Mysterious degradation in performance - Lyen/Transmag

[ttobiassen]

More details below, but basically: Lots of hard driving for about 60 minutes of a 75 minute endurance race with pretty consistent performance and current peaks of about 55-60 amps, despite very hot controller and motor temps. IR gun registered 150-170F sustained temperature for motor and controller. Last 15 minutes saw a large degradation in performance, but ALSO a significant decrease in current draw, at 35-40 amps. Symptoms persist in testing a couple weeks later, even after everything has cooled down. Peformance now is smooth, not juddery or weird, but feels similar to if we had programmed a lower current limit OR as if you're driving the vehicle at 75% WOT.

Tried troubleshooting the throttle, swapped a different one in, checked it with multimeter: 4.3V source from controller, throttles return between 3.8 and 3.9, which seems about right. Tried increasing the current limit on the XPD software and saw a commensurate increase in peak amps, but not back to normal and still feels like driving at 75% WOT.

Theories? My biggest concern was that over the long period of heat soak that we damaged the motor in some way, possibly partial demagnetization or partial short or windings (ie, like losing copper winding density). The 3 phase wires still ohm out the same as each other, and the same as the backup/identical motor we have. It bears noting that this race has its share of crashes and bumps, so keep that in mind though we didn't have anything too nasty this race.


Equipment setup:

Transmag 532 brushless motor at 36V, geared low
Lyen EB318 controller, got it from forum member Lenk42602, originally built-to-order by Lyen for a 48V Lightning Rod's upgrade GNG mid drive
cheapo thumb throttle

Programming (via XPD):

55A battery current limit
150A phase limit
98% speed limit (to ensure PWM, as per some discussions I've seen on the forum)
NO REGEN
Everything else pretty standard (throttle blowout protect, etc)

Circumstances leading to symptoms:

75 minute endurance race of electric go karts on small track. Lots of stopping and starting. 3 speed transmission means pretty high level of abuse to motor/controller setup. Sustained controller temps of something like 150-170 F and motor temps about the same.

 

[e-beach]

How many peak amps are you drawing? Are they the same as before? You didn't mention a ammeter, do you have one hooked up to your system?

On the other hand, heat does demagnetize magnets. 170° fahrenheit is where some neodymium magnets begin to lose magnesium. You could have killed your motor, but before we go there, you should go through your entire system first.

Starting with the basics,
all are your wires and connectors ok?
Batteries fully charged and in good shape?
controller delivering rated amperage and voltage upon WOT?

What does you Cycle Analyst, or Watts Up, or inexpensive RC watt meter tell you?

Let us know.

 

[ttobiassen]

The peak amps mentioned in my post are displayed via a WattsUp, so even if they're off a little bit because of RSHUNT values, they're self-consistent and have been pretty close in quantity to what the XPD programming says it should be, at least until BEFORE the change occurred. NOW we see only 35-40 amp peak under identical programming where we saw 55-60 before.

I will contact Roller (the forum member who sold me the Transmag motor) to enquire as to what magnet type they use, but they seem to build things robust and cheap for their applications (lawnmowers and snowblowers), and if it's a ferrite magnet setup then I thought the demagnetize temp was higher than for neodynium.

I have 5 separate sets of 36v batteries, so it's definitely not the batteries, PLUS the WattsUp is showing a Voltage Sag that is not that bad relative to the current we're drawing, ie we've had much worse in the past even whiless things were behaving correctly.

Still pretty stumped as to what's going on... I guess I should contact Lyen to see what his guess is.

 

[John in CR]

Check the no-load rpm. If the magnets weakened from getting too hot that will increase the no load speed.

It doesn't sound like the controller. They either work or don't work. Did maybe something short a 3 speed switch wire? 3:1 phase to battery current is WAY too high, especially during torturous use. High ratio's can only be tolerated by systems in low stress use...flat land riding with only occasional hard accelerations.

98% speed setting is BS. What Zombiess found it testing while looking at an O'scope is that those controllers never get to 100% duty. For a while he drew the wrong conclusion, ie that 100% speed shouldn't be used, but he backtracked from that. Whoever is perpetuating that myth is wrong.

Heat is our primary enemy, and long periods of hard use (constant accelerations or long climbs at low speed are the two hardest) require special considerations for heat. The relatively modest temps you mentioned sound to me like you're getting good air flow, so the insides of the units are much hotter.
The route to durability in hard use is over-sizing things, and that's the route to better efficiency and cooler temps too. A hot motor makes about 30% more heat than it does at the start of the race. because copper resistance goes up with temperature.

The simple fact that it needs a multi-speed gear box means your motor isn't up to the task. Didn't every shift upward feel like crap due to the torque reduction? Take the weight and cost of the gearbox and linkage and invest it in more motor and controller instead for far greater efficiency and performance, and that will get maintenance down to chain/belt issues.

What is the goal? An electric system should be designed for it's use. It's not like a gasser engine made of metal where the parts can withstand extreme temperature and can tolerate running at their performance limits for an hour. The harder the use, the more you have to de-tune an electric from its performance limits. How much distance did it go in the 75 minutes? You say lots of starting and stopping, but was it really full stops? Assuming not, what was the general minimum speed through the turns and max speed on the longest straight? How many watt hours were used in the 75min? What's the total load pushed around the track, cart and driver?

You said "endurance race", but was this just an experiment or a competition? If a competition, what are the rules?

 

[dogman dan]

I've cooked off two external shunts for my CA over the years, both in use at higher amps, 40 amps controllers. You might just have a shunt in that watts up that's now a nice resistor.

Could be many other things too, but take a spin without the watts up, and see if it feels stronger. Also, have a good hard look at your battery to controller connectors. Maybe it's not just that simple, but it's possible it is.

 

[ttobiassen]

Hey guys. I'll try to fill in the rest of the details for you.

The vehicle in question is an entrant into the Power Racing Series, or PRS, in which you take the body of a children's electric ride on toy like a PowerWheels and use it as a template to build an electric go kart. You must stick to the size of the body, or if you build your own body, it must look like you are driving a children's toy when you're done.

Because this is a racing series, there are lots of rules. Not too many, I think, but many. The main ones to know about are HEAVY BUDGET RESTRICTIONS (which is why we run lead acid batteries), VOLTAGE RESTRICTION (36V nominal limit, because the organizers read some OSHA document that terrified them of DC over 50V for safety reasons. They'd soil their pants on Endless Sphere), and POWER LIMIT (ie, competitors must use a provided, slow blow fuse with a rating commensurate with an approximately 2 HP output). So, this is why we are stuck with running 36V when the transmag would love 48-72, this is why we run big lead batteries, and this is why we run a transmission (Because with a momentary peak of perhaps 4 hp and 2hp continuous, the gearing does seem to help. Also helps you be flexible with the different track conditions. Plus it's fun! We made it into a stick shifter.).

The race conditions for the endurance race are 75 minutes of hard racing, with constant shifting (ie, keeping the motor in the torque range, which is hard on it), at speeds of 5mph in slow corners to 20 mph on straights. Battery changes are every 15 minutes, so we pretty much always have a pretty fresh pack. Our WattsUp is a 100A rated cheapo unit but has been trustworthy so far, showing that we are running 35-40 amps AVERAGE over a 15 minute period. We don't usually write down the W-H number, but thinking about our AH's used (not counting the Peukert effect, damn SLA's)... we use probably 10-15AH in a 15 minute period, which would be something like 300-500 W-H per 15 minutes, or 2000 W-H over the course of a race.

Attached you'll find pics of our drivetrain AND the overall kart (with drift wheels, though we have only run those one time).

Dogman: Bypassed the Wattsup, and got the same no load speed as before.

John in CR: DING DING DING! I think you got this one. Tested the no load RPM (had to disconnect the rest of the drivetrain), and I'm getting a no load RPM of 4900 at 38V, or a kV of about 130rpm/volt. Now, I didn't write down the kV we got from this motor to begin with (and I remember it being a BIT higher than the rating), but I definitely remember it being MUCH closer to the manufacturer's spec of about 72rpm/volt, so I think our magnets are fried... Lame.

So, I never expected to toast magnets without frying windings, particularly because the motor has such good cooling. I have an extra motor lying around, but what are my options:

1. Swap in new motor (the obvious option). Though I was hoping to save it for a different project.

2. Run the same motor but raise the current level of my programming. Right now, if I tell it to do 55A battery limit, I see 35-40A peak. If I tell it to do 70A, I see 45-50 amp peaks. Can I just tweak it to get the performance in line with expectations? Or will it grenade my controller or motor? Probably. I may try this.

3. Replace the magnets. As I recall from this thread (https://endless-sphere.com/forums/viewtopic.php?f=30&t=50066), the magnets are secured with hardware, and so they might be easy to swap IF I CAN GET THE OEM MAGNETS. If not, I may be able to use other magnets, but it's kind of a weird shape. Thoughts? I'll PM Roller (the guy who works at TransMag) to see if OEM magnets are a thing I can get easily.

Thanks for your help guys and I'll continue to update this thread as I find out more. Please lend me your thoughts and ideas if you have any.

 

[ttobiassen]

Bonus round 1: found this great reference for BH/Demagnetization curves. Very believable that demagnetization is our problem. https://www.kjmagnetics.com/blog.asp?p=temperature-and-neodymium-magnets

Bonus round 2: Our Facebook page has lots of pictures of our silly little go kart, in both Batmobile and School Bus 'costume'. FYI. https://www.facebook.com/thearustocrats

 

[e-beach]

If it is your magnets, and new ones from the manufacture are not available, or for some reason are not an option, there is an outside chance that you can re-magnetize them by applying 2 to 2.5 their original Gauss. That can be done by using several other neo magnets and attach them to your old magnets. If your magnets are flat on the back side, shouldn't be too hard unless they are also epoxied in place. Just be sure not to apply mechanical shock to the neo's as neodymium is brittle and can crack. Also, don't get your fingers in between the magnets so you don't get pinched.
n50 neodymium is not prohibitively expensive. A couple of those might do the re-magnetizing trick.

With my big n50 neo's I keep a separator between them (hard foam) for ease of separation when I need to separate them.

 

[e-beach]

Also, bit on what happens when a magnet is broken.....

snip.......If a bar magnet is cut in half, it is not the case that one half has the north pole and the other half has the south pole. Instead, each piece has its own north and south poles. .......snip
https://en.wikipedia.org/wiki/Magnetic_monopole

 

[John in CR]

With budget being a limitation you can still do lithium and save weight, but you need to source salvage packs. Recycling cordless power tool battery packs, getting some battery modules from a crashed Nissan Leaf, Chevy Volt, etc, or even a buttload of laptop batteries are where we get them. If you have to add up all the lead batteries used in the race against your budget, then lithium could prove cheaper. Are the battery stops mandatory for everyone? If not then you could do just one stop or no stops and pick up some time.

Your motor is an inrunner, so with the magnets in the center it's hard to cool them. Start with figuring out how much heat you're generating, and at the same time it gives you an idea of current limits (the one killing you is phase current, because that's what the motor sees). Measure the phase-to-phase resistance of the motor. Copper heat losses are current squared X phase resistance, and factor in the increase in resistance with temperate (say a 30% increase from your room temperature measurement). You can calculate core heat losses, which vary with rpm, by measuring the no load current, and at max rpm that's the heat from that source.

In addition to lowering the phase current limit to reduce heat, look at using a reduction in the gearing to reduce the load, which gains back torque loss of the reduction in phase current. Do some acceleration times in high gear as you reduce the phase current limits, because at the 3:1 phase/battery current limit, it wouldn't surprise me if you're pushing the stator of that little motor into saturation. If that's happening you're pumping juice with limited torque benefit.

Then look at heat dissipation. Get cool fresh air flowing to the motor, not hot air coming from down near the pavement. You say the motor has good cooling, but I don't see that it's opened to draw fresh air through the motor, so it's easy to improve. Openings in both end caps with a centrifugal blade at one end to draw fresh air through the motor will make a big difference.

Fresh cool air will help the controller too, and assuming it's closed up under the body with little air flow that can be aided with computer fan(s) blowing on the case where the screws are that hold the mosfet heat spreader bar to the case. I've also had good results drawing fresh air through the controller with air dams to force the flow at the mosfet rail, where most of the heat is generated in the controller. I use 40mm centrifugal fans for better results than with common radial fans that work poorly with flow restrictions.

I hate to be negative, but I don't see that bicycle geared hub holding up for long. Start planning for a replacement transmission. The voltage limit leaves you under-powered with that motor, so a tranny makes some sense, but 2 speeds is all you need. Look at Thud's dog clutch gearbox as an option, or a simple Retro-Direct 2 speed where changing motor direction gives you the 2 gears. Since you have no need to roll backward R-D is an elegantly simple solution using bicycle freewheels and 2 chains to make it work.

Too bad about the 36V limitation, because otherwise I could turn you on to better efficiency and easier cooling at that power level using a higher voltage in the same price range. At 36V and 4hp peak I'd have to consider going with an 80-100 size RC outrunner, though the Lyen controller may be an issue. A revolt 100 with hall sensors and wound to a Kv of 49 might be just the ticket, and have high enough inductance for the Infineon to handle it.

 

[roller]

Looks like you guys are on-target with over-heated magnets.

The definitive test (particularly since you have 2 "identical" motors) is NLS (no-load speed) test. Same supply/controller, nothing (significant) attached to shaft .. . record full-throttle speed /volts / amps. At NLS, each should be on the order of 70-74 rpm/V. If the suspect motor is significantly faster, it is most likely magnets overheat.

(Beware running a motor w/ OH mags -- the magnets do NOT necessarily demag "evenly", with just faded fields -- they can have generated polarity bands/reversals at varying widths, etc -- this can give fits to a hall sensor controller.)

Also - visual inspection - with rear endcap removed -- you would likely see(and/or smell!) brown/burnt coil insulation, if that were in any way culprit.

At (only) 36V, you are not able to gain full effectiveness of the shaft-fan cooling, . . . You get the highest heating at lowest shaft speeds, and may reach max of only (an unsustained) 2500 r/m. I agree with John in CR regarding additional fans (muffin or similar) which can run at full-flow all the time (or when switched on).

Housing temp is not a good indicator of coil/magnet/air temp. I assume mag temps are around internal air temp, so if you can keep air temp down (increase flow), you're doing well.

These mags are most likely 155 C rated (300 F !), so if you got there, you're pulling hard on this motor. (fyi, this coil insulation holds until about 180 C)

At 150 phase amps, you're probably "tickling" saturation. You might not notice a torque/power reduction by reducing phase current to ~130 +/-

I generally like the "idea" of transmissions, sometimes more than their reality (they seem to me to meet their efficiency specs about as well as RC outrunners) I can see how it's necessary to get your vehicle speeds down . . . If you're able to do NLS testing, it would be instructive to also see the V-A-r/m, and Wheel rpm (propped off ground). How much load is this transmission putting on by itself?

PM en route -- let's see if we can get you going again.

 

[dogman dan]

Well, not surprised it was not the shunt. But so easy to do that test, worth a look at it. Was just hoping it could be just that easy.

Also not surprising you lost your magnets without losing the winding. I've done that too, on a brushed hub motor. Glue melted, and magnets detached. After gluing them back, it ran nice and hot, and no power. Next to go was the brush holder. That hub still has seemingly good stick. I use it to position stuff for welding. But it's done as a motor.

another brushed hub, I melted the brush holders and demagnetized the magnets in the same go. This time, the magnets got so weak they would not stick to the fridge.

That cured me of brushed bike motors, and I went on to toast brushless motors. Usually with them it's just the halls that go, unless you push it so hard that flames shoot out the cooling holes you in the cover. I had one motor blow the halls, and have all the solder inside melt. But connected back up, ran fine. Black winding, but still running. Motor still in use, but I gave it away, not sold it.

 

[ttobiassen]

General follow-up:

I just hooked up the spare 532 I had, and it fires up with a kV of 72rpm/volt, just as expected. So, given my competition is less than a week away, that's the good news.

The WEIRD news is that the magnets I received in the mail are just barely the wrong size. The old magnets are 4mm thick and fit into a 20mm slot/keyway, and the new magnets are 5mm thick and would fit into a 22mm slot/keyway. This is why I hooked up the spare instead of fixing the toasted-magnets-motor; I'd have to mill/4thAx the rotor to accomodate the other magnets. I can do that, but I'm trying to avoid it Any guesses as to the magnet mismatch problem, Roller?

As an aside, the spare motor not only has a 5mm keyway (the other motor was smooth) but ALSO has the rotor venting I saw in some pictures, both are features that the toasted-magnet-motor didnt have. Odd.

On the topic of transmission drag: The no-load current of the fresh motor is 2.5A, and with the drivetrain attached (2 chain reductions with a tranny in the middle, lots of 'flywheel' weight in the form of the live axle and tires) I get only 3.5A. So not much drag in the end, really.

I'll follow up after competition with results! Should be a great race. The final thing I'm considering doing before the competition is venting the back motor cover for more airflow.

 

[e-beach]

Any thoughts on using the new magnets to try to re-magnetize the old magnets? As I understand it, neodymium only magnetizes one direction so check the poles before you try it.

 

[ttobiassen]

Looks like Roller is going to help me out with the magnet issue, so that's cool. Meanwhile, though...

The replacement motor is in, the kV check results in 70-74 rpm/volt like it should, but... We're still slow. With the old, nearly identical motor, when it was new it was REALLY impressive. It would have taken much more (instantaneous) power than we were throwing at it. And I don't buy that the new motor would have a substantial power decrease, so my problem is elsewhere. Symptoms and theories below.

Symptoms:

Test1: Taking JohnCR's advice, we started conservatively and with a 2X phase/battery ratio of 100ampPHASE/45ampBATT current limit with a 0 second hard current limit scheme. Result was a slow ass go kart pulling only about 30-32 amps max (sounds a lot like 1/3 the phase limit, doesn't it?). Confused.

Test 2: Bump it up a big. 130/45, ie, going for the same battery limit for a phase/batt ratio closer to what we were running before. Result was a moderately faster but still slow go kart with maybe a 35 or 40 amp limit.

Test 3: GIVE IT THE JUICE. To diagnose WTF is going on, we bumped it WAY up to 150/65, and the result still only pulls about 45amps max (about 1/3 the phase limit).

The weird thing through all this is that the drag on the system is really high. We used a long alley where we could get pretty much up to the max speed of one of the gears on the go kart and the current draw was just BARELY going down to 20amps at the end of the run. That's about 750W just to cruise along at like 17mph.

My primary R&D drivetrain was on another project that used a 2808 9C hubmotor with a 12FET Lyen controller at 16S LiFePO4 (50V) set to a 30A limit, with roughly a 2.25:1 phase/batt ratio. That system would get you going up to about 25 mph, at which point the current draw would drop down to 8-12 amps while coasting. That's 400-500W, but at a significantly higher aerodynamic drag and speed.

The free-air-testing of the drivetrain showed very little drag on the system (2.5 amp no load of motor alone vs 3.5 amp no load of connected drivetrain spinning in air). The only difference MECHANICALLY *could* be one of two things. 1) Our pillow blocks are introducing much more drag while loaded because of off-radial loading of the bearings or 2) Our 3spd shifter is failing in a way that produces drag ONLY UNDER LOAD and not during the no load test.

The weirdest thing is, though, that if we were having lots of mechanical drag, why wouldn't the system just max out the amp limit each time? With the old motor, I had the controller at like 65 or 85 amp battery limit, and we would occassionally hit close to that too. If the system is ELECTRICALLY the same but with more mechanical drag, why wouldn't it just RAM into our current limit and stay there, even if the go kart now FELT slow because of the effective loss of power to the ground?

At least we're going to have a functioning go kart for our competition this weekend, but we're going to be the laughing stock (not really, people are chill in this racing series) because of how much slower we are than usual :/

 

[ttobiassen]

Two quick notes:

1) To test the *potential though unlikely* idea that the 3spd is introducing mechanical drag UNDER LOAD but not at no load, we do have a drop in replacement jackshaft (simple, just a bicycle axle with the correct sprockets on it) that would prove whether or not the IGH is to blame. We can try that.

2) I do not specifically recall whether we were using a 0, 1 or 3 sec current overrun parameter last year, which *could* explain why were able to hit peaks near our BATTERY LIMIT but that the load would quickly drop down to PHASE/3 amps continuous. Does that make sense?

 

[ttobiassen]

Well, the go kart is back to being fast! I just had to set all the "wait, aren't we supposed to leave that alone?" controller parameters to "t-minus 5 minutes to meltdown" for it to work. Not sure what's going on....

The "Speed at full throttle" setting, which historically I've been warned not to use and experimentation has tended to prove that warning out on hubmotors I've tried it with, can be set up to 120% speed at full throttle. My impression of this setting was that it is forward timing the motor. I accidentally used an old troubleshooting parameter set that had the 120% speed programmed in, and things feel better! Perhaps this is somehow correcting the symptom we were having of the system always feeling like it was at 80% WOT.

The other symptom we were having was that programming the controller to, say, a 50A current limit would result in a 35A limit in practice. So, now we're running a 75A battery limit with a 125A phase limit (the controller is only rated to 45A and the fuse we run for the PRS series is 40A), and we get a 45-50A peak. So, again, I'm having to trick our controller into doing what we want, but it seems to be working.

I'm thinking the RShunt in our controller somehow got cooked, OR our 5V logic for the throttle is toast, OR BOTH.

Crossing my fingers on this one. We're fast enough that tire wear will be a big concern even if we don't cook our drivetrain. I give it 50/50 chance we burn out our controller or motor this weekend, but it IS the world championship, so GO BIG OR GO HOME!

 

[Lenk42602]

Good Luck Thorin!

Be sure to send me all your video footage again!

Len

 

[e-beach]

Good luck for sure!!!

On a side note, I did once beef-up my wiring and connectors and had a noticeable increase of power to the motor.

What I am saying is check out any possible "choke-point" in the system and if you can, beef it up.

 

[ttobiassen]

Our go kart did great! We had some SERIOUS sprocket wear and tire wear problems, but none of those is a big surprise for us. The TRANSMAG performed great, as did the Lyen controller, at battery peak currents of about 60A and a continuous draw of 40 A for the whole weekend. I also checked the kV of the motor after we got back home and it's still exactly the same as it was when new, so we didn't demagnetize this time. I'd like to machine a new back "bell" for the motor so that I can still support the bearing but add a large 120/130mm fan so that the motor can be cooled even at 0 rpm. Right now, because of the use of a bicycle component as a jackshaft, as soon as you let off the throttle the motor slams to a stop, killing the internal cooling.

Everyone is always telling me the Nexus 3spd IGH is going to die eventually, but so far I'm impressed. We've now used it for two seasons of races, or 1000s of laps of aggressive racing with several shifts per lap, and there's no sign of it changing its behavior. Other people in the Power Racing Series are using a Nexus 8spd with success as well.

They hand out 3 trophies each season: Moxie Cup (most 'moxie' points for audience appreciation of awesomeness/silliness), the Chapman Cup (for most race points) and the Tesla Cup (most overall points). We won the Moxie Cup last year, and the Chapman Cup this year, so now we just have to win the Tesla Cup next year!

Thanks for all the help and ideas, everyone. Endless-Sphere continues to be an awesome community, and one I'm thankful for and proud to be a part of.