MES-DEA TIM600 explosion! Help!

Stiive said:

Dear All,
I am currently developing a formula spec electric car as part of a university project. Our team is using a MES-DEA TIM600 inverter and has recently plugged it up to our EVE M2-AC30-L motor.
Unfortunately when trying to spin the motor for the first time using the set-up instructions in the TIM600 inverter manual, there was a large explosion and flames from the inverter! This was encountered when doing the test to verify the number of poles of the motor and check that the 3 phases were correctly wired in... The motor never spun.

I spoke to another group who had used a TIM600 and they said when enabling the traction for the first time while following the set-up instructions, a large unexpected "squeal" came from the inverter, we had a similiar issue with a small popping noise and some light coming from the inverter making us think we had blown the inverter or a fuse. We turned the invereter on later and everything seemed to work fine. We then later began the test (about 2 steps later than the original 'pop' in the instruction manual) when the large explosion happened

[Edit cut bit out to keep post under 10k Chars]

Does anyone have experience with these controllers and tell us what possible causes might be of the explosion? Do the tests even work or are they the root of the cause? Possibly it was just a faulty inverter?
Is there anyway we can test our 3-phase induction motor? would hate for that to be the reason and therefore blow up our replacement inverter as well!
[Edit cut bit out to keep post under 10k Chars]
[email protected]

Click to expand...

Just an update of what happen and where they got up to.

I was able to spend some time with the team at their workshop and looked over the system and motor and controller information.

The Motor data
Motor type: M2AC30L
Service: S21H
P: 30kW
V: 210V
I: 105A (Could be 108 ?)
RPM: 5000RPM
Freq: 175Hz
RPM Max is 7000RPM

(This suggests V/Hz of 210/175 = 1.2 V/Hz)

And the Test report supplied with the Motor showing the same serial number has a table of RPM,TORQUE,POWER. Tested sing a TIM400 and 96 volt DC Supply.

"speed":"torque":"power"

"rpm":"T (Nm)":"P (kW)"

0:119.3:0

500:118.4:6.2

1000:117.5:12.3

1500:116.1:18.2

2000:115.3:24.1

2500:112.5:29.5

3000:101.4:31.9

3500:87.2:32

4000:75.4:31.6

4500:65.4:30.8

5000:56.1:29.4

5500:46.2:26.6

6000:36.2:22.7

Click to expand...

This data plotted looks like:


From this the V/Hz can be estimated assuming they ran at Peak power at Max Vac of Vdc/√2 = 96/√2 = 65 Vac

The Mech RPM 3000RPM so the Hz is Nmech x 2.¶/60 x 2/Poles [Edit: oops should be Hz = Nmech x 4/2 x 1/60 = 3000 RPM x 2/60 = 100Hz]
So 3000 x 2x3.142/60 x 2/4 = 157.1 assuming 4 pole [ Edit:100hz at 3000 RPM]

So V/Hz = 65/157.1 = 0.413 [Edit:65/1000=0.65V/Hz]

This much lower than the name plate data suggests.

But also, in the graph you can see the rounded power curve indicating that at 3000 RPM to 5000 RPM the power doesn't drop off fast like in the store graph image:


So the inverter is over-volting the motor and increasing the frequency up to 5000 RPM but limiting the current to with in the controllers capability.

So considering the test data supplied wth the motor, if the Vac at 2500 RPM is estimated to be 1/2 the Vac at 5000 RPM
the Volts would be 65/2 VAc at 2500 RPM

So the V/Hz would be 32.5/ (2500 x 2.¶/60 x 2/4) = 32.5/65.4 = 0.496 V/Hz [Edit:Hz is 32.5/(2500/30)=0.39 V/Hz]

So what is the correct V/Hz to stop inductive saturation in the motor?
The inverter doesn't have a specific parameter for V/Hz it uses the Nominal Motor Parameter data entered in.

So using
Motor type: M2AC30L
Service: S21H
P: 30kW
V: 210V
I: 105A
RPM: 5000RPM
Freq: 175Hz

Gives V/Hz = 210 / 175 = 1.2 V/Hz
But if the "V: 210V" is really Vdc
then the max Vac would be 210/√2 = 148
So the V/Hz would be Ksat = 148/175 = 0.845 V/Hz

After much fiddling around the Nominal Values V/Hz of 0.62 was showing slight buzzing so backed off to have 0.6 V/Hz by setting the Vnom back to 105Vac the motor passed the C41 test checking motor UVW connection and encoder feedback and Number off poles.

The Motor data that passed C41 was:
Motor type: M2AC30L
Service: S21H
P: 30kW
V: 210V (entered as 105Vac)
I: 105A
RPM: 5000RPM (Stator = 5250 = Fnom x 60*2/Poles so Slip RPM is 250RPM)
Freq: 175Hz
RPM Max is 7000RPM
(Poles 4)
(Encoder 64Pulses/Rev with corrected AB connection)
(Note: V/Hz = Ksat = 105/175 = 0.6)

So what is the Motor capability?
The test they did with a TIM400 using 96Vdc Supply showed 30kW
So if they ran at 3000 RPM and got 100Nm the Power would be
P = T.w where w is the mechanical RPM in rad/sec w = N 2¶/60
Tw = 100 x (3000 x 2¶/60) = 31kW which makes sense.

The motor info on the shop.electro-vehicles.eu for the
M2-AC30-L

http://shop.electro-vehicles.eu/shop/details.asp?prodid=EVE02&cat=0&path=47,60

This shop site data shows 59kW at 3000 RPM with 188Nm
This calc's to P = Tw = 188 x (3000 x 2¶/60) = 59kW

The site info has

Main Features:
Nominal voltage: 288VDC
nominal speed: 3600 rpm
top speed: 8000 rpm
nominal power: 30Kw
peak torque: 200Nm from 0 to 2500 rpm
alluminum construction
Liquid cooled or air cooled options available
Protection class: IP 67
SKF Bearings and encoder

Click to expand...

This suggests V/Hz of 288/√2 / (3600 x 2¶/60 x 2/4)
For V/Hz = 203 / 188.52
giving = 1.07

Using the Vnom as Vdc rating is confusing as Vdc/√2 may not be the capability of the inverter. The PWM generation of 3 phase may be limited to less than the Vdc on √2 as there are Volt drops in the IGBTs and sage in the Bus due to large currents. So the Vac generated may not have been the best possible of 203 but only 190 say,
then the V/Hz would be 190 / 205.2 = 0.99

If the Online Vnom was supposed to be VNom = 288Vac (not Vdc)

Then V/Hz would be 288/188 = 1.53 V/Hz

So the Motor data supplied doesn't line up with the observed V/Hz of 0.6.

But the Motor Passed the C41 test with using the Nominal Voltage of 105 Vac and Nominal freq of 175Hz.

So what happened next, the motor was run through the C42 test and passed. This was done with Nominal speed of 1000RPM. This was repeated a few times to check repeatability.

Then the Chain was attached to drive the axles with no wheels.

The speed was limited to 80% of 500 RPM instead of 5000 RPM or even 1000RPm, to keep speeds low and the ramp accel time for test was P121=4sec the default)

The C42 test runs through
- phase 1 - with no movement
- phase 2 - Motor moves at creep speed
- phase 3 - Motor spins up to 80% Nom Speed
- phase 4 - Motor spins up to 80% 16 times

During the phase 3 the Motor Spun up to the reduced speed setting then after about 6 seconds, when the motor would start to slow down the Inverter failed (Another IGBT popped) and of the battery back fuses blew.

So what would cause this?

My thoughts are the there is an issue that running the inverter with a 82 cell Kokam pack, deceleration of the motor caused the return current (de-magnitization current) to over volt the IGBT capability of 650V (absolute max from the spec sheet found on line for the device).

This is protected by the Snubbers (WIMA FKP 1uF) and main Bus Capacitance 45uF x 6 = 275uF.

The Battery back can't be relied on the absorb this energy as the Safety contactors could be released or the Fuses could blow disconnecting it.

So all spike protection needs to be linked to the Controller DC Bus to absorb Turn-Off spikes on the Bus.

As the inductor magnetic field decays the current freewheels via the IGBT diodes onto the BUS, charging the capacitance available. The snubbers are small at 3 x 1uF and the 275uF from the 6 x 45uF caps is small too.

This decay energy must go somewhere. And the bigger the bus capacitance or accumulator, the absorption of this energy will stop it shooting up beyond the IGBT limits.

If there was no BUS capacitance or battery connection then the motor terminal Voltage will go so high that an arc through the air will occur.
And the humidity in the last few days would not help either.
It would be better to be in an area with below zero temperatures causing humidity to negligible.

The first Controller failure was very dramatic and Arcing did occur causing melting of thick bus bars.

So operating at lower battery voltages and with a motor with lower inductance and operating at lower currents would all help reduce the size Turn-Off voltage spike.

Often motor controller have Ultra fast Zener Diodes (Transorbs or other brands) across bus and to Motor Terminals. These are selected to turn on before the IGBT limit is exceeded. This would protect the IGBT by absorbing the energy released by the Motor and stray capacitance and magnetising energy.

Operating at below 300V would keep the Stray capacitance from over-volting the IGBT and would also allow more delta Voltage for the Capacitors to absorb the energy from the current charging the capacitor.

So adding a larger Bus cap externally like a 400V 450V surge 4700uF Electrolytic would provide a store for released energy from Bus Votlage at 82 x 4.2 = 344 to 450Vdc

so U_C at 344Vdc is V^2 x C/2 = 278 Joules

And at 450Vdc is 475.875 Joules allowing (475 - 278=) 197 J to be absorbed.

Compared to (275 + 3) uF on the Bus providing up to the 650V max of the IGBT limit would be 59.3 - 16.3 = 43 Joules.

The inductor energy of U_L = 1/2 L x I^2 could be used if the motor inductance could be evaluated.

If some of this sounds wrong please reply or PM stiive with any questions or recommendations or thoughts.

7¢, Ken

Hi Ken,

Nice report So I read it and then was going thru for a second read. I have to do some other things today, but I'll be thinking about this. Thanks

Anyway for now, I noticed this:

7circle said:

The Mech RPM 3000RPM so the Hz is Nmech x 2.¶/60 x 2/Poles
So 3000 x 2x3.142/60 x 2/4 = 157.1 assuming 4 pole

Click to expand...

When working with RPM, Pi does not enter into the conversion, like when using rad/sec.

Electrical frequency (Hz) = mechanical frequency (RPM) * (#pole pairs / 60 seconds per minute). So 3000 RPM is 100 Hz, synchronous frequency.

Hope to get back into this, perhaps this evening. Yeah, Saturday nights are real exciting around here

major

Hi Major Thanks for that pick up on my Hz calc.

I stuffed to bits that made the calc look correctish...

I edited the post above.

The team still needs to sort out the problem.

So any comments woul dbe helpfull

A VDR was put across the DC bus terminals to kick in at 350V there were 3 in parallel to allow 300A at 710V. Just so they would nlt blow and make a mess inside the case.

A pair inseres of Electro Caps at 4700uF / 450 Surge where also linked in on short leads to the Bus terminals.

A precharge and post discharge circuits with 500 ohm where used as well.

The only changes I can think of still are to wire the batts to the Caps then to the bus, add VDRs or Tranzorbs to the phase terminals on the IGBT modules.

The inverter controller software lloks like it was setup to be used as a wind turbine controller, so there are extra parameters that make things extra confusing.

The Zebra battery backs that Mes-Dea build may have capacitors in them to help them operate.

I am worried that the DC bus at 350 Vdc is just too high for the 600V IGBTs and using 290V pack would be a losser fit for the IGBT rating.

We could replace these IGBT with higher rated Modules.

7circle said:

We could replace these IGBT with higher rated Modules.

Click to expand...

Hi Ken,

You could do this. Likely you have 600V rated modules now. And it is possible that 1200V modules would have the same form factor. They would have different characteristics so I am unsure that gate drivers, dead bands and such carry over without modifications. I did a rather large inverter years ago for a 350V battery and had to use 1200V modules because the high current devices were unavailable in 600V. I always thought that was a blessing in disguise because we had a poor bus design and some snubber problems, yet the IGBT modules held up.

I did not know this used the Zebra battery. Don't know much about that. Is it actually built to accept regen? And to get the inverter tuned and set up, can you not just turn regen off? Ones I've used can. Then they just coast. If for some reason, you get an overvoltage condition, the control faults out and shuts down and protects itself. I am surprised your inverter doesn't do this. Which leads me to believe this is not your real problem.

As far as capacitors go, all inverters of this nature of which I am aware will have sufficient DC bus capacitance internally. In my opinion, adding external capacitors and voltage suppression devices will do little good. Like putting band aids on the body without knowing where it is bleeding.

I am also wondering why the dealer or manufacturer has not been brought in to deal with this.

I am worried that the DC bus at 350 Vdc is just too high for the 600V IGBTs and using 290V pack would be a losser fit for the IGBT rating.

Click to expand...

I was recently working with an inverter using 600V modules and we took the battery up to 400Vdc. We had no problems, well, at least, not in that area But does bring to mind a test. Can you borrow a lower voltage lead acid or lithium battery for a test?

Regards,

major

major said:

...I did not know this used the Zebra battery. Don't know much about that. Is it actually built to accept regen? And to get the inverter tuned and set up, can you not just turn regen off? Ones I've used can. Then they just coast. If for some reason, you get an overvoltage condition, the control faults out and shuts down and protects itself. I am surprised your inverter doesn't do this. Which leads me to believe this is not your real problem.

As far as capacitors go, all inverters of this nature of which I am aware will have sufficient DC bus capacitance internally. In my opinion, adding external capacitors and voltage suppression devices will do little good. Like putting band aids on the body without knowing where it is bleeding.

I am also wondering why the dealer or manufacturer has not been brought in to deal with this.

I was recently working with an inverter using 600V modules and we took the battery up to 400Vdc. We had no problems, well, at least, not in that area But does bring to mind a test. Can you borrow a lower voltage lead acid or lithium battery for a test?

Regards,

major

Click to expand...

We found the spec on the Modules fe Frontrunnners from a division of Fuitsu
http://www.fe-frontrunners.eu/Semi/Datenblatt/IGBT/2MBI600U2E-060a.pdf

The dealer has been communicated with by Stiive, but no solution found.

They tested the motor with a TIM400 at 96Vdc to 15kW and up to the RPM I showed in the above graph.

The Cart uses 82 30Ah Kokam Cells for a 82 x 3.6V = 295V Nominal and at full Charge at 82 x 4.2 = 344.4V

I can see that the chemistry could have confused the situation if LiIronPOOOO was used, max pack voltage would have been 82 x 3.2V/cell =262V. And Max at 3.65V/Cell is 299.3V

The Inverter Controller has a section P106 - P109 that are settings for DC Bus voltage that are Over
P t 106 Minima tensione del Bus in continua 180.0÷500.0 400 Volt
P t 107 Massima tensione del Bus in continua 300.0÷1200.0 760 Volt
P t 108 Soglia freno ON 300.0÷1200.0 730 Volt
P t 109 Soglia freno OFF 300.0÷1200.0 710 Volt



So should the Inverter be rated to 400V if it uses 600V IGBTs?

Thanks Tesseract (Jeff)

Stray inductance and stray capacitance are hard to quantify without looking at it closely on a scope. This still can be done accurately, but normally needed by a DIY conversion.

To help stop spikes MOV's were fitted (not VDRs ). Three K375 types that can take 100A at 700Vdc each to abasorb pulse energy where fitted on the Bus terminals.

The 20mmx8mm bus bars will cause some stray inductance on the DC side.
With the three WIMA FKP 1uF on each IGBT pos and neg Screw head and the low ESR 6 x 45uF caps linked to the bus seperately. The internal setup has minimal inductance between the CAPs and the IGBT's.
The circuit was improved by reducing the battery side cabling to be shorter and closer. Before the shunt for DC side current measurement was removed reducing the cabling to be even tighter, there was lots of ringing and sag seen at the battery end of the connections.

A extranal capacitor was added with 300mm flying leads. I would prefer the battery to be wired to this capacitor then link to then link to the internal bus, but the cap was mainly added to capture the freewheeling currents when the diodes are conducting onto the bus pos/neg.

Even after this was improved and the circuit was linked up with direct connections to the battery pack and the "T" link to the 2250uF/800Vdc Cap, the IGBT's blew in a fully tested and working inverter, the W phase top and bottom that then popped the battery fuse as it should.
So where are the spikes coming from?

I was more concerned with the phase terminals. They have freewheeling diodes to the Positve and Negative bus connections inside the modules, as per normal. When the IGBT turns OFF the Diode needs to turn ON and allow the current to freewheel through the motor inductance.

The AC side, the IGBT terminal that is connected to the motor phase, has no protection. Adding Capacitors would only cause current spikes, but perhaps more MOV's (Metal Oxide Resistors) would help protect them.

As the inverter apparently works fine with other motor battery combinations, and this system uses 30Ah Kokams, that have a large capacitive effect to the chassis, that may be contributing to the battery side capacitance. Perhaps this needs to be considered.

The DC Bus is floating ideally and is monitored by a special circuit for leakage to the chassis.

The Inverter case has thick earthing links to bare metal contact points that are not structural.

When someone asks me whats wrong that isn't too electrically minded I say it could be like buying an Amp and Speakers that blow the Amp because the speakers are too big. I hope I'm wrong.

I'm surprised someone who has the same motor controller setup has not commented on how they got their system running.

I still think the V/Hz for this motor needs to be below 0.6V/Hz.
The Motor name plate has 210V at 175Hz giving 1.2V/Hz that squeals like a pig (Swine) when C41 test is run even at a very low test current. Suggesting saturation.

The motor may have been damaged in transit and have a air gap or other issue.
But we have had the motor passing C42 with no chain running at over 2000RPM.

Hopefully a similar system can be compared to. The parameters can be saved and emailed to us if some one is up to it.