MOSFET protection

[sam9]

While the use of a switch diode - zener chain between output and gate for MOSFETs is pretty common and appears in many schenatics and app notes, a couple of writers say it may be too slow to protect the devices in very high power or in double-die based amps.

I recently saw a reference (in a context other than an amp) to use uni-directional TVSs as "fast zeners". Is anyone familiar enough with TVSs to comment on whther this would have any potential for speeding up the protection circuit described above? I'm not even sure which direction a uni-directional TVS conducts -- does it reverse conduct like a zener or do you orient it in the same direction as a switch diode. I couldn't gleen this from datasheets as they naturally assume the reader already knows what he/she is doing - which is not the case in this instance!

BTW, regarding the "too slow" problem, I'm assuming it is the zener that limits the speed of reaction not the switch diode (usually 1n4148). If I'm wrong-headed on that, kindly mention it.

Any suggestion (including fugedaboudit) will be appreciated since figuring out the answer by blowing up a bunch of double-die Exicons is unappealing.

[ChocoHolic]

Hi Sam!

Speed, will probably not be the drawback of the common
zener protections. Zeners are fast.
(When clamping some 100ns pulses I did not find
any overshoot, even when looking to the first 10ns.
This was a 39V/0.5W diode.)

I would not expect TVS to be faster.
TVS are designed to clamp voltages, even at high peak loads.
So they can handle impressive current pulses.
Zeners are designed to stabilize voltages, allowed current pulses
are still high, but by far less than in TVS.

To limit the gate drive of a MosFet audio output stage, you
will probably never touch the current limits of the zeners.
But there are many other points, which can cause trouble
in such a simple protection circuit.
The zener+1n4148 is clamping the voltage across Gate-Source-and-source-resistor. By this the max. current through the MosFet is limited.
Following issues exist in these limiting behaviour:
- Tolerances of the zener and of the MosFet can easily lead to quite different max. currents.
- The thermal behaviour of the zener does not match at all to the
behaviour of the MosFets.
- It is a simple current limiter, which does not take care about the
SOA of the transitors. ( At high Vds the allowed currents are less ! )

Only quite sophisticated protection circuits, or heaviest output stages (which appear somehow oversized and slow down the system) will result in a bullet proof amp.

BTW:
A link about some TVS basics.
http://www.protekdevices.com/appnote...20note%202.pdf

Bye
Markus

[Upupa Epops]

Mosfets must have both protections, voltage and current, 'cos by complex load haven't it the same phase. Example of this protection circuit is at attach.

[Workhorse]

Hi Upupa Epops,

Mosfets must have both protections, voltage and current, 'cos by complex load haven't it the same phase

Yes its true in the case of complex load, but the circuit you have attached serves only as foldback VI limiter which in my opinion is not worth in case of high power[>1500WRMS] amps.
Since mosfets are more robust than bipolars but they are much prone to high voltage gate transients.
A unique type of Vds and Ids sensing circuit must be implemented to protect the mosfet along with a input attenuator operating in proportion with specified designated criteria of sensing limits of both Ids and Vds .
In case of parallel mosfets zeners must be mounted directly on Gate to source terminals of each and every mosfet to clamp high voltage transients.
Single Zener on the driver side is not adequate, because there are some situations occuring when the gate driver voltage is low but the transient voltage at the gate of mosfet[isolated by gate resistor] is enough high to cause device failure.
In professional high power amplifiers large number of mosfets are paralleled with zener clamping voltage set aroung 5 to 6 volts[compared to 12V in traditional designs] in order to provide adequate protection. This ensures low gate drive and less current through individual mosfet which inturns ensures device safety in low impedance load driveability.

regards,
kanwar

[Upupa Epops]

Hi Kanwar, in my opinion we can't use zeners directly on gates, 'cos in case, when you have low value serial resistors on gates, you can destroy driver. In my amps I'm using 39 R as gate resistor, which is neccessary for high speed driving of mosfet, so for 1500 W, where you need minimal ten mosfets in parallel, it give only 3.9 R , which must be driven by driver.

[Workhorse]

Originally posted by Upupa Epops
Hi Kanwar, in my opinion we can't use zeners directly on gates, 'cos in case, when you have low value serial resistors on gates, you can destroy driver. In my amps I'm using 39 R as gate resistor, which is neccessary for high speed driving of mosfet, so for 1500 W, where you need minimal ten mosfets in parallel, it give only 3.9 R , which must be driven by driver.

Hi buddy,

We are using 68ohm as gate resistors, 8 parallel mosfets , driver idle current =100mA , driver bleeder resistor =39ohms, still no sign of driver damage.
the criteria you are following for the gate resistor is not adequately specified .

[Upupa Epops]

Have you higher voltage at driver then on mosfet rails ?

[Workhorse]

Originally posted by Upupa Epops
Have you higher voltage at driver then on mosfet rails ?

our driver rails are 12 Volt higher than output rails.