[PSIG]

You may as well switch to a logic level FET with a good avalanche rating. . . .

+1, agreed on all that, as I've described the base issues here before. However, the adapter is to attache to existing boards, including a large number with installed STP non-logic FETs. Legacy support. It is for that reason, IMO, the adapter should be the source of solutions, including this one. Changing FETs to appropriate drive voltage also does not address the drive capability as-described in my last post (and more). I think both of these issues could be addressed with a single drive/interface solution. Any alternatives that could do the same with legacy support?

David

[JHolland]

I don't think there is a problem with turn on speed, the last thing that you want with injectors is really fast switching because that leads to ringing and horrendous noise issues. Limiting the inrush current using the gate resistor is quite normal, however I would usually aim to start with a maximum gate current of around 500mA. Heat losses during switching aren't usually a problem with High-z injectors due to the relatively few switching events.

[PSIG]

Agreed on most of that; but to put this into context (and mostly for others reading), the issues needing focus are not pure speed, but Rds (on-resistance) causing the heat. This current capacity example of an STP MOSFET helps to show the problem, where 3.3V is totally unsuitable for gate voltage, 4.5V works to partially turn-on, but limits the current it can handle (in this example about 3A). However, at gate voltage (provided by a driver) of 10-15V, the current handling goes up from 3A to well over 100A for short pulses, and a similar effect for constant current:

MOSFET Gate Volts vs Amps_sm.jpg (151.63 KiB) Viewed 10618 times

While a gate resistor is sometimes used, it is applied to the minimum resistance required to reduce ringing in the gate circuit (not current-limiting), such as in the TC442x datasheets, generally recommending 8 to 18 ohms. So, no arguments for the how, but reinforcing the why a gate driver is typically required in most MOSFET circuits, and definitely on a 3.3V drive to 10V+ gate device application for legacy support. As we cannot easily add this to existing boards, adding them to the adapter in order to support almost anything makes sense to me. FWIW

Other resources that may be helpful:
What is the purpose of “MOSFET driver” IC's
Using MOSFETs As General Switches
Matching MOSFET Drivers to MOSFETS - Microchip technology 00799a and 00799b
Add-Ohms MOSFET Guide/Worksheet

David

[JHolland]

The gate resistor is also used to slow the turn on/off time of the FET to prevent ringing on the drain which is much worse from an EMI aspect because it is directly transmitted down the wiring, often snubbers are used as well.
I would always use some kind of driver anyway as a buffer to minimize the switching currents in and out of the processor. The problem with many dedicated gate drivers in an automotive application is the limited supply voltage. The TC types are 20V rated, the MOV clamps at around 43V so the gate drivers are toast if you get a load dump or you are at 24V (jump start - 66 seconds max.). Where I have used 'proper' gate drivers they have been auto spec with built in diagnostics - too complex for Speedy really, I've used a simple PNP/NPN push/pull stage to drive FETs, works well and 60V BJTs are cheap, you just need to watch the power dissipation during load dump and 24V operation. For 10V drive you could set up a separate 10V rail, maybe just limit the voltage using an R and a zener.

[PSIG]

Would the "clamped" STPs avoid that load dump on the gate and TC output? Looking for future changes to board designs rather than legacy at this point.

David

[JHolland]

No Vbat is going to rise to around 43V. Any clamping or avalanching must be above that level or the loads will turn on. The ECU should operate properly during load dump although loss of emissions functions is allowed. Speeduino as it stands has a problem there - the 2940 shouldn't really be used in an ECU.

[pazi88]

No Vbat is going to rise to around 43V. Any clamping or avalanching must be above that level or the loads will turn on. The ECU should operate properly during load dump although loss of emissions functions is allowed. Speeduino as it stands has a problem there - the 2940 shouldn't really be used in an ECU.

"Designed also for vehicular applications, the LM2940N and LM2940C and all regulated circuitry are protected from reverse battery installations or 2- battery jumps. During line transients, such as load dump when the input voltage can momentarily space exceed the specified maximum operating voltage, the space regulator will automatically shut down to protect both space the internal circuits and the load."

[JHolland]

the space regulator will automatically shut down to protect both space the internal circuits and the load."

Exactly, ideal for general purpose automotive electronics but not for an ECU, your ECU suddenly turns off and the vehicle stops - a situation considered highly dangerous and a definite type approval failure.

[pazi88]

the space regulator will automatically shut down to protect both space the internal circuits and the load."

Exactly, ideal for general purpose automotive electronics but not for an ECU, your ECU suddenly turns off and the vehicle stops - a situation considered highly dangerous and a definite type approval failure.

So it's better to fry ecu internals and stop the engine in that way? As far as I know speeduino has the mov and diodes as well to protect the regulator and other components on the board from over voltages.

[KLAS]

Exactly, ideal for general purpose automotive electronics but not for an ECU, your ECU suddenly turns off and the vehicle stops - a situation considered highly dangerous and a definite type approval failure.

i assume thats why the factory ECU of my car has a 5 pin regulator, a L4947R, with reset and delay, if i understand that correctly?