Disadvantages of higher Battery Voltage

[Alan B]

There are places where it is useful to push the limits.

FETs are not terribly expensive, heavy, or large. There is little advantage to be gained in running them at their voltage limits. It probably means the wrong winding motor is being used, or the wrong gear ratio.

Luke's ebikes leave others in the dust without pushing the FET voltage limits.

 

[Punx0r]

Bill's right about the distribution of true capability of each component within a batch. The 100V rating is the minimum standard to be met by a minimum percentage of all the components made. If you're happy to sort through a batch and reject the weaker ones you can end up with a set that will withstand 100V + transients.

Maybe it would be easier to do this sorting on the bench though rather than in a controller on the bike

 

[MadRhino]

There are places where it is useful to push the limits.

FETs are not terribly expensive, heavy, or large. There is little advantage to be gained in running them at their voltage limits. It probably means the wrong winding motor is being used, or the wrong gear ratio.

Luke's ebikes leave others in the dust without pushing the FET voltage limits.

Luke is willing to build much heavier than I.
I build the max performance I can get with a 70 lbs bike, using only bicycle components.

You said it, mosfets are cheap. They are not equal, and when you have a good set in a well built controller, running 24s is reliable. I don’t mind if I have to destroy components during the test and tuning period that follows the build, if I can save a few pounds. Running 24s is letting me use a lighter battery to achieve the power.

I understand those who are building heavier, willing to use big batteries, 0 gauge wires, motorcycle wheels and brakes, for they are building what they like and/or need. I have a street commuter running on 20s, because city cimmuting doesn’t require the extra speed and power I could pull of it. and it is above my preffered weight anyway, with luggage and accessories facilities that are convenient to commute.

That is why there are so many different bikes, because different rides are making different requirements.

 

[billvon]

You are pushing this at tad too far, with ‘poor design’ and ‘constant failures’.

OK.

I stand by my statement that using components outside their ratings is poor design. Pretty much every component failure I have looked at over my career has been due to someone, either intentionally or unintentionally, using a component outside its design envelope. I have no doubt that you can get away with it quite often - but a good designer designs things that are guaranteed to work, not things that "if you're lucky, will get you there."

You can, of course, do whatever you want.

 

[MadRhino]

You are pushing this at tad too far, with ‘poor design’ and ‘constant failures’.

OK.

I stand by my statement that using components outside their ratings is poor design. Pretty much every component failure I have looked at over my career has been due to someone, either intentionally or unintentionally, using a component outside its design envelope. I have no doubt that you can get away with it quite often - but a good designer designs things that are guaranteed to work, not things that "if you're lucky, will get you there."

You can, of course, do whatever you want.

Design has to meet a purpose, with requirement and priorities. Poor design for one purpose can be the best for another, according to each different priorities. My sport bikes may be poor design for one who need long range transportation at avg trafic speed, just like their ideal commuter would be a very poor design by my requirements. Those who are building with reliability of initial components as a priority, are willing to sacrifice a lot of performance and suffer a lot of extra weight. There is ‘designed to last as is’ and ‘ designed to win today and easy to mod tomorrow’. Both have their reason to be.

 

[billvon]

Design has to meet a purpose, with requirement and priorities.

Exactly. And it has to be done rigorously.

There is, of course, a way to accomplish what you describe and have it be a good design. You can requalify the parts. You can do it yourself, but the better way is to work with the manufacturer to re-bin the parts to your new specs. Of course, first you have to understand what those new specs are, how to identify parts that will likely meet them, how to create test methodologies to ensure they do meet them, and what specs you can allow to degrade. Once you do all that, you can get FETS requalified from (say) 100 to 110 volts - and then you will have a good design that will work at the higher voltage. You can get FETs that survive higher avalanche energies and have lower Rds on values. Of course, you might have to sacrifice gate charge, turn on time or gate thresholds, but as long as you understand your design well it's easy to evaluate whether those changes are acceptable. You will end up with a solid design that works well, is reliable and meets its goals.

There is also a way to do it and have it be a bad design - just keep changing parts until it works.

 

[MadRhino]

Manufacturers will not make mosfets especially for me. Those 4110 that they make are good for me, just not all of them, so I have to find and match a set that is good. Much easier than startin to argue with manufacturers.

Of course, I could use bigger, heavier controllers and save time and money. But weight and size are high in my scale of priorities. It is just like tires, that I am willing to buy by the dozen to save weight. Once I know how long it does last safely, I find no problem keeping inventory and replacing it as preventive maintenance.

 

[Alan B]

Selecting FETs saves a couple ounces for a couple of volts. Choosing windings would be more productive.

 

[MadRhino]

Selecting FETs saves a couple ounces for a couple of volts. Choosing windings would be more productive.

We are not talking ounces, but pounds. A faster winding at lower voltage would require a bigger controller, bigger wires and connectors, maybe a bigger battery. I am at the limit of the power I can get from a 11.6 kv motor in a 25.5 OD wheel before frying it in my 10 miles ride. More potential power is not as important to me as keeping the weight and balance of the bike for it is tuned near perfect for me now. Extra power is useless if I the bike is not riding good enough to use it. My next bike will sure be different, each one is, and every time I work to pull the best out of it. I am not working for more power, but for lighter weight with about the same power and handling that I have now. Where I am now, saving a few ounces can be expansive, or dangerous. My next try is to mod a lighter frame with carefully placed CF reinforcements, to give it the lateral stiffness of a bigger one.

The bike I sold this year (making a new one does require the space of one gone) had been my best mountain ride for years. It was mod many times and finally, I had to build a new one to beat it because I had reached its limits. It had a slightly faster acceleration that I have now, but I do the lap faster with the new one. The balance between power/weight/handling is a very sensitive part of design and tuning. What is lost in acceleration can be less than what is gained in cornering, or braking.

 

[Alan B]

Lowering voltage by 10 to 20%, increasing current by 10 to 20% == pounds, really??

Same motor weight.

Same number cells, rearranged.

Bigger wires, ounces not pounds. Maybe a couple of ounces. Wire lengths are super short on a well designed ebike.

Better Controller, ounces not pounds. Same controller might even do the job. Or six more FETs.

Do whatever you like, but these claims are suspect.

Enough of this diversion. Have fun.

To the rest, don't exceed 80% of the FET voltage rating. Just not worth it.

 

[Punx0r]

80% is a good guide for those seeking a reliability. MadRhino has made it clear he prioritises performance over reliability. His is more of a racing philosophy where you hand-pick parts and push them until they survive just long enough to do what you need. You maximise the density of a controller by running it at it's voltage and current limits.

It's two very different approaches and neither is "wrong".

 

[markz]

I have been thinking of this as I use 36V to reduce my speed but I realize now that the pack is using more discharge current.
I decided to limit the throttle and go up to 13S or 15S and install the appropriate resistor to achieve the proper LVC.

Here are just some easy numbers picked.
Lets say your controller maxes out at 1000W due to the current shunts inside the generic controller.
Same number of parallel strings for same Ah.
60V x 17A = 1000W
50V x 20A = 1000W
40V x 25A = 1000W
30V x 33A = 1000W

 

[John in CR]

I find that the fun factor increase of running higher voltage for a given system far outweighs any detriment. 31s with the smallest wheel I could find with over 100 foot-pounds of torque with a pretty flat curve up through 80kph due to current limiting, and a top speed around 115kph depending on SOC fits my commuting needs like a glove. The extended swingarm and low CG means I can launch as hard as I want with no worries about flipping over. If I did it again the only change I'd make is to go to 32s or 33s for even more passing acceleration at higher speeds, not to extend the top end, which I rarely use.

 

[MadRhino]

Yep. Power ebikes are a drug.

 

[Mywpn]

You are pushing this at tad too far, with ‘poor design’ and ‘constant failures’.

OK.

I stand by my statement that using components outside their ratings is poor design. Pretty much every component failure I have looked at over my career has been due to someone, either intentionally or unintentionally, using a component outside its design envelope. I have no doubt that you can get away with it quite often - but a good designer designs things that are guaranteed to work, not things that "if you're lucky, will get you there."

You can, of course, do whatever you want.

Your comparing the normal safe every day person to someone that wants to push the boundaries of how quick and fast they can go. The lack of design and availability in the high performance scene has only one other option and that’s to push the boundaries on what’s actually available. This isnt careless nor poor design its choice of freedom and there interests

 

[markz]

140V that must be some large ass controller with some big ass capacitors in there.

31s with the smallest wheel I could find with over 100 foot-pounds of torque with a pretty flat curve up through 80kph due to current limiting, and a top speed around 115kph depending on SOC fits my commuting needs like a glove.

the only change I'd make is to go to 32s or 33s for even more passing acceleration at higher speeds, not to extend the top end, which I rarely use.

 

[larsb]

If you give a motor voltage way above what it was designed for, it might arc through the insulation of the windings. We are talking 2x+ here.

This is not a problem, most ES applications are 0-150V systems, not even close to the limit of the winding wire and motors can normally not be overvolted 2x, then the rpm will be too high or the winding was a poor choice to start with.

Question is not so much about 36V-->72V or similar but what power do you want out of your system.

After having tried to optimise a winding i think that higher voltages can be said to be better in the 5-10kW range (and up?) as it is easier to get a wire wind done with good copper fill for those kinds of stator sizes when aiming for a lower kV.

 

[fechter]

From a motor/controller standpoint, you can get similar performance over a wide range of voltages by building things accordingly.

To me, a bigger issue is the battery. Higher voltage means more cells and more failure points and harder to protect and maintain balance.

 

[larsb]

Isn't it the same battery issues no matter what voltage?
You want xxx Wh range, it's either x*y batteries or y*x batteries to get there

An 18650 cell is roughly 10wh, You want 1000wh, it's 100 batteries, arranged in the way that suits the rest of your system.

parallell batteries are harder to protect since bms takes care of all the series cells but doesn't balance the parallell cells in each group separately.
Or how do you mean?

 

[fechter]

In a big 18650 pack, all the cells in a parallel group will be forced to balance, so you only need to deal with the series groups.
These days it's not hard to get a BMS that handles 24s, but beyond that things start getting really expensive. Since the BMS has more channels, there are more things that can fail. You also have to be a lot more careful with things over about 60v so you don't get zapped.

You can imagine pushing things to the extreme. A gigantic 1s or 2s pack with wires the size of a garden hose would be easy to balance but have more losses in the wiring. At the other extreme, something like a 100s, 1p bunch of 18650s you could use skinny wires and IGBTs in the controller. But if any one of the 100s cells has a problem, the whole pack is bad. And don't touch it. Imagine a 100s BMS too.

Somewhere in the middle is a sweet spot. You can make arguments for pushing toward either extreme but it usually boils down to cost effectiveness.

 

[larsb]

I have somewhere a 18650 test report from Nasa that concludes that parallell battery strings should not be more than 5 to keep the failure risk low enough for their liking

It's a risk even though the other batteries will try to balance the group if there are many cells involved.

 

[liveforphysics]

If i was building an EV to take on top fuel dragster records, I would run 20s.

If I was building a solar car for maximum efficiency, I would run ~10-20s.

If I was building an ebike to run at say <5kW, I would run ~10s-20s depending on controller/motor pairing.

Once you master high current interconnects and harnessing, you start to love amps and see voltage bringing you drama with corrosion and management hassles in return for nothing (or a dream that you're system will end up lighter).

 

[Sunder]

Once you master high current interconnects and harnessing, you start to love amps and see voltage bringing you drama with corrosion and management hassles in return for nothing (or a dream that you're system will end up lighter).

Hi LFP,

I asked this before in another thread, but I don't recall anyone answering. What do you need to do for corrosion management in a 144v nominal system?

I selected that voltage before this discussion came out as the highest of the retail level equipment, and until 2 weeks ago, when I finally ordered the controller (144v @ 400A... Or 72v @ 700A?) I still doubted my choice. I'm committed now, but this bike is going to be a daily rider. The last thing I need is to have high voltage cables flailing everywhere because the contact corroded enough to break under vibration...