracket driving a NuVinci CVT. It would make for a fine, powerful infinitely flexible and theoretically quite efficient system.
[/quote]
Unfortunately he died suddenly not long ago so isn't around to help out with it, but BobC
https://endless-sphere.com/forums/memberlist.php?mode=viewprofile&u=18616
here on ES developed something like that, and has threads here about it
https://endless-sphere.com/forums/viewtopic.php?f=28&t=27132
https://endless-sphere.com/forums/viewtopic.php?f=2&t=75357
more threads he started here
https://endless-sphere.com/forums/search.php?keywords=&terms=all&author=bobc&sc=1&sf=titleonly&sk=t&sd=d&sr=topics&st=0&ch=300&t=0&submit=Search
and his YT channel
https://www.youtube.com/channel/UC7hJHcgt3FxBGqd16PHvRpQ
Whereswally606 is his friend and might have (or be able to get) more info on the drive details if they're not in the thread(s).
 
amberwolf said:
The BMS is the least user-tending to deal with, but you have to make sure and get a reliable one with a good design, or you can end up with overdischarged (or overcharged) cells depending on how you design and use your pack.


Individual cell monitors and an MPU programmed to monitor them might be simple to use, but probably not to design and setup. ;) Similar things have been done here on ES over the years by various members, though there's no index of them. :/

Wow. Lots of stuff there; but real tough for a non-electronics person like me to sift and evaluate; much less construct.

The reason I like the $0.10/battery chip + RsP approach is that the wiring/soldering is 27/36/45 whatever repetitions of a very simple 2/3 component circuit. All the complexity of monitoring, level setting etc. is moved into software which I'm very comfortable with. An RsP is cheap and could monitor 27 batteries many times a second and maintain a delta-V rate for all of them over time whilst logging everything. Throw in a cheap thermistor per and it could track temps as well.

Of course, I'm probably living in the rose tinted world of my mind and in the real-world, the complexities would become apparent; but first things first. Can I establish a relationship/trust with a Li battery manufacturer to the point I'm confident to actually commit money to the idea.



Unfortunately he died suddenly not long ago so isn't around to help out with it, but BobC here on ES developed something like that, and has threads here about it
https://endless-sphere.com/forums/viewtopic.php?f=28&t=27132
https://endless-sphere.com/forums/viewtopic.php?f=2&t=75357
more threads he started here
https://endless-sphere.com/forums/search.php?keywords=&terms=all&author=bobc&sc=1&sf=titleonly&sk=t&sd=d&sr=topics&st=0&ch=300&t=0&submit=Search
and his YT channel
https://www.youtube.com/channel/UC7hJHcgt3FxBGqd16PHvRpQ
Whereswally606 is his friend and might have (or be able to get) more info on the drive details if they're not in the thread(s).

It's a crying shame about bobc; and really annoying that photofuckit have screwed his (and many others) legacy with their mercenary tactics.

Nice to see that my fantasy -- which as far as I'm aware I'd never seen described anywhere -- might actually be viable.

It'll be interesting to see what if any response I get from Wido and a couple of others I've emailed over the next few days; and if the numbers -- assuming I get any -- make sense.

I've been looking into the physical containment problem and it looks like a 3D printed nylon casing would be sufficient and cheap solution. There are plenty of companies that will run off a part from a cad file; some 3D printer owners will do it for cost plus as a way of offsetting some of their outlay; and there are some maker workshops not far from me that will let you use their's for very reasonable fees.

I'm still looking for wisdom regarding the actual requirements; but assuming I can establish some kind of conversation with a supplier, they ought to be able to provide the information; if I can find one that has good English translation skills.
 
Buk___ said:
Wow. Lots of stuff there; but real tough for a non-electronics person like me to sift and evaluate; much less construct.


The reason I like the $0.10/battery chip + RsP approach is that the wiring/soldering is 27/36/45 whatever repetitions of a very simple 2/3 component circuit. All the complexity of monitoring, level setting etc. is moved into software which I'm very comfortable with. An RsP is cheap and could monitor 27 batteries many times a second and maintain a delta-V rate for all of them over time whilst logging everything. Throw in a cheap thermistor per and it could track temps as well.

You'll have to design the circuits so each channel is electrically isolated from the others. If I'm reading he datasheet correctly, the part you link only has a max of 12v on the output pin, so while the open-collector style would let you theoretically just hook up all the output pins so any of them going active (grounding) would turn off the output of the pack for undervoltage on any cell, you'd actually have to be careful to not put more than a 12v potential on there. I don't know without drawing up a circuit diagram what the results would be of connecting these together from seriesed cells, but some of them would be more negative than others, and some would be more positive than 12v relative to others, which could invite unexpected and unwanted current paths.

You might have to use isolating transistors or optocouplers or something on each chip's output pin before connecting the output of those isolators to a common shutdown pin.

Or to connect the output of the isolators separately to pins on an MCU (or a mux going to an MCU bus).

Then you'll have to make a power source for the MPU (r-pi, etc), or tap off a cell or set of them, and so on.

It's not MPU-based, but you can look up the threads on the various Fechter-Goodrum BMS versions; I think the last was called Zephyr. There are also other BMS development threads here on ES; Methods designed and built some HVC/LVC boards (mostly meant for the common RC LiPo packs as a fire-prevention, but usable for other cells too).

It's not impossible but it is complex, whereas the existing BMS boards do all that for you--just wire it to the cells and use teh battery as a module you don't have to worry about (as long as you start with good cells spec'd for your usage scenario, use a known-reliable BMS that's spec'd for the cells' capabilities so as not to abuse the pack).

So, for "economics", the topic of the thread, it'd make more sense to use an existing BMS than go thru the design work and DIY of developing your own BMS out of individual chips and parts, unless you just really want or need features that don't exist yet or cost too much in others.


Personally I run my main packs "naked", as I'm not using the cells anywhere near their capabilities, and they're good quality cells, so they tend to stay well-balanced.

I have another couple packs that came with a BMS, and one of them (from EM3EV) works well, the other (from Luna) has issues with the BMS--I suspect they are design issues rather than faults, but don't have another to compare with directly.


If a cheap, well-designed, reliable monitoring system could be made, there's probably people that won't use a typical BMS that would use that system.

Mostly they use the RC hobby types of monitoring systems, as those are cheap if not well-designed or reliable. ;)
 
Speaking of BMS'
Wouldnt it be great if Justin sold BMS'

Oops forgot to add about batteries. I was thinking what my C-rate charging was and I thought about it 16A charging 24Ah = 0.70C and I hate waiting around when its Guerilla charging. 1 hour min or 1.5hrs to top it off. I can add an extra 12Ah to make it 36Ah pack then who knows how long I will be waiting around. Thats why I bought 2 more Meanwell HRP-300-24's I think they are. 2 identical charging setups. 16A @ 36Ah = 0.45C but split packup then 0.9C each. Keep original 24Ah, split pack, 2 charging setups = 1.40C. Keep the other 12Ah for backup spare.

Thought I'd share that, dunno why.
Guess its more money spent on chargers.

Gotta tell all the newcomers
DO NOT SKIMP ON THE CHARGER
 
Buk___ said:
amberwolf said:
Buk___ said:
If money was no object, I'd be looking to wrap one of these 1.5kg 500w motors around a harmonic reducer drive into a custom bottom bracket driving a NuVinci CVT. It would make for a fine, powerful infinitely flexible and theoretically quite efficient system. Maybe when I win the lottery -- which as I've never yet bought a ticket :?

But even then, I still need to buy batteries to run it.
Sounds great, but often, what you get in practice doesn't match the theory. Every time you gain something, you lose something. A Nuvinci is heavy and inconvenient. For a normal electric bike that gets you up hills and lets you have long rides in the country, it's difficult to beat a simple rear geared hub-motor with a controller that has decent PAS control. For me, money is no object. I've probably spent £30,000 trying to find e-bike nirvana, which, for me, presently is a Q128C motor, 48v 11.6Ah battery and a 15A sinewave controller. They cost about £600. The only improvement that I can envisage in the not too distant future is a lighter battery. I've tried just about every type of bike, every type of motor and controller, gear systems, suspension, etc.
 
amberwolf said:
You'll have to design the circuits so each channel is electrically isolated from the others. If I'm reading he datasheet correctly, the part you link only has a max of 12v on the output pin, so while the open-collector style would let you theoretically just hook up all the output pins so any of them going active (grounding) would turn off the output of the pack for undervoltage on any cell, you'd actually have to be careful to not put more than a 12v potential on there. I don't know without drawing up a circuit diagram what the results would be of connecting these together from seriesed cells, but some of them would be more negative than others, and some would be more positive than 12v relative to others, which could invite unexpected and unwanted current paths.

You might have to use isolating transistors or optocouplers or something on each chip's output pin before connecting the output of those isolators to a common shutdown pin.

Or to connect the output of the isolators separately to pins on an MCU (or a mux going to an MCU bus).

Then you'll have to make a power source for the MPU (r-pi, etc), or tap off a cell or set of them, and so on.
(About to show my very low level of IC/electronics understanding here.)

My thinking is/was:

The RsP and all the detector chips would be powered from a single separate battery/power supply (5 or 3.3v). This supply would share a common ground with the batteries to be monitored.

Each of the detector chips output's would be connected to a different one of the RsP's GPIO pins -- possibly via a cheap diode.

The detectors have a 0.9V threshold, so the monitored voltage is tapped from a resistor divider to scale it to the required level.

(Now this is where it gets dodgy :) )

If the lower resistor in the divider was a variable pot, you can adjust the tap percentage and thus tailor the threshold to be detected.
If in place of that lower resistor, you had a say 8-bit DAC driven by 8 of the GPIO pins of the RsP, then the RsP could adjust the division rate of the voltage dividers.
By setting an 8-bit value to the DAC, it can the scan each of the 27 outputs from the detectors and see which if any of the batteries are over the level set to the DAC.
It can then step up the DAC (say) 0.1V and scan again. Rinse and repeat.

In this way, by cycling the DAC to adjust the level detected, and cycling the input pins, it can monitor not just a single go/nogo value from each battery, but build a continuous graph of the instantaneous voltage levels on all of the batteries on an ongoing basis. Thus monitoring not just the overall heath of the combined battery state, but of the individual batteries and the parallel banks also. In theory, it could do this during both charge and drain.

With 27 individual cells and 8-bit DAC, that leaves 5 pins to control charging, cutoff, and warning lights.

Any more detailed information than healthy/possible problem/shutting down NOW, would be obtained by plugging a mobile phone into a USB port, or potentially via bluetooth or wifi.

I know there is a lot of detail missing, much of which I could probably work out (eventually), and some which I probably haven't thought of which I couldn't; but I *think* that the idea is viable??

(Thanks for your continued feedback.)

Cheers, Buk
 
d8veh said:
Sounds great, but often, what you get in practice doesn't match the theory.

Fair enough. It was just a little wishful thinking anyway.

d8veh said:
for me, presently is a Q128C motor,

Am I right in thinking that you experimented with the XiongDa 2-speed motor for a while?

If so, have you any comments/cautions/pointers to further info, on it as my choice of hub?
 
Question: You refer to 27 cells...but 27 cells in series for most of the typical Li types is going to be very high voltage (over 100v!), and unnecessary for most bicycle stuff. Will actually exclude using most of the available existing ebike controllers and kits.

If you mean 27 cells to build the pack from some of which will be in parallel, and some of which will be in series. If it's 9s3p, then that makes more sense, though most existing controllers would need 10s for 36v (or 13s or 14s for 48v / 52v), so you'd need 30 cells.

If there's some other pattern you're thinking of, you'll have to say or draw it up.



Generally people wire up all the parallel groups so they are monitored as if they were a single large cell; it simplifies monitoring wiring and electronics.

But you could instead wire up all the series sticks of hwoever many single cells, then wire these sticks in parallel only at main + and -, and then be able to monitor every individual cell.

There are advantages and disadvantages to each way, and there is quite a lot of discussion about it in various battery build threads here on ES, with the decision up to the pack builder. It's a lot of reading (and looking around), if you're interested in that.




The problem with resistor dividers is that you now have a resistive system constantly draining your cells, and the cells won't have equal drain so they become unbalanced at one end of the pack more than the other.

The higher the resistances you use, the smaller the drain, but the more susceptible to noise the system becomes. (not as big a deal inside a battery vs a computer, but still a potential problem).

Even with the DAC idea (presumably controlling a voltage-controlled resistor?) I'm also not sure how you can use a single resistive ladder to scale *all* the voltages to be within the range of the detectors, relative to the common ground of the pack and detection system. I think you will have to use multiple separate dividers like a stairstep, to divide the most positive voltage with one pair from the main pack + to ground, then the second with another pair from the next cell down's + to ground, etc.

Doing that will mean the lowest cells will have a lot lower resistance to ground than the most positive, and so will drain (unbalance) a lot faster than the cells more positive than they are.


YOu can prevent the drainage by inserting switches (electronic or mechanical) between all the resistors, and only turning them on when you're going to check a cell, but that adds a lot of complication and extra circuitry and controls (unless it's all manual switches), and cost, and time to design and workout bugs.


If you're envisioning something different, draw it up and attach it here, and those with experience can help. :)


Buk___ said:
The detectors have a 0.9V threshold, so the monitored voltage is tapped from a resistor divider to scale it to the required level.
If you use detectors that have the right threshold to start with, for your cells' LVC, you can eliminate the resistive dividers and all that compensation.

But you'll still probably want to put each one's output as isolated from the others and your MCU's input, so there isn't excessive (or reverse) voltage potential across things, or shorted cell groups, or cell drainage from the resistor current.


Another issue the idea you've proposed doesn't take care of is rebalancing the cells--it just monitors them. So you'd need to either add a set of circuits to deal with that, or manually balance with external resistors to drain down high cells to match low ones, or charge low ones up with single-cell chargers (or an adjustable lab PSU).


Overclocker has a thread going about a 6-channel single-cell charger (which could be expanded to a full pack's worth of cell groups); you could look at that for one example (there are many options documented around ES).

You can look at the other BMS designs here on ES to see what I'm talking about.




BTW, keep in mind that I'm not an engineer, just a mostly-observant hack. ;)
 
amberwolf said:
Question: You refer to 27 cells...but 27 cells in series for most of the typical Li types is going to be very high voltage (over 100v!), and unnecessary for most bicycle stuff. Will actually exclude using most of the available existing ebike controllers and kits.

If you mean 27 cells to build the pack from some of which will be in parallel, and some of which will be in series. If it's 9s3p, then that makes more sense, though most existing controllers would need 10s for 36v (or 13s or 14s for 48v / 52v), so you'd need 30 cells.

I was thinking and have sketched 3 banks of 9. When I mentioned "banks" earlier, I was referring to a bank of 9 in series, with 3 banks in parallel giving me 33.3V @ 13.2Ah. Being conservative with the voltage is in keeping with my philosophy of making cheaper stuff last longer by under running it rather than trying to extract ever last ounce as most seem to. It also fit with the RsP's 40 GPIO pins.

For the monitoring circuits, I was basing my thinking on a combination of the "Multi−Rail Supply Undervoltage Monitor with Power Good" circuit (on page 21 of the NCP-302 datasheet if you still have it around.) and the "Microcontroller Systems Load Sensing" (page 19), and trying to daisy chain many of the latter in a similar fashion to the former.

But I see that I was being stupid and only looking at the battery circuits and monitoring circuits in isolation. Once you start putting two different sources of power into the same circuit, none of it makes sense any more. :(

So much to read and learn and not enough engrained knowledge (just enough to be dangerous) makes for a disconnect between thought processes and reality. Thanks for bringing me back to earth. Even with the bump. :)

I got a response from Wido, but its pro former. Heavy on somewhat mangled English and over-polite, but containing no real information whatsoever.

I still think that softpack lithium batteries are the way to go. The 27 of the Wido type 19 (705896) (a different model to the one I originally mentioned) would allow a 33.3V@14.4Ah to be packed into a unit 216mm x 100mm x 60mm and weighing 2.5kg. That's including allowing for 1mm of nylon between every cell and a 2mm nylon outer case. Possibly over-engineered. Add 25mm to the length and another 250g for a 10p3s 37V@14.4Ah.

Another variation would be a 10p2s using their type27 8559156 to give a 190mmx160mmx60mm 37V@18Ah in 3.5kg; but without being able to get pricing information and shipping costs...

All of which is for naught unless I can find a reliable supplier. I'm guessing that the numbers I mentioned in my email to Wido are simply not enough for a company their size to consider. There are lots of other, smaller companies that might be interested, but then you're into the game of trying to find one that is what they claim to be.

Anyway, its obvious that I am now wasting your time; so don't feel the need to respond further.

Once again, thanks for pulling me back to earth. I'll go away and do a lot more reading before I bother you guys here further.

Cheers, Buk
 
Buk___ said:
Once again, thanks for pulling me back to earth. I'll go away and do a lot more reading before I bother you guys here further.
Not really a bother; we generally like to help those willing to learn. :)

Sometimes those without too much "knowledge" to get in the way of their imaginations come up with good ideas...sometimes there's a critical bit they don't know about yet that ruins it. :( But it's often interesting to work it out anyway. :)

(it's the ones that ask questions but then are bound and determined to do it their way even when all the answers point the other way...what was the point of asking then? ;) Or those that start a bajillion threads about the same thing, or won't tell anyone anything about the project so there's insufficient info to base even guesses on.... Those can be a bother. :/)




I do recommend sticking with "standard" numbers of series cells, because it gives you more options on what to use, not only for controllers (which usually have preset standard LVC (and sometimes HVC) limits that work better with those standard numbers of series cells, but also for pre-built BMS units if you go that way.


You don't have to charge the cells up to their max voltages, if you want to keep some of the voltage-based stresses down, and potentially lengthen the lifespan of the pack, but it will reduce the amount of usable Wh you get out of the pack. (which you can fix by adding more parallel cells if you have the space/etc).
 
Back
Top