6
\$\begingroup\$

So I wish to switch a current of roughly 3 kA on for short periods of time roughly every second. I have created a circuit shown below which I believe will be sufficient but after past failures I would really like to check with people who might know what they are doing. The first image shows the current through the loosely named "load" R11, but is the load nontheless, while the second image shows the heat generation in watts in the MOSFET. This appears to be at about 90 W normally but under switching hits 1.4 kW which seems very high. This event only occurs for 160 ns (288 nC/1.8 A) but is that enough to make it go boom? If that is going to be fine then the rest of the circuit should live I would think. It will be mounted onto an aluminium water cooling block so should stay cool, I am just mainly concerned about those massive spikes during switching. Should I be switching them even faster? Currently 1.8 A going into the gates but I am not sure how to increase that. If it is fine then yay if not any input about improving the performance of this circuit would be good, before I go and blow up 6 more MOSFETs.

Some extra detail: V1 controls the driver, V2 is the high current power supply for the circuit and yes is a battery designed to do that.

P.s. Here's the data sheet of the mosfets I am using: https://www.vishay.com/docs/79764/sqjq140e.pdf and if anyone wants to tell me "even though they say they can handle 700 amps they actually can't it's a lie", here's a similarly if not worse spec'd mosfet doing exactly what I am looking for, seemingly without a heatsink: https://www.youtube.com/watch?v=qrktIIbV8Qc&ab_channel=Nexperia so I know it is possible. Data sheet for that if you're interested: https://au.mouser.com/datasheet/2/916/BUK7S0R5_40H-2392070.pdf

enter image description here

enter image description here

\$\endgroup\$
13
  • \$\begingroup\$ The MOSFET in your diagram does not appear to be one that you have linked. \$\endgroup\$ Commented Sep 19, 2023 at 7:51
  • \$\begingroup\$ My software does not have the same mosfet but the one I chose has the same RDS on and similar gate capacitance :) \$\endgroup\$ Commented Sep 19, 2023 at 7:54
  • 1
    \$\begingroup\$ Node names on graphs are meaningless to anyone reading this question. Name nodes by labelling please. \$\endgroup\$ Commented Sep 19, 2023 at 7:56
  • 4
    \$\begingroup\$ The MOSFETs will never turn on at the same time because of the practical imperfections and device-to-device imbalances. So, one MOSFET will take significant amount of current compared to others (maybe the total current as a worst case) for a short amount of time. You need to take this into account (Check SOA and Rth graphs). Ideally, you need to select a MOSFET that is able to switch the total current, and then you can think about paralleling. \$\endgroup\$ Commented Sep 19, 2023 at 8:11
  • 2
    \$\begingroup\$ One other thing to consider is the spikes: Since you are switching very high amount of current, any small inductance (can be in order of fractional nanohenries or even tens of nanohenries) coming from the connections will create a good amount of spike across the MOSFETs when they go off. If the spikes are high enough then they can be destructive for the MOSFETs even though they are rated for 3.5 times the supply voltage. Spikes also depend on dV/dt i.e. slowing the switching down (increase rise/fall times) reduces the spikes but also increases losses. Put a 3nH inductor in series and simulate. \$\endgroup\$ Commented Sep 19, 2023 at 9:15

1 Answer 1

Reset to default
2
\$\begingroup\$

No issue with the per device continuous current limits. These devices are tough and can deliver the performance you quote.

However, as noted by others, you have quite a number of issues to deal with if this application (paralleling) is going to work reliably.

  • Total gate drive current requirements
  • Maximum peak current during switching due to threshold differences
  • Inductance in gate drive (limiting rise time, and causing gate spikes)
  • Inductance in power circuit (leading to switching spikes at turn off, with need for snubbers)

The devices you reference have a pulse current limit of 1820A for < 300us pulse and 2% duty cycle. This is high, but if one takes all the current during switch on will likely be exceeded...

There are a fair number of app notes on paralleling FETs, including this one from TI, which highlights particular challenges related to layout and inductance, and this one from Infineon with a particular focus on dynamic current imbalance.

I would suggest you look carefully at the reference designs, in particular some of the protection components for the effect of inductance, and the layout constraints.

I would strongly suggest you start with a single FET design for switching a lower scaled current pulse, and carefully observe actual switching waveforms on gates and load.

Once you have worked out the issues with that design, produce a multi-device setup, and work up to the full design in steps (1 FET, 2 FET, 4 FET, 6 FETs with scaled loads).

This will allow you to debug bits of the problem at a time, rather than trying to do all at once (which is almost certainly going to let out a lot of "magic smoke").

Edit: This article from Infineon goes into much more detail on the pitfalls of paralleling FETs, in particular the troubles with current sharing due to device variation. It includes some excellent graphs/traces showing the extend of problems you can expect.

\$\endgroup\$
4
  • 1
    \$\begingroup\$ So looks like theres a couple ways I could do this. If I increase the load resistance to 0.00667 then max current at any one time becomes (12/0.00667=1800) roughly 1800 amps right. That gives me a <300us window to turn on all MOSFETs, which I think is achievable. Otherwise if I can experiment with scaled loads, probably cut lengths of wire for such low resistances, and see if I can get pairs working with 3 driver circuits. If such a thing is possible then my max current over <300us becomes 3600 which I can easily work with, if that's how all that works. \$\endgroup\$ Commented Sep 19, 2023 at 23:36
  • 1
    \$\begingroup\$ Other than that I need to read up on inductance within high speed switching circuits and see what protection I can add. I never thought I would need protection from 1 loop of wire but I guess the sheer amount of current here is the problem right? Thank you for your comment :) \$\endgroup\$ Commented Sep 19, 2023 at 23:38
  • 2
    \$\begingroup\$ That all makes sense. Be aware that the 1820A is from the "Absolute Maximum" section of the datasheet, so view it as a hard limit not a guide, and some headroom would help reliability.. Do update with experiences once you have a working solution. \$\endgroup\$ Commented Sep 20, 2023 at 6:20
  • \$\begingroup\$ alright it's been a while but it works, had it switching 1500 amps reliably, it does get a bit hot though. \$\endgroup\$ Commented Oct 21, 2023 at 1:08

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.