Master power switch for E-bike ideas

[eccerr0r]

Yeah pretty much your observations are correct. And you probably don't need to technically worry about most of the SOA graph, because it was meant for people PWMing. Same with capacitance - it's for PWM. If you're only using it as a bulk on/off switch you only switch every few minutes, you only need to deal with continuous (DC) ratings.

And yes these are real devices, there are always tradeoffs when you make MOSFETs. Sort of these four items: high voltage, high current, low capacitance, low price, PICK TWO. Not three, TWO.

(Technically capacitance and price should be directly correlated as an unfortunate side effect, so maybe you can take out one of these two. And then again higher voltage or current are also necessarily increases price.)

[Dannyx]

Yeah, price is definitely a factor in this case: given it's still a goddamn N-channel FET at the end of the day, the price is RIDICULOUS, hence why I only have one shot at getting this right. Kill it and I'm crying for sure

My biggest concern is the startup inrush limit part of the circuit and what values those components should have, particularly the resistor and the capacitor, as I'm not sure what inrush curve I should expect from two 470uF caps in parallel inside my motor controller (so close to 1000uF total) at 50v+ for a fully charged battery.

[stj]

Vds is the maximum voltage between drain and source.
gate capacitance will relate to drive current(inrush) and maximum switching speed.

[eccerr0r]

Depends on the ESR of those capacitors, should be able to calculate the inrush...

[Dannyx]

Originally Posted by stj

Vds is the maximum voltage between drain and source.

Correct. I was curious how this "operates" during turn-on/off and IN the actual on/off states, hence why I said I'm interested in the "practical" side of things, with my humble meter. A scope would probably be better suited for the job, but hey....whatcha gonna do ?

[Dannyx]

Not sure about ESR, but I DO know there's two of them in there, each 470uF. I'm giving THIS a read.

[eccerr0r]

Add an inductor in series with the line
I doubt you really have to worry too much, the caps are probably insignificant compared to the inrush when the motor starts turning.

[Dannyx]

Originally Posted by eccerr0r

Add an inductor in series with the line

This would open up a WHOLE other discussion

Originally Posted by eccerr0r

I doubt you really have to worry too much, the caps are probably insignificant compared to the inrush when the motor starts turning.

Yes, but at that point the FET is fully on, so dissipation is low, at least I hope so. Think I'll go with a 4.7k resitor, a 10uF cap and a 15v zener on the gate. The resistor will be dropping around 38v for a fully charged battery in this case, which peaks at 53v by my observations.

[Dannyx]

I finally gave LTSpice a try and tried simulating my circuit. Here's the weird part: if I use a blank "nmos" model for the transistor, I get the curves in the first picture. This is clearly wrong, because the cap current (red) is WAY too small there - doesn't looks particularly realistic. The voltage at the gate (blue) IS acting more like what you'd expect though.

If I try replacing the "nmos" model with an ACTUAL transistor from the database, I get the second plot: the current just spikes when Vgs is around 3.6v then immediately drops back down. Even if I increase the R/C combo, it doesn't smooth out that spike any further... It obviously doesn't have my exact transistor, so I chose one which comes fairly close in terms of resistance and Vds. I have no idea how to use LTSpice yet - I just clicked on stuff, so I'm probably doing something wrong still

[eccerr0r]

depends on your default model. Looks reasonable though, you should expand out the waveform.
TBH the "real" waveform is what to expect, and not much you can do about it, though you can slow it down even more by messing with the rc delay but it'll slow turn on time even more...

[Dannyx]

Out of curiosity, I tried hooking up the caps directly to the power supply in LTSpice and of course the current went off the chart to like 350a at startup, so I'd say around 1.4a is reasonable there

[eccerr0r]

Now that's a really low ESR capacitor... Then again it is an ideal capacitor, and ideal capacitors have 0 ESR.

[Dannyx]

Originally Posted by eccerr0r

Now that's a really low ESR capacitor... Then again it is an ideal capacitor, and ideal capacitors have 0 ESR.

Can't remember if at that point I tried using "generic" caps or some of the "real" models the program offers...

I may be overthinking it, but that FET is one expensive part like I said Seems I can't smooth out that spike any more, at least not without exaggerating the startup time AND possibly driving the zener out of regulation because its current would be too LOW at that point.

Another thing I tried was placing a resistor across the FET (D-S) so it charges the caps FIRST and THEN the FET shunts it. I was happy with the simulation results, but then it became obvious that there'd be nothing would stop me (or someone else) from attempting to ride the bike with the switch in the 'off' position, so all the power going through that resistor and I can't think of a way to keep this from happening, unless I complicate the circuit even more and use that slow rise time as a switch for a secondary command, like cutting power to the control panel of the bike....dunno....

[eccerr0r]

You can't ride the bike with a 1K resistor there though you probably should go 100K or so, just to cheapen things.

[Dannyx]

Originally Posted by eccerr0r

You can't ride the bike with a 1K resistor there though you probably should go 100K or so, just to cheapen things.

I was thinking of a low-ish value, like 100ohms, which WOULD allow the display to fire up, making someone think it's ready to go, but the whole system would collapse as soon as the motor tries to kick in. It would also burn a ton of power...like 50w or so. Not sure if practical in the long run...

[stj]

just use a contactor to short the battery and blow them to hell
/jk

[eccerr0r]

Originally Posted by Dannyx

I was thinking of a low-ish value, like 100ohms, which WOULD allow the display to fire up, making someone think it's ready to go, but the whole system would collapse as soon as the motor tries to kick in. It would also burn a ton of power...like 50w or so. Not sure if practical in the long run...

Or add more capacitors and they can watch the screen fade away as they discharge... or sooner if they try to engage the motor.

[kaboom]

Originally Posted by stj

just use a contactor to short the battery and blow them to hell
/jk

Watch it there, Kreosan!
/thread

[Dannyx]

Glad to see people find the topic funny around here

I went with 100v version after all. Should have it somewhere the following week.

[Dannyx]

Just remembered this bike has a "power enable" pin in the cable which connects the display to the controller. I could tap into that to act as my power on signal to turn the FET on after the caps have already charged through the resistor idea. That way, it becomes physically impossible to take off on JUST the resistor(s) by mistake. Problem solved