Re-wind of a turnigy 80/100 (Now-tutorial w/Video)

[liveforphysics]

As I mentioned previously, both of the Prius BLDC motors run immersed in light oil and this doesn't seem to increase losses to an appreciable degree, or else I'm sure Toyota would have come up with a different way of doing it, as they were intent on making the system as efficient as possible.

I think an experiment is called for here, because we're speculating about viscous and swirl losses that may not be as great as we think they are. I have a spare 5330 motor that's already fitted with long leads. I'll have a go at running it no-load in a bucket of light oil tomorrow and see what happens. IIRC I have some 10W-30 synthetic oil around somewhere, probably a bit too thick for the job, but at least it'll give us a data point. The only snag with this test is that the motor is a low Kv wind (around 88) so I won't be able to wind it up to a massive rpm.

Jeremy

Right on Jeremy! I predict you will want to wear clothes you don't mind getting a little oil-speckled.

If you could measure the no-load power at say 3-4 different RPM points, we could get a pretty good idea of the nature of the drag function.

You mention the Prius tranny/motor having oil in it. On the dyno with a normal automotive transmission, we see around a 10% power loss in a FWD car's tranny, and you can distinctly see the linear component of the loss function (gear surface friction) taxing you roughly a flat-rate tax on your torque, and a second component which increases at the square of RPM, which is the fluid drag component from having the bottom side of the gears dragging through the oil in the sump (windage loss). It starts out with only gear-face friction being noticed, and then the fluid drag overcomes it at higher RPMs.

This is why some modern race transmissions and rear-ends actually dry-sump the oiling now, so the fluid level sits below the bottom of the gears, and a little pump sprays oil to the critical locations, and just avoiding this fluid drag is worth some time in the quarter mile. It's become the standard for NHRA prostock now, and cars that don't do it generally can't be competitive (due to everything else in the class being almost identical, and first and last place being separated by about 0.05s.)

 

[Thud]

Luke,
as always impressive drag calculations...but say there are no edges in turbulence, magents perfectly potted in epoxy, a new end cap on magnet ring w/no egdes. just a nice orifice to flow some fluid through?

something like this:


Jeremy,
I have to believe the stock open-end bells are going to create some drag but that can easily be handled with custom parts...(I know a guy) easy to make the magnets concentric with a roll of paper & some epoxy...

Now Luke,
Do the calculations on the continuous power ratings of my 6-turn double 80/100 now 133-Kv in wye...with active cooling...tell me what you need. I am really curious because I would love to use a pair of these for a 3-season commuter, built for two.

 

[Jeremy Harris]

Here's a picture of the current Prius transaxle with its two BLDC motors:


The whole thing is immersed in light oil. There's no gearbox as such, as the motor on the right (MG2) is directly connected to the front wheels via a gear reduction train, so is always directly driving (or regeneratively braking) the car. The motor on the left (MG1) is the control motor generator. It spins at whatever speed and direction is required to match the ICE input shaft speed to the speed of MG2, even if the ICE is stationary (so no need for any clutch). In effect it forms a constantly variable transmission with the ratio variation provided by electrical power circulation around MG2, MG1 and the battery pack, via the inverter module. Mechanically it's very simple, really just a differential with the front wheels and MG2 connected to one shaft, the ICE connected to another and the control motor generator, MG1, on the other shaft. Simple and clever, the sort of stuff the Japanese excel at.

If the losses are as high as 10kW I'll be very surprised. It's handling a total power level of around 134hp, with peak electric motor power of maybe 50kW and around 258lb-ft of total torque, but at relatively low input rpm, a bit over 5000rpm max. Most independent reviewers seem to think that the total transmission losses are around 3 to 5%, including the electrical loss in the inverter CVT function. This would suggest that the viscous, swirl and frictional losses could be as high as maybe 2 or 3kW, as we already know that the inverter cooling system can handle around 2 to 3kW. I suspect most of this loss will be frictional, in the gears of the transmission, rather than viscous and oil swirl loss.

Jeremy

 

[Ludo91]

Thankyou jeremy your test will for sure give us a solid point to continue the discussion from.
If we cover the magnets with someting like "dutch tape" to make the inner surface of the rotor smooth, putting the whole thing in oil makes even more sense

 

[liveforphysics]

Luke,
as allways impresive drag calculations.....but say there are no edges in turbulance, magents perfectly potted in epoxy, new end cap on magnet ring w/no egdes. just a nice oriface to flow some fuild through?

Yes, proper surface prep can certainly help reduce the rotor drag by a large amount, in my experiences with windage loss from oil on gears, we don't have the luxury of removing the features from the gear face to reduce drag, but with a rotor you can. Perhaps to a perfectly reasonable amount, it's a boat load of variables to try to estimate with an degree of accuracy. Jeremy's test will be worth more than anything I could do with a pencil.

Now Luke,
Do the calculations on the continuous power ratings of my 6turn double 80/100 now 133kv in wye....with active cooling.......tell me what you need. I am really curiouse because I would love to use a pr of these for a 3-season comuter build for 2.

To know continuous power, you simply need to know the inefficiency, and the max rate you're able to take away heat, and make them balance. With liquid cooling, the fastest way to know the rate you can take away heat is to simply measure the flow rate and the delta-T on the fluid. This is going to be best gathered experimentally. Something like putting the stator in the oven set to 350f (or whatever), having all the coolant in a single tank with the pump you plan to use setup, and all at a known temperature. Take the stator out of the oven, measure the temp, hook up your lines, use the same pump you plan to use to push that known volume of fluid through the stator into a plastic bucket on the ground. Measure the temperature of the coolant and the stator right afterwards. For the sake of an easy estimation, just average the before and after stator temps together and call that the stator temp, then measure the increase in temperature of the fluid, and the volume of the fluid you pushed through it, and the rate of the flow through it, and the type of fluid used, and I can tell you the rate your fluid cooling can extract heat from the stator.

To come up with the inefficiency value, you're going to need either sweet-talk Biff (who is a super motor expert) into modeling it for you, or the best option, which would be to gather that information experimentally with a dyno.

 

[phyllis]

Luke,
as allways impresive drag calculations.....but say there are no edges in turbulance, magents perfectly potted in epoxy, new end cap on magnet ring w/no egdes. just a nice oriface to flow some fuild through?

something like this:

Nice

Et al: I think water is much lower viscosity than any lubricating oil, and lube we don't need?
So, water will have less drag.
I like the idea of a noncondensing boiler as Luke have said. I had been thinking about a large diameter hub drive with only a squirt of water in the bottom, and an overpressure valve. The water will continuously spread the heat to the inner walls, and in case of massive overpower, nothing will get significantly above 100C.
I did a little simple calc of heat loss. Unless I'm mistaken, the majority of heat loss at high power is radiation, unless there is lots of forced air. so, buff or paint your motors black for higher emissivity.

 

[texaspyro]

Ethylene glycol has twice the thermal conductivity of most oils. Water is 4-5 times better. For gas cooling helium and hydrogen are your friends

 

[phyllis]

right! and helium also makes the system lighter!!
And has less friction than air. It is also funny, but hard to contain.

 

[texaspyro]

The viscosity of fluids can change immensely with temperature. Between 0 and 100C water viscosity decreases around 6 times.

 

[liveforphysics]

The viscosity of fluids can change immensely with temperature. Between 0 and 100C water viscosity decreases around 6 times.

Yes, this is true.
Water is really an excellent cooling fluid.
Water's low boiling point is the only weakness for a motor coolant, even if we make a pressurized system, the pressures get pretty nutty at the temps our motors are perfectly happy operating at. Some boiling point increasing additives can help this quite a bit, perhaps enough.

 

[etard]

Water, huh? Maybe you could just set up a mister that runs off a small pump motor using the same signal as your throttle so that it sprays more water when you are gunning the throttle and just turn off below half throttle? Might work for the death race at least. Instead of 2 stroke stink, you will be leaving a trail of steam!


Hate to change to a different discussion, so you can PM me if this is off-topic (not really sure what the topic is at this point). I need to rewind a Hacker motor and it is smaller than this Turnigy, it's the Hacker A60-16 L here: http://www.espritmodel.com/index.asp?PageAction=VIEWPROD&ProdID=4419

It's a 60mm by 80mm long, can anyone give me a recommendation on wire size and how much to keep it near or lower than the current KV of 168? I was thinking about taking all the wire off and then weighing the resulting nest of wire and then just ordering double that weight in 18 awg. I can't get my head around the formulas to figure out what I need. Thanks for any help guys.

This is the wire I am thinking of ordering:
http://cgi.ebay.com/Magnet-Wire-18-Gauge-AWG-400-Feet-Enameled-Copper-200C-/170602555136?pt=LH_DefaultDomain_0&hash=item27b8b46700#ht_683wt_1139

 

[olaf-lampe]

What about circulating fluid in the inner part of the stator? that`s were the shaft is... probably not enough exchange surface, right?

What about using the inner tube as return path for the heated fluid? You need to find a way pass the bearings, though.
-Olaf

 

[Jeremy Harris]

Right on Jeremy! I predict you will want to wear clothes you don't mind getting a little oil-speckled.

If you could measure the no-load power at say 3-4 different RPM points, we could get a pretty good idea of the nature of the drag function.

I've just done a quick-and-dirty check, by dangling a motor by its leads in a jug of oil. The test was a bit harsh, as the oil was fairly thick (10W-30 synthetic motor oil) and also a bit cold (around 7 deg C). The no load power at 20% of maximum rpm, in free air, was 16 watts. When immersed in oil this rose to 24 watts and the oil didn't seem to get swirled around a lot. At max rpm the no load power in air was around 36 watts but in oil it shot up to around 180 watts. At max rpm (around 3500) the oil was getting very aerated from the holes in the end bell swirling it around. There was a pronounced critical rpm where the oil suddenly started to get dragged around a lot, after this is was hard to see what was going on because of the aeration. The jug was around 2/3rds full with oil, as full as I could get it without it getting flung out by the motor. The oil depth was around 3 times the diameter of the motor, at a guess, so I had it well immersed.

This motor has a smooth inner bore to the stator, as it's one I've filled with epoxy and machined back, for marine use. It's also been rewound with 18g wire. I think the motor may have behaved very differently without the holes in the end bell stirring the oil up, as one problem was clearly aeration and oil swirl from the end of the motor. Even after just a few seconds running at full revs the oil was like whipped cream.

No photo's, I'm afraid, as I didn't fancy getting the camera covered in oil.

I think this experiment proves that oil drag is going to be a significant problem to overcome. I'll admit to be surprised at just how much power the oil absorbed. An interesting observation was that the motor didn't have an appreciable torque reaction, even though it was absorbing mountains of power. I was just dangling it on its leads and it was sitting there soaking up nearly 200 watts. This leads me to believe that viscous shear between the rotor and stator might account for the majority of the extra power absorbed. It may be that changing to a much lower viscosity oil will help, as would warming it up a bit. I may repeat the experiment in an hour or so when the oil's a bit warmer (I have the jug sitting on a radiator to warm up at the moment).

Another thought is the effect that oil may have on the magnetic circuit. The magnetic permeability of oil may well be significantly different to that of air, although I've no idea what effect this might have.

Jeremy

 

[Thud]

Hey Jeremy,
Any thoughts on trying it with just water? (I have run mine submerged to wn a bet) also thoughts about taping up the end bell of the can to reduce the blunt edges in the fluid stream...just thinking out loud.

I have a couple of motors I could easily modify with bondo to repeat your tests...

I am really interested in seeing the drag in a motor optimised to run submerged...given the air gap I wouldn't have guessed the drag to be so dramatic in that area...I was looking at the blunt edges of the end bells acting like a loaded prop & expect that effect to be dramatic.

 

[Jeremy Harris]

To be honest I don't fancy trying it with water, as it'd mean stripping the motor down and drying out the bearings afterwards, plus I'm not sure how well the wires to the Halls are sealed - my guess is that water might well creep in there and start corroding things.

As I mentioned above, the really odd thing was the lack of torque reaction. If the drag was coming from the holes in the end bell, then I'd have expected the motor to be writhing around like a snake on the end of the wires. It just stayed pretty much where it was, whilst sucking all that extra power. The air gap on this motor is pretty small, just under 1mm, which may be another reason for the increased viscous drag. I'm pretty sure it has to be this causing the big power increase, as I can't see how else a motor could draw that much power without any appreciable torque reaction.

Filling the end holes would probably help the aeration problem a lot, as they really seemed to stir things up a lot. I don't think you really need to circulate oil right through the motor for effective cooling, it's probably enough to just provide another heat path out through the can. Oil is around 6 times more thermally conductive than air, even though it's still around 3.5 times less thermally conductive than water. It looks like the thermal conductivity of oil is fairly independent from its viscosity, so if we could find a very light, low viscosity oil, the losses would probably come down a fair bit. Ironically, what we really want is an oil that's not much good as an oil, i.e., one with a thin film thickness. My guess is that if we can find one that will happily withstand counter rotating in the air gap with minimal viscous shear then the immersed motor idea might work well. Coupled with a chain reduction drive in the same housing it could make for a neat drive unit.

I've been doing some digging around and it seems that the Prius transaxle oil is special, but that probably has more to do with electrical conductivity issues and compatibility with the motor insulation than anything else. Independent tests show that it typically has a viscosity of around 25cSt at 40 deg C, dropping to around 5cSt at 100 deg C, so is more viscous than I expected it to be. It makes me wonder how big the viscous shear losses are in the Prius motors. My guess is that they must be quite high, unless there's some fiendishly clever Japanese trick in there to reduce them.

Jeremy

 

[John in CR]

Isn't the thermal conductivity just a small portion of the tale, since especially in the case of air we're looking at a turbulent flow?

Oil has about double the heat capacity of air and about half that of water. I'd rather blow 2kg of air through the motor than 1kg of oil, especially when you have to exchange the heat from the oil to the outside air. You'll need the blower anyway, since you won't get sufficient natural air flow through your heat exchanger. My vote would be to bring the outside environment through the motor for direct heat exchange.

Let's call the motor 10kw and the oil pumping losses only 3% or 300W, which is likely to be optimistic. 300W can blow tremendous amount of air. eg a little 20W server blower can push 32cfm. That alone is almost 1kg of air per minute. Use a pair of those to blow full time to keep the motor nice and fresh during normal duty, and then kick on a 300W ducted fan for extreme duty.

Ask Luke what kind of air those ducted fans move. Not only would the open air system be easier, but on demand instead of full time losses means you can step that blower way up in power and still be more efficient overall, since it would be used very little. Don't forget some of those losses with the liquid pumping rig directly reduce motor power, while the blower route can come 100% from the battery and take 0 from the motor. Wasn't the whole idea to maximize power.

What about the weight? By the time you add all the liquid cooling components and the liquid itself, weight and space-wise you're better off with a 2nd motor, or at least a bigger one.

I can see wanting to liquid cool a sealed motor, but not one set up so nicely for air cooling, and Thud's space efficient winding makes it even better for pushing crazy amounts of fresh air through his motor. The more favorable delta T between the stator and ambient air compared to the stator and warm oil may even give air cooling the edge in maximum effectiveness in this particular case.

 

[Ludo91]

nice idea would be to anlarge the free space between the mounting palte and the can to allow more space for air to come out of there and put and rc dutched fan (as someone mentioned before) somehow seealed to the other end of the motor to have A LOT of air pushed thru the motor when it`s need

 

[Thud]

John said:

By the time you add all the liquid cooling components and the liquid itself, weight and space-wise you're better off with a 2nd motor, or at least a bigger one.

Thats it in a nutshell.
But this thread is all about polishing a turd...I have 6 of these motors in one size or another now & after tearing them all down...there is something wrong with each one of them. Don't get me wrong, they are a great buy, but for an e-vehicle application there is much attention to detail missing from QC at the factory. (not that I would pay more for a better one...these are "Just" toy airplane motors being re-purposed after all)

If we can add 80% + to the continuous rating of the motor without doubling the weight its a win (at the very least a push)...I was really fishing for a rough estimate of the potential power boost...if I could push 7-10kw continuous with this tiny package that would be a goal. A bigger motor is a better answer...if one can be built light enough for the application. I gues its really a matter of getting the absolute most out of 1 set up & complexity be damned.

I don't share the view that outrunners are "ideal air cooled" motors.(at least when not in an airframe) The limited internal space & having that magnet can surounding & reflecting heat right back to its source is a burdon on these, not an asset. The skirt bearing is a pretty effective seal at 1 end of the can.

The external stator of the Astro's have far more surface area exposed to cooling air, uninhibited.
& a small cetrugiful pressure blower to blow enough air through the motor would require a subtantial amount of power also & be nearly as complex.

 

[Miles]

You also have to take into account the fact that, for the same high torque level, a larger motor will be more efficient - that might save you some battery weight.....

 

[Thud]

Mostly moot points for sure, Miles.
Its all about high-torque thin-profile Axial-flux motor builds after this death race nonsense
Always better to run a motor in its efficient range than create subsystems to end run (maybe-non existing) problems. Especialy since we don't have paralell systems to tap into like the Automobile applications have (active cooling or a reservoir of oil to access).

But a closed system with just thermal siphon transfer & a small finned reservoir for expansion. Any idea what the added thermal mass would do to the "time before critical temprature" is achieved?

looks like I have a lot of thermal dynamics to study in my free time when it arrives

 

[liveforphysics]

I don't share the view that outrunners are "ideal air cooled" motors.(at least when not in an airframe) The limited internal space & having that magnet can surounding & reflecting heat right back to its source is a burdon on these, not an asset. The skirt bearing is a pretty effective seal at 1 end of the can.

The external stator of the Astro's have far more surface area exposed to cooling air, uninhibited.
& a small cetrugiful pressure blower to blow enough air through the motor would require a subtantial amount of power also & be nearly as complex.

Not at all my friend. The magnet is the device that needs to be sheltered and protected from high temperature more than the windings. As a system, it is the temperature limiter for the motor. Putting the magnets inside means needing high-temp magnets like an Astro uses, which cost it a lot of potential torque.

Also, for a given diameter motor, the outrunner has the highest surface speeds between rotor and stator for a given RPM, which also enables it to exchange temperature very well between the stator and the outside air when properly vented.

I remember reading some tests done with various motors on RC groups, and the running outrunner motor was able to shed heat something like 4-5x faster than the running inrunner motor of similar dimensions. This was a fairly simple test, something like heat both motors of the same weight to the same temp in an oven and run them and watch how fast one cools off than the other.

If you consider that only by making contact and heating outside air that heat can be removed from the motor (for these air-cooled motors), then it pretty quickly makes sense that spinning the outside surface the can contact the air is a massive advantage.

Sometimes it's helpful not to view it as building motor cooling, but to think about how you would heat the air in the room around that motor the fastest, as that means the best cooling motor (for motors with equal efficiency at least).

 

[John in CR]

The reason I talk about it being perfect for forced air cooling is because the mounting plate end of the motor is stationary and is already mostly holes. Since both ends of the motor are open, the stationary end cap gives you and easy route to force air in. Put a housing over that end and blow air into the motor. The biggest air gaps are right there between the windings, so that's where most of the air will flow, right where you want it, and the air exits the spinning end.

Of course there's little stuff to do along the way to maximize the flow, but they don't change the simple basic approach.

 

[liveforphysics]

The reason I talk about it being perfect for forced air cooling is because the mounting plate end of the motor is stationary and is already mostly holes. Since both ends of the motor are open, the stationary end cap gives you and easy route to force air in. Put a housing over that end and blow air into the motor. The biggest air gaps are right there between the windings, so that's where most of the air will flow, right where you want it, and the air exits the spinning end.

Of course there's little stuff to do along the way to maximize the flow, but they don't change the simple basic approach.

Yeah, I totally agree. If you're going to spend the watts, forced air is a pretty good approach.


Or, if you guys remember my cooling setup, I used conduction.

 

[johnrobholmes]

Astro used Cobalt magnets because he started with them before Neo magnets were cheap and cheerful. Plus his biggest buyer for many years installed the motors into fully enclosed cases with no heat transfer, so the high temp ratings were key to get the device where it needed to go. Why change a good thing when the design is well tested and you are getting old?

As to the cooling, why would it make more sense to have the magnets on the outside? The copper makes the heat, and it will move faster via conduction if it has a direct and short path out. I would love to read the thread on the outrunner VS the inrunner on shedding heat. I would wager that the outrunner had a fan built in or at least lots of venting while the inrunner was sealed. Inrunners can be built with integrated fans as well, but it is not very common in the toy world for some reason.

 

[liveforphysics]

Its pretty simple JohnR.

You can only transfer heat to the air you can contact.

A spinning can in air, even if its perfecly smooth, is constantly accelerating the air at and near its boundry layer shear area. This makes a local low pressure area, and throws the accelerated airway from the can while drawing higher pressure stagnant air towards the can. This is true even for a perfectly smooth surface (like a tesla turbine blade).

A non spinning can can't do this. It has to rely on on passive convection air motion, which is super pathetic.


Why magnets on the outside? It's a no brainer. You want the most heat sensitive part to be in the coldest place. The windings can safely get WAY hotter than the magnets.
Also, any outrunner with an open can is going have a ton of air exchange with outside air, fan or no fan. This is because you have a ton of uncontained high speed turbulent air inside (which is perfect for cooling), and the local eddys and pressure gradients from the velocity differences are always going to be kicking air in and out randomly.