Another RG250 with Colossus Motor

[SplinterOz]

Arlo1

The 24Fet I have has been well built by Steveo, it has considerable build up on the power and phase traces, but I don't think the 26fet will fix the other issue regarding the throttle response.

 

[Arlo1]

Arlo1

The 24Fet I have has been well built by Steveo, it has considerable build up on the power and phase traces, but I don't think the 26fet will fix the other issue regarding the throttle response.

Have you done the resistor? mods to increase the throtle resalution?

 

[SplinterOz]

While the controller is away for repair, I am going to progress two things. 1) the battery pack 2) a throttle mode for the controller.

With the battery pack I have built a very rough mockup of what I want.
Yesterday and today I spent some time putting my very poor woodworking skills to the test. I have mocked up the battery holders for my battery pack to confirm my design and to check I can hold 120 cells in this setup.


Battery pack mockup by Splinter, on Flickr


Battery pack mockup by Splinter, on Flickr

The setup has three (3) plates where the final will have four. Two outer ones form the end of the double width portion of the batteries (80 cells in total) with holes in the inner two to support these batteries. The inner plates will also terminate the single width of cells that are below the frame and above the motor.
I was inspired by a dirt bike on ES that looked like this...

BatteryExample by Splinter, on Flickr
(Sorry I could not find the original source)

Maybe a video will help.
[youtube]y0W1B_jvPZc[/youtube]

Throttle mode will be fetchers mode that toolman2 already has the parts for.

 

[SplinterOz]

I have been thinking for a while what I would be doing with gauges for the electric bike. Nothing I looked at seemed to be the right solution. The original cluster was worn and tatty (it is a 25 year old set) and included a speedo, tacho and temp gauge. Also included was the usual array of lights.

(Note not my picture but very close to the condition of my gauges.

Now the electric bike needs a little bit of a different setup for one thing with a single fixed gear the tacho is not important. However I need is a State Of Charge (SOC) indication to I can tell how full my battery pack is and a current draw meter so I can keep an eye on how much punishment I am putting the battery pack and motor under. They will have to be clear, concise, accurate and attractive. I could get two gauges to replace the temp and tacho, one showing SOC and the other showing current draw however this still leaves me with a tatty look and a mismatch of gauges as no one produces a white faced set of SOC or amp gauges with yellow needles.

I have settled on two items to replace the complete cluster, the first is a Acewell ACE-3963 Motorcycle computer. This is a complete digital instrument cluster in one simple package.

http://www.s3performance.com.au/ace...-speedometer-with-temperature-gauge-p-38.html

I am pairing this with the ZEVA Fuel Gauge Driver Plus http://www.evworks.com.au/?product=INS-ZEVA-FGDP which will drive the tacho as an amp meter, the fuel gauge as a SOC and also provide an adjustable “low fuel” warning light. This fuel gauge drive uses a hall effect amp meter to detect current flows up to 1200Amps (both directions) however all of the circuit is driven from the 12volt system keeping it totally isolated. Simple POTs and dip-switches are used to set it up.

 

[SplinterOz]

Looking into the Sevcon Gen4 controllers and have found (with the help of others) a good manual.
http://electricboatdesign.com/wp-content/uploads/2010/11/Gen4-Product-Manual-V3.0.pdf
Complete with wiring diagrams and programming.

Pages 5-10 and 5-11 cover the Hall sensors in detail. However nowhere in this do I see a maximum RPM limit...

 

[Biff]

Looking into the Sevcon Gen4 controllers and have found (with the help of others) a good manual.
http://electricboatdesign.com/wp-content/uploads/2010/11/Gen4-Product-Manual-V3.0.pdf
Complete with wiring diagrams and programming.

Pages 5-10 and 5-11 cover the Hall sensors in detail. However nowhere in this do I see a maximum RPM limit...

There is no Maximum RPM limit, it depends on the inducance of the motor and the control loop gains and such, but it is common start to get control loop erros (and shut down events) around 600Hz motor electrical frequency. With the low inductance of the colosus motor I don't think you will get much above 600Hz without having the controller shut down randomly. The only way to figure out what the sevcon Gen4 will really be able to do with the motor is to hook it up and tune it as best you can.

-ryan

 

[SplinterOz]

Ryan,

Thanks for the info... and thanks for the feedback on the motor's thread.

Based on your calculations on that thread it looks like up 4000rpm would be no problem for the Sevcon but above that it may get rough with the PWM not being high enough frequency to excite each coil evenly. I presumed that meant the the switching between coils is happening faster than the PWM frequency.

I also understand this "may not" be an issue as at higher revs we are applying higher average voltage and the PWM is having less effect. It may be possible to tune the controller to the motor and get good results. My problem is that there is no Sevcon distributor in my country any more and that means buying the controller and programmer and then learning it myself. This is a little risky for me.

I know the guys building the motor have used an Infineon based controller and a Kelly controller. I know my 24 fet Infineon can handle this motor smoothly up to 6,300rpm. I also know it will have a very hard time driving the low inductance windings of this motor to give me the low down torque. I know the guys building the motor got a Kelly controller to drive this motor mounted in a VW beetle and drive it around town. I don't know the max revs they got it up to.

I also have a 2nd problem with the Infineon based controllers, they have the wires hanging directly out of them rather than terminals to connect to. This is an issue as in Australia the rules I have to follow to get the bike street legal require me to use Orange cables for voltages higher than 60v. This would mean rewiring the controller or something.
3rd problem is the controller is a speed based controller and trying to control the torque delivery by hand is hard!

Based on the above I am thinking of getting a Kelly controller that peaks at 500 amps to the motor (250amp continuous). With the high speed option is will handle 70,000 erpm (motor rpm * poles) and has 16.6kHz operating frequency... however I am sure this is the microprocessor not the PWM. Maybe it will have the same issues?

Thanks

Tony

 

[Biff]

I believe you can just put orange heat-shrink over the cables and that will satisfy the regulation.

The high speed Kelly should be good. The one thing about the Kelly is that I don't think it does field oriented control, so torque will drop off as RPM increases, unless you use a "Hall Advancing" type of system with your own version of Field Oriented Control. The 16.6 kHz is probably the PWM frequency. If Kelly runs 16.6kHz PWM and go up to 70,000 ERPM (1.16kHz) that is still good for more than 10PWM cycles per electrical cycle which is the rule of thumb for decent control. I would hope their processor is running at least 20Mhz, 1000x faster than the PWM frequency.

-ryan

 

[SplinterOz]

I believe you can just put orange heat-shrink over the cables and that will satisfy the regulation.

Based on my reading of the Vehicle Standards Bulletin 14 Version 2...

All wiring in the vehicle connected to a HAZV battery pack (either positive or negative), or containing HAZV relative to the chassis of the vehicle, must be coloured orange even when installed within orange conduit.

The wiring must be "coloured" orange not just be "wrapped" in something orange. The Engineer working with the EV group here has confirmed that interpretation.

The high speed Kelly should be good. The one thing about the Kelly is that I don't think it does field oriented control, so torque will drop off as RPM increases, unless you use a "Hall Advancing" type of system with your own version of Field Oriented Control. The 16.6 kHz is probably the PWM frequency. If Kelly runs 16.6kHz PWM and go up to 70,000 ERPM (1.16kHz) that is still good for more than 10PWM cycles per electrical cycle which is the rule of thumb for decent control. I would hope their processor is running at least 20Mhz, 1000x faster than the PWM frequency.

-ryan

Ok in the next week or so I am going to put the money down on the Kelly.

 

[Arlo1]

The orange wire is about the dummest shit i have ever herd! I would be ok with orange shrink tube but come on! I would just orange shrink wrap it anyway and say its orange underneath but dont cut it open to see because its over 60v it will kill you

 

[SplinterOz]

Ok simple update on the instrumentation... it works... it works well and very easy to setup.

[youtube]rqj6wFIjPww[/youtube]

 

[jonescg]

"...and that's about all we can do before it gets cooked" :lol:

Seems brushless motors require pretty beefy controllers...

 

[SplinterOz]

"...and that's about all we can do before it gets cooked" :lol:

Seems brushless motors require pretty beefy controllers...

Well... yes, especially driving 22Nm on that little motor... and guess what we did next

We didn't cook it but it did go faulty. Now I have ordered a 72volt 500amp (peak) Kelly controller for my bike.

 

[SplinterOz]

After coming home today with my motorcycle frame ready to do some work on the wiring and general layout, I found waiting for me 3 boxes. In one of the boxes was my centralised minibmsto look after my batteries, this is a very neat little set of boards. The second was a radiator, reservoir and pump all build into one neat package from SwiftTech. This setup should be able to dissipate around 1000 watts of heat from my little motor.

Last, and definitely not least, is my new Kelly Controller. This controller is good for 500amps motor current for 1 minute and 250 amps continuous. Based on the motor spec (approx 0.124 kt) that should give a peak torque of 62Nm (approx 46 ft-lbs) more than enough to get me rolling. I performed a quick test with Mark (toolman2 on endless-sphere) with his smaller outrunner motor and we easily pushed it to 300amps to the motor (for a short time… the motor got a little hot) while applying a significant load.

 

[jonescg]

Nice! I love getting boxes in the mail. Especially e-moto stuff

 

[AussieJester]

The second was a radiator, reservoir and pump all build into one neat package from SwiftTech. This setup should be able to dissipate around 1000 watts of heat from my little motor.

Would be extremely impressed if it managed half that considering the cpus they are designed to
keep cool produce only 100-200watts of heat maximum... Think you need a larger heat exchanger buddy or
two of them. I run 2 similar rads in my pc and two pumps that produce higher flow rates, my cpu
manages a poultry 120watts (overclocked ~1000Mhz too)... Just cant see the setup you have working on that beast of a motor...
Hopefully i am wrong... Was right about the Infineon so im 1 for 1 so far LoL...

best of luck

KiM

 

[SplinterOz]

Nice review of the heat exchanger here... http://martinsliquidlab.i4memory.com/SwiftechMCR320-Review.html
Half way down this page is this graph.

Showing a 13 degree temp delta from air to water with a 1000watt thermal load (with the right fans). I may be mis-reading the chart but I think the exchanger should be fine.

 

[jonescg]

If it takes enough heat out of the controller to lower it's operating temperature more than what air alone would do, then it's surely better than nothing. Plus with this mounted in the front of the bike there is plenty of airflow to keep things cool. On a chilly Canberra morning I don't think it will have any problems.

 

[Arlo1]

Its for taking heat out of the motor not the controller right?

 

[jonescg]

Ah, yes it is.

Yeah, that will no doubt generate a lot more heat than a controller would. EV Works sells a slightly bigger radiator and separate pump: http://www.evworks.com.au/index.php?product=MIS-DAV-WRC

3/8 bore means a higher flow rate, so that could be an option?

 

[SplinterOz]

Yep taking heat out of the motor for this... I could develop a cooling plate for the controller as well but I don't think I will need it. I have seen the radiator on ev-works it is actually smaller (Overall dimensions: 226 x 64mm) compared to the Swiftech (128 x 429mm) but smaller pipes. The Swiftech I have but this is just a neat tidy package... 12volt pump, radiator and reservoir all in one package and not too expensive off ebay.

Time will tell if it will be enough, if I push the motor hard enough to even need it.

 

[x88x]

Well, that radiator is actually quite a bit smaller than the Swiftech one you got (like, ~6x smaller dry volume) and it has aluminum fins instead of copper fins on the Swiftech. It may have a higher fin density than the Swiftech one though, since the Swiftech is designed to work with relatively low airflow. Either way, if that one from EV Works will do the job, the Swiftech will be more than enough. I know in computers the flow-rate and tubing diameter really doesn't matter that much because once the coolant reaches equilibrium, neither of those has much if any affect on the coolant temperature. Normally PC's aren't putting out more than a few hundred watts of heat, but I imagine the same principals would apply. According to the stats, as long as the coolant does not pass 60 degrees C, it should be fine. And I think with the greatly increased airflow on the motorcycle instead of the low airflow normally used in liquid cooling computers, it should be fine dissipating the heat levels you're looking at.

 

[SplinterOz]

x88x, yea I just noticed that and edited my post above. I think it should work fine.

 

[AussieJester]

That heat exchanger is crap CHriS tis 'tube' style rad with ali fins no where near as efficient as the rad
Splinter is useing. The Swiftech stuff is top class double pass rads with copper fins and tanks
Alumnium v copper = copper wins for better heat transfer .Watercooling is more effective dissipating
heat than air alone is no doubt, long as the heat exchanger can cope with the load if it can't
it soons become less effective as there is a layer of hot water seperating the rest of the motor
rather than alumnium which acts as a heat sink.... There has been
a shit load of tests done on all the h20 PC cooling gear rads, pumps, flow rates and the difference
lapping blocks and some cpus even conclusion on tube sizing though is 7/16 and 1/2 in size
tubing make squat of difference....this is all in pc application mind you
dumping the heat into it Splinter will be doing it will definitely be interesting too see
how it goes, something i have wanting to do for awhile now but couldn't
justify the extra weight, on a motorcycle though its not so bad as a e-bicycle...if your
pumping 1000w through it continuously i still think it might struggle..
.. those tests were done using fans forcing air through at a constant rate, unless you plan on fitting
fans to the rad when mounted on the bike? your air flow won't be even close to the air pressures in
in the test application. I really do hope it works, I don't mean to sound negative
about the setup, fingers crossed for you Splinter give it shiiiIiIiiT

KiM

 

[jonescg]

I have seen the radiator on ev-works it is actually smaller (Overall dimensions: 226 x 64mm) compared to the Swiftech (128 x 429mm) but smaller pipes.

Ah, sorry didn't read the dimensions of it; 429 mm long! Where's it going to fit?