Sizing FETs

[Michael Ferguson]

There was a discussion about sizing FETs at Wednesday's SIG meeting. A statement something like "my motor driver uses 20A and these FETs are rated for 30A" was made, which led to a whole discussion of FET sizing, and how datasheet numbers for FET current are completely fictional for real world current capacity. Here is the tutorial that I promised to send a link to: http://www.mcmanis.com/chuck/robotics/projects/esc2/FET-power.html

-Fergs

[KM6VV]

Nice paper Chuck!

Alan KM6VV

> --

[Wayne C. Gramlich]

All:

Some quick comments about high current H-bridges:

1) Most club robots do not need them. Indeed, the only robots
that really need them are probably competition grade battlebots.
If you think you need one, you should post your robot goals to
the list to get some feedback. I doubt that anybody that was
at the SIG meeting last night needs a high power H-bridge.

2) If you do need a high current H-bridge, you should seriously
consider just buying one. The HB-25 from parallax is a
reasonable place to start. Be forewarned that many of the
high current motor drivers out there have great spec. sheets,
but when you actually try to get them to perform at spec., they
do not cut the mustard.

3) Chuck has forgotten more about high current H-bridges than I will
ever know. That said, the URL below is just scratching the surface.
The heat sink design is just as important as the electronics design.
Any reasonable high current H-bridge should have a hard current
limit circuit. (Running the current sensor into a microcontroller
A/D is probably not going to be fast enough.) Also, there should
be a thermal sensor on the heat sink so that the H-bridge can be
turned off for thermal overload.

Regards,

-Wayne

> --
> You received this message because you are subscribed to the Google
> Groups "HomeBrew Robotics Club" group.
> To post to this group, send email to hbrob...@googlegroups.com.
> To unsubscribe from this group, send email to
> hbrobotics+...@googlegroups.com.
> Visit this group at http://groups.google.com/group/hbrobotics?hl=en.
> For more options, visit https://groups.google.com/groups/opt_out.
>
>

[jason d]

For the high current limit in #3:
 
  I haven't tested this myself but I've seen others run a jagged sawtooth looking trace from the FET.  The idea being that it increases the resistance of the trace just enough to cause it to vaporize in a very high current situation.  The downside is that the PCB is toast but it should save everything else on the board.

[Bob Smith]

jason d wrote:
> I haven't tested this myself but I've seen others run a jagged sawtooth
> looking trace from the FET. The idea being that it increases the
> resistance of the trace just enough to cause it to vaporize in a very
> high current situation. The downside is that the PCB is toast but it
> should save everything else on the board.

I was thinking of adding an automotive spade type
fuse per H-bridge. That way I can size the fuse
to match the rest of the system. Also, automotive
are cheap and easy to find.

Which do you think is better, an automotive fuse
or the relying on the internal current limiter in
the H-bridge controller (if it has one)?

thanks
Bob

[Wayne C. Gramlich]

On 01/27/2013 11:17 AM, jason d wrote:
>
> For the high current limit in #3:
>
> I haven't tested this myself but I've seen others run a jagged
> sawtooth looking trace from the FET. The idea being that it increases
> the resistance of the trace just enough to cause it to vaporize in a
> very high current situation. The downside is that the PCB is toast but
> it should save everything else on the board.

Jason:

Why bother to vaporize the PCB to save the PCB components? You are
going to throw the PCB away along with most of the components anyhow.
It does not make much sense to me. The copper trace resistance is
strictly a function of length, so zig-zagging the copper trace just
saves board space. Lastly, there are many poorly designed high
current H-bridge boards out there. I wonder whether the board you
have with the zig-zagged trace is any good.

If you want a fuse, just install a fuse. However, the FET's will
frequently blow before the fuse does. You are much better off
designing the system to detect over current and shut down. It
better shut down fast, though.

I do not want to discourage people from designing a high current
H-bridge, just be prepared to blow some of your boards up as you
figure stuff out. It is called the school of hard knocks.

Regards,

-Wayne

[Bob Smith]

Wayne C. Gramlich wrote:
> If you want a fuse, just install a fuse. However, the FET's will
> frequently blow before the fuse does. You are much better off
> designing the system to detect over current and shut down. It
> better shut down fast, though.

Thanks, Wayne. This answers my question about auto fuses.

thanks
Bob