Pages: [1] 2   Go Down
Author Topic: MOSFET driver external circuit  (Read 1149 times)
0 Members and 1 Guest are viewing this topic.
Vancouver, Canada
Offline Offline
Full Member
***
Karma: 0
Posts: 118
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Hello,

I have a IXDD614 MOSFET driver (http://www.clare.com/home/pdfs.nsf/www/IXD_614.pdf/$file/IXD_614.pdf) and I want to use it to drive a IR hexfet power mosfet (http://www.irf.com/product-info/datasheets/data/irlb3036pbf.pdf). The mosfet will be flowing upwards of 60A through it for a period of 300ms (I am not PWMing the driver). what would I need in terms of outside circuitry to make sure the driver works properly? Like do i need decoupling caps or output series resistance and the values as well?

Thanks

Logged

Offline Offline
Edison Member
*
Karma: 116
Posts: 2205
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Check the soa in the datasheet. As you are pushing to 10w dissipation, you may want to put a heatsink on it - I would use a to3p device for that kind of current.


Logged

Vancouver, Canada
Offline Offline
Full Member
***
Karma: 0
Posts: 118
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

I have heat sinks on the mosfet and they work fine.

what is soa? and which data sheet ?

Thanks
Logged

0
Offline Offline
Shannon Member
****
Karma: 214
Posts: 12406
Arduino rocks
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

SOA = Safe Operating Area - see the graph on the MOSFET datasheet of load voltage against current - there will be contours for various
pulse widths on the graph (figure 8 on the one you posted).

It looks like you won't be able to switch more than 7V at 60A from the datasheet, the 300ms contour is about identical to DC.

The MOSFET driver is very powerful (which will help of course by reducing switching transition time).  However the driver will
need thorough decoupling right at the chip and the connections between it and the MOSFET gate/source will absolutely
need to be very low inductance or you won't get the benefit.  The wiring between the two must be short and thick and _must_
be separate from the main (60A) current path to the MOSFET's source lead.
Logged

[ I won't respond to messages, use the forum please ]

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 159
Posts: 2916
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Looking at fig. 8 of the d/s, I see about 15V at 60A for 1-msec pulse, but as MT says, you need short and
VERY heavy traces with tiny inductance, heavy wires connecting the battery/power source, bypass caps
up close on the power-in pin, and maybe a robust diode [reverse-polarity] across the load. And maybe a
fat electrolytic across the bypass cap [helps compensate for inductance in the battery wires].

However, 60A is a lot. You might instead use 3 of the same MOSFETs in parallel, then the dissipation figures
drop down about 10X .... (20A/60A)^2 = 1/9, and SOA is not a problem.
Logged

Vancouver, Canada
Offline Offline
Full Member
***
Karma: 0
Posts: 118
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Hello and thanks for the replies,

So I looked at the graph in fig 8 and I understand what you are saying about the DC current capacity of the mosfet. However I have been safely switching 30A at 12V, and that is also outside the dotted line for DC. The thing is the mosfet does not even get warm (maybe because I have a heat sink on it). This leads me to believe that the graph is not very accurate (or is accurate without the heat sink). Increasing the number of mosfets is certainly an improvement but I am still not sure if that is really needed. This mosfet has impressive characteristics and putting 2 in parallel might be overkill for this application.
My question is about the driver. What value of gate resistor do I need for a period of 300ms (again I am NOT PWMing the Mosfet). Essentially the mosfet is acting as a switch that is only on for a period of 300ms.

Thanks
Logged

Offline Offline
Edison Member
*
Karma: 5
Posts: 1730
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

You don't want any gate resistor, you want it to switch as fast as possible to limit losses
Logged

Offline Offline
Edison Member
*
Karma: 116
Posts: 2205
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
the graph is not very accurate (or is accurate without the heat sink)

It is accurate. Just that no one in this thread has been able to read it accurately.

Quote
You don't want any gate resistor, you want it to switch as fast as possible to limit losses

Bingo!

You do, however, use gate resistors from time to time for dead time management.
Logged

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 159
Posts: 2916
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

I have been safely switching 30A at 12V, and that is also outside the dotted line for DC. The thing is the mosfet does not even get warm (maybe because I have a heat sink on it). This leads me to believe that the graph is not very accurate (or is accurate without the heat sink). Increasing the number of mosfets is certainly an improvement but I am still not sure if that is really needed. This mosfet has impressive characteristics and putting 2 in parallel might be overkill for this application.
My question is about the driver. What value of gate resistor do I need for a period of 300ms (again I am NOT PWMing the Mosfet). Essentially the mosfet is acting as a switch that is only on for a period of 300ms

Oops, I misread you last time, and looked it up for 300-usec. Duh, 300-msec is much different.

However, you were saying 60A, now you're saying 30A. That's a difference of 4X in dissipation, and
that's also much different. For Rds = 2-milliohms, Pd = 30A*30A*0.002ohms = 1.8W, which is not
much when using a heatsink. OTOH, with 60A*60A*0.002 = 7.2W, you'd probably see some heat-up.


Logged

Vancouver, Canada
Offline Offline
Full Member
***
Karma: 0
Posts: 118
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

I tested the mosfet with 30A and I needed to know if it can handle 60A for 300 ms (and it can with a heat sink). From you power dissipation calculation, even with 60A passing through the dissipation is small. Also I have a heat sink attached, which make this work even better.

About the gate resistor, I tried a different driver with 220 Ohm resistor and it blew up. This driver had  +/- 9A peak current. So you would think it would work for this application, But NOOOOO!!! The part number was UCC37322P, if you are curious.
This was the reason that I choose the new beefier driver. +/- 14A and +/- 4A continuous at room temp! However I am afraid that if I dont put any gate resistor, this thing would blow up too. Share your thoughts.

Thanks
Logged

Offline Offline
Edison Member
*
Karma: 116
Posts: 2205
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Quote
However I am afraid that if I dont put any gate resistor, this thing would blow up too. Share your thoughts.

If you do put in such a large gate resistor, it will blow up. What a gate resistor does here is to slow down the charge / discharge of the mosfet, in essence, causing it to dissipate more (higher switching losses).

Very rarely you see gate resistors in switching applications. When you see them, they are typically very small (11ohm or less), and for special purposes.
Logged

Offline Offline
Edison Member
*
Karma: 116
Posts: 2205
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

This is why sometimes switching devices would blow under very light duty cycle.

That puzzles a lot of people, including a now-famous Class D amplifier designer.
Logged

Offline Offline
Edison Member
*
Karma: 5
Posts: 1730
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

The mosfet gate can basically be modeled as a capacitor, the "resistance" of the mosfet depends on the voltage on that capacitor as that voltage changes the internal structure of the mosfet basically
So adding a resistor makes it an RC circuit with a time constant, so to havd the fasting rise time(prefered in most applications) lowering the R in the equation makes it faster, hence why you don't have a resistor, the driver has a low output impedance and why you use short large traces(also because the trace basically makes it an RLC circuit), if you minimize the R&L all you need to focus on is the peak current available to charge the C, so more peak current means a faster charge time, and that ensures the mosfet passes thru its linear or "ohmic" mode as fast as possible, in high frequency situations the switching losses can easily become more than the static losses from the RDS(on) value
Logged

the land of sun+snow
Offline Offline
Faraday Member
**
Karma: 159
Posts: 2916
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

The other things that weren't mentioned are

- is this to go on a pcb?
- what connection terminal blocks are being used, if any?

Eg, if you look at trace info, the curves don't even go up to 60A. Even at 20A, they need
to be very wide and thick.

http://www.ultracad.com/ipccha1.gif
http://wiki.xtronics.com/index.php/Wire-Gauge_Ampacity
Logged

Vancouver, Canada
Offline Offline
Full Member
***
Karma: 0
Posts: 118
View Profile
WWW
 Bigger Bigger  Smaller Smaller  Reset Reset

Hi All,

Thank you for your replies.
I think there is some misunderstanding over here. I remember I mentioned that I used a driver with +/- 9A peak current capacity (UCC37322P). I used several different gate resistor values on this one:
- no gate resistor, and it blew up with not even a load attached to the mosfet (same mosfet).
- 100 Ohm resistor, it worked with no load, but was fireball with 30A load
- 220 Ohm resistor, it worked with 30A load, but explosion with 60A load
- 470 Ohm resistor, works with 30A load, but explosion with 60A load

Now I am using this beefier driver IXDD614CI with +/- 14A peak current, and +/- 4A continuous. I tried it with 470 Ohm gate resistor and 60A load. No explosion or fire, but it failed (latched on).
I need a gate resistor in place of the 470 Ohm (higher or lower, I dont care) that can handle the high 60A current through the MOSFET.
I have attached a PDF schematic of what I am doing.

Thanks.

* New Driver.pdf (13.29 KB - downloaded 15 times.)
Logged

Pages: [1] 2   Go Up
Jump to: