ISO5852SEVM: Ringing issue in high-side IGBT connected transmission line

Hi,

What supplies are you using for the ISO5852SEVM drive voltage (VDD and VEE) and where are they referenced?

Thanks,

Audrey

Hi Audrey,

I am using two separate programmable DC power supplies (TEKPOWER TP3005P) for the input and output side of the gate driver to achieve isolation. I am using the black(negative) terminals in the power supply for floating reference and to achieve  isolation. As the main power supply of the circuit I am using Spellman SL1200 whise ground is same as the overall circuit ground.  

Hi,

Thanks. In that case it sounds like you are not generating a ground loop from your various supplies. However, would you be able to show a schematic image so that I can understand clearly how the driver is connected to your supplies?

Regards,

Audrey

Hi Audrey,

Thanks for the reply. The schematic is posted below.

The emitter and VEE ground are at the same potential. The emitter is also connected with the transmission line positive side. I am not sure driving this driver in this high side configuration is a problem.

Hi, 

I am thinking if the dynamic characteristics (input, output, reverse transfer capacitance) of the IGBT are playing a role in this issue(alternate path). Somehow current is being drawn out of the HV power supply even after IGBT is switched off.

Hi,

Thanks for the additional diagram. Since the driver is isolated it is fine to use as a high-side driver. I am not sure that the reverse transfer capacitance would cause such a spike so delayed after the turn-off.

Could you provide a waveform which shows the gate driver's output voltage to the IGBT (V_GE) and the voltage referenced to ground after the IGBT?

Additionally, could you please clarify:

- Did you mean VCC instead of VEE?

- Is the ground of VCC of the driver supply also referred to the IGBT's emitter?

Thanks,

Audrey

Hi Audrey,

Thanks for the reply. I used unipolar supply for the output power supply of the gate driver EVM, so it is Vcc. In the EVM the negative of the Vcc and the emitter of the IGBT are all at the same reference as GND2(naming convention reference from ISO5852SEVM datashhet).

Please find the waveform of Vge and voltage referenced to ground after IGBT attached herewith.

Hi,

Thanks for the waveform. What is the status of S2 during this transient? Also, are you using isolated probes?

Thanks,

Audrey

Hi Audrey,

Thanks for the reply. I have not directly measured the voltage across S2 during the transients. However, voltage before and after S2 have been measured. Image attached here (yellow and blue trace refers to voltage before and after S2 respectively). I am using 1.tiffTek P6015a probe and TPP 1000 probe for measurement. the channels of the oscilloscope are ground referenced.

Hi,

Thanks for the valuable suggestions. I will try to implement the ideas. In addition, I am trying to simulate the behavior of the test circuit in LTSpice. At present I am using a simple gate drive emulator to drive a MOSFET/IGBT in the simulation. Part of the schematic is attached herewith. Is there any detailed LTSpice model available for the TI gate driver I am using? Any suggestion (component addition, connection) for detailed gate driver design in the schematic is much appreciated.

Hi,

All simulation models for ISO5852S are located in the product folder here. The passives surrounding the circuit are shown in the Application and Implementation section of the datasheet.

Thanks,

Audrey

Hello Wasekul, 

I was monitoring this thread and noticed that it has been a while since we heard from you. 

I will therefore assume this issue is closed and mark this thread as resolved. If so, please press the green button or let us know if you have further questions by further posting on this thread.

Regards,

-Mamadou

Hi,

The problem is actually not resolved yet. I am attaching additional wave forms herewith for the sake of better understanding of the situation.The first two images highlight the condition of Vge and Vce during the switching processes. As you can see, there is actually two switching happening. One switching period is the portion where Vge is high (IGBT is on) and another switching point is the instant where Vce goes from low to high (at that moment, PCSS is triggered and charge stored in the transmission line before PCSS is transferred to the right side of the PCSS). I2 is the current through the transmission line after the IGBT. In first two images, you can see the spike in Vce and voltage across transmission line right after PCSS triggering. However, the third picture shows a different configuration where there is no substantial spike in voltage waveforms. In this configuration, the transmission line is directly connected with the HV power supply without the IGBT/MOSFET. Hence, the spike has definitely something to do with the IGBT/MOSFET. I have tried to increase gate resistance, put freewheeling diode across the transmission line before the PCSS to dampen the ringing. However, it stays the same with IGBT/MOSFET included. I will test with negative Vge shortly to prevent any possibility of false turn-on. 
CREE MOSFET over IXYS IGBT reduces the oscillation peak a little bit and which might be do with lower input capacitance. 

4.tif3.tif5.tif

Hi,

Thanks for the reply. I have conducted the experiment using bipolar supply (Vcc=15 V, Vee= -5 V) without making any difference in the voltage spikes. Therefore, it seems and reinforces the prior notion that false turn-on is not the responsible mechanism for the spikes. I will try putting RC snubber and maybe a ferrite bead (to act as resistance in oscillation frequency range) before the drain of the power switch and also before the PCSS. If I consider the PCSS as the low side switch, the switching node is between the power switch and the PCSS. Considering this, should I put the RC snubber across the PCSS as it might be the responsible switch for the voltage spikes?

Thanks in advance.

Hi,

Yes, you could see if the RC snubber across the IGBT alone will work. Otherwise, it may help to add the snubber across the other switching device.

Regards,

Audrey

Hi,

I have tried minimizing the oscillation amplitude by putting a snubber (R=12 ohms, C= 10/18/28 pF) across the IGBT and the PCSS as well independently. However, I have not seen any tangible improvement. Do you think the selected avlues of the RC snubber components are ok? For optimized snubber design, it is required to add an external capacitor in parallel with the switching device such that the oscillation frequency is halved. The problem here is, only the first cycle of the oscillation is distinguishable. After the first big peak, the oscillation dampens and becomes negligible if I add external capacitor. Therefore, I couldn't estimate the exact required value of the C of snubber in this method.

I am thinking, if the very high slew rate of the PCSS switch is making it very hard to dampen the oscillation whether I put a snubber across it or the IGBT or not. 

Thanks in advance.