IRF Direct Fets How to solder??

[Oznog]

Deja vu- I remember saying the exact thing about the IRF3703, amazing disbelief, and then saw the pkg was limited to 75 amps. As such I don't know why they bother to list a spec that the part cannot handle.

What are you trying to switch? There are some super hexfets that switch large currents and they're fairly cheap. There are some in SO-8 pkgs that can handle 16 amps constant current each due to their amazingly low rds-on, without a heatsink, for around $1. Of course their low rds-on won't help with switching losses and they don't have as much heat dissipating capabilities to handle switching loss heat.

 

[Optikon]

Deja vu- I remember saying the exact thing about the IRF3703, amazing disbelief, and then saw the pkg was limited to 75 amps. As such I don't know why they bother to list a spec that the part cannot handle.

What are you trying to switch? There are some super hexfets that switch large currents and they're fairly cheap. There are some in SO-8 pkgs that can handle 16 amps constant current each due to their amazingly low rds-on, without a heatsink, for around $1. Of course their low rds-on won't help with switching losses and they don't have as much heat dissipating capabilities to handle switching loss heat.

I think he is doing a synchronous rectifier..

He might be in a worse situation with these parts efficiency wise. The switching losses could conceivably be higher with these parts than with say the SO8 pack you mentioned.

 

[williB]

Deja vu- I remember saying the exact thing about the IRF3703, amazing disbelief, and then saw the pkg was limited to 75 amps. As such I don't know why they bother to list a spec that the part cannot handle.

What are you trying to switch? There are some super hexfets that switch large currents and they're fairly cheap. There are some in SO-8 pkgs that can handle 16 amps constant current each due to their amazingly low rds-on, without a heatsink, for around $1. Of course their low rds-on won't help with switching losses and they don't have as much heat dissipating capabilities to handle switching loss heat.

I think he is doing a synchronous rectifier..

He might be in a worse situation with these parts efficiency wise. The switching losses could conceivably be higher with these parts than with say the SO8 pack you mentioned.

Yes..sort of .. It is the one from earlier, the one we simulated ...
i etched a circuit board ( for the direct Fets ) and found out They are real hard to solder..

 

[Oznog]

How much current and voltage are you trying to switch? And what sort of device is this in?

 

[williB]

They're here..!!!
they just arrived..!!


sorry

How much current and voltage are you trying to switch? And what sort of device is this in?

does, personal generator, ring a bell..
if not .. about 20V .. & somewhere a little south of 10A..

 

[Oznog]

I'd say you may not have the best part right there.
Go Digikey, search "mosfet", select "SO-8, 8-SOIC, 8-SOP".

Or look at this page:
**broken link removed**
**broken link removed**
**broken link removed**

Handling 10 amps is no sweat with most of these.

You can get SO-8 fets with an rDSon of only a few milliohms, and they're way simpler to mount. Don't even need to drill a hole. You'd think a SO-8 wouldn't be able to handle power due to the smaller leads alone, but you parallel 3 leads on one side and 4 on the other which gives a LOT of current ability. They have a very low junction-to-lead thermal impedance so a wide trace can dissipate a surprising amount of heat!

And look at switching losses. The IRF3703 is a very large gate charge. Don't underestimate the energy lost in charging and discharging that capacitance. And higher capacitances mean lower rise and fall times for a given driver, which can increase switching losses.

I have a bunch of IRF7811W SO-8 that I use, 0.009 mOhm rDSon, 14 amps continuous with no heat sink, and the gate charge is 18. And I use them because they're cheap. $0.86/ea in small quantities. Gate charge on the IRF3703 is 209. So it's more that 11 times slower to switch on the same driver, and even if your driver is big then you're still spending 11x more total energy to charge that cap per cycle.

Due to cost, size, and mounting issues, if I needed to switch 20 amps I'd greatly prefer 2, 4, whatever of these over a TO-220 or anything with a heatsink. Smaller, easier, and cheaper especially if the TO-220 would need a heatsink.

 

[Optikon]

I'd say you may not have the best part right there.
Go Digikey, search "mosfet", select "SO-8, 8-SOIC, 8-SOP".

Or look at this page:
**broken link removed**
**broken link removed**
**broken link removed**

Handling 10 amps is no sweat with most of these.

You can get SO-8 fets with an rDSon of only a few milliohms, and they're way simpler to mount. Don't even need to drill a hole. You'd think a SO-8 wouldn't be able to handle power due to the smaller leads alone, but you parallel 3 leads on one side and 4 on the other which gives a LOT of current ability. They have a very low junction-to-lead thermal impedance so a wide trace can dissipate a surprising amount of heat!

And look at switching losses. The IRF3703 is a very large gate charge. Don't underestimate the energy lost in charging and discharging that capacitance. And higher capacitances mean lower rise and fall times for a given driver, which can increase switching losses.

I have a bunch of IRF7811W SO-8 that I use, 0.009 mOhm rDSon, 14 amps continuous with no heat sink, and the gate charge is 18. And I use them because they're cheap. $0.86/ea in small quantities. Gate charge on the IRF3703 is 209. So it's more that 11 times slower to switch on the same driver, and even if your driver is big then you're still spending 11x more total energy to charge that cap per cycle.

Due to cost, size, and mounting issues, if I needed to switch 20 amps I'd greatly prefer 2, 4, whatever of these over a TO-220 or anything with a heatsink. Smaller, easier, and cheaper especially if the TO-220 would need a heatsink.

I think you mean .009 Ohm!...

'7811s sound like great parts.. I'll have to try one sometime.

..everyone has their favorites! :wink:

 

[williB]

They look nice but the Max Gate to source voltage is only 12V..
my Voc is allready 20V , and when i add more coils to the generator this value will surely rise..

 

[Oznog]

They look nice but the Max Gate to source voltage is only 12V..
my Voc is allready 20V , and when i add more coils to the generator this value will surely rise..

Most of the SOIC-8 FETs are Abs Max +/- 20V, same as the IRF3703. Plenty of those are in the milliohm range.

You need to build a better driver. You can't directly drive the gate of a IRF3703 with 20v either, the absolute max voltage is 20v. You should also plan for spikes to be twice that and you can't ever exceed the absolute max gate voltage. 10v is enough to turn on the gate as much as it ever needs, higher voltages only mean more energy is wasted in the gate.

 

[williB]

sigh , last night i breadboarded the fet bridge rectifier.. and it still doesnt work as well as a diode bridge rectifier..
i am using some fairly hefty diodes..btw..and when i was cramking/charging two NiMH batteries the diodes were warm to mildly hot , with 1A going through them..whereas the fets werent even warm ( although i wasnt getting nearly as much current through them..) about 0.8A and the battery voltage was affecting the conduction of the fets..which wasnt in the simulation , I think ..
As soon as i get autocad hooked to a laser printer I can make another board for the 3703's ..

 

[Oznog]

What was your schematic? As I recall your schematics kept using an incorrect principle of source voltage polarity rather than current direction.

 

[williB]

You're not actually only trying to charge a 2.4v battery with a 12v source, are you? If that's the case there's already plenty of voltage available and the standard fwb makes a lot more sense here.

I agree , its a lot simpler ,and it works well enough .. It may not be perfect , but what is? besides now i can concentrate on constructing the
personal generator.. 8) look for it in your local Wallmart in a couple of years.. :lol: