Ansaldo Motor

[johu]

I think if the motor is so sensitive you should make use of the udcnom parameter sooner or later. Otherwise your tuning is off on battery sag.

[piotrsko]

Not saying anything, but FWIW: if it's delta wound wouldn't it use more motor current? I am under the impression delta is for low voltage. Actually sounds like the pulse width to the motor is rather off

[Jack Bauer]

So I had a chance to do a little more tuning today. This project is turning into a lot more work than I expected / wanted

The motor really does not like low rpm. it just wants to suck amps and max the current limit while not delivering much by way of torque. Now, when it gets moving you can feel the power come in. At 3k rpm it pulls like a train but getting there is a pain. I'll shoot some video during the week which should make more sense and keep on tuning. Of course the Tesla 10kw charger I fitted also decided to try my patience but that's for another thread

[Jack Bauer]

Making progress :

[arber333]

Making progress :

Hi D

What happens if you give it more boost parameter? I have a rather low inductance motor too and it demands sharp minimal slip 3Hz at rather high boost setting to start smoothly. Like 7000 boost at 380Vdc! If i provide less boost motor stalls at low throttle and i have to add like 1/4 more throttle to take off. Current is close to max then. But as soon as the car gets moving its seems like releasing a rubber cord. Motor starts to accelerate quickly. I would remove some throttle then , but then it stalls again and back and forth. I think this behaviour is what caused so many broken clutches. Flywheel would compensate rocking but poor clutch got banged up. Once i went to direct link i could feel when boost is enough and everything is good now.

EDIT: I use UDCnom parameter extensively and voltage sag never bothers me now. This is useful especially now in winter.

A

[Jack Bauer]

Thanks Arber. What you say seems quite familiar. Here is some data on the original Ansaldo motor and inverter :
http://www.batteryvehiclesociety.org.uk ... 0-60kW.pdf

I went for a longer drive today including motorway. Here are the main findings :
Still very little torque at sub 1000RPM. Have to use first gear when pulling out onto a main road and even then is quite sluggish. Only with increasing rpm can I add throttle without hitting current limit. This is the main problem I need to solve.

Hill starts are very tricky requiring first gear and very gentle throttle to pull away.

Even at higher rpms adding throttle gives better but still sluggish performance.

If the car encounters a hill, rpms back down and attempting to add throttle to maintain speed hits the current limit.

[arber333]

Like i said you probably need way more slip for takeoff.

Try using 3Hz for a fslipmin and develop from there. I have seen designers use variations on 3 there. Like 3Hz or 3.3Hz or 3.6Hz etc...

Also since you have 200Vdc you should consider adding more boost. Try 7000. If there is too much boost inverter will throw desat.

Change udcnom to 160V motor should feel softer there so you can use more minimal slip.

You could upp the minimum amps. Now motor should have a kind of rubber effect with such small amps for start. Maybe you dont even get across rotor resistance and you have to add too much throttle. I use 21 for minimum amps. Car starts really good.

With such a large slip on start inevitably there is some jolt when going over 0A threshold. But with good coupling and large car the mass should dampen the surge.

Also i noticed you use 70 for your deadtime!!! I find that way too small. There is a possibility of shootrough with DIY drivers. If you use original inverter drivers it should be even slower. I use 115 on Ampera inverter and 185 with semicron Skyper 32 drivers i use in Mazda.

[johu]

I just thought of another systematic tuning method if all else fails:

Set boost=500
Pull the handbrake, go into manual mode. Be ready to shut down the inverter should the car decide to take off.
Set ampnom=100, fslipspnt=0.5

Now increase boost until you hit your desired current limit. Try to quantify how hard the car pulls against the hand brake
Now decrease boost back to 500, increase fslipspnt by 0.3Hz and repeat. If 500 is a stupid starting value, increase it.

At some point there has to be low end torque without tripping. Whatever fslipspnt gives you the best starting torque at lowest current is your fslipmin.

By using this procedure you eliminate the uncertainty of using the throttle pedal.

@arber: I'm using deadtime=63 for DIY drivers and SKM400GB12 (1200V/400A) with no shoot through.

[Jack Bauer]

Another day, another tune. Good news is we now have zero rpm torque. Enough to break free of the handbrake in 1,2 and 3rd gear so big improvement. also, the car will now move off using only 400-500W of battery power. Parameter file attached.

Now the bad news. I simply cannot get linear acceleration. I now have very nice initial starting acceleration to about 10mph. As I then add throttle to accelerate I hit a wall in the shape of the current limit. Backing off throttle is then required resulting in very poor acceleration. If I can get through the "wall" into a higher rpm band then I do get reasonable power and can hold speed on a hill climb etc.

So the bad news is this makes the car unsafe to drive on the road. Banging my head on a brick wall.

[Peter]

Hi Damien. Which version of software are you using ? I am trying to mimic what is happening but may be on different software possibly ?

[Jack Bauer]

v4.9 but its the motor that is particular to my case.

[Peter]

I understand motor is different but if your parameters give similar effects then maybe its a good test. Have you tried 4Khz as it says on the datasheet? Thinking about parameters, if you put the same into the drift car and see what happens as its your favourite Enova motor ? Asked Johannes about iacmax too, you have that set lower by another 50 amps today, wonder if that has any affect on currents and throttle, waiting for Johannes to reply now

[arber333]

I just thought of another systematic tuning method if all else fails:

Set boost=500
Pull the handbrake, go into manual mode. Be ready to shut down the inverter should the car decide to take off.
Set ampnom=100, fslipspnt=0.5

Now increase boost until you hit your desired current limit. Try to quantify how hard the car pulls against the hand brake
Now decrease boost back to 500, increase fslipspnt by 0.3Hz and repeat. If 500 is a stupid starting value, increase it.

At some point there has to be low end torque without tripping. Whatever fslipspnt gives you the best starting torque at lowest current is your fslipmin.

By using this procedure you eliminate the uncertainty of using the throttle pedal.

@arber: I'm using deadtime=63 for DIY drivers and SKM400GB12 (1200V/400A) with no shoot through.

You know my position about deadtime, i am not budging! I rather leave some empty space in PWM than burnt IGBTs in my bin. God knows i did lots of them at one time... Since then my car runs great and inverter is doing a great day to day job.

Nice procedure for testing minslip on new motors. I think i will try it soon .

Tnx

A

[arber333]

Another day, another tune. Good news is we now have zero rpm torque. Enough to break free of the handbrake in 1,2 and 3rd gear so big improvement. also, the car will now move off using only 400-500W of battery power. Parameter file attached.

Now the bad news. I simply cannot get linear acceleration. I now have very nice initial starting acceleration to about 10mph. As I then add throttle to accelerate I hit a wall in the shape of the current limit. Backing off throttle is then required resulting in very poor acceleration. If I can get through the "wall" into a higher rpm band then I do get reasonable power and can hold speed on a hill climb etc.

So the bad news is this makes the car unsafe to drive on the road. Banging my head on a brick wall.

Why dont you try 10Hz as minimal slip like the dataplate suggests? Use 160Hz as fweak point? That way you taper off initial surge. Also you can use lower voltage for udcnom to keep the rotor excitation feel softer. We have a small motor that demands 7.4Hz to start comfortably. It is a small motor though...

[Jack Bauer]

Thanks Arber. Will try tomorrow. Here is a video from today :

[sonachan1]

Hi Damien. I watched your latest video last night. It got me wondering if the overcurrent events you're experiencing are real. Maybe there's something going on with scaling or logic in the firmware? Could Johannes make a special cut of the firmware with some logging of relevant variables built in which you could use to analyse the problem in more detail? Having some internal data captured might help anyway?

[arber333]

Hi Damien. I watched your latest video last night. It got me wondering if the overcurrent events you're experiencing are real. Maybe there's something going on with scaling or logic in the firmware? Could Johannes make a special cut of the firmware with some logging of relevant variables built in which you could use to analyse the problem in more detail? Having some internal data captured might help anyway?

I use the same firmware version and last month when i tried to program another motor i got similar results. I am sure its just bad parameters. Otherwise it wouldnt work at high RPM.
I had similar results with my motor back when it was wound for 75Vrms and i used 200V to run it!!! I had to use fweak 185Hz so it wouldnt trip oclimit at start. I think too low "boost" is coupled with small "fslipmin" and too small "ampmin". Rotor needs lot of boost voltage to overcome resistance and it needs to strech magnetic field to develop torque. Ampmin also has to be higher to help when boost effect subsides and current moves the rotor in slip mode...

BTW JB do you feel your motor is heating a lot? When i had slipmin set to 2.4Hz my motor was heating up quite a bit at higher RPM. The moment i set it to 3Hz and provided 7000 boost heating subsided and motor actually likes to run at 5kRPM.

[Jack Bauer]

Problem solved Details to follow.

[Jack Bauer]

Thanks everyone for all the help and support:)

[johu]

Finally, congratulations! Good thing I showed you the limiting code. Seems it really needs to be communicated more clearly what it actually does.
Give us the parameters!
I just allowed uploading with json extension, no need to zip anymore.

[Jack Bauer]

Parameters attached. I still couldn't upload the .json file

[johu]

Argh have to enable everything twice... Lets test

params_ansaldo_e39_driving_10_1.json

(1.25 KiB) Downloaded 186 times

So field weakening at 4000rpm, sounds sweet What power can you draw from the battery?

Regarding a comment from the previous video, I'd say scalar control FTW

[Jack Bauer]

So far I'm seeing about 45kw max battery power but I think I can get more just needs some fine tuning.