Alan B
📗 Minor legend
16x4 is very close to the Cromotor's 18x4, and the MXUS 3000 is 16x4 as I recall.
-
Hello ES! We could use some help to get us past the finish line on building the new knowledgebase for the forum.
Can you donate? Please see our fundraising page. Thank you!
Science, Physics, Math, & Myth
- Thread starter Kingfish
- Start date
teslanv
🏍️ Giga poster
The MXUS 3000W V2 is indeed a 16X4T wind.Alan B said:
What I am trying to determine is if it makes any sense to go with a 21X3T wind, if I am bound by controller current limitations. If I understand correctly, higher phase currents will put more stress on the FETs. Is that correct? If we are talking about an 18FET IRFB4110 controller, is the 16X4T the optimum wind?
Alan B
📗 Minor legend
The FETs operate at phase current, so it is a concern. But the implementation and drive are very important, the number of FETs doesn't tell the whole story. A good 24 FET controller can handle several hundred phase amps, while a lesser one will burn.
I find the Cromotor produces plenty of power at 18x4 and 120 amps phase current (80 amps battery), but with a sine wave controller at 18S it tops out at about 32mph with 17" moped tires a little over 23" diameter, which is excellent. Field Weakening can push it to much higher speeds but I haven't bothered. It can be pushed much harder.
I find the Cromotor produces plenty of power at 18x4 and 120 amps phase current (80 amps battery), but with a sine wave controller at 18S it tops out at about 32mph with 17" moped tires a little over 23" diameter, which is excellent. Field Weakening can push it to much higher speeds but I haven't bothered. It can be pushed much harder.
MXUS 21x3T needs 1,33 x current at 0,75 x voltage compared to 16x4T. If phase wires are also upgraded by the factor 1,33 than the motor will produce almost exact same torque and efficinency. Sure, FET losses are higher by factor 1,33 but how many additional watts heat might that be??
IMO its better to run a faster motor than push a slow motor with flux weakening to higher speed (like almost everyone here does). I will swap out Cromotor on my streetbike for 3T MXUS because i believe it will stay cooler at 80-90kmh, but mainly because it weighs less. The Cro needs Adaptto OVS 3 to achieve this speed while the MXUS will do it without at hopefully higher efficiency
IMO its better to run a faster motor than push a slow motor with flux weakening to higher speed (like almost everyone here does). I will swap out Cromotor on my streetbike for 3T MXUS because i believe it will stay cooler at 80-90kmh, but mainly because it weighs less. The Cro needs Adaptto OVS 3 to achieve this speed while the MXUS will do it without at hopefully higher efficiency
Alan B
📗 Minor legend
Another way to look at it is that 33% more FETs will be needed. However adding FETs becomes less effective, so more FETs will be needed by another factor as current division is not equal. So 50% or more FETs must be added. I've had many 24 FET failures on the 18x4 turn motor, so perhaps a 36 FET would be recommended for 3 turns. Cranking up the current on the same controller won't get parity, that could always have been done on the higher turn motor, and won't result in the same reliability or safety margins. Lower inductance and resistance require better controller design which may mean moving to much more expensive controllers, preferably with actual phase current measurement and control.
Also watch the eRPM and motor losses if RPM increases are planned, lower pole count motors are designed for efficiency at higher RPM.
Also watch the eRPM and motor losses if RPM increases are planned, lower pole count motors are designed for efficiency at higher RPM.
liveforphysics
📜 Mega legend
Alan B said:
Alternatively, you could just embrace the improved FET technology available today.
Remember, copper and iron aren't likely to make any magical substantial improvements over time. Silicon power switching density gets improved by someone every few months.
http://www.ti.com/product/csd19506kcs
teslanv
🏍️ Giga poster
liveforphysics said:
So Would this be as easy as swapping out the FET's in a standard Xie Chang controller to a newer more current capable model of MOSFET, or would it require a redesign of the controller to accommodate the newer FETs?
zombiess
🏍️ Giga poster
teslanv said:
The die are more capable than the legs and bond wires which feed many modern power MOSFETs. Losses are more involved than just conduction which are the losses often referred to on this site when talking about RDSon.
The package and its ability to shed heat are a limit on many MOSFETs. Data sheet specs need to be taken in the context of the real world.
I think the Rdson vs the voltage are the most important specs. If you have a mostfet with 2x the rdson its going to produce more heat might not be 2x more heat but its more so the rdson is very important you still need to look at all specs but I don't ever bother with FETs with high rdson because they have no hope feeding the current the lower rdson FETs do.zombiess said:
Alan B
📗 Minor legend
The Xie Chang controllers probably won't benefit all that much from the better FETs since they don't drive the FETs properly. There will be some benefit, but not the full improvement. New controllers with proper design are required.
liveforphysics
📜 Mega legend
Alan B said:
I think they would still greatly benefit, but yes, you're definitely right about needing the next generation of controllers to have real gate drive design rather than the barely functional disasters on a board that are the industry norm.
A lot of this stuff just requires spending an extra $10-20 in parts per controller. Then you get real FET drive and real FETs. It's a no brainer, the only reason it's not currently the standard practice is because the origin of most ebike related parts for the last decade has been imported Chinese domestic market stuff that has been "value-engineered" into rubbish on the borderline of functionality by the folks who keep demanding they cost less.
For hobby/enthusiast level controllers where the user is going to commute with it everyday, spending an extra $20 on the controller isn't a big issue IMHO, particularly when the return on that $20 is reliable and smooth power-wheelies at every stoplight.
John in CR
📜 Mega legend
$20 more for a controller is fine. Where I have the problem is when the little aluminum box of electronic bits cost more than the motor it drives.
John in CR said:
Just putting it in perspective, in most industries the price of the motor controller is roughly comparable to the price of the motor it drives. Even in the uber cheap world of RC motors and controllers this holds true more or less, so I don't see why you'd expect ebikes to be that different. I once thought it seemed pretty crazy too, but the actual expenses involved in making a 3 phase controller are comparable to making the motor that it drives. There's no moore's law for mosfets, heavy copper PCBs, electrolytic capacitors, etc. Microprocessors have gotten much cheaper and more powerful, but that's only a small part of the BOM cost of a good motor controller.
John in CR
📜 Mega legend
justin_le said:
Just because it's commonplace doesn't make it right. I can drive a $600 retail motor with over 20kw peak input reliably with less than $400 retail of controllers, and I can drive it to over 12kw peak for under $300 worth of controller. Once you start talking about more common ebike power of under 6-8kw peak then controllers get down into the double digit prices. Before the middle of last year those prices were the domain of dumb inflexible controllers and only available at the lower end of the power range, but those days are over, and we're talking about controllers much more feature rich and flexible than even the more expensive controllers have been up until now.
Motors are made with expensive copper, high quality expensive steel, and rare earth magnets. Controllers have a bit of copper, ever cheaper and more power silicon chips, and PROGRAMMING. It's the programming that has made them cost more than I believe they should, and thanks to the Chinese the days of expensive controllers are soon to be over.
John the best thing for you to do is go build both a motor and controller and see how long each takes and how hard they are to build then see what the costs are of each the motor and controller.
From experience. Anything with quality, controller wise you will be EXTREAMLY lucky to get it as cheep as the motor. You have been VERY lucky and some what spoiled to find the deals on almost brain dead next to junk (that works well for your application) controllers to keep you going. But the GOOD controllers will take some money to build even on a mass scale. This will be like big screen TVs and computers as the price will be chased down with volume production and people looking to find a better way but it will take time in the same time motors will become cheaper....
From experience. Anything with quality, controller wise you will be EXTREAMLY lucky to get it as cheep as the motor. You have been VERY lucky and some what spoiled to find the deals on almost brain dead next to junk (that works well for your application) controllers to keep you going. But the GOOD controllers will take some money to build even on a mass scale. This will be like big screen TVs and computers as the price will be chased down with volume production and people looking to find a better way but it will take time in the same time motors will become cheaper....
John in CR
📜 Mega legend
Arlo1,
The reason I disagree with you and Justin regarding pricing is because the ebike and escooter factories share my view about controllers, and finally at least some of the big controller factories are making significant changes that will benefit us all.
The reason I disagree with you and Justin regarding pricing is because the ebike and escooter factories share my view about controllers, and finally at least some of the big controller factories are making significant changes that will benefit us all.
johnrobholmes
🏍️ Giga poster
John, you of all people shouldn't brush off the complexity of a motor controller. The lower end of dumb and semi dumb controllers will be raised over time as hardware technology trickles down and new softwares become old, but prices aren't artificially high because of a group of greedy vendors. The best software in the world can't make shitty hardware perform better, and fantastic hardware needs fantastic software to run right. And how do the cheapest chinese companies make the best software? They steal it or wait until it is obsolete and discarded. Or they can develop it themselves and charge the price it takes to recoup development which is why we have the prices found in today's controllers.
I can also confirm that RC motor controllers are very similar to motor prices in volume manufacturing, for a given power level. It also takes a highly skilled (highly paid) person or team to engineer a motor controller whereas motors have plenty of pre-engineered stators ready to roll of the lamination suppliers machines. Less up front engineering cost on the motors, more raw materials cost.
So if motor controllers should be so cheap, go ahead and make yourself one motor and one controller from scratch and count up all your time and inputs doing it. If you come out less than 10x the development cost than the motor on your controller you should open your own company. It takes me about 7 grand of labor to fully engineer a motor from nothing. Power engineers have told me 2000 hours into a 100% new controller design is not out of the ordinary, and at the base rate of $35 an hour to hire a smart and fresh power engineering grad thats right at $70,000 without taxes and benefits. Contract labor for power engineering in the US can be anywhere from $120 to $400 an hour. The same will apply in any country at the price of their currency and labor market, there is a huge labor disparity between developing a motor and developing a controller.
I can also confirm that RC motor controllers are very similar to motor prices in volume manufacturing, for a given power level. It also takes a highly skilled (highly paid) person or team to engineer a motor controller whereas motors have plenty of pre-engineered stators ready to roll of the lamination suppliers machines. Less up front engineering cost on the motors, more raw materials cost.
So if motor controllers should be so cheap, go ahead and make yourself one motor and one controller from scratch and count up all your time and inputs doing it. If you come out less than 10x the development cost than the motor on your controller you should open your own company. It takes me about 7 grand of labor to fully engineer a motor from nothing. Power engineers have told me 2000 hours into a 100% new controller design is not out of the ordinary, and at the base rate of $35 an hour to hire a smart and fresh power engineering grad thats right at $70,000 without taxes and benefits. Contract labor for power engineering in the US can be anywhere from $120 to $400 an hour. The same will apply in any country at the price of their currency and labor market, there is a huge labor disparity between developing a motor and developing a controller.
teslanv
🏍️ Giga poster
Coming back to the Myth a little here:
Is it fair to say that for a given motor design (And negating any controller inefficiencies), if you were to compare a 16X4-Turn motor to a 8X8-turn motor, could you accurately say that:
For a given battery voltage at 100% Throttle, the 16X4-Turn motor will spin exactly twice as fast as the 8X8-Turn motor and for a given battery and phase current profile (amperage) the 8X8-turn motor will produce twice the torque as the 16X4-Turn motor.
Or to say this statement a different way:
It takes twice as much voltage for an 8X8-turn motor to spin at the same RPM as a 16X4-turn motor, and it takes twice as much current for the 16X4-turn motor to produce the same torque as the 8X8-turn motor.
I am not trying to propagate any more myth here, just trying to confirm that what I understand about the relationship between a motor's winding, and Speed and torque are accurate.
Is it fair to say that for a given motor design (And negating any controller inefficiencies), if you were to compare a 16X4-Turn motor to a 8X8-turn motor, could you accurately say that:
For a given battery voltage at 100% Throttle, the 16X4-Turn motor will spin exactly twice as fast as the 8X8-Turn motor and for a given battery and phase current profile (amperage) the 8X8-turn motor will produce twice the torque as the 16X4-Turn motor.
Or to say this statement a different way:
It takes twice as much voltage for an 8X8-turn motor to spin at the same RPM as a 16X4-turn motor, and it takes twice as much current for the 16X4-turn motor to produce the same torque as the 8X8-turn motor.
I am not trying to propagate any more myth here, just trying to confirm that what I understand about the relationship between a motor's winding, and Speed and torque are accurate.
i believe the 8x8 only will produce twice the torque at stall or at very very low rpm and than torque drops very fast because it max out at only half the RPM compared to a 16x4. this has to do with back EMF (the voltage the motor pruduces if you would spin it "by hand" at given RPM), phase resistance -> therefore voltage drop in windings.
or with other words the 8x8 only draws max power for a very short time because it reaches much faster top speed or max RPM.
or with other words the 8x8 only draws max power for a very short time because it reaches much faster top speed or max RPM.
Combine the above with the fact that controllers do some sort of 'from zero RPM' current limiting and you end up with an even shorter blip of torque.madin88 said:
[EDIT] Sorry, my comment was assuming the same controller on each wind.
liveforphysics
📜 Mega legend
teslanv said:
Anytime the motors with differing Kv in your example are at the same RPM getting the same input power, they will be making exactly the same amount of torque output, as well as the same amount of heat produced to make that torque as well as identical efficiency and battery current etc.
In your example, the motor with half the turns gets twice the RPM range of the other of course.
