DRSSTC IGBT selection

[Zilk]

Hello,

I am currently planning my first DRSSTC. I already read a lot on kaizerpowerelectronics.dk and loneoceans.com etc. but I still got some questions/ I do not really know if I am missing something and want a second opinion.

My goal is a system with a 350-450A primary peak current at a frequency of around 120kHz.
Given the calculations on kaizerpowerelectronics.dk, I chose this IGBT for a half-bridge configuration:

IXYN100N65C3H1 https://www.mouser.de/ProductDetail/747-IXYN100N65C3H1

It says optimized for up to 60kHz switching, but thanks to soft switching it will do the 120kHz two, right?
From a thermal perspective  it should be fine (at least if i did the calculations right)
Is this a good choice?

For the balancing Capacitors i wanted to use something like this:
https://www.mouser.de/ProductDetail/80-C4BSMBX4200ZALJ
I wanted something with screw terminals, low ESR and high current capability. Is this suitable for the job? (Vbus = 325V, MMC = 225nF)

To balance them I calculated around 40Mohm for the resistors, seems a bit high to me? 
I used this formula from Mads for it R = n * Vm - Vb/(0.0015*C*Vb)


would be glad if someone with experience could look through this, before I blow up my money.

Kind regards

Kai

[Mads Barnkob]

I think you should do the tedious calculations for the heat dissipation, 120 kHz is not super high for a fast IGBT like that, but its 0.25 °C/W thermal impedance rating could pose a small problem.

I have pushed my 60N60 miniblock IGBTs to 500A in stable operation, I think I blew up a bridge at some 600-700A, so a maximum of 500A seems reasonable for a IGBT like IXYN100N65 that has a Icm rating of 450A. Maybe you could push it to 600A, but the thermal impedances is much higher for modern switches than the good old ones, less material in them for cheaper production.

Your balancing resistors does sound high, what acceptable voltage ripple did you calculate it with?

[Zilk]

Thanks for your answer,

you are right, I noticed a mistake in my thermal calculations and now its pretty close. With very good cooling i might get away with it, but i rather dont risk it.
This one seems to be better:
IXXN110N65B4H1 https://www.mouser.de/ProductDetail/IXYS/IXXN110N65B4H1?qs=%2Fha2pyFaduhKndUMJlgd3%2FiNl2KfCOt55lqtHEWNkDnLagFRw9Mx%252Bw%3D%3D
To be sure I will just go through the calculations.

For D I'll go with your extreme example with 0,06 and a 200us on-time.(should be well below that)
From the Z(th) diagram I took (because there is no 0.06 line) next worse one D = 0.1 and got a thermal impedance Z(th) of 0.03 K/W.
So now I am bit confused, in your guide is written that I now have to multiply this value with R(th)? but this would result in a far to low thermal resistance, so I assume that just have to take that value as R(th)? So R(th) applies for a constant burden and Z(th) for pulsed operation?

If i go on with the 0.03 K/W, a max Tj of 80°C and a Tc of max 50°C, I get a Pdiss of 1000W (which seems way to high for me).
For Pcond its 450A * 4,5V * 0,06 = 121,5W.

With a dI/dt = 340 A/us and Tdon + Tri = 66ns and Toff + Tf = 189ns , I get Eon = 1,3mJ and Eof = 2,3mJ , taking the 24V Vg into account its 2,25mJ.
and finally Fmax = (1000W-121,5W)/2,25mJ = 390 kHz, which i think is way to good.
If I use R(th) = 0.2W/K instead i get 13 kHz.

Is this all wrong?

For the balancing resistor calculations I took the same voltage values like your example since I use the same Bus Voltage.
So Vm = 180V, Vb = 320V  , C = 2uF --> R = 2*180V-320V/(0,0015 *2uF*320V) = 41Mohm.
I think in the Example you used this for a Voltage doubler (6000uF), so its clear where this high value is coming from.

I know this is a lot work I am asking for, so really thank you for helping.

Kind regards

Kai

[Mads Barnkob]

Thanks for your answer,

you are right, I noticed a mistake in my thermal calculations and now its pretty close. With very good cooling i might get away with it, but i rather dont risk it.
This one seems to be better:
IXXN110N65B4H1 https://www.mouser.de/ProductDetail/IXYS/IXXN110N65B4H1?qs=%2Fha2pyFaduhKndUMJlgd3%2FiNl2KfCOt55lqtHEWNkDnLagFRw9Mx%252Bw%3D%3D
To be sure I will just go through the calculations.

I have not looked at this part, but be sure to watch the gate charge, the IGBTs might just get harder to drive right when you step up in specifications.

For D I'll go with your extreme example with 0,06 and a 200us on-time.(should be well below that)
From the Z(th) diagram I took (because there is no 0.06 line) next worse one D = 0.1 and got a thermal impedance Z(th) of 0.03 K/W.
So now I am bit confused, in your guide is written that I now have to multiply this value with R(th)? but this would result in a far to low thermal resistance, so I assume that just have to take that value as R(th)? So R(th) applies for a constant burden and Z(th) for pulsed operation?

0.03 K/W is your thermal impedance corrected for low duty cycle, continue with that.

If i go on with the 0.03 K/W, a max Tj of 80°C and a Tc of max 50°C, I get a Pdiss of 1000W (which seems way to high for me).
For Pcond its 450A * 4,5V * 0,06 = 121,5W.

With a dI/dt = 340 A/us and Tdon + Tri = 66ns and Toff + Tf = 189ns , I get Eon = 1,3mJ and Eof = 2,3mJ , taking the 24V Vg into account its 2,25mJ.
and finally Fmax = (1000W-121,5W)/2,25mJ = 390 kHz, which i think is way to good.
If I use R(th) = 0.2W/K instead i get 13 kHz.

Is this all wrong?

I can see what mean by confusing. I mention RθJC/Rth (steady state resistance, no correction for duty cycle) that in your case is given as 0.25 ℃/W, but it is Zth(j-c) we want and can read out from the Zth(j-c) graph for our calculated/estimated duty cycle.

Pdiss of 1000W is not unreasonable for a very low duty cycle and thermal impedance. (its rated for 600W in hardswitching)

Since you went with D=0.1, you should also use that for Pcond instead of 0.06, drops down to 355 kHz then.

I need to clean up, stream line and not use different examples in the IGBT calculation article.

For the balancing resistor calculations I took the same voltage values like your example since I use the same Bus Voltage.
So Vm = 180V, Vb = 320V  , C = 2uF --> R = 2*180V-320V/(0,0015 *2uF*320V) = 41Mohm.
I think in the Example you used this for a Voltage doubler (6000uF), so its clear where this high value is coming from.

I know this is a lot work I am asking for, so really thank you for helping.

Kind regards

Kai

You do not need balancing resistors on small film capacitors. Voltage is shared according to ratio of capacitance, there is hardly any difference on those and still they are likely to be within spec and even if not, metal film capacitors have a higher transient voltage resistance compared to electrolytic capacitors. That is why you get that high number, your capacitance is very low.

[Zilk]

Again a huge Thanks Mads for taking your time.
Your design guide is really good for getting into the subject and I learned a lot from it.

I am very happy that the calculations were almost correct, now I can go on by myself.
Today I winded the secondary, I'll go trough my designe once again and if everything seems to be fine I will order the parts.

Once I am able to show some results, I will post them.

Kind regards

Kai