I'm looking at a high power mosfet. The datasheet makes the statement above. If it is not meant to be put on a PCB, are they saying they'd rather have it shaking around inside of some sort of circuit box?
"Not recommended for Surface Mount Application"
- Thread starter mbferguson
- Start date
Part: IRFP4868PbF
Datasheet: https://www.infineon.com/dgdl/irfp4868pbf.pdf?fileId=5546d462533600a40153562c9c822023
Would it be unwise to wire the mosfet to other components soldered to a PCB?
That FET is in a TO-247 which is my favorite package for high power when the device needs to mounted to a heatsink. It is not suitable for surface mounting.
If this is for your ultrasonic pulser, you don't need a FET anywhere near that big. Something with an ON resistance of an ohm or even more would be OK. You are looking at a peak current of probably no more than 4 amperes, for a very short time and a very low repetition rate. One ohm would drop 4 volts and dissipate 16 W during the pulse. That isn't a big proportional loss - about 2% if the probe is 50 ohms. The average power dissipated in the FET would be a very small fraction of a watt. You could halve the loss with a 500 milliohm FET.
Total gate charge is important here because you want to get that FET turned on as fast as you possibly can. Slow turn-on could easily waste more energy than high ON resistance. Power MOSFETs consist of a huge number of tiny FET cells all connected in parallel. For a given general cell design and voltage rating, the way you get lower ON resistance and current handling is by adding more cells. Each cell has a certain amount of capacitance associated with the gate, so more cells means more gate capacitance and more total gate charge to turn the part on. This means selecting a FET with lower ON resistance will make it harder to turn on fast than a FET in the same technology with higher ON resistance.
I'll try to have a quick look later for some suitable FETs. In the mean time, I'd suggest looking for devices in the 200 milliohm to 1 ohm ON resistance range.
The "D pak" or TO-252 package is a surface mount small power package that is not hugely popular but enough so to be a reasonable option to consider. Newer FET designs with very low gate charge are likely to be in lower profile surface mount packages that are more or less variants of the SO-8 package.
If this is for your ultrasonic pulser, you don't need a FET anywhere near that big. Something with an ON resistance of an ohm or even more would be OK. You are looking at a peak current of probably no more than 4 amperes, for a very short time and a very low repetition rate. One ohm would drop 4 volts and dissipate 16 W during the pulse. That isn't a big proportional loss - about 2% if the probe is 50 ohms. The average power dissipated in the FET would be a very small fraction of a watt. You could halve the loss with a 500 milliohm FET.
Total gate charge is important here because you want to get that FET turned on as fast as you possibly can. Slow turn-on could easily waste more energy than high ON resistance. Power MOSFETs consist of a huge number of tiny FET cells all connected in parallel. For a given general cell design and voltage rating, the way you get lower ON resistance and current handling is by adding more cells. Each cell has a certain amount of capacitance associated with the gate, so more cells means more gate capacitance and more total gate charge to turn the part on. This means selecting a FET with lower ON resistance will make it harder to turn on fast than a FET in the same technology with higher ON resistance.
I'll try to have a quick look later for some suitable FETs. In the mean time, I'd suggest looking for devices in the 200 milliohm to 1 ohm ON resistance range.
The "D pak" or TO-252 package is a surface mount small power package that is not hugely popular but enough so to be a reasonable option to consider. Newer FET designs with very low gate charge are likely to be in lower profile surface mount packages that are more or less variants of the SO-8 package.
I had been under the guise of magical thinking when I forgot about rise and fall times. The rising/fall times for the datasheet I posted were pretty big compared to my 100ns target. I don't think that means I wouldn't still be able to have a pulse width of 100ns, but it would surely make me change the point at which the mosfet would need to be turned off. Many of the parameters had the condition of Id = 42A as well...
Last edited:
Here's an outrageously long url that should get you to some probably-suitable FETs at Digi-Key, sorted by ascending gate charge.
I used 1 ohm as the upper cutoff for ON resistance. Notice that they are mostly at the top of the table when sorted by gate charge.
The FQP9N30 looks like it might be a reasonable compromise of ON resistance and gate charge.
The turn-on times all tend to be specified with 25 ohms in series with the gate an at a drain current equal to the stated current rating. If you drive more current into the gate (e.g. via 5 or 10 ohms), you can turn it on faster.
https://www.digikey.com/products/en...y=&ColumnSort=2262&page=1&stock=1&pageSize=25
I used 1 ohm as the upper cutoff for ON resistance. Notice that they are mostly at the top of the table when sorted by gate charge.
The FQP9N30 looks like it might be a reasonable compromise of ON resistance and gate charge.
The turn-on times all tend to be specified with 25 ohms in series with the gate an at a drain current equal to the stated current rating. If you drive more current into the gate (e.g. via 5 or 10 ohms), you can turn it on faster.
https://www.digikey.com/products/en...y=&ColumnSort=2262&page=1&stock=1&pageSize=25
It's probably because mounting on the PCB limits the amount of power that the MOSFET can dissipate, since a PCB sucks as a heatsink. Without a proper heatsink, the MOSFET's life can be severely cut short.
By the way, where is "United State"? Is that somewhere in India?
By the way, where is "United State"? Is that somewhere in India?
Zimbabwe actually.
I think I found the holy grail of mosfets.
SiHP6N40D by Vishay
Conditions for the following -->
Vdd = 400V
Id = 3A
Vgs = 10V
Rg = 9.1Ω
It has:
Low gate charge -->
Qg = 9nC (typ) 18nC (max)
Qgs = 3nC
Qgd = 4nC
Low rise/fall times -->
td(on) = 12ns (typ) 24ns (max)
tr = 11ns (typ) 22ns (max)
td(off) = 14ns (typ) 28ns (max)
tf = 8ns (typ) 16ns (max)
So if all were typical I could get as low as 45ns for the total delay. Other chips had minimum of 100+ns. My voltage won't be at 400V, so I'm wondering which direction that would shift the delays.
Low Ron --> 0.85Ω (typ) 1.0Ω (max)
Since you mentioned I could get even better delay times by using a faster current, and we expect my pulse to be somewhere <=4A, maybe I could get better than 45ns.
Should I close my eyes and order it without thinking anymore?
http://www.vishay.com/docs/91498/sihp6n40d.pdf
Another good option with similar parameters as above: STD9NM40N
https://www.st.com/content/ccc/reso...df/jcr:content/translations/en.DM00067012.pdf
I sooner believe the reason is that radiator demanded is too large, thus the pcb FR4 is not capable to carry the thermal flux. But for keeping the radiator PLUS transistor hanged into paches glue, that is too massive mechanical stress. Thus, why to not mount in the dip manner??
