Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

[SOLVED] MOSFETs handling high current within short duration

Status
Not open for further replies.

Zak28

Advanced Member level 2
Joined
Aug 19, 2016
Messages
579
Helped
6
Reputation
12
Reaction score
6
Trophy points
18
Activity points
4,681
What in particular characteristic do MOSFETs have (perhaps a specific item in a datasheet) when the intention is to employ them to conduct lots of current in a short time frame (as in the case for capacitor discharge) to do it safely?

I did some searching around and I believe its avalanche energy but its rated in milli juels even for high power FETs so I Im still unsure.

Greatly appreciated.
 

Avalanche rating is the current it can take at Vds=BVdss
for some period, expressed in Joules (V*I*t) most often.
In the capacitor discharge case, V will decline along the
Ron*C profile and you'd perhaps apply the integral FET
power dissipation over the discharge time against that
avalanche rating.

Avalanche ratings' timescales are usually short, as they
are meant to deal with an inductor's flyback kick and
the current behind it (which also declines, but test
condition may have been a square current pulse for
convenience).

The avalanche Joules are what can be taken by the
die mass alone, or some small portion of the package
as well, in a thermal pulse without raising the die temp
to the point of reliability degradation. Adiabatic or
nearly so, is the conservative estimate for this. The
capacitor discharge event may differ in timescale,
depending on the cap, voltage and on resistance of
the FET. At mS timescales, you are no longer that
close to the adiabatic approximation but any error
will be "upside" (more tolerant due to more of the
package thermal mass being "accessible" (thermal
time constant vs electrical time constant).

An SCR could be the right answer for this application
as you only have to provide the trigger and the load
supplies the remainder of the drive, turnon can be
fast and turnoff happens when the load has been
fully discharged. Of course you need to be sure the
trigger happens only when you want it. SiC SCRs
can stand off beaucoup voltage. Silicon SCRs are
available up to kV and kA if you want something
with a hockey puck form factor.

You may find SOA curves that offer DC, mS and uS
range versions, if the mfr cares about applications
support in detail.

Run a SPICE simulation and print integ((V(d)-V(s))*Id)
with your planned initial cap voltage and the FET SPICE
model, and see what mJ number you get.
 
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
This subject is on my mind right now as well as I'm concepting a design which requires fast cap discharge. In my scenario I want it rapidly discharged when power is lost so I'm considering depletion mosfets for their normally-on properties (during operation the depletion will be driven off).

Here are a few quick notes:

First, you can get some information from the "Transient Thermal Impedance" charts that are usually near the end of the datasheet. For example this one:
https://ixapps.ixys.com/DataSheet/DS100185D(IXTY-TA-TP1R6N100D2).pdf

This app note from Infineon also covers dynamic thermal impedance of many packages.
"Thermal Resistance Theory and Practice"
https://www.infineon.com/dgdl/smdpack.pdf?fileId=db3a304330f6860601311905ea1d4599

I can't recall the source now and will return if I find it but I recall reading something and concluding a DPAK was more like 1C/Joule, which appears lower than the infinon app note which shows 3.5C/watt for a 200ms pulse.

One thing the app note illustrates is that at about 1 second and higher the PCB layout and its thermal mass becomes an effective sink (that's where the different footprint curves start diverging) and you should be able to find the thermal mass of copper to calculate that.

SCR's are more likely to have an A2S (amps squared second) fusing rating which will also give you a clue as to the energy they can handle (though you wouldn't want to come close to that, perhaps 1/10th of it or less).

The bottom line I believe is that standard mosfet packages like DPAK are good for single-digit joules at least, not milli-joules.

EDIT: Ah I found it. TI's appnote on hotswap applications:
https://www.ti.com/lit/an/slva158/slva158.pdf

This says D2Packs are ~1J/C, DPAK's are 0.37J/C (again this requires sufficient time to heat the package which I believe is on the order of 1 second)

EDIT2: Sorry I'll just keep going. If the mosfet can't handle the energy by itself consider pairing it with a Joule rated resistor. Wirewound resistors will often have joule ratings and some resistor families are specially targeted charge/discharge applications. Here is one example:
https://www.ohmite.com/assets/docs/res_a.pdf

And their broader high energy series.
https://www.ohmite.com/surge-high-energy/

Another good charge/discharge device is a PTC which has the feature that its self-protecting if the discharge circuit was to fire during operation. These PTC's are joule rated specifically for charge/discharge
https://en.tdk.eu/inf/55/db/PTC/PTC_ICL_in_housing.pdf
 
Last edited:
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
Depending on the details you might like RF depletion mode GaN
FETs, these can be had with some pretty good power packaging
and are a lot more "sporty" than a RF MOSFET. eGaN gets all the
press but there's a lot to look at, in RF power GaN.
 
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
You can power normal fets off the Vcc until it gets down to ~5V, logic level fets to ~2.5V to get your volts below that you will need a BJT ( 0.7V) or SCR 1.2V

As long as you keep inside a reasonable peak current for a mosfet ( or mosfets ) you will be OK and avoid die melt or bond wire melt

For very high currents you need an SCR e.g. TYN640, 40A, 600V but > 560A for 8.33mS half sine wave current shape ... TO-220

- - - Updated - - -

it makes sense to have a choke of some sort in series for an SCR to limit the di/dt initial - to protect the die, mosfets can be turned on slowly ( 1k gate R) to get a similar effect.

IGBT's and BJT's can be turned on very fast e.g. <10nS...
 
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
Hi,

to protect the die, mosfets can be turned on slowly ( 1k gate R) to get a similar effect.
Afaik with slow turn ON you will increase the thermal loss within the MOSFET thus you increase heating...

I'd say there's no way around choosing a MOSFET which is caoable to carry the high peak current... and then fast turn ON.

The series inductance is a good way to decrease the stress for the MOSFET.

Klaus
 
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
Actually - the total heating in the mosfet(s) is the same regardless of how fast or slow you turn on as, all the energy in the caps ends up as heat in the mosfets, a slower turn on gives the current time to spread out over all the cells in the die (avoids current crowding bang) and lowers the peak current a bit, 0.5C.V^2 of energy ends up in the junctions ...
 
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
The problem isn't clearly specified yet. Are we talking about capacitor energy dissipated in the transistor (no relevant external load impedance present) or is the problem about switching a load impedance? Gate voltage timing will have different effects in both situations.
 

@Easy peasy:
You are correct.. with no additional load...
I somehow assumed there is a load .. but after reviewing the thread I´m not sure about this.

Klaus
 

Depending on capacitor type, its internal inductance and
resistance (ESL, ESR) the switch transition time may still
matter to the partitioning of power dissipation. If the FET
gets switched in 100nS, and the capacitor has a SRF of
1MHz or lower (many electrolytics are -way- lower, as
they were only concerned about flattening 100/120Hz
ripple) then the FET will have gotten to its lowest-
dissipation state before the current surge makes it out
of the can.
 

Certainly if the Ron of the fet(s) is/are lower than the ESR + wiring R, then there will be a splitting of the power dissipated ....
 
  • Like
Reactions: Zak28

    Zak28

    Points: 2
    Helpful Answer Positive Rating
Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top