Owner's Review Stark Varg - First Customer Review!

[TCMB371]

Hi guys. I don't mean disrespect for Mark, but if you look back at his posts he's been pretty negative about the Varg. I would consider it borderline spreading FUD about the bike with statements like "So, it's easy for me to see something like a Varg making an 80hp dyno spike for advertising, then de-tuning it."

There are two ways I can read this:

1. It can do 80hp in a very narrow RPM range -- basically has a large horsepower "hump" that must be flattened out in software to make it nice to ride.
2. Or dyno tricks - "When I managed to weight the bike to regain traction the spike HP numbers were unreal"

Neither of those if very flattering. Both imply some sort of trickery.

I am not trying to convince Mark of anything. All I am trying to do is to refute his claims of trickery for those that are positive about the bike and looking forward to it.

I will not post dyno curves at this time (I have to OK that with a number of people first). What I can do is show the current going into the inverter. Battery was charged to 80% SoC. Voltage sagged to about 365V because of current. Throttle was opened at about 50kph and closed when the motor got to RPM limit.

Wow!!!!!!

 

[Number Six]

Yeah, no sh#t. That thing sounds diabolical, I want no part of that.

 

[Mark911]

I not sure if I see a decimal point or not. However, 1800 amps x 365v = 657K watts, that's not possible. So I assume its 180 amps. 180 x 365v = 65K watts or about 87 theoretical HP. However, current isn't the only factor. 180 amps (for 80hp) assumes all the amps is going through the Q axis with no d axis (field weaking) current. I've shown 550 amps at 72 volts on some of my conversions. Without FW, that would indicate about 40K watts or 53hp on a little QS138 with a Sevcon size 6. However, I never got over 37 real (not spike) hp as I had to apply so much d axis current to flatten/extend the torque curve that little (Q axis current) was left to make torque at higher rpms.
Again, I'll believe the advertised numbers when I see a dyno chart with the supporting logs. I'm not saying the Varg isn't very fast or has a limited powerband, I'm sure it's a step forwards compared to the Alta. I've compared the cells specs (at least the ones that everyone claims the Varg is using) to the Alta's VTC6s and when everything is normalized there's not much difference in power delivery. So 90% of the improvements must be motor/controller based. It would be a "technological breakthrough" to gain even 10% efficiency, but the power and range increases indicate a much greater level of efficiency. That's where I'm skeptical as even research/laboratory experiments haven't shown that kind of leap. If true, your company would be better off selling the tech to the many OEM automotive companies as it's probably worth billions.

With respect, Mark911.

 

[fsfs]

I not sure if I see a decimal point or not. However, 1800 amps x 365v = 657K watts, that's not possible. So I assume its 180 amps. 180 x 365v = 65K watts or about 87 theoretical HP. However, current isn't the only factor. 180 amps (for 80hp) assumes all the amps is going through the Q axis with no d axis (field weaking) current. I've shown 550 amps at 72 volts on some of my conversions. Without FW, that would indicate about 40K watts or 53hp on a little QS138 with a Sevcon size 6. However, I never got over 37 real (not spike) hp as I had to apply so much d axis current to flatten/extend the torque curve that little (Q axis current) was left to make torque at higher rpms.
Again, I'll believe the advertised numbers when I see a dyno chart with the supporting logs. I'm not saying the Varg isn't very fast or has a limited powerband, I'm sure it's a step forwards compared to the Alta. I've compared the cells specs (at least the ones that everyone claims the Varg is using) to the Alta's VTC6s and when everything is normalized there's not much difference in power delivery. So 90% of the improvements must be motor/controller based. It would be a "technological breakthrough" to gain even 10% efficiency, but the power and range increases indicate a much greater level of efficiency. That's where I'm skeptical as even research/laboratory experiments haven't shown that kind of leap. If true, your company would be better off selling the tech to the many OEM automotive companies as it's probably worth billions.

With respect, Mark911.

There is a decimal point. The video was taken a while ago and I just happened to have it on my phone. I didn't do a dyno run for this exchange. At that point (and still at today) the power curve is not perfectly flat. There is an 8% drop from max power to high RPM region. That can be tuned flat if/when we find the time to do it. Up to this point it simply hasn't been a priority. To be precise there is a gradual decrease from 196A to 180A. So at peak we have a bit over 71kW into the inverter and need 60kW on the shaft. We would need just under 85% efficiency to achieve that. I will attach a screenshot of the peak current.

About the Q/D axis stuff... Even at zero/low speed there is significant current through the D axis; if 100A is in Q then 50A is in D. At higher speed D axis current dominates. To say that D axis current does not contribute to shaft power simply _wrong_.

Now for the cells. Maybe what everybody claims is wrong...

Regards,

Fran

 

[Mark911]

There is a decimal point. The video was taken a while ago and I just happened to have it on my phone. I didn't do a dyno run for this exchange. At that point (and still at today) the power curve is not perfectly flat. There is an 8% drop from max power to high RPM region. That can be tuned flat if/when we find the time to do it. Up to this point it simply hasn't been a priority. To be precise there is a gradual decrease from 196A to 180A. So at peak we have a bit over 71kW into the inverter and need 60kW on the shaft. We would need just under 85% efficiency to achieve that. I will attach a screenshot of the peak current.

About the Q/D axis stuff... Even at zero/low speed there is significant current through the D axis; if 100A is in Q then 50A is in D. At higher speed D axis current dominates. To say that D axis current does not contribute to shaft power simply _wrong_.

Now for the cells. Maybe what everybody claims is wrong...

Regards,

Fran

View attachment 9821

Above base speed of the motor, the d axis is used to counter the back EMF generated from the permeant magnets as the motor starts acting like a generator. These back EMF forces severely limit the usable and extended RPM range of any PMAC motor. The only way (to my knowledge without something like a variable flux system) to counter the back EMF is to apply an equal current in the opposite direction, essentially the d axis is in a fight to keep the back EMF from dominating. This required opposite d axis current takes away from the total current available, thus reducing the Q axis current by the same amount and the Q axis produces torque. The battery doesn't care where the current is going (Q or d axis), it's just the total consumed by the system. D axis current can be substantial. On my 84v nominal systems running 550 total battery amps I had to allocate over 300 amps to the d axis to get the powerband flat and long. That means only 250 amps was left to generate torque. But as we all know, HP is a function of torque and rpm, but the torque decreased faster than the rpm increased, so the net affect was steady drop in hp even though the battery was still cranking out 550 amps! So although the video of your battery amps over speed is interesting, it's hardly evidence of what HP is being produced at what rpm. Dyno chart with data logs of battery amps and voltage is a more realistic snapshot of the powerband and how hard the cells are being pushed. The Alta certainly could have made 180 amps (30 amps/per string of VTC6s), but although withing spec, 30 amps relatively continuous would have shortened the cell life and generated an unmanageable amount to heat for that system. With ambient temperatures here in SoCal being close to 100f in our garages in the summer, the packs will already be heat soaked to that temp before you even line up. The narrow delta T between cell thermal breakdown (shorted life) and those ambient temps makes it very difficult to get the heat out fast enough given any passive system. I think the true test will be a hot summer day on a track like Glen Helen with a fast rider. If it makes both motos (charging in between) without any water misting systems or other such claptrap, I'll be a believer.

 

[fsfs]

Above base speed of the motor, the d axis is used to counter the back EMF generated from the permeant magnets as the motor starts acting like a generator. These back EMF forces severely limit the usable and extended RPM range of any PMAC motor. The only way (to my knowledge without something like a variable flux system) to counter the back EMF is to apply an equal current in the opposite direction, essentially the d axis is in a fight to keep the back EMF from dominating. This required opposite d axis current takes away from the total current available, thus reducing the Q axis current by the same amount and the Q axis produces torque. The battery doesn't care where the current is going (Q or d axis), it's just the total consumed by the system. D axis current can be substantial. On my 84v nominal systems running 550 total battery amps I had to allocate over 300 amps to the d axis to get the powerband flat and long. That means only 250 amps was left to generate torque. But as we all know, HP is a function of torque and rpm, but the torque decreased faster than the rpm increased, so the net affect was steady drop in hp even though the battery was still cranking out 550 amps! So although the video of your battery amps over speed is interesting, it's hardly evidence of what HP is being produced at what rpm. Dyno chart with data logs of battery amps and voltage is a more realistic snapshot of the powerband and how hard the cells are being pushed. The Alta certainly could have made 180 amps (30 amps/per string of VTC6s), but although withing spec, 30 amps relatively continuous would have shortened the cell life and generated an unmanageable amount to heat for that system. With ambient temperatures here in SoCal being close to 100f in our garages in the summer, the packs will already be heat soaked to that temp before you even line up. The narrow delta T between cell thermal breakdown (shorted life) and those ambient temps makes it very difficult to get the heat out fast enough given any passive system. I think the true test will be a hot summer day on a track like Glen Helen with a fast rider. If it makes both motos (charging in between) without any water misting systems or other such claptrap, I'll be a believer.

Your understanding is not correct. Parts of what you say are correct, but certain bits are just wrong and lead you to the wrong conclusions. First, the battery certainly does care which axis the current is going to. Second, thinking of the Q axis as the torque axis and the D axis as the field/flux axis is a vast oversimplification. Sorry, but your conclusions are simply wrong.

 

[Mark911]

Your understanding is not correct. Parts of what you say are correct, but certain bits are just wrong and lead you to the wrong conclusions. First, the battery certainly does care which axis the current is going to. Second, thinking of the Q axis as the torque axis and the D axis as the field/flux axis is a vast oversimplification. Sorry, but your conclusions are simply wrong.

Well, just don’t tell me I’m wrong and what conclusions I’m missing, explain your understanding of FOC and how the Varg is different/superior.

 

[Mark911]

Obviously, I simplified the FOC process so that more people could follow. But if you want to get into a more detailed conversation regarding FOC, SVM, MTPA, axis coupling, field weakening, voltage circles, windup, rotor angle lead/retard, PI controllers, forward and reverse park transformations, setting controller gains, or any other topic I'm all ears. It's been a couple years since I really studied these theories and processes, but it'll come back fast. Regardless of theory, I've studied countless dyno run logs which show all the principle parameters and I've seen actual evidence of what I speak. Most controllers these days have a vast array of parameters to log real time for the dyno or during actual riding conditions. Even the "cheaper" ones (not concurrently engineered with the other parts of the system as you mentioned) have logging capability and a long list of parameters to download and study. Some show the data real-time on your cell phone! You just need to know what you're looking for.

 

[Mark911]

Oh, can you please address the battery thermal issue I mentioned? How with a limited (not much) delta T between the cells and ambient temp (here in SoCal assume 100F) can you get all the heat out. From what I see, you're using the same passive thermal management system that Alta used with a few twists. You're still using only the negative end as the heat transfer path, you'll still need to electrically isolate the ends from the heatsink using a TIM (not very efficient), and you're still dependent on ambient airflow across the heatsink to draw heat from the cells to the air, again not very efficient. Your heatsink by virtue of more direct airflow is superior to Alta, but the little fins probably do more to stiffen the cover than transfer heat. Yes, the cells have a slightly higher C rating, but that probably won't be enough on a hot day, soft (sand) track, a fast rider and a long moto. In my opinion they're (the cells) are going to get hot. Maybe not enough to thermal limit (we'll see), but enough to degrade the cells and affect overall life. Alta limited total battery output, temp, charging current, LVC and HVC so they could promise 1000 charges before getting to the industry standard 80% of original capacity and I think this proved true regardless of rider or conditions. What's the Varg promise?

Again, with all due respect to a fellow engineer. Mark911

 

[fsfs]

Well, just don’t tell me I’m wrong and what conclusions I’m missing, explain your understanding of FOC and how the Varg is different/superior.

I didn't say the Varg has some different/superior/magical vector control scheme. What I said is that current into the D axis does not only weaken the flux in the air gap. It also generates torque and therefore power. Current into the D axis is not wasted. Here, have a look at the torque and power curves to a Brusa HSM1-6.17.12 motor driven by a Brusa DMC524 inverter.

​

A first breakthrough: BRUSA components selected

Things are starting to move in the right direction... BRUSA components After I had the chance to meet Mr. Brusa in person (owner of br...

s80ev.blogspot.com

For continuous operation, we have constant torque to 5k RPM and then constant power to 12K RPM. The power is not perfectly flat but it is damn close. It is quite similar to the alta dyno run curves, isn't it?

So how is this possible? It is possible because the motor is designed to do that. It has saliency. This means that D axis current not only weakens the air gap flux, but also applies torque to the rotor. The motor must be designed for this. The inverter must also know how to control the motor in this region of operation. Above 5K RPM the inverter is applying to the motor the maximum voltage possible for the given battery voltage. It is effectively in saturation. The motor is being controlled by the commutation angle. This can be tricky and every motor I have tuned required a slightly different strategy. I am not talking about just changing PI regulator constants and such. I am talking about modification to the vector control algorithm to achieve stability. The transfer from above to below the flux weakening knee can also be tricky.

Now if you take some QSxyz motor and connect it to some Sevcon... First, I doubt the QS was designed with flux weakening in mind (has no or minimal saliency). Designing for efficient operation in the constant power region means you need to make some sacrifices in other areas. Mother nature doesn't give you something without taking something else away.

The Brusa power curves look like they do because the motor was designed to do that. The mathematical model in the inverter corresponds to how the motor behaves. It is this way because the people who made the system worked together and knew what they were doing. Same goes for the Alta. Same goes for the Varg. That is why I stressed the closeness of integration between motor and inverter design at my company. It is what we do. If you could slap together inverter A and motor B and get those results then we would not have a business model.

Someone who slaps together inverter A and motor B and then thinks that that system is pretty close to optimal and a bespoke motor and inverter system cannot do significantly better is just being ignorant.

Regards,

Fran

 

[Mark911]

So you're saying you don't use surface mount magnets? Imbedded magnets (the QS uses a V shape), something with some saliency; less ripple/cogging force and the possibility to do some reluctance switching, etc? That would be totally different than the Alta. Rotor design and the corresponding stator design are in my opinion still in the wild-wild west era. Meaning, there's people in garages (and labs) who can by intent or accident, make significant gains in whatever area they're pursuing. It's like porting was in the seventies. Everyone had their own theory, but nobody truly understood what was happening (I still don't think they do, even with Muli physics simulation programs like ANSYS). Anyway, yes I agree, these require more sophisticated control algorithms.
However, I think that many of the controller manufactures are doing a darn good job generating the code to self characterize whatever motor they're connected to and be pretty efficient, even down to multiple strategies for several different rotor/stator types and configurations. After that, it's not (at least for the few people I know) a matter of just running whatever "nominal" values are in the software. We spend hours on the dyno and on the track fine tuning the controller. The Sevcon has probably 200 parameters that can be changed (the software actually has many, many more, but most don't apply for what we're doing.
Of course, if you program in more amps than the battery can support via voltage drop you're going to fault out. That's one of the reasons run logs are so important. Good SVM algorithms help but aren't typically changeable. I like to see an oscilloscope shot of your waveforms at max output. That could tell me a lot.

We've kinda worn out the motor control stuff, so why don't we focus on the Battery tech, specifically thermal Managment as I indicated in my last post.

Again, respectfully Mark911