Do Mosfets Share Current?
- Thread starter cohberg
- Start date
They do share, though never perfectly. If you have a 15 FET, 5 each phase, then two are for one half of the bridge, and three are for the other. So at weakest point, two of them each take half the current.
There are a few threads about controller design that discuss current-sharing among FETs, if you want some technical details.
There are a few threads about controller design that discuss current-sharing among FETs, if you want some technical details.
liveforphysics
100 TW
Also, your FETs don't see battery current. Your controller might be pulling 60amps from the battery, yet your FETs might be seeing 250amps of phase current.
liveforphysics said:
in that case shouldn't my controller (with fake 4710 fets rated at 75 continuous amps) have exploded and produced a unwanted crotch lipo fire
i'm sorry i know that this is elementary but shouldn't phase current just be 3 times battery current?
No, no one shoudl PM questions that aren't personal, because otherwise no one else can learn anything from the discussions.cohberg said:
You'll probably want to do some research on how a controller is designed; there's several places on ES with at least partial schematics of them, including the various controller development threads.
But basically, the FETs in a brushless controller are setup with a top FET and a bottom FET in a "bridge", with phase windings wired between the center connections of the FETs, usually in a triangle (delta). So current flows thru the top fet of one bridge, thru two phases, and then thru the bottom fet of another bridge.
Some controllers dont' use the same number of FETs for top and bottom legs, for various reasons described in other threads. So the smallest number of FETs is what really matters when determining current-sharing on any particular phase.
I think what we need to add to this common statement to make it much clear is that the MOSFET current isn't there "the whole time" while the battery current is. That's how you can have "more current" flowing through the MOSFET than the one flowing through the battery.liveforphysics said:
post-fix: "phase/motor" out, "MOSFET" in
John in CR
100 TW
Njay said:
I don't really know how controllers work, but the FETs for one side of each phase are connected to the positive wire from the battery, and other than the big caps, those FETs are the only thing connected to the battery positive, so how can you say the fets don't see battery current? I understand PWM causing current multiplication for the phases, but don't some of the FETs see battery current? BTW, which ones are "high side" and "low side" ?
I made a mistake previously, sorry guys. Motor current can be constantly much higher than battery current. What I said about "the whole time" applies to MOSFET current. But let me try to clarify this using simulation.
Think of a simple case: one MOSFET turning a coil ON and OFF (the coils could be a motor, a phase, etc).
The blue line ( -I(V1) ) is the battery current, the green line is the coil (let's call it "motor") current ( I(L1) ) and the red line is the MOSFET current. While the MOSFET is ON, it "drives" the current through the motor, so both currents are equal (red = green); the current grows while the MOSFET is ON, because coils don't let current change suddenly.
Now the MOSFET turns OFF, and so it's current goes to zero (red line down); however, because we have a diode and because the coil wants to push it's current until the magnetic field collapses (once again, it doesn't let current change sunddenly), this current flows through the diode while decreasing. It doesn't fall to zero because we don't give it time, we turn the MOSFET back ON in a few moments.
Battery current oscillates, but is never higher than some 5A and, If you do an average on it, it's only 1.58A. However, motor current is around 16A. And MOSFET current, when MOSFET is ON, is 16A also, but only when ON.
Now, as you can see, in one of the cycles I have marked a brown area which is between battery (blue line) and MOSFET currents (red line). If battery current were equal to MOSFET current, that is, if the battery were providing the MOSFET current, both red and green lines should be underneath each other. But that's not what we see, we see that MOSFET current is way higher than battery current. There is this "extra" current, which is the area I colored brown and put a question mark on. Where does this current comes from? From the capacitor! The capacitor provides this "extra current", and then it re-charges while the MOSFET is OFF, with the current provided by the battery in the violet colored area with the question mark. The MOSFET ON current is the sum of the battery current and capacitor current.
The battery current would be more constant with a bigger capacitor, the 1000uF I picked is too small.
Hope this helps!
p.s.: Data about the simulation is: battery 24V, 25mOhm internal resistance; coil is 250uH 100mOhm; MOSFET being driven by 10KHz PWM with 10% duty cycle; capacitor has 1 mOhm impedance.
p.s.: I can provide switcherCAD's simulation file
p.s. Did some English fixes
Think of a simple case: one MOSFET turning a coil ON and OFF (the coils could be a motor, a phase, etc).
The blue line ( -I(V1) ) is the battery current, the green line is the coil (let's call it "motor") current ( I(L1) ) and the red line is the MOSFET current. While the MOSFET is ON, it "drives" the current through the motor, so both currents are equal (red = green); the current grows while the MOSFET is ON, because coils don't let current change suddenly.
Now the MOSFET turns OFF, and so it's current goes to zero (red line down); however, because we have a diode and because the coil wants to push it's current until the magnetic field collapses (once again, it doesn't let current change sunddenly), this current flows through the diode while decreasing. It doesn't fall to zero because we don't give it time, we turn the MOSFET back ON in a few moments.
Battery current oscillates, but is never higher than some 5A and, If you do an average on it, it's only 1.58A. However, motor current is around 16A. And MOSFET current, when MOSFET is ON, is 16A also, but only when ON.
Now, as you can see, in one of the cycles I have marked a brown area which is between battery (blue line) and MOSFET currents (red line). If battery current were equal to MOSFET current, that is, if the battery were providing the MOSFET current, both red and green lines should be underneath each other. But that's not what we see, we see that MOSFET current is way higher than battery current. There is this "extra" current, which is the area I colored brown and put a question mark on. Where does this current comes from? From the capacitor! The capacitor provides this "extra current", and then it re-charges while the MOSFET is OFF, with the current provided by the battery in the violet colored area with the question mark. The MOSFET ON current is the sum of the battery current and capacitor current.
The battery current would be more constant with a bigger capacitor, the 1000uF I picked is too small.
Hope this helps!
p.s.: Data about the simulation is: battery 24V, 25mOhm internal resistance; coil is 250uH 100mOhm; MOSFET being driven by 10KHz PWM with 10% duty cycle; capacitor has 1 mOhm impedance.
p.s.: I can provide switcherCAD's simulation file
p.s. Did some English fixes
liveforphysics
100 TW
cohberg said:
All wrong.
Njay said:
Win for you Njay. Glad to have you here.
Controllers are like 3 bi-directional buck-converter circuits using the motor as the inductor. The only load the FETs see's is phase current, the only things that see battery current are the shunt, the traces leading to the source of the low-side fets and the drain of the high-side fets, and whatever ripple the caps suffer through in the ringing loop on the PCB to FETs and battery to PCB/cap loops.
liveforphysics
100 TW
cohberg said:
My FET's blow all the time.
It ALL depends on the motor, as it's the R and L of the bucking circuit. If you hooked your controller to a colossus motor with 2mOhm of phase resistance and just a couple uH of inductance, then even at like 10battery amps, your phase amps can be in the hundreds, and your controller explodes in seconds (as happens to everyone who tries to power a colossus motor). If you hook the same controller to an 8x8 9C or an X5305 or something (motors with plenty of resistance and inductance), the 10amps of battery current might be maxing at 30amps phase current, and rapidly tapering off as RPM builds (causing BEMF to make it seem like your pack is a lower voltage), and everything is happy as can be. Or your 60amps battery current might only be 100amps of phase current etc.
Glad to be here among guys like you!liveforphysics said:
LFP's 250A was only an example. You probably don't know how much "peak" current your MOSFETs are seeing. Unless your controller does current limiting, which a good controller must do to protect MOSFETs, battery and motor. In that case, the controller will turn OFF the MOSFET when the current reaches a certain maximum value. In some controllers you can configure that value.cohberg said:
liveforphysics said:
That cleared it up. Thanks Luke!
If i wanted to max out my controller could i measure phase current and battery current simultaneously and get a somewhat linear plot? but you mentioned that it tapers off. hrmm.
I guess i could just do a progressive shunt mod and keep running tests, but i risk smoking fets. Maybe it is time to buy those 4115s
Njay said:
I have a cheapie Hua Tong from china (33 shipped) so i don't know if there is limiting. But would measuring the phase current (don't matter what phase right?) tell me roughly how much current is going through?
liveforphysics
100 TW
Measuring phase current accurately and fast enough to be meaningful to a Mosfet is pretty tricky stuff. You can do it with a current probe on a scope pretty easily, but that's a $200 probe, that a controller is not going to be equipped with.
If you use Shunts (which are dirt cheap), then you've got a bit of a nightmare reading the tiny voltage of a badly bouncing circuit. Many/some of the hall sensor type sensors (allegro etc) are too slow to catch the spikes/overshoot.
Kelly controllers read across the RdsOn value of the shunts, which then means they auto-de-rate themselves as temp increases and the RdsOn goes up. The down side is, that means you can have 2.5x LESS torque on a hot controller than a cold one... Which is also a bit wonky.
If you use Shunts (which are dirt cheap), then you've got a bit of a nightmare reading the tiny voltage of a badly bouncing circuit. Many/some of the hall sensor type sensors (allegro etc) are too slow to catch the spikes/overshoot.
Kelly controllers read across the RdsOn value of the shunts, which then means they auto-de-rate themselves as temp increases and the RdsOn goes up. The down side is, that means you can have 2.5x LESS torque on a hot controller than a cold one... Which is also a bit wonky.
liveforphysics said:
yeah that was the first thing that i thought might be a problem but i'm an undergrad and i have access to a scope
I really know how think of these "pwm pulses" do i treat them as a continuous load or can i kinda integrate over a set time and treat that as the amps its pulling?
liveforphysics
100 TW
cohberg said:
On the motor side, only the voltage has the pulses, and nothing "feels" voltage. The voltage getting chopped up and sent to the motor can be a 20% duty cycle PWM squarewave, and the inductance filters that into a nice smooth current form with a tiny bit of ripple, and the current is ALL that makes the FET heat, motor torque, motor heat (at least while at low speeds), etc.
John in CR
100 TW
To answer the previous question about the 240A and no smoke, that's because that current is split between multiple FETs on each phase. No one's 6fet controller is doing 100A/240A for a single fet to see 240A, at least not that I know of.
NJay,
Now I'm thoroughly confused. I get it that the phase currents are higher than battery current, but not why they are. How can those tiny diodes on a controller's circuit board handle the high currents? My thought process was that with the aid of the capacitors that the battery current is more constant, which seems to agree with your post, and the phase current is so much higher because it flows only while the fet is on, which also seems to agree. The green line of high current through the diodes has me thoroughly confused.
WRT to easy to drive motors with higher L and R, I thought they changed the shape of the tall thin rectangular box of motor current while FETs are on to round off those sharp square corners, because the motor inductance slows the current flow in the ON state. Then the hard to drive motors such as colossus allow a much more square shape and the current flow spikes much higher before the controller can react and turn the FET back off since resistance and inductance are so low. NJay and LFP please help me understand where I'm off.
John
NJay,
Now I'm thoroughly confused. I get it that the phase currents are higher than battery current, but not why they are. How can those tiny diodes on a controller's circuit board handle the high currents? My thought process was that with the aid of the capacitors that the battery current is more constant, which seems to agree with your post, and the phase current is so much higher because it flows only while the fet is on, which also seems to agree. The green line of high current through the diodes has me thoroughly confused.
WRT to easy to drive motors with higher L and R, I thought they changed the shape of the tall thin rectangular box of motor current while FETs are on to round off those sharp square corners, because the motor inductance slows the current flow in the ON state. Then the hard to drive motors such as colossus allow a much more square shape and the current flow spikes much higher before the controller can react and turn the FET back off since resistance and inductance are so low. NJay and LFP please help me understand where I'm off.
John
The green line is the current through the coil, and in the zone where the MOSFET is OFF, it's also the current through the diode, because the diode is the only current path available to the coil while the MOSFET is OFF. Without that path, the coil would ramp up voltage, reaching the MOSFET's breakdown voltage which makes it turn ON and let the coil discharge its current through it (so called "avalanche current"). The capacitor smooths the battery current (that is, it tens to make constant the current that is requested to the battery; and also avoids wiring inductance spikes) because it slowly takes current from the battery (while MOSFET OFF) and then provides it fast to the power stage (MOSFET + coil) (while MOSFET ON).John in CR said:
I don't know of what diodes are you referring to but, when diodes are used, let's say in brushed motor controllers, they are "big" diodes in TO220 cases. In brushed motor controllers that use a half-bridge (like a phase control circuit in a brushless), and in the brushless controllers I supposed you are used to see in e-bikes, the top MOSFET can do the same job of the diode, turning ON while the bottom MOSFET turns OFF or just by staying OFF and letting the current flow through their internal intrinsic diodes (more power dissipation). Sometimes the designer also puts smaller, very fast diodes in parallel with the "top MOSFET"; this is just to help the current flow at the start and avoid voltage spikes, because while for example MOSFET internal diodes are normal, "slow" PN junction ones, these external are fast schottky ones. They only conduct a tiny amount of current for a tiny amount of time while the big ones "get ready".
In a coil with lower inductance, the rate of raise of current is higher, which means, compared to the simulation, that the top green line during MOSFET ON would have a higher slope, it would climb faster and therefore higher in the same amount of time (that the MOSFET is ON).John in CR said:
Keep in mind that the simulation is on an invented motor, in an invented situation (we don't know current RPM, load, etc). Typical look from a real motor may be different, curves with different slopes, but this simulation tried to show the functional behavior (FET turns ON and current flows here, the OFF and diode conducts, etc), which is the same on a real motor. I can take another simulation of the same circuit with lower coil inductance so you can see the difference.
I'm not sure if I made it clearer, I'm not yet seeing exactly what's confusing you.
whatever
100 kW
- Joined
- Jun 3, 2010
- Messages
- 1,297
its probably come up before but here is a pic of a mosfet with an iron bead around one high amp leg, from a quick search it seems this plays a similar role to capacitors to stop ringing and get rid of transient voltages. Can anyone elaborate further and do you guys think it would be useful to be using these?
It occurs to me that its a bit similar to an inductor.
Leg with iron bead has different hole location on the board to allow for the bead
this fet is off an mppt
It occurs to me that its a bit similar to an inductor.
Leg with iron bead has different hole location on the board to allow for the bead
this fet is off an mppt
liveforphysics
100 TW
That's a ferrite bead on the gate leg. It increases the inductance, and while voltage may bounce as it's entering the miller plateau on the FET begins to conduct (which locally pulls up the ground plane, causing the bounce), the field stored in that bead collapses, and rather than bouncing through the conductive region multiple times (ringing that causes massive high resistance partial-conduction losses on the FET), it carries holds it saturating the miller plateau until the inductive bounce on the board that would have caused the ring to stabilize (we are talking nS time window here).
In other words, it helps avoid ringing, lowers conduction transition losses, and allows the FET to use it's heat budget to carry a couple more amps, or to run a bit cooler.
You generally only see that stuff in 2 situations. A band-aid fix on a crap circuit that they called in an expert to help with, or a circuit designed by an EE master with a very poofy and fuzzy grey beard.
In other words, it helps avoid ringing, lowers conduction transition losses, and allows the FET to use it's heat budget to carry a couple more amps, or to run a bit cooler.
You generally only see that stuff in 2 situations. A band-aid fix on a crap circuit that they called in an expert to help with, or a circuit designed by an EE master with a very poofy and fuzzy grey beard.
whatever
100 kW
- Joined
- Jun 3, 2010
- Messages
- 1,297
there is a good article here
http://www2.microsemi.com/micnotes/APT0402.pdf
i didn't notice but there is another identical mosfet on the other side of the board in parallel, so it seems to be there to stop the ringing of the paralleled fets. Nice to know for those paralleing fets.
http://www2.microsemi.com/micnotes/APT0402.pdf
i didn't notice but there is another identical mosfet on the other side of the board in parallel, so it seems to be there to stop the ringing of the paralleled fets. Nice to know for those paralleing fets.
