Pages: [1]   Go Down
Author Topic: Arduino to 74HC595 to IRF9520?  (Read 785 times)
0 Members and 1 Guest are viewing this topic.
Offline Offline
Newbie
*
Karma: 0
Posts: 21
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

Is it feasible to use an Arduino to shift data out to a 74HC595 shift register, then latch it to turn on and off a P-Channel IRF9520 MOSFET (https://www.jameco.com/webapp/wcs/stores/servlet/Product_10001_10001_670629_-1)?  Maximum sustained current flow will be around 10A, but typical current flows likely to be much less than that.  Also, duty cycle is only 8%.

Also, with a MOSFET, no current limiting resistor need be applied at the base because no current flows, correct?  So the output from the 595 can be connected directly to the gate of the MOSFET, right?

Thoughts?  Thanks!
« Last Edit: March 06, 2011, 03:00:18 am by mechengr » Logged

0
Offline Offline
Shannon Member
****
Karma: 162
Posts: 10496
Arduino rocks
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

In a word, no.

Firstly the IRF9520 is not a logic level device - it needs a 10V or more gate drive so some interfacing between the shift reg and MOSFET would be needed, but more fundamentally its no good for 10A.  It might take 1.5A before overheating and perhaps 6A with a big heatsink, but 10A would have it dissipate 60W!

You don't say what voltage you want the MOSFET to control nor why you need a P-channel one.

Quote
Also, with a MOSFET, no current limiting resistor need be applied at the base because no current flows, correct?  So the output from the 595 can be connected directly to the gate of the MOSFET, right?

Not really - the gate of a MOSFET is like a capacitor - a current flows until its charged up - power MOSFETs have capacitances around 1nF to 50nF or so.  For high-frequency switching you typically put a few amps into the gate to make it switch fast enough. A current-limiting resistor is often used to prevent high-frequency oscillation.  For low-speed switching this is less important - you can drive a logic-level FET direct from logic if you don't want high-speed switching and can cope with switching times of several us.  A current-limiting resistor of 100 ohms or so will lessen the stress on the 595 output pin.  Incidentally if the current through the source-drain circuit is fluctuating there will be current in and out of the gate since the gate-capacitance depends on the drain current.  A lot of people don't realize this.

You seem to require a logic-level FET with a Ron of 0.05 ohms or better - its only the older MOSFET designs that have such high Rons, its not hard to find N-channel MOSFETs with Ron of 0.01 ohm and P-channel with 0.03 ohm....  Remember heat dissipated is I-squared-R.  The current rating for a MOSFET is its least useful figure of merit since its always a power-dissipation limit with infinite heat-sink!
Logged

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

Offline Offline
Newbie
*
Karma: 0
Posts: 21
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

In a word, no.

Firstly the IRF9520 is not a logic level device - it needs a 10V or more gate drive so some interfacing between the shift reg and MOSFET would be needed, but more fundamentally its no good for 10A.  It might take 1.5A before overheating and perhaps 6A with a big heatsink, but 10A would have it dissipate 60W!

Sorry for the confusion.  The 10A draw would be pulsed (8% duty cycle), so I think it would be okay, right?  The data sheet says maximum pulsed drain current is 27A (https://www.jameco.com/Jameco/Products/ProdDS/670629IR.pdf).  Looks like the old Jameco link on my first post didn't come through correct.

You don't say what voltage you want the MOSFET to control nor why you need a P-channel one.

The MOSFET is the source for 144 LED's (12x12), which will be controlled by a TLC5941.  The current limiting resistor will be set at probably ~70mA per channel when pulsed, ~20mA per channel when steady-state.  The forward voltage per LED is ~3.4V and they are all in parallel.

Since the TLC5941 sinks current, the MOSFET has to source the current (each panel of 12x12 needs to be switched on and off at 8% duty cycle).  My understanding was that P-channel MOSFETs (and PNP BJT's) usually source current while N-Channel MOSFETs (and NPN BJT's) usually sink current.

Quote
Also, with a MOSFET, no current limiting resistor need be applied at the base because no current flows, correct?  So the output from the 595 can be connected directly to the gate of the MOSFET, right?

Not really - the gate of a MOSFET is like a capacitor - a current flows until its charged up - power MOSFETs have capacitances around 1nF to 50nF or so.  For high-frequency switching you typically put a few amps into the gate to make it switch fast enough. A current-limiting resistor is often used to prevent high-frequency oscillation.  For low-speed switching this is less important - you can drive a logic-level FET direct from logic if you don't want high-speed switching and can cope with switching times of several us.  A current-limiting resistor of 100 ohms or so will lessen the stress on the 595 output pin.  Incidentally if the current through the source-drain circuit is fluctuating there will be current in and out of the gate since the gate-capacitance depends on the drain current.  A lot of people don't realize this.

That's pretty interesting.  I didn't realize that's how MOSFET's work.  I am looking to switch the MOSFET's at around 240 Hz.  This is probably overkill for what I am looking to do, but it should eliminate flickering of the LED's.

You seem to require a logic-level FET with a Ron of 0.05 ohms or better - its only the older MOSFET designs that have such high Rons, its not hard to find N-channel MOSFETs with Ron of 0.01 ohm and P-channel with 0.03 ohm....  Remember heat dissipated is I-squared-R.  The current rating for a MOSFET is its least useful figure of merit since its always a power-dissipation limit with infinite heat-sink!

Do you have any recommendations for part numbers that might fit my requirements?  I guess something like a high current, logic-level FET (per your recommendations)?  Thanks for your help!
Logged

Offline Offline
Newbie
*
Karma: 0
Posts: 21
View Profile
 Bigger Bigger  Smaller Smaller  Reset Reset

What about the TIP120?  I can drop the current for each LED down to fit the 8A pulsed limit.  Then, the base input current would only be 8mA.  The only issue is that it's an NPN transistor, so the voltages at the transistor can get funky, but we have plenty of these transistors around the labs.
Logged

Pages: [1]   Go Up
Jump to: