MOSFET gate PCB design question

[rafeh1]

how about some more ideas.

1. make the mosfets modular. so others can make mini pcb with new mosfets as they come out
2. send power in and out on common metal rails mounted below pcb.
3. input dc pwr and output motor connectors cables directly screwed on metal rails
4. modular mosfets pcbs are screwed on the rails

pwr _________________________________ rails
gnd__________________________________

_____________________________________M0
_____________________________________M1
_____________________________________M2

this way if a mosfet goes u simply unscrew it and put on a replacement. also if better fets come out they can be plugged in. A better arduino comes out just plug it in. you want higher voltage current just plug in the fet for that voltage current.
Arduino blue tooth streams live telemetering data to ur smartphone .

new arduino has bluetooth or wifi or gsm which can provide anti-theft and location alarms.

 

[rafeh1]

one more thing
use differential opamp to measure voltage drop across 2" of power line metal and use that to sense current use or use hall sensor. Do not use in series resistor

hall effect current sensor

http://www.digikey.com/product-deta...FF-T/620-1320-ND/2042745?WT.mc_id=PLA_2042745

design in both and then play around and see which is better. programmable differential opamp can have huge range..

 

[HighHopes]

rafeh - all good ideas. in fact, zombiess Cadillac motor drive already incorporates this.

ps. it is harder than it sounds to "weld" copper busbar to a PCB wide copper trace. any poor connection will become the hot spot ruining all that hard work. to improve on this (i.e. to better control the quality & variation) i think would be for DIY'er to discover the toaster oven reflow method, but that is a conversation for another day.

 

[rafeh1]

rafeh - all good ideas. in fact, zombiess Cadillac motor drive already incorporates this.

ps. it is harder than it sounds to "weld" copper busbar to a PCB wide copper trace. any poor connection will become the hot spot ruining all that hard work. to improve on this (i.e. to better control the quality & variation) i think would be for DIY'er to discover the toaster oven reflow method, but that is a conversation for another day.

i am advocating no pcb to bus bar but rather screw on to metal bar like used on auto battery terminals used to extract 500a for 30 seconds or 50 to 100 a continuously..

 

[Futterama]

Do the traces between MOSFET driver and each gate have to be the same length for the induvidual MOSFET gates?
Or can the traces vary in length if only they are connected in a star configuration?

I'm having a bit trouble keeping the driver-to-gate distance low when adding the next 3 parallel MOSFETs. So I'm considering a slightly different design but need to know whether I need to connect the gates in star configuration.

 

[HighHopes]

if this were my design i would consider putting a resistor between your 4-point node and "driver", probably in the range of 5 to 15ohm but depends on your overall gate driver design.

this is sort of an expansion on the argument of how to do the perfect layout for delivery of power. the main issue with parallel mosfets is current sharing when they turn ON. so if the gate lengths are a little difference the questions is if it is different enough to make a difference to the mosfet's turn ON power sharing .. or not.. or .. yes makes a difference but in this particular case, not enough of a difference to worry about. so you need to be able to answer these questions before you can decide what to do. keep in mind there is tolerance on the resistors that is probably much bigger than the traces themselves. 1 ohm resistor.. is this a 1% tolerant resistor? should it be 0.1% ?

then the next question.. when you have parallel mosfets you can see that between the gates of each mosfet is 1 Ohm + 1 Ohm = 2 Ohms. is this enough to dampen any oscillations low enough that it is not going to cause operation proplems?

 

[ricardo81]

@Futterama: I think the paralleling of the high power MOSFET is very important and it is true that for matching the MOSFET should be operated in linear region and it is a good idea to do it in PWM configuration because you can adjust the input power by adjusting the width of pulse to keep the MOSFET in linear region and avoid it to going in the saturation and thus helps in matching the parameters of MOSFET.

 

[bearing]

To me, matching MOSFETs seems like a last effort. You should be able to make a design insensitive to the gate threshold and such.

 

[HighHopes]

you can make it insensitive, absolutely, and this is my thinking as well. the fact is your multiple parallel mosfet IS sensitive to parameter mismatch and WILL blow up if not taken into consideration. let's say your choice is to do nothing, no analysis, no clever dynamic compensation design as alluded by ricardo81 , no bench testing to confirm matching transconductance. even here you can get a gate driver that operates just fine with doing no additional effort, just a standard gate driver from an app-note. only thing to realize is you will not get anywhere near the rated current out of this in a reliable way. your 300A inverter just became 150A rated. at 150A the zero effort multiple mosfet driver is reliable, so you get what you wanted. but if you want full also the full rated performance, then you must consider the world the mosfet lives in which is different than ours.

 

[bigmoose]

Note to futterama: how the sources of your paralleled FETs come together is of critical importance to how the gates drive also.

 

[Arlo1]

To me, matching MOSFETs seems like a last effort. You should be able to make a design insensitive to the gate threshold and such.

bearing... I think this is relative. There is a lot of things that make it a good idea and the only thing that makes it a bad idea is it takes some time in my case I matched up 24 TO247 fets in a couple hours and compared to the 60hr + it took to build it all and get it running that's not a lot.

As well when you have a small amount of pass though or ringing or something that can damage the fets its distributed to all fet in parallel more evenly so gives the whole thing a lot better chance. On top of that the goal is to keep the thing Cold and its the heat that helps mosfets self balance in a paralleled system so the colder they are when you push them hard the more out of balance they will be. When trying to switch them on fast as well they will not have as much time to self balance...

I think for toys and ebikes fet matching is no big deal but if you want performance and reliability its a must.

 

[Futterama]

Another important little tip is to mount a 600W transildiode at the gate to prevent a mosfet breakdown conducting backwards and detroying the whole controller. Should be fitted as: Gate - Resistor - Transil - Gatedriver.

Transient Voltage Suppressors are new to me, so what ratings should I look for?

Uni- or bidirectional?
Reverse stand-off voltage?
Breakdown voltage?
Clamping voltage?
Peak pulse current?
Reverse leakage? (I guess lower is better)

Edit: Sorry, a bit of seaching and reading answered most of it:
http://en.wikipedia.org/wiki/Transient-voltage-suppression_diode

So, they are kind of like a super fast zener diode? They should be connected from gate driver output to GND, right?

From what I understand, the clamping voltage should be lower than the voltage allowed on the gate driver output and the maximum reverse standoff voltage should be higher than the voltage driving the FET gate. Is that correct?