Why 50V-60V is considered the optimum battery voltage?

[liveforphysics]

The new generation of high-capacity / high-discharge cells had started me thinking about the viability of smaller, lower voltage battery packs combined with more robust cabling for mid-range, mid-power bikes (commuters and grocery-getters) but I don't have the EE background to know if this was more than just wishful thinking. Sounds like it could be reasonable.

The new cells available today are amazing, but it's no additional burden on the cells to do a low voltage system vs high voltage system. If you have 50cells and a 1kW bike, it's going to be 20w per cell, and identical current per cell no matter if its configured say 5s10p or 50s1p. The difference is the 5s pack is just way less to monitor.

 

[Alan B]

There are interesting engineering trades to make here. For example, using a 10 cell or a pair of 5 cell BMS chips as opposed to needing 12 or 18 FETs to handle more current. BMS ICs are much cheaper than Power FETs and heavy traces are not cheap, nor are heatsinks, and the size of the controller is driven more by current than voltage, which has packaging costs and fitment to a bicycle issues.

I suspect that 6-12 FETs, 60V and moderate current is in the optimum neighborhood these days. When they come out with much better FETs that are cheap, that might change, but at this point the BMS chips are very cheap compared to the power FETs and extra copper.

 

[liveforphysics]

I suspect that 6-12 FETs, 60V and moderate current is in the optimum neighborhood these days. When they come out with much better FETs that are cheap, that might change, but at this point the BMS chips are very cheap compared to the power FETs and extra copper.

This has been the general consensus for the last 10 years, meanwhile MOSFET RdsOn has dropped by ~5x.

 

[Alan B]

Since FET heating is I squared R, a reduction of 5X is only an improvement of 2.2x in current for the same FET heating, so now a 6 FET controller will do about what a 12 FET used to do. Cost of BMS chips has probably dropped more than 2X as well, and copper has probably gone up 2X in cost. Intrinsically a BMS chip could be a lot cheaper than it already is, whereas FET cost is probably not as fluffy. The FET resistance is already in the neighborhood of the traces, connections and conductors so lowering the one component doesn't translate directly to overall improvement.

I don't see any 90+ N-m hubmotors that a 6 FET controller can easily drive at 18V to 28 mph. As motors get below 4 turns it would seem the end-turn efficiency losses would become an increasingly larger fraction and restrict the motor's efficiency.

It seems that a current modern 6 FET controller (like the PhaseRunner) runs a 10 turn motor quite nicely (at 60-72V), but a 4 or less turn would need a lot more current.

I built my first ebike with lots of copper to reduce losses. The bulk of high current copper connections isn't a great fit for a bicycle.

There are many inter-related issues. It will be interesting to see what actually becomes available.

 

[liveforphysics]

Seems like I should just build a kickass high efficiency sporty performance Ebike running 6s to show it can be done.

It's no end turn penalty, and with good packaging layout it's going to be a few ounces more copper in harnessing.

 

[diff_lock]

Seems like I should just build a kickass high efficiency sporty performance Ebike running 6s to show it can be done.

It's no end turn penalty, and with good packaging layout it's going to be a few ounces more copper in harnessing.

Where would you find a (hub)motor with a kV high enough? Or would you build a setup with a reduction?

 

[Alan B]

That's always the best way - show it.

It would be easier with a reduction, plenty of high KV motors. Hubmotors not so much.

 

[999zip999]

A 6s battery @ 25.2 volt fully charged and deliver 400 amps. 1t motor. These is what you talking about ?

 

[liveforphysics]

Seems like I should just build a kickass high efficiency sporty performance Ebike running 6s to show it can be done.

It's no end turn penalty, and with good packaging layout it's going to be a few ounces more copper in harnessing.

Where would you find a (hub)motor with a kV high enough? Or would you build a setup with a reduction?

If I added a reduction I would be missing the point of efficiency through simplicity.

My last handful of pack designs have all been for vehicles pushing >500w per volt of battery pack, with a few over 1500w per volt. It always seems a bit funny to me in the Ebike world to see all the added headaches of >10-20s packs for delivering a few kW of power a foot or two to the motor.

 

[imidacloprid]

Its all about finding a balance between wire size and desired max./conti. power AND available hub motor turn count and desired max. speed.
For FOUR times Ampere the copper conductor diameter need to be doubled.
After doubling the diameter, the new area is equal to FOUR times the original area.

 

[thunderstorm80]

I must say I disagree with you.
400A? What wire gauge and transistors are you using?
It's WAY HEAVIER than the weight of the BMS you saved.
I also see how low voltage seriously limits my power output. Here is a current-example I deal with:
I tune my new Phaserunner with my 9C+ 2706. The phaserunner can keep up with steady 40A of phase-current. For pulse regen braking you need way more than that, but if the phaserunner is already in it's thermal rollback mode, you get up to 40A or less depends on it's temperature. If instead I've had a 2707 or better - the H3525, I could have get way more braking torque with 40A of phase-current, with the price of having to increase the battery voltage.
The phaserunner can deal with up to 90V, so why not take advantage of that? You can then maximize your power density. That means: Use 21S NCA cells - 75.4V. (88V hot off the charger)
I also understand the phaserunner's heat (and any other controller) is only dictated by the battery&phase currents, so if you can raise your voltage - you can reduce the wasted heat on the controller. (for the same power output)
Of course you mostly save weight and heat on all your wires, with exception of the motor windings as those have the same copper weight. (for the same motor series)

What do you think?

I am also now in dilema of how big the 21S should be. I think about 21S5P which is 75V 15Ah and 5.5Kg. (with BMS).
That should be able to cope with 20A of battery-current regen for short period, and it has a nice capacity and not too much weight.
But since both of my motors can draw together a peak of 4KW, I wonder if this is enough: For example, I would like the option of going fast (60Kph) with the 2706 when I need to, and that speeds drains energy really fast.
I can take a higher weight pack of course - but what do you think is the maximum battery weight it's recommended to place on a full-suspension mountain bike? I currently place a 4Kg A123 battery on the seatpost rack, which can handle up to 15Kg, but I am sure you shouldn't cross 8kg or so.

I will also ask this in terms of dimensions: The 21S5P pack is offered to me as rectangular 40x11x7.5cm, which I think is a good maximal size that can fit almost any frame.
Do you think I should choose a different rectangular geometry, to increase the versatility in case I change bike? (no triangle packs, please)

 

[liveforphysics]

If you work designing multi-thousand-amp systems as a day job, making a couple feet of 100-200A bussing seems like a walk in the park.

I do understand it seems like something tricky if your experience had been limited to working with low currents.

 

[spinningmagnets]

Luke, you're making a persuasive argument. Suppose I was willing to go down as low as 12S, 10S, 8S, or even 6S, but very high amps (when the trend has been to go up to 20S?). Would I buy an 18-FET controller without FETs, buy them myself, then solder them in? What new FETs are the cutting edge for lower volts and higher amps? If not a common and avaiialble 18-FET, something from Kelly? I am inexperienced with higher-amp controllers.

A 5P pack of Samsung 25R cells is readily available (no need to special-order), and could provide 100A safely (and I suspect a light bicycle would accelerate fast enough that it would never draw the full 100A)

#420 chain? something similar to this picture, but not these exact components (just a pic I had handy, for conversational purposes)

 

[thunderstorm80]

If you work designing multi-thousand-amp systems as a day job, making a couple feet of 100-200A bussing seems like a walk in the park.

I do understand it seems like something tricky if your experience had been limited to working with low currents.

Sorry. Point taken!
Still I wonder - what motor you use that has only a 1t count?
And what do you think about my battery idea, or would you plan it otherwise?

 

[imidacloprid]

Searching on E-S i found here https://endless-sphere.com/forums/viewtopic.php?f=31&t=76109&start=50 conversation about T1 hub motor.
Circuit have custom wind a T1.5 hub motor. But no more info than a no-load test.

 

[liveforphysics]

Suppose I was willing to go down as low as 12S, 10S, 8S, or even 6S, but very high amps (when the trend has been to go up to 20S?). Would I buy an 18-FET controller without FETs, buy them myself, then solder them in? What new FETs are the cutting edge for lower volts and higher amps? If not a common and avaiialble 18-FET, something from Kelly? I am inexperienced with higher-amp controllers.

Infinion optimos5 has some impressive offerings.

Also, I'm not saying 20s isn't the ultimate for a high performance application due to the MOSFET power density per package unit being better in the 100V fets than 150v or 75v range at this time.

I just personally have come to appreciate fewer series cell count strings and embracing current. I find copper/aluminum bus cross section to be more reliable, because every added bms wire and connection and components is added points of failure, and making power bus terminations to handle just a few hundred amps and last forever is cake, and publically available to order MOSFETs have been released that can do compact 6 fet designs for 6s pack voltages and handle a few kW easily.

http://www.infineon.com/dgdl/Infineon-IRL40SC228-DS-v01_00-EN.pdf?fileId=5546d462566bd0c701567ece08d03664

If you are chasing the pinnacle of power density, by all means go with 20s.

If you just want a couple kW for a bicycle it seems like a bit of unneeded complexity from my own perspective, but I understand the concerns of folks who have a lot of experience working in say 5Amp to 50Amp systems and seeing first hand what hundreds of amps of current do to cabling and interconnections designed for 5-50Amps. If cabled and terminated correctly it gets no warmer and wastes no more power and the lower system voltage makes no impact on batter/controller/motor system efficiency if using the right parts and designed around embracing current rather than always trying to avoid it.

If there's one common trend I've seen in all my EV builds, it's that current is the friend of EV performance, not the enemy. Properly sized wires and busing and interconnect terminations are not things prone to malfunction.

Wouldn't the Inductance from a 1T hub motor kill any controller?
Never have seen someone sell a 1T hub motor.

It's been the same difficulty it was a decade ago, but today we have MOSFETS capable of higher switching speeds with lower losses and affordable high current measurement abilities have increased at least 100x in performance and motor-controller brain chips have 10-100x faster processors to control the parts.

I bet it's cheaper today to make a motor controller that does ~200A phase current and capable of say 7-8s battery voltage and capable of controlling current in a 1T winding than it cost 10 years ago to build one that did say ~40-50A and could only control current in high inductance motors.

 

[thunderstorm80]

Suppose I was willing to go down as low as 12S, 10S, 8S, or even 6S, but very high amps (when the trend has been to go up to 20S?). Would I buy an 18-FET controller without FETs, buy them myself, then solder them in? What new FETs are the cutting edge for lower volts and higher amps? If not a common and avaiialble 18-FET, something from Kelly? I am inexperienced with higher-amp controllers.

Infinion optimos5 has some impressive offerings.

Also, I'm not saying 20s isn't the ultimate for a high performance application due to the MOSFET power density per package unit being better in the 100V fets than 150v or 75v range at this time.

I just personally have come to appreciate fewer series cell count strings and embracing current. I find copper/aluminum bus cross section to be more reliable, because every added bms wire and connection and components is added points of failure, and making power bus terminations to handle just a few hundred amps and last forever is cake, and publically available to order MOSFETs have been released that can do compact 6 fet designs for 6s pack voltages and handle a few kW easily.

http://www.infineon.com/dgdl/Infineon-IRL40SC228-DS-v01_00-EN.pdf?fileId=5546d462566bd0c701567ece08d03664

If you are chasing the pinnacle of power density, by all means go with 20s.

If you just want a couple kW for a bicycle it seems like a bit of unneeded complexity from my own perspective, but I understand the concerns of folks who have a lot of experience working in say 5Amp to 50Amp systems and seeing first hand what hundreds of amps of current do to cabling and interconnections designed for 5-50Amps. If cabled and terminated correctly it gets no warmer and wastes no more power and the lower system voltage makes no impact on batter/controller/motor system efficiency if using the right parts and designed around embracing current rather than always trying to avoid it.

If there's one common trend I've seen in all my EV builds, it's that current is the friend of EV performance, not the enemy. Properly sized wires and busing and interconnect terminations are not things prone to malfunction.

Wouldn't the Inductance from a 1T hub motor kill any controller?
Never have seen someone sell a 1T hub motor.

It's been the same difficulty it was a decade ago, but today we have MOSFETS capable of higher switching speeds with lower losses and affordable high current measurement abilities have increased at least 100x in performance and motor-controller brain chips have 10-100x faster processors to control the parts.

I bet it's cheaper today to make a motor controller that does ~200A phase current and capable of say 7-8s battery voltage and capable of controlling current in a 1T winding than it cost 10 years ago to build one that did say ~40-50A and could only control current in high inductance motors.

Considering my basic soldering tools and my home equipment, I would prefer to stay with the highest voltage and the lowest current possible. On a long distance touring trip, you really don't want to worry that one of your connectors might come wiggly and melt. (if you used high current)
I would prefer to spend money on a good controller (Phaserunner), and on a high quality BMS, and have my mind in peace.
I can also temporary increase my cruise speed if needed, without sacrificing efficiency due to field weakening.
My experience with high currents have mostly been very negative - I already witnessed the Grinfineon 72V40A starting to melt the Anderson phase connectors towards the 2706, before I switched to the Phaserunner, that practically solved that issue.
You are right low voltage/high currents make things more simpler, but as someone wrote - copper prices are on the rise while BMS/Mosfets are declining.
15 years ago I used low voltage SLA and a brushed DC motor, and had to replace melting connectors on a regular basis, not to mention the heavy gauge wires. We move forward, not backward - And the Phaserunner is the proof to what the future holds. That's only my opinion, of course

 

[liveforphysics]

Solder = fail for high current. Solder conducts like crap.

https://www.amazon.com/1-25-16mm²-Japanese-Strength-Terminal-Non-Insulated/dp/B00A71FA98

(If you have higher current needs it only takes going to this hydraulic crimper)

https://www.amazon.com/gp/aw/d/B00HJXG3KM/ref=mp_s_a_1_3?ie=UTF8&qid=1474362968&sr=8-3&pi=SX200_QL40&keywords=hydraulic+crimping+tool&dpPl=1&dpID=41CIqOMNs7L&ref=plSrch

With that crimper and these brand/style lugs and shrink:

http://www.nichifu.com/buttsni.htm

http://www.nichifu.com/olc/SmartCat52fb.html?az=73474777&type=list&tablewidth=675&font=verdana&fontsize=1&tablealign=right&stripecolor=CCCCFF&ptno=D-8A+D-14A+D-22A+D-38A+D-60A+D-70A+D-80A+D-100A+D-150A+D-180A+D-200A+D-325A&show=unit_price+part_number+Wire_Size+Stud+W+C+L+B+D+S+Tools+Std._Packing+Material+UL/CSA_Listed&header=Tubular+Terminals&template=http://www.nichifu.com/template.htm

Your joints will be cold places on the wire because that crimper can crush them into a homogenized cold forged section of copper that has flowed together.

The cover with some of this brand of very robust glue lined shrink.

https://www.amazon.com/Dual-Wall-Adhesive-Lined-Shrink/dp/B00RC36FWA

You can make yourself 200-400A interconnections this way for ~$0.50-$1 each that will not get hot or fail before your bike wears out and the adhesive lined shrink keeps it weather robust.

If you attack 200-300A current busing with techniques good for 25-50A, you're not going to have a lot of luck. If you go at it with appropriate sized cold forged crimps it's no longer an obstacle or issue at all.

 

[footloose]

... that crimper can crush them into a homogenized cold forged section of copper that has flowed together.

LFP -- I was a happy man with a full life when I woke up this morning.
Now I am lusting after a new crimper!

 

[Cephalotus]

50V DC is safe to touch, 72V DC could kill you. Not very likely but possible.

Also sparks will burn your contacts more quickly at 72V DC.

 

[Alan B]

The NFPA 70E safe electrical work rules regarding voltage were set for 50 VAC. They kept the same values for DC for simplicity, according to one expert instructor who is involved with the committee. They are planning to change the DC rule to 100 VDC while the AC rule will remain at 50 VAC. It is always best to avoid touching even these voltages, but much stricter work processes are required when these voltages are exceeded.

Note that 50 VAC and 100 VDC are not "safe" if the skin is not dry and undamaged. Penetrating the skin, for example, makes even a few volts potentially fatal.

Wearing a set of clean and undamaged leather gloves provides significant protection from these voltages, as well as protection from KFF burns (kentucky fried fingers) from arc plasma. Always prudent to wear gloves and eye protection when dealing with live circuits.

The energy in the spark is 1/2 * C * V^2. So it is a function of voltage squared. Higher voltages do take a toll on the connectors when good size capacitor banks are involved.

 

[Alan B]

The hydraulic crimpers are excellent, I have a lower cost one that came from ebae, the one Luke linked looks even nicer.

The many tons of force doesn't seem to quite fuse the strands into one piece of metal, but it is very close to that, and conductivity is excellent.

I have been using the low cost hardware store house wiring electrical crimp tubes (rather than butt crimp tubes), and overlapping the input/output wires and crimping over them with the tube, then covering the joint with the glue-lined shrink tube. These are extremely solid connections. This technique reduces the number of joints the current must cross and shortens the connection for less bulk. The current does not need to flow through the crimp tube as the wires themselves are in direct compression contact.

Here is one 4P section of the Borg's Multistar Harness:

 

[liveforphysics]

They are planning to change the DC rule to 100 VDC while the AC rule will remain at 50 VAC. It is always best to avoid touching even these voltages

I've taken 116vdc across the chest at least once a week for the last 6 years. Never had more than an uncomfortable tickle if my hands are especially sweaty, otherwise a faint tickle.

 

[thunderstorm80]

In my current 79.2V system (86.4V hot of the charger), I can feel a tickle if I am closing a circuit with my fingers. Nothing more.
The sparks on plugging are a serious threat - I solved mine by using a single pole household circuit-breaker rated at 40A.
I always plug everything first, and the circuit-breaker is the last to turn on. The arc upon connection makes a strong bang sound with a beautiful visible arc, and after several years - I think I only needed to change it once.

Regarding dangerous voltages: At higher voltage (for example 220DC vs 220AC) DC current is more dangerous - because it latches your muscles and you can't let go, while the AC goes on/off 100/120 times per second.
That's why the threshold for muscle latching uncontrollably is agreed to be 30mA, and that's the trigger current of all anti-electrocution household devices. (That measure the difference in the current between the live and the neutral wire). For DC, that figure would have to be much lower.

 

[Alan B]

Understanding of electrical hazards has changed over time (and many old understandings are not actually correct, or they are correct in a certain limited circumstance but not in general). Latch-on occurs with AC also, and as it turns out 60 hz is extremely dangerous to the heart due to triggering of fibrillation, just a few milliamps can shut down the heart. 120 VAC kills more than any other voltage/frequency because people don't take it seriously. Best to avoid becoming a conductor at any time and use good work practices lest you forget one day when working with 240 VAC or 2+KV DC.

Note that AC breakers are not designed to handle the DC arc and may fail to operate correctly when needed. These days there is no reason to use household AC breakers as there are low cost AC/DC rated breakers available manufactured for the solar industry. These have magnetic DC arc quenching gear in them and are much more suitable for ebikes.