Blown MOSFET on motor control board

[Strider17]

Gentlemen:

I am working on a project that has a proprietary motor
control board that is used to control 6 DC motors on a medical
device. During development and in the field there have
been situations where the motors have been connected
wrong, wired wrong, accidental shorts etc which has
fried the MOSFET on our control board which results in
major ($) rework. The EE's have
said there is no way to protect them because a fuse,
etc, is slower to blow than the MOSFET. Question:
Is there a way to protect a semiconductor device from
such problems? This seems to me to be a problem
that someone must have encountered before. Please be
kind with your replies, I am an ME not an EE. Thanks.

The more you know, the more you
know you don't know....

 

[VisiGoth]

For a long term solution you might have your systems engineering group add the requirements for operation into a miswired load.
A short term solution should involve your vendor, especially in the medical industry.

 

[itsmoked]

Often MOSFETS can take severe overloading(including shorts) long enough for a micro to turn them off. In sketchy applications I include current monitoring to the fets and do just that. Once you *bother* to monitor this parameter you can do lots of neat stuff. Light signal the user that the "OUPUT 4 OVERLOADED".

 

[Strider17]

VisiGoth: I cannot just blidly change requirements
without knowing there is a viable solution. Can you
give me the name of a device, circuit, or method that
I could refer to as an option? Thanks

itsmoked: It REALLY smoked. Thanks for the tip. See further info below.


Additional info just acquired:

There is a 33MHz PIC micro on board already(!). There is
also a current limiting circuit that senses over-current
situations and cuts off the FET. It looks like it
uses a voltage comparator across .1 ohm resistor to gnd.
It will not (obviously) detect a motor short to gnd. Could a similar circuit across a series resistor work? Or...??

Many thanks for any help....




The more you know, the more you
know you don't know....

 

[fsmyth]

Add a circuit or change the board to high-side
current monitoring. This essentially now
watches the supply instead of the load only.
An additional circuit is the best of both;
now you can monitor the load and the supply.
Plus, it is easy to test: just short the output
terminals.
<als>

 

[itsmoked]

Yes Strider17 you put a small value like sub 1 Ohm (.1) on the motor to ground point and then the current signal is positive and easy to monitor and use. If the circuit won't work then you need to do high side monitoring which is a bit more of a headache.

This is all if you are starting with a design. If you have a busted design that needs quick work then the easiest solution is those very nice "protected low side drivers" or "protected switches".

The:

http://www.irf.com/product-info/datasheets/data/ips042g.pdf

is one of my favorites. Pay attention to how you signal them though!

 

[felixc]

Look at the IR3310 to 3312. They're an example about current-sensing MOSFETS that can yield a quick turn-off circuitry.

 

[VEBill]

"...sub 1 Ohm (.1) on the motor to ground point and then the current signal is positive and easy to monitor and use."

One characteristic of monitoring the return current (as described above) is in the case where the high side output is shorted to ground there is zero return current and the system will report zero.

Suitable software design can detect that occurance (ouput high but return current zero, hmmm...) and take the appropriate measures.

 

[benta]

Why not use protected MOSFETs?
Infineon HITFET and PROFET comes to mind.

Benta.

 

[VisiGoth]

Strider17,
There are several good ideas presented above, most showing that you might be able to sense with a resistive element. If your parts can take it you could even consider using a relay in the output, especially if you can not afor any more drop than you have. If you can live with a drop, then there are the foldback cicuits to consider.
I still think it is a good idea to give your systems engineer the responsibility for the next design. A sys eng job is to make sure the requirements are realistic before giving them out. They will of course consult with the experts in the field in order to do this.

If you can address just the unsreasonably high current as the salient indicator of the problem you could consider a current sense transformer. If the rate of change of current is too fast and the value is too high you would trip the disconnect circuit, possibly using the existing control circuit to do it.

 

[Strider17]

VisiGoth:

Thanks for your reply. All suggestions above seem
to have validity. I (as an ME) would never attempt
a new design or redesign, but I do want to effect
a change. I am trying to convince the EE's and
management that there are alternatives to smoking
control boards which slows the whole project. I
guess this is a political problem. Thanks to
you and all others for the ammunition to try and
change this situation.

The more you know, the more you
know you don't know....

 

[killabyte]

Usually you use mosfets when in power audio amplifiers, those mosfet banks handle heavy loads up to 120 amps peak, and 40 amps program.

the circuitry is protected (of course) for shorted loads because is very common to have a broken, shorted or damaged cable.

the protection scheme find in almost every amplifier is just a fuse. some uses fast blown fuses other use slow blown fuses.

mosfets are equalized through a veri low value resistence which helps a little when short circuits happen.

is common to find fryed mosfets, but is most common to find blown fuses and save mosfets,

ragards

killabyte

 

[electricuwe]

Some comments on the solutions discussed above:

-It's not necessary to add a resisitive element for the purpose of overcurrent protecion for a Mosfet

-If one of the devices having overload protecion, like Infineon PROFETs or HITFETs fits your requirements: take it

-if not: Mosfets can be protected by using the Rdson of the Mosfet as sensing element. Monitor the drain voltage starting 1..2 us after turn on and turn off within about 10us if it is not as low as it should be. Usually that function is implemented in hardware.

If you provide more data (voltage, current, type of Mosfet used) I can poovide further hints.