Lateral FET current limiting

[Neutrality]

How important is current limiting for Lateral FETs? Since they, due to their negative tempco at drain current above 100-300mA are sort of sell protecting, is current limiting really needed?

If so, what would be the best way to implement it?

[Neutrality]

It has been suggested that the Vgs should be limited to 10V and then use some sort of thermal sensor to turn off the amplifier when the HS reaches 70-80 Degrees Celsius.

That could be implemeted with some sort of transistor arrangement that turns off the drivers when reaching 70-80 Degrees Celsisus. Maybe shunting the voltage from the VAS to the 2 BJT drivers to ground would do the trick? Just 2 transistors, one on each base input of the drivers, then connected to the thermal sensor.

Btw, using an output stage with BJT drivers for the Lateral FETs.

EDIT : Something like attached schematic.

[Ian Finch]

Depending on the use (ie as personal use, commercial or for retail sale etc,), there may be no need at all for output limiting with lateral Mosfet output stages.

I have built, used and then sold around 20 Mosfet amps in the power range of 50-200W over the many years since Hitachi introduced the parts. None have had limiters and none, to my knowledge, ever failed due to overcurrent - such is the reliability of the thermal self-limiting in conservative AB amplifier designs.

For personal use, I can't see any reason to fit limiters but there will always be crazy loads and fault conditions which suggest something more positive and fast acting in commercial high power applications. I guess we're not considering that use and can rely on the main drawcards of laterals, which is inherent short circuit immunity, load sharing and extreme ruggedness.

[Neutrality]

Originally Posted by Ian Finch

Depending on the use (ie as personal use, commercial or for retail sale etc,), there may be no need at all for output limiting with lateral Mosfet output stages.

I have built, used and then sold around 20 Mosfet amps in the power range of 50-200W over the many years since Hitachi introduced the parts. None have had limiters and none, to my knowledge, ever failed due to overcurrent - such is the reliability of the thermal self-limiting in conservative AB amplifier designs.

For personal use, I can't see any reason to fit limiters but there will always be crazy loads and fault conditions which suggest something more positive and fast acting in commercial high power applications. I guess we're not considering that use and can rely on the main drawcards of laterals, which is inherent short circuit immunity, load sharing and extreme ruggedness.

Private use only, 3 Pairs of 16A, 250W, TO-264 Lateral FETs in the output.

Approximately 125-135 Watts into 8 Ohm with 48 Volt supply rails for the driver+output stage.

Going to drive floorstanders with 8 ohm nominal impedance, dropping to about 4 Ohm somewhere in the low bass area.

Another thing, now that you mentioned that you have built over 20 Lateral FET amps through the years, I have a question for you. Do you use source resistors when paralleling several pairs of Lateral FETs or do you just match them?

[Neutrality]

Originally Posted by Neutrality

It has been suggested that the Vgs should be limited to 10V and then use some sort of thermal sensor to turn off the amplifier when the HS reaches 70-80 Degrees Celsius.

That could be implemeted with some sort of transistor arrangement that turns off the drivers when reaching 70-80 Degrees Celsisus. Maybe shunting the voltage from the VAS to the 2 BJT drivers to ground would do the trick? Just 2 transistors, one on each base input of the drivers, then connected to the thermal sensor.

Btw, using an output stage with BJT drivers for the Lateral FETs.

EDIT : Something like attached schematic.

Well, my brain wasnt working properly when I thought of that.

I could just use the thermal sensor circuit to mute the amplifier input instead. So much easier. Just 1 transistor on the input, shunting it to ground would do the trick.

[wahab]

As already said , the protection circuit must be connected to the drivers
bases not directly to the fet s gates....
Relevant circuitry must then be implemented to limit the vas current.

As connected in your schematic , the drivers would not survive if ever the
protection start to conduct.

[Neutrality]

Originally Posted by wahab

As already said , the protection circuit must be connected to the drivers
bases not directly to the fet s gates....
Relevant circuitry must then be implemented to limit the vas current.

As connected in your schematic , the drivers would not survive if ever the
protection start to conduct.

Then my question is, why?

EDIT : Ah, think I understand now. If the protection starts to conduct in some sort of fault condition, the current the drivers will try to deliver could reach alot higher levels than what the drivers are biased at. Right? Basically I could fry my drivers. However, wouldnt the most likely scenario be the zeners going up in smoke first?

The protection is there to protect from gate-source voltages reaching more than 20 V. Here the voltage is restricted to about 10V.

VAS current is limited to about 10 mA, both LTP and VAS uses a CCS.

[wahab]

Originally Posted by Neutrality

Then my question is, why?

However, wouldnt the most likely scenario be the zeners going up in smoke first?

When overbiaised , diodes goes short circuits...

[Ian Finch]

Originally Posted by Neutrality

Another thing, now that you mentioned that you have built over 20 Lateral FET amps through the years, I have a question for you. Do you use source resistors when paralleling several pairs of Lateral FETs or do you just match them?

A good question, because these designs were similar, using Video Driver BJTs in a VAS with Current Source and various buffer stages plus 1-3 Mosfet output pairs. They date from the 1980s-90s and used simple 0R22 source resistors without matching, as the necessity with BJTs is much reduced here by current sharing. However, some folks will want to match anything that produces audio, in an effort to minimise distortion. I've compared a similar design with the same but matched TO3 devices and not heard a worthwhile difference when driven hard into inefficient 4-way speakers - perhaps occasionally clipping.

Bob Cordell has some interesting comments in his book regarding Mosfet matching because of particular concerns with Hi-fi at lower levels. Given the enormous potential cost of matching these expensive parts, it's some relief that he trusts "same tube" batch consistency, since just the threshold voltage can then be used to determine transconductance.

Unfortunately, I can't comment on amplifiers without source resistors. I am only sure of having listened to a few inexpensive automotive types.

[Neutrality]

Originally Posted by Ian Finch

A good question, because these designs were similar, using Video Driver BJTs in a VAS with Current Source and various buffer stages plus 1-3 Mosfet output pairs. They date from the 1980s-90s and used simple 0R22 source resistors without matching, as the necessity with BJTs is much reduced here by current sharing. However, some folks will want to match anything that produces audio, in an effort to minimise distortion. I've compared a similar design with the same but matched TO3 devices and not heard a worthwhile difference when driven hard into inefficient 4-way speakers - perhaps occasionally clipping.

Bob Cordell has some interesting comments in his book regarding Mosfet matching because of particular concerns with Hi-fi at lower levels. Given the enormous potential cost of matching these expensive parts, it's some relief that he trusts "same tube" batch consistency, since just the threshold voltage can then be used to determine transconductance.

Unfortunately, I can't comment on amplifiers without source resistors. I am only sure of having listened to a few inexpensive automotive types.

Hmm, so one could do without maching with 0R22 source resistors?

Also, I have Cordells book and I did read that part, but although FETs from same batch should usually be fairly close one cant be certain without matching.

Matching expensive Lateral FETs could get really, really expensive. At something like 7.8£ for the Exicon 16A, 250 W devices from Profuisonplc matching 2 sets of 3 pairs of N channel and 2 sets of 3 pairs of P channel FETs would not be cheap.

Have to think about this, either no matching, but 0R22 source resistors or matching and then a smallar source resistor of 0R1.