Hello there, So I was using a l298n for motor control but it had a very high voltage drop for high currents and I need something more accurate to gather data in order to identify the paramters of dc motors for later study so I though of a power mosfet bridge.
This makes me wonder about some fundamentals of the mosfet.
I want to use low logic level mosfets and as I reserched I should have p-channel on top and n-channel on bottom and that I understand.
So the n channel with Source connected to Ground and with an input Vgs = 5v for example will saturate the n mosfet and I can supply any voltage at the drain from zero to the Max voltage they give. I know that -5v as Vgs also works the same.
So here I use 0 or 5v to control it.
Now for the P channel.
If I want a logic level P channel How would that workout ?
If I have a motor that needs 20V I can see that connect Vgs to ground will make it conduct. but if I put 5v on the Vgs it will still conduct because theres this 15v diference Vgs.
On the other hand I can connect Source to Gata via a 10kohm holding it off and with a transistor connect gate to ground to turn it ON.
Thats how I usually see it implemented.
But now my really question....
-What is the point of the P channel being LOGIC LEVEL if I cannot control it just with the 5V
-Other thing is that P-channel for me looks like an N-channel but connected the other way around, is it ?
-If I connect the load the the source of a n-channel mosfet will it act as a p-channel or only if I switch the source by the input?
With P-channel, you will use a lowish-value resistor to pull the gate to 20V to turn it off. Use an NPN transistor (or an N-channel MOSFET) controlled by Arduino to pull the gate Low to turn it on.
You'll have to watch the P-channel's VGS to make it is not exceeded by pulling it low.
Standard MOSFET is okay when the voltage being used is over 5V and the extra transistor in front is needed.
That is why a transistor is used for the p-channel mosfets.
You can not use a negative voltage to a gate to a n-channel mosfet. It won't work.
You also can not use a load at the source of a n-channel mosfet, unless you have a higher voltage than 20V and there is a very special reason for it (trust me, there are no special reasons).
The p-channel is the other way around. The source is connected to the 20V power supply, the drain to the load, and the Vgs (gate voltage) is the difference between the 20V and the gate. It is easy to remember, since the Vgs is important for n-channel and p-channel, therefor the drain is always connected to the load.
You can also use a special mosfet driver chip. Those are designed to drive the gate of a mosfet. When you use such a chip, you don't need 'logic level' mosfets anymore.
I did some research but this little things that I couldn't understand, as for the n-channel not working for negative voltage I read in some places that the voltage range would go -Vgs + vgs
So I am wondering the lowest voltage supply I can use to power a motor and the mosfets(the p-channel ones) still function near to what is mentioned on the datasheet for the voltage they use to test the parameters, usually 10v.
For example this one: http://inmotion.pt/documentation/sparkfun/COM-10349/FQP27P06.pdf
They test it at Vgs= -10v and if I use a 7V source It will be some what Ok for a vgs(th) of -3, right ? but I will have more Rds and a lower maximum ID right ?
Vgs(Vth)= -2v -4v;
Other thing about the p-channel mosfet is:
the max Vds is -60 and max Vgs is -25v - The max 25v on the gate will limit my source voltage to 25V ?
Lets say I supply it with 50V (60V max) then to turn The p-channel mosfet off I would need to pull the 50V to the gate resulting in a burn out ?
So they can be running with 50V and no problem but It would make it impossivel to turn them off with 50V at the gate as it would burn them right ?
I am sorry if not clear, I just want to have sure that what I learned is true so I don't burn this up ;D
Yes, that is right. A negative voltage to a n-channel will keep it 'off', and that mosfet breaks down at -20V.
Those drivers are too expensive ? But I don't like that schematic :~
I like a circuit like this:
I have also seen circuits with optocouplers.
For a p-channel mosfet in a h-bridge is indeed a problem that the Vgs could be beyond its limits.
You could add a protection zener diode.
When you look at the figures in the datasheet, you might think that it is okay to have the mosfet only 50% on. But that is not okay. The mosfet will get hot, and that changes its characteristics. It might also get too hot. You can even calculate everything to be sure that it will work, but when you try it for real, you see that it won't work. As a rule of thumb: when you use a mosfet as a switch, make sure it is always full on or completely off.
"Negative" in this case is relative only; Vg is negative compared to Vs to turn a P-channel on. Vg is positive compared to Vs to turn an N-channel on.
Nothing is actually negative with respect to Gnd. Don't let that throw you off.
The "negative" was a question about the n-channel IRLR8743PbF mosfet.
In that datasheet it says: Absolute Maximum Ratings: Vds 30V, Vgs +/- 20V.
The +/- 20V could be confusing.
I have to thank you very much for the time you put aside to answer this questons, The bridge that I sent its in a way that I would only need 2 pins to control the direction and 1 for pwm (also more expensive).
So the zener diode would be in series with the 10kR in order to drop some of the supply voltage, didnt think of that .
I'm thinking to use a PWM of 32KHz, that is like 31.25uS, and if the mosfet has a rise and turn off time of 100ns would the pwm really make it that hot ?I will apply pwm to one of the mosfet for each direction, the N-channel ones. I ONLY need it to work during 10 seconds which is the time for applying each signal I don't need it to work for minutes straight.
I just need a bridge where I can use pwm and doesn't have a big voltage drop as the l298n has. I might have peaks of 8A current when motor is responding to a step wave for instance from 12v to -12v.
With the l298 I had a really big voltage drop for bigger pwm frequencies and it was just not working for me to get the real data from the motor.
After I want to get the velocity and current measures and with the signals I apply I will get the motor parameters and model it in simulink. In My project the goal is to make this plataform where I can use it with different motors so that someone using it can quicly follow some easy steps to get the motor parameters with a basic linear model and also non-linear for more precision.
Having this in mind I though that Mosfets with low RDSon would be the best for this type of application.
Edit:
About those diodes in paralel with the mosfets, usually they come with the mosfet package? or it is good practise to add some of my own? Because I always see them in the datasheets(The motor will be changing from full speed to full reverse speed, I will lay them like I did for the l298 (http://www.laurentkneip.de/steppers_files/bipolar_circuit.jpg) Do they need to be shocky ones, Flyback ? In some schemes they use Ultra-Fast Avalanche Sinterglass Diodes.
With all these ideas in mind I am thinking, are the logical levels the best option ? I just need a very low RDSon and probably a switching capacity than can easy handle 32khz of pwm, I will be inputing signals to the motor, of 10hz maximum not more because I am reading velocity data and current at 100hz.
-As in what matters the p-channel I can supply a max supply corresponding to the max Vgss,(add a 5v zener and it can handle extra 5v).
-Now for the lowest supply I could input would be recomended to be at least the same they use it to calculate parameters such as Rdson. in this case at least 10v I would know it will function as in the datasheet, correct ?
This is all I need and I will start to choose mosfets for the bridge
The L298N can be used with the Arduino PWM of 500Hz, and for 12V toy motors.
You forget one thing: the gate capacitance.
The gate of a mosfet is like a capacitor. Because of that it is slowly charged. How slow depends on the circuit of course. So the rise time is a slower than what is written in the datasheet.
When you want 32kHz PWM, you better use one of those special mosfet driver chips. Some of those chips have an output that combines a normal transistor and a mosfet to blast a big amount of current into the gate. I have used such a chip, but I have to search for it to see which one.
I will tell you that the l298n at 60% PWM(corresponding 14.4v) gives me:
120rpm at 490hz it rotated even faster, some how I am getting more voltage 17-19V
50rpm at 7.8khz the bigger the frequency the bigger voltage drop across l298
0rpm at 32khz
The motor should rotate at something around 100rmp at 14.4(VA voltage at the motor terminals) acording to the KE velocity constant and experimental data with regulated source. But I get diferent velocity values for diferent pwm frequencies so I cannot use this to determine motor parameters as they will be diferent for every frequency. I just want a bridge that is not affected this much by the pwm because I need precise values.
and at 60% I should get a voltage drop of 14.4V across the motor. Instead I am getting around 8 at the motor and 6.4 at the l298N 6.4V.... (at 7.8khz pwm) My current is 0.2A I gess the l298 everytimes the pwm raizes from 0 to 255 creates current spikes that are attenuated at the l298.
This is the reason I want to do it with mosfet and not with transistors. Because I tried with a regulated voltage source and all the values match up but not if I use pwm to achieve the same values. the motor I was testing can go as high as 11A if I block it.
I can have some voltage drop, but not this much...
For fast PWM I'd go for proper H-bridge design - MOSFET driver chips and all n-channel
on the bridge. High-low MOSFET drivers bootstrap a supply to driver the upper n-channel
switches above the supply voltage (in fact the supply floats with the output line).
It is indeed important to switch fast with fast PWM - any circuit with 10k pullup
resistors on the gate is going to fail bad at 30kHz speeds, you want, perhaps 10 to
100 ohms gate resistance (and most driver chips already have that sort of output
resistance).
My suggested reading list starts with the HIP4081A datasheet, since that's a full H-bridge
MOSFET driver, but usually you'd just have a couple of 1/2 H-bridge drivers (there's a
lot of choice for these).
This approach also decouples the motor voltage completely from the logic
side, since the bootstrapped supply floats the motor voltage can be anything from
1V to 600V (depending on the driver's limits!). They do need a 12V supply (low
current) for the gate drivers, however, but that means logic-level devices are not
needed.
I read that N-channel mosfet wouldn't work well with load on the supply pin, has I see on the shematic the top mosfet will have the load on the supply pins but the driver you sent me does work that out with the BHS and AHS pins right?
If so and this lets me use only N-channel mosfets it will save me some trouble.
I am using phase-correct to achieve 32khz (prescaler 8 - 16Mhz) As I see the pwm input there is a NOT gate for the pwm input so as I am using the phase-correct that would work if I input the other fase oc1B to get the "NOT pwm"? . I am using atmega1280.
What is the 2.5A peak current for the driver? I though the gate didnt riquire much current.
Gates need loads of current, but for a very short time - they need to charge up
fast to get fast switching. Big MOSFETs can have 100's of nC of charge, needing
amps to charge them on a 100ns timescale.
For this reason the bootstrap capacitors and decoupling capacitors need to low
ESR types (good ceramic MLCC's, surface mount).
