Data logging details?

[johnrobholmes]

Pat,

Can you shed a bit of info on how you log the current in the ICE style controllers? A hall effect sensor? Maybe you measure the resistance or voltage drop across some components? Can't say or else I would have to be killed? I have some haters that are saying the numbers are not even close to a quantification, only an indication of the change in current.

[BrianG]

I believe it's a HE sensor. The shunt resistor needed for 100A+ would be rather large, and difficult to size so that it can measure high currents without too many losses, yet still have accuracy at lower currents.

[johnrobholmes]

That is what I thought, but double checking is always better than just spouting off. I will wait before tearing mine apart in search for the little bugger

[BrianG]

It shouldn't be too hard to find. Just look for a TO-92 or TO-220 style/size transistor looking device that has a pair of much larger metal leads that are inline with the main battery + line.

[Pdelcast]

Originally Posted by johnrobholmes

Pat,

Can you shed a bit of info on how you log the current in the ICE style controllers? A hall effect sensor? Maybe you measure the resistance or voltage drop across some components? Can't say or else I would have to be killed? I have some haters that are saying the numbers are not even close to a quantification, only an indication of the change in current.

Shunt - -believe it or not.

For example, on the Mamba XL (which is the same hardware as the ICE-200) we use 4 x .0005 ohm shunts. (for a total resistance of .000125 ohms)

There is an instrumentation amplifier that amplifies the signal.

The absolute accuracy is OK -- +-10% normally. The relative accuracy is very high.

The reason we didn't use Hall sensors is we wanted good response time for ESC protection. Hall sensors are relatively slow, and response times can be too slow to offer ESC protection in short circuit events.

Shunts are very fast -- limited only by the bandwidth of the instrumentation amplifier. But, they are more expensive (because of the high cost of a good instrumentation amplifier...)

Thanx!

Patrick

[BrianG]

Wow, color me surprised!

That instrumentation amplifier must have some SERIOUS resolution to be able to detect the voltage across the shunt at lower currents. At 1A, that's only 125uV.

By the way, how are you determining what is a short circuit vs just a really high current surge from a tall geared setup, especially since you are looking for very fast response times (those burst currents can get really high for a couple ms)?

[Pdelcast]

Originally Posted by BrianG

By the way, how are you determining what is a short circuit vs just a really high current surge from a tall geared setup, especially since you are looking for very fast response times (those burst currents can get really high for a couple ms)?

Pretty easy actually -- we just set maximum PWM values based on the integral of the amplitude of partial throttle currents. This limits maximum power dissipation on the transistors.

[johnrobholmes]

Thanks for the detailed info. Good enough for me :D


So what is the controller and shunt resistance of the ICE HV 160 (no wires)?

[Pdelcast]

Originally Posted by johnrobholmes

Thanks for the detailed info. Good enough for me :D


So what is the controller and shunt resistance of the ICE HV 160 (no wires)?

360 microohms for the FETs, and 125 microohms for the shunt. Copper resistance is around 50 microohms.
~22 watts total I^2R controller losses at 160A. (or approx .45 watts per FET) This doesn't include switching losses. Switching losses depend on the commutation speed of the motor -- but a good "middle of the road" estimate is approx. 8 watts of switching losses at full throttle. (Partial throttle switching losses are higher -- because the PWM losses are combined with the commutation losses.)

At 50V, 160A, full throttle, this gives an efficiency of ~99.63%

(note: the losses on the controller is about the same as the losses in the ~25" of 8ga wire attached to the controller... 8 ga wire is approx. 600 microohms per foot -- total losses in the wires at 160A is approx. 32 watts... and the numbers are even better on the ICE-200 and MM-XL, because the FET resistance is only 213 microohms...)

Pretty damn good, I think.

[brushlessboy16]

Holy high efficiency batman!

I really wish all manufacturers could be as detailed as castle! We need more patrick's!


That efficiency number is rediculous, When is castle going to comine out with a full size car controller? my 2001 kia sportage could use it!

[BrianG]

Nice! Now all you need to do is get rid of that pesky copper and get better FETs.

[Pdelcast]

Originally Posted by BrianG

Nice! Now all you need to do is get rid of that pesky copper and get better FETs.

Better FETs are coming all the time... It's amazing what the FET manufacturers are doing...

Just three years ago, the best FET available at 60V in an SMT package was about 6 milliohms -- today it's about 3 milliohms. By this time next year, we are hoping to be at 2 milliohms. Awesome stuff happening in the FET world.

[johnrobholmes]

Sounds like I need to start running 00 cables on my ebike.

We do need another Patrick, or at least Patrick needs another Patrick I will work on my cloning machine, and you work on my 100v sensored controller for 3hp continuous. We can race to the finish.

[Pdelcast]

Originally Posted by johnrobholmes

Sounds like I need to start running 00 cables on my ebike.

We do need another Patrick, or at least Patrick needs another Patrick I will work on my cloning machine, and you work on my 100v sensored controller for 3hp continuous. We can race to the finish.

Dude, I need the cloning machine.

[johnrobholmes]

I will need a thorough interview of you to ensure the clone is sufficient for the Pat time doubling technique.