Nice MOSFETs for class D from IR (or how they try to impose us the DirectFet thrash)

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
Eva

Eva

Member
Joined 2003
This is is a list of recent IR parts rated at 150V or more (also one 100V) and suitable for high-power high-performance class D.

Format of the list:
-----------------------
name - Vds - Id@25ºC - Rds-on(typ.) - Qg(typ) - Qrr(typ)


DirectFet parts:
-----------------------
IRF6775 150V 28A .047ohm 25nC/164nC
IRF6643 150V 35A .030ohm 39nC/190nC
IRF6785 200V 19A .085ohm 26nC/190nC
IRF6641 200V 26A .051ohm 34nC/320nC


TO-220 parts:
-----------------------
IRFZ48V 100V 36A .021ohm 42nC/41nC
IRFB4019 150V 17A .080ohm 13nC/160nC
IRFB4321 150V 83A .012ohm 71nC/300nC
IRFB4020 200V 18A .080ohm 18nC/280nC
IRFB4227 200V 65A .020ohm 70nC/430nC
IRFB4233 230V 56A .031ohm 120nC/760nC
IRFB4229 250V 46A .038ohm 72nC/840nC
IRFB4332 250V 60A .029ohm 99nC/820nC


Note how those damn ·$%&$%&$&/$"%"·$ are trying to impose us their DirectFet thrash by releasing the most suitable parts for class D only in this non-standard and unreliable format. The TO-220 parts are either too weak or too heavy. This is a big leap backwards in reliability and power handling.

Does anybody know of any 150V and 200V "medium sized" TO-220 parts from IR competitors with low gate and reverse recovery charges?

I only know these from Fairchild:

TO-220
--------------------
FDP2572 150V 29A .054ohm 26nC/169nC
FDP2552 150V 37A .036ohm 39nC/242nC
FDP2532 150V 79A .016ohm 82nC/327nC

Power-56 (oh no!! more custom package trash??)
---------------------
FDMS2672 200V 20A .077ohm 30nC/238nC

Those 150V parts from Fairchild are really nice, and they also have nice 100V parts:

TO-220
---------------------
FDP3682 100V 32A .032ohm 19nC/90nC
FDP3652 100V 61A .014ohm 41nC/45nC
FDP3632 100V 80A .008ohm 84nC/120nC


I hope this information to be useful, since I see too much class D made with stone age MOSFETs and ultra slow body diodes.
 
Eva

Eva

Member
Joined 2003
Workhorse said:
Evita you seems to be very fascinated from the specs offered by DirectFets but at the sametime you are equally frustated that these specs are not available in your favourite TO-220 packages....:D :D :D ;)
Click to expand...

This is unfair marketing policy. They make the best dies available in a custon package to force you to migrate to a platform that can't be later replaced by anything else from competitors. Fortunately we have the 150V Fairchilds in TO-220 and these are very good too.


Workhorse said:

I think this also the same Stone-Age Mosfet:D :D
Click to expand...

Indeed!! :D
 
C

ChocoHolic

Member
Joined 2003
I am also a little bit unhappy about available MosFets, but I think there is a good reason for the DirectFet geometry. Parasitic inductances. With TO-220 you cannot overcome the issue of undesired voltage peaks in the gate drive at high di/dt in the D-S path. With only three pins available you always add the inductive peak of the source lead to the gate-source signal. And the geometry of TO-220 has quite some inductance...
The directfets are much more compact, less inductive.

...but for some reason IR missed to offer two source connections, which would allow to bring the correct gate-source signal directly to the chip...
Did they simply not think, when they designed the new package or do the have a reason to mix signal and power across a long common path?
The second connection to the source is standard since ages for high power devices.
Why not applying this for modern small and fast devices?
ooohps I am just seeing that my comlain is valid for the IRF6665, but the IRF6643 has two source pins! Nice !!
I would say IR is doing it right with the Direct Fets.....

I have some direct Fets on hand but during playing around I came to the conclusion that they are mechanically not rugged enough for my current DIY attitudes.
Hm..., but the IRF6643 is really looking nice....
 
Eva

Eva

Member
Joined 2003
It's the opposite, Markus, those 7.5nH of inductance that each TO-220 MOSFET has in its source leg help you a lot in controlling di/dt in hard switching applications. The smaller inductance of DirectFets shifts all the parasitistics a couple octaves higher so that you no longer see them with the oscilloscope but all the ringing is still there ant it's radiated much easier as it gets above 100Mhz.

See the oscilloscope pictures that I have just posted in my other thread. There are also some comments on DirectFets on this thread made by someone that has been experiencing unexplainable failures with them in some particular layouts. Apart from the ridiculous ruggedness, their main pitfall is that they lack a proper thermal pad in order to handle heating peaks produced by avalanche and class D body diode hard switching.
 
P

Photonicube

Member
Joined 2004
Direct Fets do have there pit falls but, to me the package is early in its life, and there are many kinks to work out, as pointed out by eva and others. From what I have heard from IR they are working on the problems that we are experiencing. As for thermal pad size if you look in here they are planning to have a large can form factor. my hope is that is package will help with instantaneous heating in high power applications.
 
Eva

Eva

Member
Joined 2003
What is it? They intentionally forgot to include the 1.3mm thich, 10mm wide and 12mm high copper thermal pad (and heat spreader) that every TO-220 MOSFET has in it! Damn! Copper is too expensive nowadays... (Yet we can't live long without it, like transistors :D:D:D )

An externally hosted image should be here but it was not working when we last tested it.


BTW: This 150V device from NXP (formerly Philips?) seems like a very nice one too.
 
C

ChocoHolic

Member
Joined 2003
Eva said:
It's the opposite, Markus, those 7.5nH of inductance that each TO-220 MOSFET has in its source leg help you a lot in controlling di/dt in hard switching applications. The smaller inductance of DirectFets shifts all the parasitistics a couple octaves higher so that you no longer see them with the oscilloscope but all the ringing is still there ant it's radiated much easier as it gets above 100Mhz.

See the oscilloscope pictures that I have just posted in my other thread. There are also some comments on DirectFets on this thread made by someone that has been experiencing unexplainable failures with them in some particular layouts. Apart from the ridiculous ruggedness, their main pitfall is that they lack a proper thermal pad in order to handle heating peaks produced by avalanche and class D body diode hard switching.
Click to expand...


I definitely like your gate drive wave forms :cool:
And I will for sure examine the ferrite bead in the gate, from theory it is making sense to me.
Regarding the inductance in the source in order to avoid unsaint di/dt, I have my concerns. In some circuits a ferrite bead in the source is nicely calming down the system (...yes, I know some PFC where we are using this method, too.) In other circuits you might build a high power RF-oscillator, i.e. in my 1kW PFC any ferrite bead in the source was really poison. IMHO to control the di/dt by limiting the du/dt slope of Ugs should better be limited in a way that does not strongly depend on Ids, otherwise you might often struggle with a parasitic active oscillator function....
 
C

ChocoHolic

Member
Joined 2003
Of course I do fully agree on your view that smaller L will shift the resonances, but not avoid them. Right, by principle it cannot avoid the resonances. Any L will form resonances with the capacitances.
But at higher frequencies the Q will be less and also dampening seems to be more effective by the losses of the ceramic chip capacitors...

Don't bother about my USB scope which is blind above 20MHz. A nice guy borrowed me one of his 10000 scopes (OK, may be just 1000, I did not count... :D ) and now I can see up to 350MHz... The resonance of my current learning proto for reverse recovery issues of the body diodes is around 30MHz.
 
C

ChocoHolic

Member
Joined 2003
Again many thanks for listing several interesting devices:

In fact I see two more interesting points:
- Compared to older devices the spliting of the gate charge is much more fortunate. I am seeing quite nice low reverse transfer capacities even in heavy devices, while the higher % of the gate-source capacity should help to keep the Ugs more smooth.... ...I would expect shorter length of miller plateau and less trouble with miller feedback ringing... Looking promising...
- There is no max dv/dt specified. And i.e. IRFB4321 is stating explicitely the use in hard switching applications!

Does this mean that the body diodes are now designed in a way that there is really no dv/dt limit and may be they even might offer soft recovery characteristics??
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom