The MOSFET does not operate PWM and only turns ON/OFF

Thread Starter

Kim-JiHoon

Joined May 3, 2020
133
Hello,i made a simple circuit.



bbb.PNG



and this is


aaa.jpg

Yellow : Q9's Gate sig

Green : PC817 Num1 pin sig

the problem is 'PWM duty is not applied'
1~99 duty = 99 duty (only ON/OFF)

this is my tried

1. change R20 to 330
2. change R25 to 1~10K
3. change R26 to 0~10K
4. chagne D13 removed
5. change R27 to 1~47K

all same result
 

Ya’akov

Joined Jan 27, 2019
9,269
Well, that gate waveform is definitely largely the capacitive discharge. But there are clearly other influences. Have you tried an R27 of a much lower value to see if the waveforms changes? I mean like 100Ω, so it is very low resistance to 0V. Just meant a s a troubleshooting step.

Someone else might immediately recognize the problem, I am just trying to work it out logically.
 

Ya’akov

Joined Jan 27, 2019
9,269
If you wired the circuit exactly as in the schematic the FET will now be dead, as it directly shorts the 24V supply.
So it does. I was so focused on the gate-related portion I didn’t even notice that. Ouch. Good thing I am not working on mission critical systems…
 

Thread Starter

Kim-JiHoon

Joined May 3, 2020
133
Well, that gate waveform is definitely largely the capacitive discharge. But there are clearly other influences. Have you tried an R27 of a much lower value to see if the waveforms changes? I mean like 100Ω, so it is very low resistance to 0V. Just meant a s a troubleshooting step.

Someone else might immediately recognize the problem, I am just trying to work it out logically.
yes i tried 1K ~ 47K
the diode is heating up too much
I think the diode is broken

but i have nothing suitable
I'll buy a new one and try it
 

ci139

Joined Jul 11, 2016
1,898
Last edited:

Ian0

Joined Aug 7, 2020
10,277
Now you know how slow optoisolators are.
From the datasheet: maximum rise time 18us. Seems about right to me.
IRLR7843 will switch 160A and has the sort of gate capacitance one would expect from a FET that large: 4.4nF
Combined with the 10k resistor, it gives a time constant of 44us. Waveform Also seems about right.

There are isolators designed to drive MOSFET gates such as HCPL3120.
 

Thread Starter

Kim-JiHoon

Joined May 3, 2020
133
http://www-vlsi.es.kit.ac.jp/thesis/papers/pdfs/COMPEL2017_yamashita.pdf

(( depending the opto type https://www.vishay.com/docs/83590/fastswit.pdf#page=11 ))

See Fig.3 . . . you may want to move the 15V zener to the collector node , assuming 1V drop on PC817 : \(24-15-1=8\) volts to drive the mosfet

+ you may want to revisit your opto configuration and maybe the mosfet's channel polarity https://www.farnell.com/datasheets/73758.pdf#page=8
i think it seems a little difficult for me. bu i'll read it slowly

Now you know how slow optoisolators are.
From the datasheet: maximum rise time 18us. Seems about right to me.
IRLR7843 will switch 160A and has the sort of gate capacitance one would expect from a FET that large: 4.4nF
Combined with the 10k resistor, it gives a time constant of 44us. Waveform Also seems about right.

There are isolators designed to drive MOSFET gates such as HCPL3120.
i think i don't need to isolators . motor is just 24V/2.1A and this is not an environment where noise filters should be applied.

so i'm thinking of trying N-BJT instead of optocoupler. things like 2N2222, 3904..
 

Ian0

Joined Aug 7, 2020
10,277
i think it seems a little difficult for me. bu i'll read it slowly


i think i don't need to isolators . motor is just 24V/2.1A and this is not an environment where noise filters should be applied.

so i'm thinking of trying N-BJT instead of optocoupler. things like 2N2222, 3904..
Do the job properly - get a LOW SIDE DRIVER such as MCP1402. There are hundreds to choose from.
 

Papabravo

Joined Feb 24, 2006
21,325
I think you are approaching the design of circuits in the wrong way. It seems that you are throwing components onto a drawing in the hopes that you might get something to work. In the process you are revealing that you have no deep understanding of fundamental device behaviors and operations. Might I suggest that you study and build some things that are known to work so that you can analyze and understand them.

EDIT: The yellow trace in the original post shows two flat areas. The short one is on the rising edge and the long one is on the falling edge. These are called "Miller Plateaus", so named because of the Miller Effect. What is happening is a dramatic change in the gate capacitance with respect to the other nodes of the device. On these plateaus the device is neither on nor off, it is temporarily stuck. A great deal of power is dissipated and lost in these conditions. How you drive a MOSFET is important, and an active pullup with a passive pulldown, like you did, is probably the single worst way to do that.

https://en.wikipedia.org/wiki/Miller_effect
 
Last edited:

tindel

Joined Sep 16, 2012
938
If you’re driving a 24V computer fan or equivalent, then you are making this way more difficult than it needs to be. Get a logic level FET with a lower gate charge and drive the input with your microcontroller or whatever. Connect the fan+ to the 24V rail, connect fan- to the drain of the FET, and you're off to the races. Don't forget to add a back EMF diode too. I can't imagine that you need such a large transistor (161A! is quite a lot). You probably need something that will work with 1-2A if you're PWMing a small fan. If you have a huge fan, then you just need a switch to turn it on and off - doesn't make sense to PWM a big fan I suspect.

10kHz seems a little fast, but not crazy. Maybe slow it down to 100-1kHz or something get it working slow at first.
 

Thread Starter

Kim-JiHoon

Joined May 3, 2020
133
If you’re driving a 24V computer fan or equivalent, then you are making this way more difficult than it needs to be. Get a logic level FET with a lower gate charge and drive the input with your microcontroller or whatever. Connect the fan+ to the 24V rail, connect fan- to the drain of the FET, and you're off to the races. Don't forget to add a back EMF diode too. I can't imagine that you need such a large transistor (161A! is quite a lot). You probably need something that will work with 1-2A if you're PWMing a small fan. If you have a huge fan, then you just need a switch to turn it on and off - doesn't make sense to PWM a big fan I suspect.

10kHz seems a little fast, but not crazy. Maybe slow it down to 100-1kHz or something get it working slow at first.
LR7843 has Very Low RDS(on) at 4.5V VGS
i already did that [CU PWM SIG] - [FET's gate ]
but FET get's too hot over 40 ℃
it's too hard for the MCU's I/O pin to onnect FET's gate directly
that is the reason why i'm doing this.
i think i have to fully open FET's gate.

i have motor controller modules. so analyzed and copied the circuit.
most were made up of 555 Timer ICs

and i found some modules maed by optocoupler.
I think you are approaching the design of circuits in the wrong way. It seems that you are throwing components onto a drawing in the hopes that you might get something to work. In the process you are revealing that you have no deep understanding of fundamental device behaviors and operations. Might I suggest that you study and build some things that are known to work so that you can analyze and understand them.

EDIT: The yellow trace in the original post shows two flat areas. The short one is on the rising edge and the long one is on the falling edge. These are called "Miller Plateaus", so named because of the Miller Effect. What is happening is a dramatic change in the gate capacitance with respect to the other nodes of the device. On these plateaus the device is neither on nor off, it is temporarily stuck. A great deal of power is dissipated and lost in these conditions. How you drive a MOSFET is important, and an active pullup with a passive pulldown, like you did, is probably the single worst way to do that.

https://en.wikipedia.org/wiki/Miller_effect
i tried fully off the mosfet. so i cahnged pull down resistor 1K~47K

and added pull down resistor optocoupler's base.

but it didn't fully off. so it cause of parasitic capacitor?
 

Ian0

Joined Aug 7, 2020
10,277
I’d never use a MOSFET that large without a gate driver.
If the drive signal comes from a microcontroller, is it capable of sourcing and sinking 50mA?
Assuming a saturation current of 4mA, it is still going to take 12.5us to charge or discharge the gate (50nC/4mA), which might be OK for occasional switching but will result in noticeable power dissipation above 4kHz.
 

MisterBill2

Joined Jan 23, 2018
19,474
In designing a PWM circuit the very first specification to verify is the switching speed. Then the current and voltage ratings. Then the driving circuit must be able to support the switching rate with enough current for the required switching time. Then there is a chance of the circuit working.
 
Top