Electronics > Projects, Designs, and Technical Stuff

Wiring/driving Si2305DS SOT-23 P-mosfets correctly

(1/4) > >>

3dgeo:
Hello,

I'm implementing Si2305DS P-fets in my project – using in LED matrix on a high side to drive 45 LED at a time, driven by port manipulation with MCU ~200Hz, not PWM). I've seen people using pull up resistors on a signal line (I added them as well), but is there a need to use them If I drive fets with MCU pin which will be set as HIGH/LOW and may float only when MCU powers on? I assume 10K resistors (I've seen people use 1K) will be OK due to small package and low charge (total gate charge 10-15 nC, input capacitance 1245pF)? I tested them with my hobbyist oscilloscope and 1 LED – with 10K output rising is clean and sharp, but falling is a bit "rounded", I'm guessing with 45 LEDs falling will be way "sharper" (sorry for not providing falling timings)?

Fets datasheet:
https://www.vishay.com/docs/70833/70833.pdf

Thanks in advance.

spec:

--- Quote from: 3dgeo on January 20, 2019, 06:00:16 pm ---I'm implementing Si2305DS P-fets in my project – using in LED matrix on a high side to drive 45 LED at a time, driven by port manipulation with MCU ~200Hz, not PWM). I've seen people using pull up resistors on a signal line (I added them as well), but is there a need to use them If I drive fets with MCU pin which will be set as HIGH/LOW and may float only when MCU powers on? I assume 10K resistors (I've seen people use 1K) will be OK due to small package and low charge (total gate charge 10-15 nC, input capacitance 1245pF)? I tested them with my hobbyist oscilloscope and 1 LED – with 10K output rising is clean and sharp, but falling is a bit "rounded", I'm guessing with 45 LEDs falling will be way "sharper" (sorry for not providing falling timings)?.

--- End quote ---
Hi 3dgeo,

I assume that the port driver (MCU) is operating from 5V rather than 3V3, and that the drain source [edit #1] of the PMOSFET is connected to the 5V supply line.

The effective gate capacitor will be much higher than you think because of the high value of the PMOSFET's parasitic capacitor from drain to gate (CDG).

Because your PMOSFET has such a low minimum gate threshold (VGS[th]) of 450mV, it would be wise to have a pull-up of 1k8.

Also, put a 22R directly on the gate terminal of the PMOSFET, using as short a lead as possible. Then connect the pull up to the 22R.

And connect a 470nF minimum ceramic X7R capacitor (physically large, not surface mount) from the PMOSFET source terminal to the 0V line as close as possible to the PMOSFET.

You would expect the PMOSFET drain voltage to rise fast and fall in a mess due to the LED (+ resistor?) load. If you would like to speed the fall, connect a 220R (or even 100R) from the drain of the PMOSFET to 0V.

3dgeo:

--- Quote from: spec on January 20, 2019, 06:44:59 pm ---I assume that the port driver (MCU) is operating from 5V rather than 3V3, and that the drain of the PMOSFET is connected to the 5V supply line.

--- End quote ---

Thank You for your replay!

DRAIN connected to 5V??? Shouldn't DRAIN be connected to load and SOURCE to 5V??? I'm sooo confused now....  :-//
Funny as it is, today in my schematic and PCB design I flipped all sources and drains cos I was connected them wrong myself, drain was on 5V and source on load, but examples and tests shows that source should be connected to 5V and drain to load. Am I going nuts?   |O :D
Yes, power line is 5V.

Can you give a bit more information on how you got these numbers and why these components have to be used? I want to learn, not only implement and forget it :) Sadly, I can't add not SMD components, I only have max 4mm above PCB.

P.S. low side of the LEDs will be PWM with TLC59116, it has constant current limit, so no need for external resistors, don't know if that makes any different...

spec:

--- Quote from: 3dgeo on January 20, 2019, 08:00:07 pm ---DRAIN connected to 5V??? Shouldn't DRAIN be connected to load and SOURCE to 5V??? I'm sooo confused now....  :-// 
--- End quote ---
:palm:


--- Quote from: 3dgeo on January 20, 2019, 08:00:07 pm ---Can you give a bit more information on how you got these numbers and why these components have to be used? I want to learn, not only implement and forget it :)
--- End quote ---
First the PMOSFET gate resistor. That is a gate stopper which discourages the PMOSFET from oscillating of its own accord, typically from 50kHz to 20MHz. The gate stopper plays no part in the fundamental operation of the circuit.

The gate of a MOSFET only requires voltage to control the drain current, so it follows that to turn the PMOSFET on and off fast, you need to have a fast voltage change between its gate and source. But the PMOSFET has a large real and virtual capacitance between its gate and source, so to generate a fast signal at the PMOSFET gate you need to charge the gate capacitance up and down fast, and that requires current, the more current the faster.

But there is a further complication with the gate of the PMOSFET. It only turns off at 4V2 so the MCU driving the PMOSFET gate cannot provide much current to discharge the gate capacitor and turn the NMOSFET off. This is where the pull up resistor comes into play: it discharges the gate capacitor to 5V and turns the NMOSFET off relatively fast.

When it comes to turning the PMOSFET on, the MCU has plenty of voltage swing, probably down to 0.5V. The PMOSFET, thus turns on at 4V2, but the MCU carries on down to 500mV which provides a large drive to charge the PMOSFET gate capacitance and turn the PMOSFET on fast.

Moving on to the drain of the PMOSFET. As stated, the PMOSFET has a nice big turn on gate drive so there will be bucket loads of current pouring out of its drain. That current will turn LEDs on and charge up capacitances as fast as hell.

But, when the PMOSFET is turned off (even fast) everything on its drain is just left hanging in space with no discharge path to 0V: the drain pull-down resistor provides that discharge path.

How were the particular values chosen: by precedent and  compromise.  The gate stopper was by precedent and compromise- I know that a resistor of around 10R to 1K is OK for gate stopping, and would normally have chosen 100R, but 100R would have slowed the PMOSFET gate drive too much, so I went for 22R as a compromise.

The pull-up resistor was also chosen by compromise. On one hand I wanted it to have a low value for fast PMOSFET turn-off, but on the other hand the minimum value was limited by the current sink capacity of the MCU.

The pull down resistor was also chosen by a compromise between speed (low value)  and limiting wasted current (high value).

In general, for speed you need to charge and discharge capacitance fast, and  to do that you need to source and sink high currents. To illustrate this, there are MOSFET gate driver chips which can swing 15V and source and sink 9A (see link below).

And finally, design in any field, is very much about optimizing compromises.


--- Quote from: 3dgeo on January 20, 2019, 08:00:07 pm ---Sadly, I can't add not SMD components, I only have max 4mm above PCB.
--- End quote ---
OK, use SM, but 1uF if you can.


--- Quote from: 3dgeo on January 20, 2019, 08:00:07 pm ---P.S. low side of the LEDs will be PWM with TLC59116, it has constant current limit, so no need for external resistors, don't know if that makes any different...
--- End quote ---
No- it will be fine, but be sure to fit the decoupling capacitor.

http://www.ti.com/lit/ds/symlink/ucc27321.pdf

3dgeo:
Thank you, it's a bit more clear now.

Well, this will probably make no different but I'll switch to Si2305CDS
To be fair, I didn't expect that I have to add so many components  ;D I already made place for pullups, I may even squeez in 22R, but I don't have grounding in that part of a circuit  :scared: You said "fast", I don't think 200Hz is fast, in worst case scenario what can happen if I skip 22R and caps? :) I'll do some more testing tomorrow to find out its oscillation at 200Hz. I'm also using bunch of 10K resistors, so it would be really nice to use those as well.... I'm not pissing away all your effort here, I just wonder, what if?  :popcorn:
Did I understood circuit correctly:

Navigation

[0] Message Index

[#] Next page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod