Stark Varg

You may be wondering what the Stark Varg has to do with trials and the EMs.  Nothing and everything.  It is obviously not a trials bike, but I expect it will be setting a very high bar for all future electric motorcycles.  I also hope EM pays attention to it.

When I first heard of the Varg, I could not help but start reverse-engineering it on paper.  I wrote about the bike to friends in December of 2021 and intended to use that analysis here.  But as I double-checked things, I found specifications on that were not available a year ago.  They helped immensely, and I made refinements to my analysis where required.  Otherwise, most of the information comes directly from the manufacturer's website and screen captures from their promotional video.

Stark Varg promotional video

Aaron Colton testing the Stark Varg in Barcelona, Spain


Before getting started with my technical dissection, I want to impart the sense of enthusiasm I got from talking with Aaron Colton about his testing of the Stark Varg in Barcelona, Spain.  Aaron said that all the current 450 MX bikes were also present at the test.  Each magazine's representative test rider was encouraged to flog their personal favorite mount around the test track prior to trying the Varg.  That's confidence.  Every single tester was able to lap faster on the Varg than their personal favorite MX bike.  That's astonishing!


Stark claims the Varg makes 80 horsepower and costs just $13,590.  And, if you can get by with only 60 HP, it's $1000 cheaper!  A year ago the introduction video almost made it look like an April Fool's joke and left me hoping it was not all smoke and mirrors.

The 80 HP rating is at its peak.  I'm guessing the continuous power would be considerably lower – otherwise, things overheat.  For a dirt bike, that should not really be a limitation.  If you initially buy the 60-HP version, it is firmware-upgradable to 80 HP later.

The entire bike weighs 110 kg (242 lb).  As I recall, my YZ450F weighed 99 kilos, but that was dry.  So the Varg is certainly in the ballpark weight-wise.

For reference: 1 watt = 1.34 horsepower and 1 newton-meter = 0.7375 pound-feet.


The motor weighs 9 kg (19.8 lb) and resides inside a carbon fiber sleeve.  It is rated at 60 kW (80.5 horsepower).

Based on its form factor, I'm guessing it's either a permanent magnet AC motor or an induction motor.  Permanent magnet brushless DC motors have a more “pancake” shape.  You can get a good sense of scale in this screengrab.

The video claims “no noise.”  This makes me think the controller likely produces sine waves.  

Stark Varg motor

Stark Varg's  motor, controller, primary gear, and structural frame member 

Gear Reduction

Stark claims a maximum rear wheel torque of 901.2 Nm (664.6 lb-ft) and a maximum countershaft torque of 262.9 Nm (193.9 lb-ft ).  

This implies a secondary ratio of 3.43:1 (901.2 / 262.9 = 3.43).

The claim is 80 Nm at the electric motor and 262.9 Nm at the countershaft.  This implies a primary gear ratio of 3.29:1  (262.9 / 80 = 3.286). 

This equates to an overall gear reduction of 11.27:1 (3.43 * 3.29 = 11.27). 

Counting sprocket teeth (about 14 front and 50 rear) in early photos lead me to believe the secondary ratio was 3.57:1, so they may have changed the rear sprocket to 48T. 

It's not visible in this photo, but the countershaft sprocket appears to be below the motor.  So I'm thinking the right side is a gearcase (looks like a sight glass for oil level) that contains the primary reduction gears.  This would cause a familiar torque reaction as almost every ICE bike has the same motor rotation.  You can see that the motor/controller package is structural.

Notice also that you can see what appears to be a pair of connections for cooling lines in the motor/controller/structure photo.


The display and programming facility is a cell phone.  This is brilliant!  The computing performance per dollar of a cell phone is better than anything else on the planet, and there are a lot of people who know how to write apps for them.  Of course, this makes wireless communication with the controller trivial.  The adjacent screengrab is from the promo video.  Notice the maximum rpm of 14,000.  (Later specified to be 14,200.)

Screen grab showing editable power curve.

Estimating a Dyno Chart

I estimated a dyno chart by first making a paper printout of the “Edit Power Curve” screengrab shown above.  I assumed the upper curve represented the maximum possible power available.  I then electronically digitized that curve every 1000 rpm and entered power and rpm numbers into a spreadsheet that calculated torque and drew the dyno chart. 

It was difficult to read the peak power number from the original screengrab (it looks like 56 kW).  I used 80 HP (60 kW) instead.  This is not super-critical, as the shape of the curve and approximate torque values are what I want to illustrate.

The curve starts at “zero” (really 38) rpm and ends at 14,000 rpm.  Torque is calculated in pound-feet.

My initial graph resulted in insane amounts of torque at the lowest motor speeds (like 117 lb-ft @ 650 rpm).   Later, in the published specifications, I read that the maximum motor torque is 80 Nm (59 lb-ft).   So I limited the torque for the dyno chart to 59 lb-ft until about 6000 rpm, where it begins to fall naturally.  You can see the result below.

Note that because the Y-axis graduations are numerically the same for horsepower and torque, the graphs cross at 5252 rpm.

We are not at all used to seeing torque plots like this for internal combustion engines!  Firstly, it starts at nearly zero rpm. Secondly, it is flat or falling for its entirety.

Calculated Dyno Chart, torque limited to 80 Nm (59 lb-ft)

Performance Prediction

All models are wrong.  Some models are useful. – George Box (British statistician)

I wondered how the Varg would do at the dragstrip so I entered its data into a program I wrote long ago that models vehicle acceleration.

The model is fairly straightforward.  Both John Robinson (Motorcycle Tuning, Chassis) and John Bradley (The Racing Motorcycle, A Technical Guide for Constructors) have presented the method I used.

The basic idea is that in order to model a vehicle's acceleration and/or top speed, you first must construct a set of “cascade curves.”  These curves show the thrust available at the rear wheel in each gear plotted against ground speed.  This information is derived from a dyno chart, gearing, and tire circumference.  The total resistance to forward motion from aerodynamic drag, rolling drag, drivetrain drag, and inertia is estimated.  As long as there's more thrust than drag at any given speed, the vehicle still has the power to accelerate.  Acceleration is computed by rearranging Newton's F = ma.  

I had to modify my program to deal with the Varg's single-speed “gearbox,” but the changes were simple – so I don't think any new errors were introduced.

My assumptions are:  

This model calculated a quarter-mile run of 10.76 seconds at 122.7 mph!  The top speed was 123 mph. 

For comparison, the model predicted my optimally-geared KX500 supermoto's performance as running the quarter mile in 11.73 seconds at 102 mph.  The predicted top speed was 103 mph.

I'm pretty sure the Varg would eviscerate my old KX500 – at least with the power turned all the way up to 80 HP.

Screen grab showing various user-configurable operating parameters.

Edit Power Mode Screen

The adjacent screengrab from Stark's promo video shows the power mode editing screen.  The controller provides programmable regenerative braking and traction control.  But more interestingly, it provides a patent-pending “virtual flywheel weigh,” which is separately programmable for both acceleration and deceleration.

I email Stark Future asking about the patent-pending “virtual flywheel weight” but never got a reply (despite them answering other questions).

Prior to even asking, I did a patent search for US applications but found nothing.

If you search “virtual inertia” you will see it's not a new idea.  It is presently being applied on the flip side of things in the field of power generation.  Traditionally, power generation is done with heavy rotating machinery which inherently has massive inertia.  However, as more and more “inverter” power systems that have little or no inertia come online, maintaining a constant frequency becomes more difficult.  I found a good 2020 master's thesis: Control of Voltage-Source Converters Considering Virtual Inertia Dynamics that considers various algorithms.

Likewise, I found a 1988 patent (US 4758967) by the Ford Motor Company that deals with simulating the inertia of an entire vehicle when testing internal combustion engines on an electric dynamometer.  This obviates the need to have unmanageably large spinning flywheels in a production environment.

Both of these ideas boil down to controlling the rate of change of torque versus time (dT/dt).


Specs indicate the 6048 Wh battery pack weighs 32 kg (70.5 lb).  You can see the green circuit boards at the top right, which are for the battery management system.  The orange cylinder is the main contactor.

Stark says the battery contains 400 cells in a 100-series, 4-parallel  configuration providing 360V nominal, and 420V maximum.

I counted 8 x 10 plus 8 x 12 places for cells in the adjacent photo.  That's only 176 places per case-half, so there must be 24 places for cells outside the view.

Stark Varg battery case

6048 Wh and 400 cells imply 15.1 Wh per cell.  That is a really high rating for an 18650 cell.  Cells may be optimized for a high discharge rate (which would be needed to deliver 80 HP) or for high storage capacity – but not both simultaneously.

A friend who is currently designing with lithium-ion cells at work thinks the cells are more likely 21700-size, as Tesla has made that the de facto standard for high-performance work.  An example of an extremely good high-discharge-rate cell in that size is the Samsung INR21700-40T.  It has a capacity of 4000 mAh (at 3.6V that's 14.4 Wh) with a maximum discharge current of 35A.  It weighs 70 g.  

If we use 400 of those cells, we still only get a bit over 5.7 kWh with a total cell weight of 28 kg. (Recall the specified overall battery weight was 32 kg.) 

The higher the voltage, the more efficient the power delivery will be due to minimizing I² * R losses. 

A 35 amp maximum discharge rate multiplied by 4 parallel groups is 140A.  This gets us very close to the motor's 60 kW rating (140A * 420 volts, fully charged, makes 58.8 kW).  As the battery pack voltage drops, the maximum power deliverable is reduced.  At the nominal 360 volts rating, it's 50.4 kW (57.5 HP).

Do these calculations mean anything?  Only that, the battery pack must be very, very good.

North American 120 VAC Charging

The charger is rated at 3.3 kW.  Stark claims it will charge the battery to full in less than 2 hours.  They say it's “delivered with a plug that fits in your normal power socket.”  

I don't know about you, but I can't get 3.3 kW out of my normal power socket.  A North American 120 VAC outlet is typically 15A (1800 W) and that's assuming nothing else is drawing power.  

Even a 20-amp circuit is only good for 2.4 kW.  Let's assume the charger is 93% efficient (that's pretty good and what the ePure Race can do).  So  1800 W * 0.93 = 1674 W.  At that rate, fully charging a 6 kWh battery pack is going to take over 3.5 hours.


If you could actually use all 80 HP (60 kW) continuously, a 6 kWh battery would be depleted in 6 minutes (1/10 of an hour) – assuming 100% efficiency.


It's brilliant to offer 7 spring rates for rider weight ranging from 65kg (143 lbs) to 100kg (220 lbs).  Likewise, you have a choice of an 18″ (enduro) or 19″ (MX) rear wheel.  Another innovation is to offer a rear brake option that can be ordered as conventional foot-operated or left-hand-operated (since there is no clutch).

So far, the biggest downside to me is that it's going to be assembled in Spain.  I'm not impressed with the build quality of a typical trials bike – definitely not up to Japanese or Austrian standards.

April 2023 Update

Stark Future announced customer bikes were beginning to ship as of April 25th, 2023.  Furthermore, the battery is being upgraded from 6.0 to 6.5 kWh at no additional cost.  (But also a significant weight gain, now being spec'ed at 260 pounds!)

Also, from a recent press release, “ updates and functions will be delivered to customers automatically Over-Air.  This will allow customers to stay up-to-date with the latest features and functionalities.”  Sounds great.  But there is a potential downside risk to this, which is losing control of the product you bought. 

Stark Future seems like a very forward-leaning company.  I would hope their OTA updates are fully transparent.  For me, this would include:

Alta Redshift Comparison

Just for the sake of comparison, I'll include the information I have about the now-defunct Alta Redshift MX bike.  A complete Alta Motors Rear Bulkhead Kit w/ Motor is shown below.  It's being offered on eBay for $2215.82 + $50 shipping.  

It would seem much of the inspiration for the Stark Varg came from Alta.  The motor looks very similar.  

Alta claimed 40 horsepower.  I have read both 14,000 and 11,500 as the redline.  I was surprised to see helical primary gears as that sacrifices some efficiency versus straight-cut gears.  But they are quieter, and that was probably an important design consideration.  I  have read that the motor is a permanent magnet AC type.  This makes sense based on its shape (and the battery voltage).

Credit: eBay seller liquid_performance

The Alta owner's/service manual is still available for download.  It is very clearly written and illustrated.  The photos are good enough that it was possible to count teeth on the primary.  19T drives 67T for a ratio of 3.526:1.  As confirmation, one reporter said 3.5:1.

It's also possible to count teeth on the countershaft sprocket at 12.  In some specifications I read, the final drive was listed as 12-53 (although one photo of a rear wheel showed a 50T sprocket).  Assuming the written specification was correct puts the final drive at 4.416:1.

I was disappointed (but not really surprised) that the factory manual offered nothing even resembling a wiring diagram.

It's pretty cool that a few motors are still available (albeit grouped with a bunch of other stuff).  The same eBay seller also had the high-power electronics (a Mitsubishi IGBT module) offered at a reasonable price.  However, there were no motor controllers (the actual “brain”) listed.  Perhaps they have all been snapped up, or perhaps none were ever made available?

Alta Battery

Alta's battery was advertised as a “waterproof li-ion 350V / 5.8 kWh.”  A YouTuber (Shea Nyquist) disassembled the battery to reveal 4 modules.

Each module comprised a 21S6P arrangement. So 21 * 6 * 4 = 504 total cells.  Each module being 21S would nominally give a 75.6-volt rating (21 * 3.6V = 75.6).  Four such modules connected in series would produce 302.4 volts.  So Alta's 350-volt rating would have been for fully charged cells (say 4.2 volts) 21 * 4.2 * 4 = 352.8 volts.

Likewise, the pack's capacity was a bit overstated as well.  Shea found Sony VTC6 cells inside.  These are 18650-size, rated 3000 mAh capacity, and have a discharge rating of 15A.  So, 3000 mAh * 504 cells * 3.6 volts per cell = 5.44 kWh.

The VTC6 cell nominally weighs 47.5 grams.  Thus all the cells together weigh 47.5 x 504 = 24 kilograms (about 53 pounds).  Each cell was bonded to a printed circuit board with fusible links Shea has estimated to be good for 50 amps.

As far as discharge goes, 6 parallel groups of 15-amp cells yield a 90-amp theoretical rating.  90A * 300V nominal = 27 kW (36 horsepower).  With a fresh charge, the battery voltage would be more like 340V, so 30.6 kW (41 horsepower).