Voltage sag (was: 1500w motor overheating on < 20A?)

[Jabotical]

[**EDIT** Turned out what I thought was motor/controller heat-related was instead my CA limiting output due to an aggressive CA low voltage cutoff combined with voltage sag in my underspec'd NCRb battery. Savvy and generous responders did provide some excellent insights and info about both motor heat and battery spec'ing]

I converted a Trek 820 a few days ago with a 1500w leafbikes motor, a C3635-NC Grinfineon controller (35A, 48v max), and a 52v 13.5ah battery.
This was my introduction to ebikes, and a joy to ride at first. On the first longish ride (a dozen miles) I noticed (and confirmed via Cycle Analyst) the power dropping increasingly low even at full throttle, as I went along.

I belatedly realized my battery pack was only equipped to source ~20A (4p NCRb cells), so the next day I reduced the current limit to that and went on another longish ride on a full charge (eleven miles in the morning and then eleven back in the evening). It did involve some (fairly minor) inclines, but as I went along the second stretch the performance got worse and worse, until the CA said I was only pushing ~5A / a few hundred watts (and it felt like it!).

In the first instance my battery got pretty hot, and I assumed that overtaxing it was the only issue. But this second time (running with a 20A limit, mind you), towards the end I felt the motor, and it was so hot it was painful to keep my fingers on it. The controller was also nearly that hot. What gives?? That motor is nominally rated at 1500w, and people here report running it continuously at much more than that. The phase wires and battery connector were cool to the touch, and the battery case wasn't very hot.

On the Cycle Analyst diagnostic screen it would show a capital V when I engaged the throttle, which I believe means the voltage was being limited? The throttle mode was set to amperage. I'd only used around 8 or 9 ah of my 13.5 by the time I limped home all asweat from having to peddle so much in the heat.

Any ideas what could be going on? It's really hot here in Phoenix, was probably around 110 F on the way home when it failed/heated so dramatically. But surely that shouldn't cause a big beefy motor to overheat like that.
If I lift the back wheel and give it a moderately good spin with my hand, it goes for 4-5 seconds before dragging to a stop. I assume that's reasonable for a big DD motor, even a reportedly well-designed one like this? I can peddle along pretty well even without power (though there's naturally some drag compared to pre-conversion -- some of which might be the added weight). The regenerative braking continued to slow it down pretty well when I engaged it.

It's very disheartening, but I'm keen to do what I need to to get it functioning properly. Any advice from all you savvy types on how to analyze the issue is deeply appreciated!

 

[spinningmagnets]

Sounds like the wrong phase and hall combination between the controller and motor. The color codes on the wires are different between one manufacturer and the next. I always recommend that the throttle, controller and motor are bought from the same vendor, so this will not be an issue, very frustrating.

Here is Fechters wire-sorting thread:

https://endless-sphere.com/forums/viewtopic.php?f=16&t=3484

 

[Jabotical]

Ok, that would make a lot of sense. Thanks! I'll check the thread out.

I thought the Grinfineon controller was supposed to auto-detect the hall settings, but maybe I accidentally did something that made it lose or misconfigure them. And I guess the only reason I believe I have the phase wires right is because the wheel moves in the right direction (which is I believe all the instructions for the controller had said to verify).

If that's all it is, I will be quite pleased! Once again, you come to the rescue, spin. You are clearly a pillar of the community.

 

[dogman dan]

Umm, how to put this diplomatically. Do you weigh over 200 pounds?

It's phoenix, essentially hell on earth this summer. Nothing can shed heat for shit, particularly on the lower humidity days. Dry air cant cool shit. People argue this one with me all the time, but ask any old hippie from the desert how that VW van ran in summer, before it rained. Even once the rain comes, 110 is freaky hot.

Stop and go city riding means you spend a LOT of your ride in the very inefficient low rpm, leaving the stop sign. This means a pretty darn hot motor and controller is completely normal after about 10 miles. The first 20 feet, you are just making heat, so that's the time to stomp the pedals.

And, FWIW, regen braking heats motors too. So that aint helping any.

I think YOU DO have a battery issue. That 5 amps draw was likely all the battery could send at that moment. It could also be a mis calibration of the CA to the shunt in the controller.

Lastly ( actually always do this first) , check your plugs. You might be running on just two phases, or something weird like that.

 

[Jabotical]

I'm glad you weighed in, dogman. From my previous lurking I'd observed you were a notable Phoenician, and I thought you might have some locale-specific insight about the heat situation.

Hehe, I'm currently about 190 pounds, which I'm sure isn't ideal, but I figured (perhaps incorrectly) that should be manageable. I know my bike's not particularly light. I haven't weight it yet, but I can pick it up just fine when I need to.

Yeah, it's crazy hot here, I know. I just didn't have any practicable experience with how these sorts of components manage their heat transfer in this weather. I mean, modern-ish cars seem to have little trouble divesting enough heat, and power tools don't shut down, etc. But I guess they don't run constantly under such loads. This week is supposed to be cooler, so I'll see if that makes any difference.

Good point about city riding. I wasn't stopping and going too much at first before the performance started getting pretty bad while going east on Deer Valley, but it was up a bit of an incline. It's so hard not to want to use the motor to accelerate and then pedal to help it at speed, but I'll try switching to a lower gear and use the pedals for accelerating.

It had occurred to me that regen braking might not be helping, I had just never come across a single suggestion that it was significant. I guess Phoenix summers are a different reality than most people here are facing with their equipment.

Ah, if I DO have a battery issue, what IS the issue? Is the battery itself overheating (like I said, the bottle case didn't feel very hot, compared to the motor and controller)? I had the CA current limit set to the advertised continuous current capability of the 4 parallel NCRB sections (~20A)... As for the CA shunt value, good thought, but it seems unlikely since it's a controller from ebikes.ca that came with the shunt value written on it.

Thanks for the plug tip. I did check the phase connections more than once while riding, and they seemed fine (and none of them were even warm). But I'll go investigate them a little more thoroughly now, in case one of them somehow isn't conducting even though it's connected or something.

 

[spinningmagnets]

If your motor is hot, and the plugs are not...my money's still betting on a wrong combo. Best bet is borrow a 6--FET sensorless controller and run 20A.

You are clearly a pillar of the community[/ quote]

I'm more like the philosophical homeless guy who won't go away, and then gets drunk and passes out in the alley, and I have to sleep on my side so I don't choke on my own vomit, but...thanks for the encouragement!

 

[Jabotical]

If your motor is hot, and the plugs are not...my money's still betting on a wrong combo. Best bet is borrow a 6--FET sensorless controller and run 20A.

Ok, that's good input. My phase wire plugs aren't huge, either -- fairly small bullets rated at ~45A My current controller can run in sensorless mode. I tried all the phase wire combinations, and I believe there were three of them that resulted in the wheel turning forward. I couldn't tell any difference in the performance among them with the wheel freely spinning, so I went back to the one with the wire colors matching (which I recognize means little). I'll try a ride in sensorless mode and see if the motor heats up as quickly. I connected the temp sensor to my CA, though of course the value is pretty hard to interpret (as the main leafbikes thread predicted), but at least I can use it to compare.

You are clearly a pillar of the community

I'm more like the philosophical homeless guy who won't go away, and then gets drunk and passes out in the alley, and I have to sleep on my side so I don't choke on my own vomit, but...thanks for the encouragement!

People always appreciate having one of those around, as well.

 

[Sunder]

I tried all the phase wire combinations, and I believe there were three of them that resulted in the wheel turning forward. I couldn't tell any difference in the performance among them with the wheel freely spinning, so I went back to the one with the wire colors matching (which I recognize means little).

You're almost there.

Out of those three, one of them will consume almost no power (1-2amp) when it spins the wheel with no load, the other will use up much more (about 5a in my experience, but will vary depending on the motor)

The one that only consumes 1-2a is the real one.

So all you need is am ammeter, or to watch the CA.

 

[Jon NCal]

I tried all the phase wire combinations, and I believe there were three of them that resulted in the wheel turning forward. I couldn't tell any difference in the performance among them with the wheel freely spinning, so I went back to the one with the wire colors matching (which I recognize means little).

You're almost there.

Out of those three, one of them will consume almost no power (1-2amp) when it spins the wheel with no load, the other will use up much more (about 5a in my experience, but will vary depending on the motor)

The one that only consumes 1-2a is the real one.

So all you need is am ammeter, or to watch the CA.

I think there are 3 of the 9 phase combinations will be equivalent and correct in sensorless operation. ie A-->B-->C = B-->C-->A = C-->A-->B. Or am I mistaken?

 

[Jabotical]

I tried all the phase wire combinations, and I believe there were three of them that resulted in the wheel turning forward. I couldn't tell any difference in the performance among them with the wheel freely spinning, so I went back to the one with the wire colors matching (which I recognize means little).

You're almost there.

Out of those three, one of them will consume almost no power (1-2amp) when it spins the wheel with no load, the other will use up much more (about 5a in my experience, but will vary depending on the motor)

The one that only consumes 1-2a is the real one.

So all you need is am ammeter, or to watch the CA.

I think there are 3 of the 9 phase combinations will be equivalent and correct in sensorless operation. ie ABC = BCA = CAB. Or am I mistaken?

That would line up with my latest experimentation. All three of them spike up to ~5A when I go full throttle with no load, then drop to ~1.2A.

I guess I'll still try a test run with one or both of the other two combos, because I need to figure out what's going on (unless it's just the ambient heat + inefficient use of DDs + me being fat, as dogman proposed -- but it seems wrong to me that a motor this highly regarded, rated at 1500w, overheats when I feed it 1000w watts for a while)
I may also try a hall-less run to see if that helps any...

 

[Sunder]

I tried all the phase wire combinations, and I believe there were three of them that resulted in the wheel turning forward. I couldn't tell any difference in the performance among them with the wheel freely spinning, so I went back to the one with the wire colors matching (which I recognize means little).

You're almost there.

Out of those three, one of them will consume almost no power (1-2amp) when it spins the wheel with no load, the other will use up much more (about 5a in my experience, but will vary depending on the motor)

The one that only consumes 1-2a is the real one.

So all you need is am ammeter, or to watch the CA.

I think there are 3 of the 9 phase combinations will be equivalent and correct in sensorless operation. ie ABC = BCA = CAB. Or am I mistaken?

True on sensorless. On sensored mode, I think it would make a difference, since you want to make sure that not only is it firing in the right sequence, but matching the corresponding sensor.

Not 100% on this, but that was my experience. Should draw it up to figure it out for sure.

 

[Jabotical]

Using one of the other phase wire combos, with the hall sensor connected (using the Grinfineon auto-detected hall mapping), it behaved very similarly. I rode it hard for a few miles, with some inclines, and before long I was being limited by the low voltage cutoff I'd configured of 47.6v (increasingly so as I went along -- i.e. the allowed current was less and less). 47.6 is 3.4v per cell -- maybe that's a little conservative? The unloaded voltage at the end was 51.8v (3.7v per cell), and the capacity used since full charge had only been 5.8ah out of the 13.5 it's rated for, so it should be right in its nominally-operating band, right?

It was also only 100 degrees out. By the end the controller was hot enough I could barely keep my hand on it, and the motor was just about the same.

Is the voltage being dragged down sometimes a symptom of motors or controllers overheating? (i.e. is the low loaded voltage a cause or effect?)

Am I wrong in thinking it unlikely that a 35A-rated controller and a 1500w motor could't handle 19A of current continously, even on a hot day? (this ride I limited it to an extra-conservative 19A, since the battery should be able to do 19.5 continous).

I guess the next thing to try is running in sensorless mode?

Though according to that thread spinningmag linked: "A correct Hall/Phase wire combination will run very smooth and draw very little current (1 or 2 amps) under no-load testing. A wrong Hall/Phase wire combination will generally run very rough and draw much higher current under no-load testing"
-- All three of the combos I'd identified, with the Hall sensors connected, draw ~1.2 amps full throttle under no load.

 

[Jabotical]

Well, it seems that aside from whatever reason the motor and controller are heating up so hot, I very likely am dealing with a damaged battery.

When I went out to ride again after letting everything cool down, it picked up right where it was before -- the CA limiting the current more and more due to the low voltage cutoff being encountered. I must have really cooked the cells the first couple of days when I rode for a little bit with the limit at I think 32A, and then 26A. By the end I was only getting a little over 8A even though the motor was merely warm. CA shows 7.5ah used out of the 13.5 the pack's rated for, since a full charge this morning.

I'd recharged the battery after the last time I'd really started seeing this behavior, so I didn't realize it was related to the depletion level. Unloaded, it still spins fine. Just can't source the current without dropping past 3.4v/cell.
I imagine I made a rather expensive mistake, not limiting the current more aggressively from the beginning, with this anemic pack (unless it was defective to begin with, which is unlikely, and I'll never know anyway). I didn't quite realize how limited it was, or I'd never have bought it -- guess that's one way to learn.

Ah well. I can't stop now that I've had a taste. I reckon I'm just going to have to take the hit and get a new battery that can give me current I demand.

Maybe the motor and controller heating up aren't actually causing me any performance problems? They're just getting hot due to the factors dogman identified? How hot is too hot/normal? (I guess I can research that on my own)
If it weren't for the reports of people pumping thousands of watts through the leafbikes 1500w motor and it not even getting warm, I would just assume its heating was normal, at this point (but maybe it's simply the difference in ambient temperature in Phoenix vs where they are?)

 

[Jon NCal]

True on sensorless. On sensored mode, I think it would make a difference, since you want to make sure that not only is it firing in the right sequence, but matching the corresponding sensor.

Not 100% on this, but that was my experience. Should draw it up to figure it out for sure

Yes, hall combo needs to be correct, but he is using Grinfineon with auto hall mapping. So the controller will remap the halls in correct sequence.

 

[spinningmagnets]

Grin products are very reliable, but sometimes even a Grin product malfunctions. If the auto-detect for hall sequence is off, they will replace it no sweat. However, we must first determine what part is off.

 

[Jabotical]

Grin products are very reliable, but sometimes even a Grin product malfunctions. If the auto-detect for hall sequence is off, they will replace it no sweat. However, we must first determine what part is off.

Yeah, is there something I should do to test that? I guess I could try to rig up some sort of test circuit with leds, like I saw on that thread you linked.

Of course, when I ran in sensorless mode (hall plug not even connected), it didn't seem to improve things any -- motor + controller still got hot, still getting too little current from the battery far too soon (which I'm really assuming must mean I irreparably damaged the battery, per my last post).

The too-quickly-diminishing amperage situation I expect will only be fixed with a new battery pack. Would still be nice to figure out why the motor's (and controller's) getting hot enough for the case to nearly burn me, though (unless it turns out that's normal behavior, in which case I guess I just have to learn to know its limits and learn to work around it!)

 

[Jabotical]

I just noticed the lunacycles listing for the battery says "This pack comes with a very advanced BMS (battery management system) which protects the pack from damage during charge and discharge. This BMS also has temperature protection to keep the pack from getting too hot during discharge and charging."

Seems like that means I shouldn't easily be able to damage the battery by over-drawing from it, no? And yet, it seems to be unable to provide more than a few AH of juice.
Is there another potential explanation for the way the loaded voltage drops below normal discharge levels a few AH in from a full charge?

 

[dogman dan]

Well, at least its not your weight.

BTW, I'm not in Phoenix, but I live in the SW desert, and know a hot motor and controller is pretty normal in summer. They will get pretty warm very fast in the weather we're having lately. I'm not completely convinced your motor is running wrong.

But I'm pretty confused that one of those three combos did not show a markedly lower amps draw running no load.

If not already, It's clear you need to talk to Grin. Colors match ought to be correct, but confirm with them.

It's also clear, your battery is no where near happy. I think all you loss of power symptoms come directly from there.

 

[Jabotical]

Well, at least its not your weight.

BTW, I'm not in Phoenix, but I live in the SW desert, and know a hot motor and controller is pretty normal in summer. They will get pretty warm very fast in the weather we're having lately. I'm not completely convinced your motor is running wrong.

But I'm pretty confused that one of those three combos did not show a markedly lower amps draw running no load.

If not already, It's clear you need to talk to Grin. Colors match ought to be correct, but confirm with them.

It's also clear, your battery is no where near happy. I think all you loss of power symptoms come directly from there.

Good, good. One never wants it to be the fault of one's weight.
Duh, New Mexico is right there on the right of every one of your posts -- whoops. Still means you get low desert cred, obviously

I would have loved for one of those phase wire combos to perform differently than the other two, as well! Yeah perhaps I should talk to Grin.

I'm increasingly of the same mind, that these temperatures might be normal for a dry 100 degree+ day? So you see similar sometimes after pushing a DD hub for a few miles?

Mainly now I'm trying to figure out if my battery's bad, or if I've just fallen prey to the difference between loaded and unloaded voltage. I blindly took advice I saw somewhere to set the CA's low voltage cutoff to 3.4v/cell, but that might be a bad idea if I'm trying to discharge the pack past more than 50% or so?

If I look at the resting voltage this morning, after having used 7.75AH, it's 50.7v. Which puts me at 44% remaining on the official voltage curve, in line with the amount of capacity I should have at that point.

I will try reducing the CA's low voltage cutoff dramatically and relying on the BMS protection, to see if I can still perform well at this charge point.

 

[nutnspecial]

I will try reducing the CA's low voltage cutoff dramatically and relying on the BMS protection, to see if I can still perform well at this charge point.

Hi, that's a good idea. Maybe you can also test the 'self learning' by swapping a few halls and seeing how it responds.

Also, you should immediately notify sellers' support teams for both items in question. They will walk you through the proper steps of info gathering and testing to determine what, if anything, is wrong for each, independently.

In my experience, the batteries tend to be nearly worthless for any substantial draw below 3.5v. You do have a bms though, so take CA out of the equation and let bms do it's thing. Also make sure the battery is fully charged and balanced. Sometimes that takes quite a while on the charger I hear. Though there should be no excuse for a high imbalance with a new quality bms'd pack.

 

[Jabotical]

Thanks for the suggestions and insight!

--- the battery behaved! ---
Without the CA limiting the voltage I went out and pulled 4-5 more AH out of it without any discernibly dimished performance, getting me to 12ah total (for a 13.5ah battery).

The resting voltage at that point was around 3.3 v/cell, which is about what's expected for around 90% depleted.
So it looks like the battery is fine, as it should be (with its BMS). This whole ordeal was just because my low voltage cutoff under load was set too high at 3.4v/cell (which might be a decent cutoff for resting voltage, for this battery).

The CA battery gauge is part of what reinforced my misconceptions, because it showed the battery as 95+% depleted when it still had ~45% capacity remaining. I'll have to see if I can calibrate that.

--- regarding motor temperature ---
It was 100 degrees (F) outside, and I went up and down a bunch of steep little hills and regen braked almost exclusively, and generally pushed it hard. At the end the motor casing was uncomfortably hot to the touch. My infrared thermometer said it was in the low 50's C (so ~120's F). From what I've been able to glean from other threads, as an internal temperature that's nothing to worry about, right? Of course that's not the internal temperature, but hopefully it's not too far off...

I really need to replace the [100k thermistor?] it came with with a 10k thermistor so I can get an accurate reading on the CA, and ride with more confidence.

 

[spinningmagnets]

getting me to 12ah total (for a 13.5ah battery)

This isn't well-advertised, but that is normal. It will get you a true 13.5-Ah if you put it on a continuous drain of about 2A, like they do in the lab when it's rated. That's just the way its done for all batteries, they're not unusual by doing that.

 

[neptronix]

If it's 110f and you're riding, and the sun is out, it's not much to create another 10 degrees of heat on the motor at all. Any motor...

If you have the default winding 4T leaf in a 26" wheel, then your battery is very much undersized for what you're asking it to do. It runs best at 30-40 amps continuous or more. Most motors will not perform well when they are being lugged. NCRBs are very weak cells, and you'd want about 60ah of them to comfortably output what the motor needs without torturing the battery.

Also if you have the 4T, it's a motor that does 40mph on 48v. When your battery is sagging extremely, you're not going to hit that 40mph figure, and as your weak pack continues to sag more and more, you're gonna get exponentially worse performance out of it. That's why you need a battery with much higher discharge.

 

[neptronix]

getting me to 12ah total (for a 13.5ah battery)

This isn't well-advertised, but that is normal. It will get you a true 13.5-Ah if you put it on a continuous drain of about 2A, like they do in the lab when it's rated. That's just the way its done for all batteries, they're not unusual by doing that.

It's cutting off early because of all of the voltage sag. The BMS trips early under load..
My rule is to run batteries at no more than 1/4th of their maximum continuous current. Then you can get 95-99% of the capacity stated, because voltage sag will be really low on any battery being used that way.

I run 25ah of 7C lipos at around 2C average ( 50A ) on my 4T leaf and get 98% of the capacity they were rated for and don't take much of a hit in winter performance because there is so much overhead in the battery's ability to discharge. Continuous current is around 41A, peak current is 80.. and the thing screams.

 

[Jabotical]

Ok, thanks neptronix! Yeah when I actually measured even the outside temp, it made me worry less (and I appreciate you confirming)

And in fact it seems straightforward enough to translate the CA temp reading with thermistor lookup tables. At the end of my last ride when the motor case temperature was low 50's C / 120's F, the CA read about 20 C for the internal thermistor:
Looks like that's around 170F / 77C.

In another thread, dogman cites 200F (93C) as an extremely safe motor temperature (which I've also seen elsewhere), with potential problems starting to hit by around 250F (121C).

Others say that 200C (392 F) is about the point where irreparable damage is guaranteed. But at 150 C (302 F) the halls sensors and such will often be toast.
Apparently the CA has the default max temperature at 130 C (266 F).

I expect I'll set the limit at 200 F / 93 C, and try to stay below 180 F (maybe even 150 F) as dogman suggests, in these beginning days of ebiking. Would rather not push my luck on my commuter vehicle