detecting a fault in an ST Omnifet on the high-side

This is the place for any magazine-related discussions that don't fit in any of the column discussion boards below.
Post Reply
Soundscu
Posts: 1
Joined: Wed Feb 04, 2009 2:20 am
Contact:

detecting a fault in an ST Omnifet on the high-side

Post by Soundscu » Wed Feb 04, 2009 2:23 am

I am designing an H-bridge motor driver using 2 L6384 drivers, and 4 VNV35N07 Omnifets. Since the Omnifets are internally protected, I hope to avoid using sense resistors for fault detection. Please suggest an elegant way to detect faults on the high-side. If a high-side VNV35N07 has clamped the gate, what is a good way to detect this, and produce a 5V logic level signal referenced to ground? Power supply to the H-bridge will usually be 24VDC, though it could be anywhere from 6V to 30V. Thank you for your time.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: detecting a fault in an ST Omnifet on the high-side

Post by MrAl » Wed Feb 04, 2009 3:12 am

Hi there,

In a word, "Hall Effect".
Really, hall effect sensors would be the best way to go here.
LEDs vs Bulbs, LEDs are winning.

User avatar
haklesup
Posts: 3040
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Re: detecting a fault in an ST Omnifet on the high-side

Post by haklesup » Wed Feb 04, 2009 4:19 pm

MrAl, How would you use the hall device? Put it next to the drain trace and use it to sense magnetic field as proportional to current then run that signal to a comparator to convert it to logic levels. A bit tricky to get the right sensitivity wouldn't it?

What high side fault do you want to detect? Overcurrent? You could always just use a current limited source and let it do the work.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: detecting a fault in an ST Omnifet on the high-side

Post by MrAl » Wed Feb 04, 2009 5:07 pm

Hi there Hackle,


Just off the top of my head, the best way is to insert the hall effect device inside the gap of a toroidal
core. Im sure there are variations with this idea, but probably the best is to wind a few turns of
the heavier current carrying wire to sense current (primary), and a much thinner winding with say 10 times
(or whatever) the turns also on the core (call this the secondary).
Now when the primary is energized with dc current the secondary is also energized with an op amp
circuit output. The op amp circuit is set up to force the magnetic flux to zero, that is, cancel out
the effect of the primary current with the (smaller) secondary current. The circuit maintains
zero flux and in doing so effectively measures the primary current, as the primary current will
be 10 times the secondary current (various ways to convert the secondary current to voltage).
The comparator circuit would follow, of course with hysteresis and probably some delay time
once it's triggered.
I guess you might call this circuit a 'flux bridge' :smile: When the flux is zero the bridge is
balanced and the primary current is a multiple of the secondary current.

Another idea might be to wind several turns around a core and use the hall sensor in the gap
again, only this time use the output of the hall sensor for the signal and use the output of
another hall sensor as offset to help compensate for temperature changes.
This would be simpler than the flux bridge circuit.

Of course the simplest is to just buy a hall effect current sensor and use that with some
comparator and delay time once it's triggered. Im sure there are lots of these sensors
out there, and the comparator is cheap. The delay is required so that the circuit does
not turn back on right away once the high current exceeds the set point and triggers
the comparator.

A more sophisticated way would be to use the output(s) of the hall effect sensor(s)
as current regulating PWM feedback. This would limit current and also regulate it at
the max level allowable. This would be like a regular switching regulator only
with hall effect sensor feedback.

What i dont know is what kind of resources the OP has at his disposal...ie, can he
order specialized cores with a pre-cut gap or does he have to stay super low
cost, or is this a hobby application, etc.

I might also add that some of the manufacturers of hall effect sensors have some
decent literature on using their hall effect devices for the purpose of sensing
current. They provide some formulas too to help with the design process.
LEDs vs Bulbs, LEDs are winning.

User avatar
haklesup
Posts: 3040
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Re: detecting a fault in an ST Omnifet on the high-side

Post by haklesup » Thu Feb 05, 2009 11:40 am

Workable but if if the OP is resistant to putting a couple low value resistors in to sense current, I don't see him signing on for a more complicated custom inductor and sense circuit This solution does give superior isolation between the supply source and sense circuit though. An inductor would likewise add a small amount of DC resistance and possibly a significant AC resistance depending on the output signal.

Can I assume the "Fault" would be an over current due to a stuck, stalled or mechanically overloaded motor (not considering component or circuit failure. A sense resistor is the most reasonable solution (relatively simple and inexpensive). Depending on how the load (motor) and transistor interact, there may be an opportunity to just sense voltage with a high impedance probe.

One way we do it (and I don't immediately see how to implement it in an H bridge) is when using an op amp to drive a signal, you can put the sense resistors on the supply pins and measure current going into the op amp rather than the power coming out. This is less likely to load the output node and alter the signal.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: detecting a fault in an ST Omnifet on the high

Post by MrAl » Thu Feb 05, 2009 1:38 pm

Hi again,


Well, sometimes designers of the higher power stuff dont want to stick in sense resistors when they
go through the trouble of using very low Rds MOSFETs as that sort of defeats the purpose.
Thus, hall effect methods are used. This is actually a fact, not an idea of mine alone.
I dont know what kind of power levels the OP is pushing here though, so i cant say for sure
one way or the other. I guess he/she would have to give us some more information.
BTW very heavy gauge wire does not add that much resistance to the circuit.
LEDs vs Bulbs, LEDs are winning.

Post Reply

Who is online

Users browsing this forum: Majestic-12 [Bot] and 36 guests