TIM600 Explosion

Technical discussion on converting internal combustion to electric
Josi
Noobie
Posts: 7
Joined: Sun, 16 Jan 2011, 02:06

TIM600 Explosion

Post by Josi »

I wrote an email to EVE, and they said that they are working on it, and I should try to start the tests with 140-150V, but that doesn't work. I'm still waiting for an answer from EVE! I think that motor is manufactured by FIMEA(italy), and EVE is just selling it! The motor ran to 7000 rpm because the potentiometer was on "full throttle"! What is a dyno?
Thanks for your comment Stiive!

Josi
User avatar
coulomb
Site Admin
Posts: 4403
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

TIM600 Explosion

Post by coulomb »

Josi wrote: What is a dyno?

That's slang for a dynamometer, a machine to measure the mechanical power from a vehicle.
Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
1.4 kW solar with 1.2 kW Latronics inverter and FIT.
160 W solar, 2.5 kWh 24 V battery for lights.
Patching PIP-4048/5048 inverter-chargers.
Josi
Noobie
Posts: 7
Joined: Sun, 16 Jan 2011, 02:06

TIM600 Explosion

Post by Josi »

Hello everybody!

So here's what happened in the last weeks!
Our motor works without making strange noises, if I keep the V/Hz ratio at 0.6, the highest possible voltage is 175V (at 175V/0.6 = 292Hz) if I go higher, the motor starts squeezing and the tests don't finish! Giving more voltage has almost no effect on the torque. Our problem is that our car runs only 60 km/h and the acceleration is not good! So we need more torque. Yesterday I had a conversation with the motor expert of EVE, he said the highest possible motor voltage is (battery voltage / sqrt(2)) in our case 240V/sqrt(2) = 169.7V! He advised me to keep the voltage at 105V and go down with the frequency in 5Hz steps, to the lowest possible value. After changing the frequency i should try the tests. He also told me that we should measure the highest possible speed of the motor when driving on the road. When we finished his advises he said he would contact an MES-DEA expert trying to get more torque out of the motor!
Yesterday I tried the tests with 105V and 145Hz, after finishing the tests without alarms, the throttle potentiometer wasn't working anymore! The potentiometer is working up to the soldering points of the 42-pole-plug of the controller, so my guess is that the analog-digital-converter is damaged!Image I also can't see any changes in the display menu (watching D50) when pressing the throttle pedal! Any ideas?
I think I will try to change the A/D-channel via software parameters and see what happens!
T2
Groupie
Posts: 121
Joined: Sat, 08 May 2010, 04:51
Location: ON,CANADA

TIM600 Explosion

Post by T2 »

It certainly looks like it would be useful if manufacturers were to specify pole pairs on the motor name plate, furthermore with the use of non standard stator windings it would helpful if V/Hz for nominal current was published and perhaps also for twice nominal current since we need to know what our vector drive should be doing when commanded to move in closer to breakdown torque during an extreme acceleration ramp.

It bears repeating that these torque curves are pretty meaningless since the torque rollover depends on the battery voltage chosen by the user.

One more thing I wonder who at MES thought it was a good idea to send curves for the TIM400 to the TIM600 users ? It doesn't inspire much confidence in end users - there are always problems in commissioning any drive.

You mentioned applying 292Hz (7000 rpm is 236Hz) and the car only reaches 60km/hr. In first gear or circa 12 : 1 ratio this would seem OK.

At the same time if the accel is poor then we need to know further what the current actually was during this ramp up. Either the battery or the motor current would be helpful.

I like the idea of overspeeding but you may want to get the rotor balanced for it. I've driven an off the shelf 1-pole pair machine with 120Hz and it squealed pretty good but stayed intact during a run lasting a couple of minutes. I've also lost control of a machine at high speed when the tach input saturated. A pulse wheel driving a magnetic sensor triggered a one-shot. When the input frequency caused the pulses to coallesce, the resulting continuous DC voltage indicated lack of feedback and the drive just took off !
Last edited by T2 on Fri, 25 Feb 2011, 03:12, edited 1 time in total.
T2
Stiive
Groupie
Posts: 295
Joined: Wed, 31 Dec 2008, 08:26
Location: Melbourne

TIM600 Explosion

Post by Stiive »

Hi All,
Thought i'd give you an update of where the Swinburne's vehicle stands with it's problems with the TIM600.
I have always been concerned that such a large capacitance is evident between the low voltage (LV) and high voltage (HV) systems. This is caused by the Dow-Kokams having such a large surface area with only a fine lamination between the cathode/anode and the aluminium sheet the battery rests on. This causes a capacitance between the cell and the chassis. This is then multiplied by the 82 cells mounted like this. The 12v LV system is also grounded to the chassis as per the competition rules. Therefore a large capacitance exists between the low voltage and high voltage systems of the vehicle.
The LV system controls the "brains" of the controller and gate drivers of the IGBT, and it is thought that the capacitance is causing noise on the LV system when switching large currents on the HV systems.
The capacitance varies at voltages upto 300V and is large enough to create a spark between the HV and chassis even though they have no other current path. The capacitance can also be felt as a contant tingle when accidently bridging your body across the HV and chassis. Even though the capacitance can be 300v, placing an LED between the HV and chassis causes the LED to blink quite brightly every second as the capcitance builds up and then discharges. It is never enough to blow the LED however.

It is also worth noting that on footage captured when a IGBT blew, an arc was seen about 25mm between the IGBT and controller casing! It is believed that the noise created by capacitance is causing the IGBTs to misfire, creating massive build ups of voltage which stress the IGBTs reverse voltage threshold.

I have had Barney from Semikron come out and have a look at our system and this was his thoughts on the situation:

"Hi Stefan

It was good to see you and your project.

Attached are the notes that I made during our meeting, as well as a
relevant section from our Application Manual Power Modules section 3.4
Power Design. You can download the whole manual from
http://www.semikron.com/skcompub/en/SID ... ication_ma
nual-193.htm

See fig 3.25 Cause of interference currents (also fig 3.23 parasitic
elements).

The MESDEA converter construction is not ideal, too much stray
inductance in the commutating loop, but I believe that your problem
relates to the capacitive noise propagation paths. Check the dv/dt
rating of the optocouplers in the gate drivers and the interwinding
capacitance of the crappy pulse transformer used to provide isolated
drive power to the gate drivers.

This will be exacerbated by the serious problem of battery capacitance
to ground and possibly motor winding capacitance to ground. There is no
series inductance in the motor leads to help alleviate this.

In addition, the Fuji IGBT packages have comparatively high internal
inductance, compounded by the poor construction of the MESDEA
controller.

All of this is indicative of the type of problems that you are
experiencing, causing noise resulting in loss of IGBT control in 3
controllers.

You have to measure the IGBT currents with a fast non-inductive
non-invasive probe such as a Rogowski coil and also measure the Vce and
gate drive voltages and then work back to find the cause of disturbance.

Best regards
Barney Meyer

Managing Director

Tel: +61 3 85615605
Fax: +61 3 95618769
Mob: +61 409 522093

www.semikron.com
barney.meyer@semikron.com
SEMIKRON PTY LTD
8/8 Garden Rd, Clayton, VIC 3168 Australia

Dear Barney,
Thanks for taking the time to come out yesterday. You have certainly made some very interesting and thought provoking points.


To produce valid test results while protecting our equipment, which of the following would you suggest;
- Reducing the DC bus voltage to about 250V. Hopefully this will limit the voltage stress on the semiconductors
- Placing a load in parallel across the DC bus. Will this effect readings?
- Placing a series inductance in the motor leads
- Isolating the 12V system from the chassis to eliminate the capacitance between the HV and LV systems

We stumbled across some other interesting information last night when watching one of our videos where an IGBT failed. In the video it can be seen what is believed to be an arc created between the IGBT and the casing. This is further verified by markings left on the inside of the casing. This arc would have been at least an inch long, which would have meant massive voltages!

If you were still planning to be around Swinburne this afternoon, I would really like to run some more information past you to gather your thoughts.

Thanks again for all your help so far.

Kind regards,
Stefan



Hi Stefan
An arc between the IGBT and the casing is a significant event. Let’s think about that problem out loud:

1.       An arc indicates a significant voltage source behind it and is usually created when current flowing in an inductor is interrupted or an interrupt is attempted.

2.       On the DC side you have a battery, some short leads (not high inductance, only microhenries), a DC fuse, capacitor external and internal. That’s quite a high energy supply with lots of capability to deliver fault current. This is typical of any high energy IGBT converter. We specifically test our converters for failures under these conditions because one day, at the end of life of the IGBTs, we are going to have a short circuit on the DC bus when the IGBTs die and crowbar the DC bus. We don’t want a ground fault under these conditions, or a package explosion. What happens is that the IGBT bond wires onto the chip act as fuses, open up and clear the fault without explosion or arcing or ground faults. I must assume that the Fuji devices work the same way because all isolated IGBTs have bond wires.

On the AC side you have leads to your motor and motor winding with high inductance. You should not have any capacitance in the leads of motor winding to ground (because of the high currents induced into the stray capacitance by the high dv/dt output of the IGBTs). The inductive motor current may not be interrupted in any way, even briefly, because this would cause destructive arcing due to high voltages. That’s why the IGBT bridge output is firmly clamped to the DC bus by freewheel diodes. During your PWM there are short instances when the IGBTs are all off (2-3µs). During this time the inductive load current commutates to the DC bus via the FWDs and the motor current is uninterrupted.



Therefore, most likely if you are seeing an arc like the one you mention, from IGBT to case, this indicates that the motor current has been interrupted. Likely that means you have had an IGBT shoot-through (top and bottom IGBTs on together – maybe due to the postulated noise interference in the control), or an IGBT has died due to an overvoltage spike and you have had shoot-through (shouldn’t actually happen with a good control system – should detect a shorted IGBT and prevent the other one from coming on to create a short). This high fault current blows open the IGBT wire bonds and suddenly you have no conducting path for the current in the motor winding hence the flashover, which possibly has also damaged your motor insulation.



Best regards
Barney Meyer"





We are now going to do some tests this week to hopefully get the car running by the weekend. The car will be on show at the Top Gear Live festival this weekend, and will hopefully be able to preform a few laps.

Some members of the AEVA and MEVIG have been kind enough to loan some IGBTs, motor inductor set and lead acid batteries to do some testing of the motor and controller. Hopefully we can clear the motors name and clarify whether the the capacitance is to blame for our problems. I have proposed the following test procedure:

"It would also be great to conclusively test whether the capacitance is causing misfires on the IGBTs. This is the test procedure i think we should follow;

1. Put Kevin's IGBTs in the CERES old TIM600 unit.

2. Put the controller phase leads onto a purely resistive load (perhaps the 3-phase load in the green lab).

3. First run the test with the controller powered by James' external battery pack and monitoring the IGBTs for misfires with Kevin's meter - If all goes well the controller will run as expected with no misifring.

4. Run the same test with the controller powered by the batteries in the vehicle with all its capacitance - We should see IGBTs misfiring, but this should cause no over voltage on the IGBTs due to a purely resistive load (minimal inductance).

5. Remove the capacitance between the LV and HV systems of the vehicle and run the test again - If we are truly blessed, there should be no misfires :)

It is possible the misfires only happen when the controller is pushing out lots of current (due to bigger fluctuations in voltage in the HV which is then linked to the LV via capacitance). I believe the 3-phase load in the green lab has a minimum of 10Ohms, perhaps we need a smaller resistance? Thoughts?"



Any thoughts or comments would be grealty appreciated!! Does anyone think we're on the right path?
Am I safe in assuming the IGBT's will be safe running on a purely resistive load? Perhaps some inductance is needed to limit di/dt, any thoughts on a ideal value?


Stefan
Last edited by Stiive on Mon, 07 Mar 2011, 06:42, edited 1 time in total.
Rgds,
Stiive
Josi
Noobie
Posts: 7
Joined: Sun, 16 Jan 2011, 02:06

TIM600 Explosion

Post by Josi »

Hi,

I finally found the failure with the throttle! After a reset to "factory defaults" the parameter C64 ("enable current control") is set to 0, but must be set to 1! I wrote an email to MES-DEA, and sended them all of the documents about the EVE M2AC30 motor, to get the right parameters.

Here's what he answered:

Dear Mr. Josi,
I never look this motor
The data about the motor that you have send me are not the nominal data and the chart are in contradiction with this. The motor have any label?
I think you must ask the data to the supplier of the motor.
If this is not possible I can estimate this data for you but without responsibility!

For enable the throttle please set C64=1

Best Regards

My answer was:
We are driving with 105V 175Hz 105A 4 poles at the moment, the tests of your software (C41, C42) pass! The motor runs very silently! But our problem is that we don't have enough torque, so the acceleration of our car is very bad! What do you think we should do to get more torque out of it? What parameters would you enter?
If I enter the data of the label, the motor is squeezing very loud, and does not turn. It also becomes very hot! We get the best results when keeping a ratio of 0,6V/Hz! So 175Hz * 0,6V/Hz = 105V! The tests pass also with for
example: 175V, 292Hz (0,6 V/Hz). If I go higher with the voltage, the motor starts squeezing again!

Best Regards,
Josi

MES-DEA answer:

Sorry for delay,
The data of the motor are incomplete!
I advice you to setup these parameter:
P60=95
P61=70
P62=121
P63=100
P64=120
After this remake the autotuning C42=3
finished the test save all!
Please send me the new RAM Data Report for check!

Best Regards

MES SA

My answer:

Hello,

with your setting the test C41 works, but the test C42=3 does not work!
In the test C42, the motor turns up the first time, then it breaks down, after that I get the errors A07 and A15! The test does not finish! My actual ram data report is attached to this email!

Thanks for your help

Josi

MES-DEA's last answer was:

Please,
set
P60=95
P40=150
P101=7500
P61=100
P62=82
P63=100

After this remake the autotuning C42=3
finished the test save all!
Please send me the new RAM Data Report for check!

Regards,
MES SA



Maybe it could help you with your Swinburne vehicle!
I talked with EVE about the motor name plate, they say all values are correct!

Our problem is the acceleration of the car! It's very poor!

The parameter P78 shows the nominal torque of the motor, it' s calculated in the test C41. I think the software calculates it with this formula:

M = (U*I*sqrt(3)*p*eta)/(f*2*pi)

M = torque[Nm], U=motor voltage[V] , I = motor current[A], p = number of pole PAIRS, eta= efficiency factor (about 0.87), f = motor frequency[Hz]

So my question is, how can I get more torque out of the motor?
I have to keep the V/Hz ratio of 0.6! The motor curve shows a torque of about 190 Nm. With 205V, 175Hz, 2 pole pairs 105A , the current would be 338A! How could they get 338A out of a TIM 400 controller? The max. output current of a MES DEA TIM 400 inverter is 280A!

Any suggestions?
Post Reply