Doing the Math

[liveforphysics]

Eddy currents is unrelated to magnetic attraction...
Its entirely about conductivity.

When flux lines move in location to a conductor of any type, it induces a voltage. If no path for that voltage to go, they circulate until they are all shorted out as heat.

[Kingfish]

Luke, hence the reason for Litz wire? My understanding was that the eddies were iron core related/induced.

I must read more... KF

[liveforphysics]

Luke, hence the reason for Litz wire? My understanding was that the eddies were iron core related/induced.

I must read more... KF

Nope. Eddys have nothing to do with the iron, other than that it conducts. But silicon steel conducts so much worse (47.2×10-8Ω·m) than aluminum ( 28.2x10-9Ω·m), it's like 1/20th of the eddy brake of aluminum. BUT! Then you have it layered in strips called laminations that are electrically insulated from each other, and positioned with the axis of field rotation, so now it can only be as much of an eddy brake as the magnetic field is able to induct a voltage across that very thin section, so unless the field is moving at a very high rate, or the flux is incredibly strong, it hangs out in that negligible losses area. For example, in a hub motor, Justin's dyno showed no x^2 parasitic drag increases clear up to 40-50mph test speeds, meaning in the normal direct drive hubmotors, eddy's weren't even on the concern map.

Also, carbon fiber is a pretty damn good conductor, and playing on the test bench this morning, I needed a >100w electrical loadbank, and ended up using a chunk of carbon fiber cloth, and kept needing to trim it down to be more narrow because the supply couldn't handle the load. Then later it was glowing red and turned into a sparkler sort of effect, went open circuit, unloaded the flyback transformer, and the cap exploded when nothing was there to suppress the voltage from climbing out of control.

For the equilivant strength/rigidity part, carbon fiber actually conducts better than the silicon steel they use for laminations.
Kevlar cloth isn't nearly as strong, but you can have magnetic fields wizz through it, around it, next to it, etc all day long and it doesn't care.

[liveforphysics]

Likewise, Biff wasn't saying the stator of the CISRO motor has any aluminum in it for the area the rotor flux moves through (it doesn't, it uses it as a center hub, the stator is entirely epoxy encased litz wire).

[liveforphysics]

It's like maybe 1/10th the amount of the design effort to make a motor with iron. However, then you've got hysteresis losses, lame cogging effects, and other lame losses. Many of the losses that vary exponentially with RPM can kinda be neglected as issues for hubmotor designs though, because they stay in the very very small part of the loss curve slope, where something like a 10,000rpm RC motor has these effects become extremely critical, and going ironless has a huge payoff (like the launch-point motor for example).

For direct drive hubmotor speeds, ironless is a TON of extra design solution work for minimal performance improvement, and adds it's own weaknesses like winding cooling issues etc.

That said... it doesn't require having custom laminations stamped, and it would end up giving you the best possible end product if you managed to get the right design, so it does dangle a tasty carrot for a DIY'er, but the carrot doesn't come easy.

[Alan B]

Interesting to see the discussion here on the Halbach Arrays. I met Klaus Halbach years ago where I work at Lawrence Berkeley Lab, he designed some of the magnetics on the accelerator I work on (the Advanced Light Source). Sometimes the world seems very small! There are a number of engineers he trained still working there. Good to see the technology in use! (Or at least proposed).

[Arlo1]

Mr Kingfish I never realized that was your purpos with this thread!
I realy Like were this is going!
I Think once we can help get some ground rules in place we can try for a easy to build/wind/modify DIY motor where people can add to or customize to suite their needs!

So I have been talking to luke about this since My trip to seattle and It took me one weekend to realize that If I want to get excatly what I want I have to build it my self!
We started talking about stator laminations and iron powder epozied stator cores till we (or maybe just I) realized Ironless is most likely the way to go!
As for Axial flux well my plan is to make it stackable! And add more stators and more rotor discs to increase the power the multi layer has a huge benifit of being able to be run in series to start moving then switch to parallel for higher speeds which will help with controler building and designing and the life of the controller it self!
I have been playing a lot in the shop with magnet wire and One thing I discovered is you can Wind your stator in a figure 8 patern and one loop of the 8 is reverse to the other then you can have 2 magnets with reverse polarity next to each other on the same side of the rotor!

In conclusion I am 100% stoked to be a part of this like I said I will build a motor from scratch soon but like Luke I keep re-designing it in my head and trying to make it better! I Think if we can aim for a 85-95% efficient DIY motor, not even realy trying to beat the peak efficiency of the current motors (although I want to and plan to) but try to make it run in the peak more often, with Carbon Fiber rotor and kevlar stator we can have a light weight quick reving efficient motor!
I have kevlar, fiberglass, and carbon cloth in my shop and west systems epoxi resin, and a vacume bagging system so I can build most the stuff cheep!

[Biff]

but is around 94% efficient (vs the 97% of the CSIRO)

that's a gulf as wide as the grand canyon.

You are right, it is a huge difference, but it depends on how you look at it, and what is important.

In terms of heat production it is significant, as 94% efficient produces 2x the heat as the 97%

In terms of power output vs power input, the difference isn't that large, at 2kW power input you would get 1.94kW out with the 97% efficient drive, and 1.88kW out with the 94% efficient drive.

the biggest difference is where the heat can be shed. With the coreless axial flux (without some fancy heat transfer system) you end up trying to get rid of all the heat between the rotors where you have a 2mm airgap on each side and virtually 0 airflow from an external source. With the iron stator, you have the ability to move the heat out to a location where you can easily pass fresh air, so the extra heat is very easy to remove.

i'm surprised & confused by the use of alu, i would have bet anything an HA demanded CF.
why isn't it causing eddy interference & parasitic heating the way it did for Thud & luke4phys??

The CSIRO motors that I have are actually not hallbach and have iron backing. The added cost and complexity of manufacturing the Hallback arrarys (and I belive using an aluminum "back iron" for structure) was not worth the added efficiency gain and weight lost, so all of the new kits they are selling are standard magnetic arrays with iron rings. The kit that CSIRO sells for around $10,000 consists of 2 rotor rings (magnets attached to iron rings), a stator (coils made of litz wire encased in epoxy (and I believe a small amount of fibreglass)), hall sensor board and 3 large inductors. Once you get the kit you need to design the rim, shaft casing and everythign else you want to actually build the motor, most of those components that my team came up with are aluminum to save weight and still be easy to manufacture. With CF parts it would be significantly more complex to manufacture, and also difficult to model the potental failurs.

There are little to no eddy currents in the rotor due to the magneic field in the rotor remaining almost constant. Where you end up with eddy currents is when you have magnetic field moving past a large piece of conductor (like Luke mentioned, CF would also form eddy currents in that situation).

Kingfish was interested in why they used Litz wire. I read the paper on the development of the motor and if I remember correctly they did it because of the PWM frequency and frequency in the motor, they wanted to reduce the losses at high frequency. But that doesn't make a lot of sense to me, since the PWM frequency is filtered out by the inductors requred to be able to actually control the motor, and the fundemental frequency of the AC driving the motor is low aswell. It does however make sense to use stranded wire to reduce eddy currents because the large (around #10) wire actually used would offer a fairly large cross section to the moving magnetic field causing eddy currents. Litz wire however is not that fun to work with, and it uses a fairly large amount of area for the insulation between all the little strands, so I would recommend using regular magnet wire for your own designs, but use a few conductors in hand to break up the large cross section of a single wire, the 100's of strands in litz wire is overkill IMHO.

For those of you interested in the efficiency of the motor, I have found this graph my team made, I put in a request for the data used to produce the graph, but I am not sure if anyone still has it. I believe this is the efficiency of the motor and controller (the controller we use is worth around $6,000 and has a maximum output of 10kW, search for tritium wavesculptor if you are interested in it). I think most of the RC efficiency numbers are also motor and controller, so if you estimate the RC controller efficiency at around 94% that would mean a system efficiency of 86% would indicate that the motor efficiency is around 91%

[Kingfish]

Friend Arlo

The goal has been to build an ironless AF motor for use with a sinusoidal controller - perhaps offered as a kit or assembly after proof-of-concept if such a thing could be worked out; though for me it has been all about faster, farther, better. If other folks can benefit from this research then mores the better. As these designs become more refined and the costs resolve into focus it begins to make sense to collude with kindred interests and partner the expenditures, but of which type shall we build? HA is the primary choice, NHA is secondary; Eric prompted me to seriously consider the option long ago (relatively speaking in Thread-Time) and I see now the merits of it. Additionally I think we need to take a look at Radial Flux design so that folks understand the differences between the two configurations, and lastly I would like very much to review a Rim-oriented (near-full radius) AF motor. This would bring us full-circle to all hub-oriented options.

Of late I have considered hybrid designs which include stacking; “which of these lends the most payback?” and that sort of stuff. The models grew more complex as I tried to refine the study until they wouldn’t render any more useful data. A workaround was discovered and have simplified the models; no more side-by-side comparisons to confuse the software application.

I have explicitly embraced ironless to get away from cogging and associated issues, and have accepted the fate of resolving winding-induced heating issues. The direction of late is to maximize the overall Flux Density per pole-phase in the hope of reducing the current thereby lessening the heating.

Eddy Currents and Coreless Stators:
According to my best reference on the subject Axial Flux Permanent Magnet Brushless Machines, eliminating ferromagnetic material in the stator also eliminates associated eddy currents, hysteresis losses, and cogging torque; by doing so we achieve a significant increase in efficiency. (See Chapter 5 in the link provided).

However as Luke mentioned there are still eddy current losses within the windings; these can become large as the frequency increases, thus we must study this well to provide the best solution. The book goes into detail about how to assuage eddies and circulars beginning at 5.8.2 on Page 172 (I own the book). I can’t tell you how many times I have read this section and each time I pick up something new. I have looked at flat/ribbon wire, Litz, and twisted; there are advantages and drawbacks to each.

• Flat appears easiest to work with but is not optimum.
• Litz couldn’t be found off-the-shelf in the size I required (and likely very costly).
• Twisted-strand does seem like the best (and least-expensive) alternative, either bought or crafted to spec, and is probably the way I shall go.

At present the Plan-D model operates between 300 to 1500 Hz (not rpm!) covering expected bicycle to rice-rocket motorcycle ranges - although in Luke’s case we should up that! Luke, how fast have you been on a motorcycle?

Carbon-Fiber:
Luke/Biff bring up a good point. Perhaps it is better to avoid it altogether and go with Glass-Epoxy sandwich in the critical area. There is time to sort this out properly.

More in a bit. KF
PS – Though I have been remiss, thanks for all the thanks; I wouldn’t want to be doing anything else. Well~ OK maybe swimming on a remote beach in the tropics with a cute petite island girl. But this is the next best thing <~slurp coconut drink>

[Biff]

Winding the Axial flux coreless motors isn't the easist thing either, so really you will need to build a jig or something to be able to get all the copper in the right place, then use a small amount of bonding agent to keep things togehter while you put it into the final mold to make the part.

One thing that I have considered for AF designs, and also for radial flux aswell would be to make the inital winidng jig out of this material

http://www.coolpolymers.com/dseries.asp

It is a thermally conductive plastic. By winding the copper on this you would be able to make a path for the heat to of the stator and also have a nice machined (or molded) jig to ensure that everything is in the right place. Once you finish the wind, I would use a VPI type technique to ensure good adheasion between the strands of magnetwire and to the plastic, then fianally put it into the glass/epoxy mold to ensure a smooth finish (minimize windage loss at high speed) and to add the structural support required for mounting and such.

You could easily incorporate cooling fins into the plastic winding jig, or a nice surface to mate to an aluminum mount/heat sink once it is out of the flux path so you could get the heat to where you can get rid of it.

The material I looked at most closely is
D5108: thermal conductivity of 10W/mK, Conductivity 2.5x10^16 ohm-cm, Tensile strength of 37 Mpa and costs $65 / pound in quantities of 30lbs. (and drops to $33 if you buy 3,000lbs )

*EDIT:
I also just looked into a thermally conductive epoxy and found this
http://www.epoxies.com/therm.htm

The 50-3150FT material operates up to 200C and has a thermal conductive of 15 btu/h/ft^2/degF/in ... which as best as I can find converts to W/mK (Watts / (meter * Kelvin) ) by multiplying by 0.144 so that would be 2.6W/mK which isn't as good as the plastic. But perhaps my conversion is wrong because I couldn't actually find any reference to btu/h/ft^2/degF/in I could only find btu*in/h/ft^2/degF which has the inch as a multiplier rather than a divisor, so I tink that the data sheet might have the wrong unit. Honesly who would use a unit like that anyways, W/mK is so much easier.

[Arlo1]

Kingfish. Luke and I discussed a liquid cooled winding meaning the winding is filled with a fluid that circulates and the windings are a small copper tube them selfs! Oh and Guess what conducts better the copper!
On another note Lukes not the only one who likes going fast

[Kingfish]

Kingfish. Luke and I discussed a liquid cooled winding meaning the winding is filled with a fluid that circulates and the windings are a small copper tube them selfs! Oh and Guess what conducts better the copper!
On another note Lukes not the only one who likes going fast

You’re not thinking Mercury are you? How about Silver, or were you thinking of an alloy?

Oil is a great for heat transfer. Also used are alcohol, ammonia, and metal salts in the case of nuclear reactors. No you have me at a loss – I could only guess what you guys have in mind!

I have spent some time dreaming about the way to cool the stator(s):

• Implied air turbine is the easiest - though in the PNW is would get dirty pretty quickly and you couldn’t take it off-road mudding.
• There are certainly capillary options across the exposed rotor faces but the cost… I think we need to be more clever.
• There is also the heat pipe option if an interface could be exploited. One in particular comes to mind, but it requires in situ testing to validate that we even need it.

Biff, thanks for the leads to these materials; there is much to consider. Do you have insights on the magnet bonding agents? I think we need to be sure that the cure-temps do not exceed the magnet ratings.

So Arlo – what is the fastest you’ve ever been on a motorcycle? (I am being serious, though a bit crazed as well). Anyone go over 120, 150 mph regularly? What about racing? What’s the speed record for an ebike (motorcycle)? We gotta have a goal, know what I mean Vern?

Charging up now, KF

[Biff]

Kingfish. Luke and I discussed a liquid cooled winding meaning the winding is filled with a fluid that circulates and the windings are a small copper tube them selfs!

Hollow copper is fairly common in large machines, I though of the same thing, but never found any source for really thick walled small diameter pipe so I gave up. Sort of related to that idea was my idea to use a sacrifical casting technique, by putting a styrofoam pattern in with the copper, then after the stator is fully assembled, use a disoving agent to remove the foam creating a channel for liquid to flow right in with the copper.

have you found a source for small copper pipes suitable to develop your idea?

Biff, thanks for the leads to these materials; there is much to consider. Do you have insights on the magnet bonding agents? I think we need to be sure that the cure-temps do not exceed the magnet ratings.

Yep, that is important. The high temperature epoxies (good to 200C) that I mentioned above (http://www.epoxies.com/therm.htm) could be used to encase the ironless stator. I have always used JB weld for attaching magnets to stuff, but the high temp epoxy could work, I also think there are some high-temp epoxies from Devcon that would work. JB is just very easy to find and I have never had a problem with it.

[Arlo1]

Haha yup mercury :p
I am a motorcycle mechanic and I have worked at 2 dealer ships who sell Ducati's three who sell Honda three who sell yamaha and three who sell Suzuki! On Ducati demo days two years ago I got to max out a 999 and a 749 on the same evening! I also got to max one of the fist hyabusas out at 198! And many many others! The first one liter bike I road was a friends r1 and I clamped I'n I'n first going over an over pass which was on a hot day so I was spinning and slightly crossed up with the front end off the ground as a cop was coming from the other way lucky there was a barrier I'n the middle of the road so I had time to disappear!
I also raced my boss through Edmonton while I road a gsxr 750 and he was on a hyabusu with his wife on the back!
Good old demo days .
As for my electric bike I have gone 85 km/h on the flat and 94 down a hill! I have big plans to set the record and let Steveo and doc go home crying lol.
I don't own a sport bike because I need my licence and although it may sound like I'm wreckless I'm not it's just not worth riding a bike you can't open up! Well unless it's a silent flat black bicycle with no plates
As for the tube biff I searched it online and it is available to a very small size.

[Toorbough ULL-Zeveigh]

The CSIRO motors that I have are actually not hallbach and have iron backing.

whew, that's a relief!
thanx 4 the clarification.

i still believe CF plays a big part in wrangling in those last few percentange points.
Luke is using the wrong kind is all.
the pyro should try the lyte stuf.

[Kingfish]

Thanks Biff on the epoxies; we should check into thermal curing as I think that is the way of the industry.

Gosh Arlo, I would think 80 km/h (50 mph) is plenty fast for a bicycle and more-or-less my target for P1 as a 2WD – and I think I can get there with an over-volted Plan-D 3-Rotor/2-Stator build. That 198 km/h (123 mph) mark is impressive though. We need to find the fastest e-cycle record, and then build a special motor just for burying it. But hey – that’s a down the road a ways: We have lots of time to dream!

However - I would perish the thought of using Mercury for cooling. I don’t even want to touch the stuff, let alone breathe it: Not practical for a lot of reasons.

If you are serious about liquid metal, I’d review the Wikipedia topic. My suggestion is NaK because at least it won’t poison you. But then you wouldn’t want to expose it to air or water either.

Might I suggest a very safe alternative is Mineral Oil: Dielectric, non-compressible, lubricating, and has nice thermal-bearing characteristics exceeding 200° C.

Or, maybe there is a biodegradable equivalent out there; wouldn’t that be peachy! Stop for a burger and charge, top off with a little Canola and away you go.

What a wonderful spin of ideas; most fun!
Cheers, KF

[Arlo1]

Thanks Biff on the epoxies; we should check into thermal curing as I think that is the way of the industry.

Gosh Arlo, I would think 80 km/h (50 mph) is plenty fast for a bicycle and more-or-less my target for P1 as a 2WD – and I think I can get there with an over-volted Plan-D 3-Rotor/2-Stator build. That 198 km/h (123 mph) mark is impressive though. We need to find the fastest e-cycle record, and then build a special motor just for burying it. But hey – that’s a down the road a ways: We have lots of time to dream!

However - I would perish the thought of using Mercury for cooling. I don’t even want to touch the stuff, let alone breathe it: Not practical for a lot of reasons.

If you are serious about liquid metal, I’d review the Wikipedia topic. My suggestion is NaK because at least it won’t poison you. But then you wouldn’t want to expose it to air or water either.

Might I suggest a very safe alternative is Mineral Oil: Dielectric, non-compressible, lubricating, and has nice thermal-bearing characteristics exceeding 200° C.

Or, maybe there is a biodegradable equivalent out there; wouldn’t that be peachy! Stop for a burger and charge, top off with a little Canola and away you go.

What a wonderful spin of ideas; most fun!
Cheers, KF

At this point the liquid wonding is just a thought! Simplicity of the build its the most important! And as I think about it the better you build it the more effiecient it is and the less heat you need to get rid of!
And the Hyabusa toped out at 198 MPH (316km/h) for that year!

[Kingfish]

And the Hyabusa toped out at 198 MPH (316km/h) for that year!

Good gawd man! Did you wear a chute?

Well to get to 200 mph we'd need to consider some serious aero to reduce current. I say let's make a plan for Bonneville! <nods>

TEAM ES of World League. Yeah, I can noodle up some ideas for cooling on that scale.

BTW – I have to ask since it appears that the space inside the hub is going at a premium: I was planning on leaving this design as Star/Wye and not Delta because of the voltage issues. Certainly there should be enough starting torque for Delta, I just don’t think we need it. Point: My preset hub design for 2-Rotor/1-Stator would have ample room for switches. However if we start talking about multi-stator & cooling that would be the first feature I’d ditch: Gimme freedom to design.

Thoughts? KF

[Arlo1]

Certainly there should be enough starting torque for Delta,
Thoughts? KF

OK so.... The torque is the same for WYE and Delta
I guess the easiest way to explain it is to solve for torque you convert watts to HP then use the rpm to solve for torque (as long as you are still in the usable rpm of the motor)!
I have had to sit and watch luke explain this to 20 or more people! And Luke it so smart he forgets to dumb it down for them! You are a very smart guy so I am sure you will understand but I decided to find a simple way to explain it!
I have to educate a lot of people with my trade so I find easy ways to help them understand like when explaining the relationship between Torque and HP I just say they are the same except HP has time added to the equation!
Hope this helps,

I am interested in building this for the Drag bike I want to make as well!

[BrushlessinSeattle]

Hey KF,
I was wondering about your choice of bar magnets. It seems like there would be a lot of flux leakage from magnet to magnet near the ID where the magnets are almost touching. Nearer the OD you have large gaps between the magnets - torque and efficiency will be negatively impacted. I know you're not keen about wedge magnets because the pole count of off the shelf wedges is only 16 and that would result in a motor which runs too fast. This is where my understanding starts to get a bit fuzzy, but cant you just run the controller at a lower frequency? Would this affect efficiency?
-BiS

[Kingfish]

Brushless, controller frequency not the issue: It’s about gearing my friend.

Imagine you are trying to climb a hill; lower gearing makes it easier. It’s the same with magnetic poles – simplistically, the more pole-pairs that you add the less power it takes to rotate the wheel. The typical windmill magnets are pre-designed for a certain speed and that is a bit fast for us hub guys. Example: The 9C hubs actually have 46 magnets/23:1 gearing. The Plan-D option I am considering uses 32 magnets/16:1 which I believe is ½ speed of the windmill, but obviously faster than the stock 9C ...which is what I want.

Now, to do the calculation properly you need to presume you want the windmill size and that you can live with the lower-starting torque by providing a proportional amount more power. Don’t worry about the controller frequency. Just run through the calculations. Your windings will be larger, or perhaps you may opt to have more winding sets per phase.

Are you considering the ¼-inch or the ½-inch thick? Do you want me to do this calc with you so you can visualize the difference? I Personally think it would be a valid exercise, and fitting for the NHA AF section of this thread. If so then I need some specs from you:

• What is the desired target speed for your ebike that you require?
• What is the size of the target wheel?
• Do you know how much power you want to use at the target speed? Example: I set my model to use 2 hp @ 30 mph with a 26-inch diameter wheel.
• What is the target Battery Voltage or Current? (Pick one or the other)
• Can you send me a link the specific magnets that you have in mind so that I have the specifications please?

Happy to Help, your Eastside buddy KF

[BrushlessinSeattle]

Thanks for the explanation KF, I like the gearing analogy. I'm still confused though. I've been approaching the problem from a Lorentz equation point of view ie for a given amperage the torque achievable by a motor at low rpms is proportional to the length of wire and the density of the flux that the wire is immersed in. Now assuming, for the sake of argument, that the 16 and the 32 pole models have the same radius, magnet area, flux density etc, we could immerse just as much wire in the 16 pole version as the 32 pole version. There would be half the number of coils in the 16 poler each consisting of twice as many turns but since the magnets are wider the coils can be wider too, and our stator width/airgap stays the same. Am I visualizing this incorrectly?...entirely possible. With an ironless stator cogging would not be an issue. When it comes to length of coil end turns....you got me there, they would be longer with the 16 pole version.
Your insights would be much appreciated. Once I get my head around this I can move on to other considerations - target voltage, mph etc
-BiS

[Kingfish]

Brushless,

On the face of it – that sounds like a good direction for theorizing. A practical approach though would avoid the presumptions and start from scratch. You wish to design around what the magnets can do for you, yes?

Define the model: Walk through the calculations until you have a concrete answer. Change ONE parameter and recalculate. Compare and contrast. Repeat until you understand the limits of the model.

Avoiding the rotational frequency differences, I am not convinced that a NHA 16-pole windmill magnet will perform the same as a 32-pole with exactly the same physical features sans width/arc-length. With a NHA you also have to account for air gap between the same-plane magnets, and I think that decreasing the number of poles would affect field strength in a non-linear manner.

Consider: I started with 20 poles (Plan-A) thinking it would do, but then had to step back from that after three iterations because I was not satisfied with the resultants. If you want to see a detailed study using the windmill magnets I suggest reading up on Shane Colton’s work.

Best, KF

[BrushlessinSeattle]

KF,
Thanks for the Shane Colton link, that's some really relevant stuff. I was a bit disappointed when I saw the table on pg 45 which compares Kt for the 3 motors. It shows the axial coming in a distant 3rd, but I punched the tabulated numbers into the equation and got a much more impressive 0.71Nm/A for the axial. Hard to believe he would miscalculate that one .......maybe I did (although I did it twice)
-Bis

[Kingfish]

BiS, that’s why I also model with FEMM, and why you’ve seen me produce so many iterations. It’s all about doing the diligence of discovery, and it takes time. The happy part of the story is that you are spending that time learning as oppose to burning through money on less-optimum paths.

The path of your discovery will be different than mine. Where ours cross and perhaps run together is where we can collaborate. As the song says… “because two heads are better than one!” ♪ ♪

Now I’m Sleepless, KF