Selecting Low RDS(on) Logic Level N-Mosfet for 3.3V Microcontroller & 12V Motor Control

Question

Hello,
I plan to make a simple PWM system to control a 12V motor controlled from a microcontroller powered by 3.3V, hence the question about a logic level mosfet with the lowest possible RDS(on) resistance, with which I can control a motor with a load current of 10-15A? Will a heatsink be necessary at this current and low RDS(on) resistance? Any example of optoisolation for 3.3V? Thank you in advance...rob

tmf · Accepted Answer

The optocoupler has one actively forced state (here ground), while the transition to 12V will take place passively through the resistor (it may be the other way around, it doesn't matter). You have a CTR of 50-600%. So in the worst case 20mA through the LED will give you 10mA at the output. It's miserable. The transistor will switch for a long time and will get hot. Driver is a simple matter - you have application notes, the ones given do not require anything practical beyond that. Consider that at 15A, even if the switching time is only 100us, at the given frequency it will be 20ms, so 5% of the whole period. And 100 Hz is not much and there may be a problem with motor control. Let's assume that for these 5% of the time the transistor will have an average RDSon of not 4 but 40 mOhm. This will give us an average of 0.45W of losses. If the average is 100 mOhm, the losses will already be 1.25 W. Of course, these are not precise calculations, but rather very rough estimates. You can roughly calculate the switching time knowing the gate capacitance and driver current. Probably the given 100 us is still very optimistic.

Tytus Kosiarski · Accepted Answer

There is also the optodriver TLP351. And then you have no problems with driving the MOSFET gate with a higher voltage and its fast switching. It is available at TME, just like the TLP250.

Oh, here: http://skory.gylcomp.hu/alkatresz/toshiba%20optocouplers.pdf is a PDF (p.63), how to use this TLP to control a transistor, admittedly an IGBT. But instead of an IGBT you will insert your MOSFET and remove the Vee voltage source.
Regards, KT

diquan · Answer

As for the mosfet, see PSMN2R7-30PL (100A / 30V / Rds (on) 0.0023 Ohm). The insulation is e.g. Optocoupler 6N137.

tmf · Answer

10-15A is not such a small motor anymore

Use some product selector, e.g. on farnell, after entering N-MOSFET, you select the appropriate tab, then you specify the required parameters and a list of transistors that meet your criteria appears.
At 10-15A, finding an N-MOSFET with an RDSon small enough to not need a heatsink can be a challenge. With a VGS of around 10V, you can easily get a MOSFET with an RDSon of 8 mOhms, which would give 2W static loss for 15A, so a copper field under the transistor would be enough, and a double-layer board with vias increasing heat conduction would be a good idea.
Ew. you can think about some H half-bridges - Infineon has excellent ones, e.g.

Added after 4 [minutes]:

In the meantime, the colleague above gave the direction to a quite interesting transistor.

robiw · Answer

23mOhm is a bit much. At 10A, this gives 2.3W of loss . Can you provide an example optoisolation scheme? robin Added after 6 [minutes]: ....OK, I've already noticed my mistake 2.7mOhm...great, but it's not in TME. I also need a transistor in a TO220 package... robiw

Freddie Chopin · Answer

Count the zeros well (; Added after 1 [minutes]: TME is not the only company in the world supplying electronics

robiw · Answer

Of course it's not...but from them I got...free...
I found:

INFINEON (IRF) IRLB8748PBF
Transistor: N-MOSFET; unipolar; logic level; 30V; 92A; 75W
4.8mOhms

PLN 2.6...

I counted the zeros above ;-)

. Any proven scheme with an optoisolator? robin

diquan · Answer

An example of optoisolation can be found here:

robiw · Answer

Even better is the IRL7833PBF...

tmf · Answer

It's just a small but... if you look at graph 3 (Typical Transfer Characteristics) you'll notice that this transistor has fantastic parameters (like the previous ones) at VGS of 5-6V, and for some 10V. Then you have nominal RDSon. You wrote about control with 3.3V, RDSon will be many times higher. Also, it would be nice to add some minidriver that will increase VGS. BTW, why do you need optoisolation? If this is really to work, you would have to completely isolate the power module electrically.

Anonymous · Answer

@robiw IRLB8748 not suitable for this application. You have a 3.3V output at your disposal and you need a current of 15A. You don't specify the frequency of operation or the solution of the control stage, but we can make some assumptions. Controlling directly from the uC is out, because you will not overload the C_GS capacitance (plateau is >3.0V). Also, in Figure 1 it is clear that for 10A current and VGS 3.0V you must have a minimum VDS ~=2V. This gives at least 20W, which forces Rthja max to be 6°C/W. It may work, but I'm skeptical. In addition, Figure 12 shows that it makes sense to apply at least 4V to the gate. The other one, IRL7833 looks better, but you need to check if you can fit into the SOAR - Figure 8, because Figure 6 shows a plateau shift closer to 4V. By supplying it with 3V, the state change will take a relatively long time, which will generate losses and heat the structure. On the one hand, it's good because the threshold VGS will drop and the system has a chance to stabilize, but the channel resistance will increase quickly and it's not a good engineering practice. If you want to provide optoisolation, why not power the gate with a higher voltage, e.g. 12V? You then have a much larger selection of transistors with lower resistance, cheaper, in different packages.

robiw · Answer

I was just wondering if it makes sense, because the microprocessor system will be powered from the same 12V voltage (through the stabilizer) as the DC motor, but with a different pair of wires ... Added after 2 [minutes]: I will rather give up this isolation, but I will use an additional transistor with pull-up to +12V. Is a typical NPN transistor enough? robin

krzysiek_krm · Answer

Hello,
I think that a good solution is to use a specialized driver: tiny, dirt cheap, easy to drive with almost any logic, available with different output currents, etc.

robiw wrote:
because the microprocessor system will be powered from the same 12V voltage (through the stabilizer) as the DC motor, but with a different pair of wires...

You should probably be a little careful not to inadvertently create a parasitic ground loop, unless that's what you need galvanic isolation for.

Regards

yogi009 · Answer

And that's the essence of this discussion.

robiw · Answer

Any specific type? I've never used ... robiw

krzysiek_krm · Answer

Because my guess is that you are going to use a transistor in a fairly simple setup n-channel, low-side so you can use for example alternatively and also: Analog Devices, Maxim, Fairchild, Linear, etc. First you should choose the right transistor. In the general case, not necessarily. If the frequency of your PWM control will be relatively high, you should rather look for a transistor that the manufacturer advertises as Excellent gate charge x RDS(on) product (FOM) or similar, on the one hand, such a transistor has quite a low channel resistance, on the other hand, it requires switching the transfer of a relatively small load, you will not have to jerk it with some monstrous driver. You could also build yourself up a bit theoretically by browsing through a handful of items titled Power Mosfet Switching.

kowal011 · Answer

Search on PC motherboards.

robiw · Answer

However, I will provide optoisolation on LTV817 pulled up to + 12V and a transistor with as little RDS(on). PWM frequency of the order of 500Hz...dow

tmf · Answer

@robiw The LTV817 itself will turn on forever. At 15A and 500Hz you'll fry the transistor, or at least get it hot. You need a driver that will reload the gate quickly. At least an additional transistor controlled from this optocoupler will be required. And preferably two as a push-pull. At 15A, even if the transistor is briefly in the transient state, the losses will be huge. Such currents are no longer forgiving of errors, so I would buy the suggested MOSFET driver.

robiw · Answer

I'm not familiar with this topic of drivers, hence the reluctance to solve it. And reducing the frequency to 100Hz will not improve the situation? Or maybe a low gate capacitance mosfet? R

robiw · Answer

I found something else like TLP250...R

robiw · Answer

TLP250 has a simpler application without additional transistors.. r Added after 26 [minutes]: Here is an example diagram:

robiw · Answer

Following your advice, I built a layout as below:

Connections:
- 12V fan between FAN-1 and FAN-2,
- 12V power supply for FAN-2,
- mass for FAN-3.

It worked for a few moments, then the C6 went up in smoke and the system stopped working, but I'm not sure it's the fault of this implementation. did i miss something? Thanks in advance...do

michalko12 · Answer

Well, there are no miracles here. C6 released a magic balloon either due to overvoltage or reverse connection of C6 or supply voltage. The supply voltage may have increased if there is a diode at the output of the power source, because the arrangement of the elements fan, D2 and C6 created a boost converter circuit. There was no power supply. R8 is redundant.

robiw · Answer

As for C6, it's clear, it was probably too low voltage (16V), but why was the system damaged? By surges? After all, there is D2 parallel to the fan. What transil do you suggest (power supply max. 14V)? robin

michalko12 · Answer

Was the fan power supply directly connected to the power supply, and e.g. the wires from the power supply to this system were thinner, longer? Diode D2 would do its job well if it was mounted close to the load.

robiw · Answer

The fan was connected to the FAN terminals and the power supply to them. The wires are about 20cm long, so it shouldn't matter...dow

michalko12 · Answer

Somehow some of the energy from the fan preferred to be deposited on C6 and then U4 rather than completely deposited on D2. Something was wrong between the D2 cathode and the FAN2 terminal, maybe the terminal itself is of poor quality, or something was badly screwed/clamped. I can't think of anything else.

excray · Answer

C6 works as an energy store. When the key is turned on, a significant amount of current flows through it. Make sure that it is a low esr capacitor designed to work with high currents in the impulse.

[Solved] Selecting Low RDS(on) Logic Level N-Mosfet for 3.3V Microcontroller & 12V Motor Control

Post #1

Post #2

Post #3

Post #4

Post #5

Post #6

Post #7

Post #8

Post #9

Post #10

Post #11

Post #12

Post #13

Post #14

Post #15

Post #16

Post #17

Post #18

Post #19

Post #20

Post #21

Post #22

Post #23

Post #24

Post #25

Post #26

Post #27

Post #28

Post #29

Post #30

Topic summary