IGBT current vs Ocurlim

[lumax]

I got an Infineon IGBT module FS800R07A2E3 (BMW I3 inverter) rated at 800A cont and 1600A peak, where should be set my Ocurlim ?

1500A???

Thanks!

[arber333]

I got an Infineon IGBT module FS800R07A2E3 (BMW I3 inverter) rated at 800A cont and 1600A peak, where should be set my Ocurlim ?

1500A???

Thanks!

I got the feeling you could be using 2x Irating for OClimit because it happens really fast. At least for ACIM motors. But you are correct in setting more conservative setting. I think you are OK for start with 1500A. I must say i use 550A limit for Leaf motor with Ampera inverter (both halfhs are 600A rated) and i think i will soon open them to 600A.
I use 3.1uS deadtime.

[lumax]

From the IGBT specsheet

Implemented collector current ICN 800A
Continuous DC collector current TF = 75°C, Tvj max = 175°C IC nom 550A
Continuous DC collector current TF = 25°C, Tvj max = 175°C IC 700A
Repetitive peak collector current tP = 1 ms ICRM 1600A
SCdata VGE ≤ 15 V, VCC = 360 V tP ≤ 8 µs, Tvj = 25°C ISC 5600A
SCdata VCEmax = VCES -LsCE ·di/dt tP ≤ 8 µs, Tvj = 150°C ISC 4000A

My motor is an IPMSM, the DC link fuse is a PEC 350A 2877-0010 (don't know the curve)
My dead time is set to 1.6us (110 unit from what I recall)

I know usually in power electronics we leave a safety factor of 2X minimum... Is this applicable since we got an hardware OC limit?
Because for now my OCurlim is set to 600A and I'm not able to turn the motor "NO LOAD" over 6000Rpm without an OCurlim fault... from the plot when I ramp it slow it draw 30 to 60A on IL1 and IL2

I got the feeling you could be using 2x Irating for OClimit because it happens really fast. At least for ACIM motors. But you are correct in setting more conservative setting. I think you are OK for start with 1500A. I must say i use 550A limit for Leaf motor with Ampera inverter (both halfhs are 600A rated) and i think i will soon open them to 600A.
I use 3.1uS deadtime.

Thanks for you feedback it's really appreciated

[arber333]

Yes 350A is sustained DC current where my Leaf motor wants to go beyond 170km/h! So i set my rpm limit to 11.000rpm at cca 150km/h. On acceleration this is another matter. If i floor it my motor after 2s allready gives 350A. Since i set the limit there acceleration is then limited to 350A from DC side with 550A on the voltage rise.

I am using Lebowski brain, but some things are similar. I really recommend you use at least 3us deadtime for testing. It might be that high side IGBT emitter tail is meeting the low side collector turning on and you are seeing some short circuit action through current sensors.

[lumax]

It might be that high side IGBT emitter tail is meeting the low side collector turning on and you are seeing some short circuit action through current sensors.

Yes, that's also what I'm thinking... The plot seem to be really good, and suddenly... Bang!!! over current

I will double the dead time and see if it improve

Thanks arber333 !

[johu]

Keep in mind he is using the openinverter drivers without desat, I think oclimit beyond nameplate current (800A) is quite risky... Overcurrent fault at 600A when steadily spinning up seems weird, maybe current sensor signals need shielding?

The overcurrent comparator does NOT detect cross conduction, only current that actually exits the phase terminals!

[lumax]

Keep in mind he is using the openinverter drivers without desat, I think oclimit beyond nameplate current (800A) is quite risky... Overcurrent fault at 600A when steadily spinning up seems weird, maybe current sensor signals need shielding?

The overcurrent comparator does NOT detect cross conduction, only current that actually exits the phase terminals!

I took special care to route the current sensor cable far from other signal (picture is the original placement of the current sensor cable but now they are 6cm apart from others), but they are not shielded cable... so that might be the problem... and the wire between the gate and gate driver are probably too long (about 4cm) RF???

[arber333]

Keep in mind he is using the openinverter drivers without desat, I think oclimit beyond nameplate current (800A) is quite risky... Overcurrent fault at 600A when steadily spinning up seems weird, maybe current sensor signals need shielding?

The overcurrent comparator does NOT detect cross conduction, only current that actually exits the phase terminals!

OH! I thought you use native drivers from BMW. I guess then it could be EMI spikes on those sensor wires. Maybe you could try some more filtering? I use 4K7 pulldown resistors on my current lines.

Also i am using my Mazda DIY inverter with Skyper drivers and just bare wires. No special shielding. But i have 3uF snubber on each IGBT and i am using overlaping DC link.
I really think we need to make that 8A modular protected driver board Johannes. I have design somewhere...

EDIT: @lumax, You really need to make your gate (GE) wires thicker and twist them. At every PWM pulse a lot of current wants to flow through there to drain/charge gate capacitor and that can create a lot of ringing. It can really mess up the gate signal. Why is there a twisted pair in the background and not on your IGBT gates?
If you twist the wires there would be a lot less ringing. Also use thicker wire. I have seen 0.75mm2 wire with 200A IGBTs in various UPS and more.

[johu]

Also the driver pairs are interchangeable. If you swap the input low/highside you can swap the output as well. Would also contribute to shorter wires.

[lumax]

Also the driver pairs are interchangeable. If you swap the input low/highside you can swap the output as well. Would also contribute to shorter wires.

T'hats actually a freaking good idea

EDIT: @lumax, You really need to make your gate (GE) wires thicker and twist them. At every PWM pulse a lot of current wants to flow through there to drain/charge gate capacitor and that can create a lot of ringing. It can really mess up the gate signal. Why is there a twisted pair in the background and not on your IGBT gates?
If you twist the wires there would be a lot less ringing. Also use thicker wire. I have seen 0.75mm2 wire with 200A IGBTs in various UPS and more.

As Johannes suggest, il will reverse the hi and low side (and use thicker wire) the small twisted pair is for the heat sink temperature and the large twisted pair is for the DC bus voltage sensor

Thank you so much guys for all these infos... I will update the design... Hope that will fill the bill