Mosfets in Parallel

[diydidi]

I have been fixing a lot of powered speakers by an Italian company called Montarbo.
What's interesting is that on some of the models they parallel mosfet output devices with no source (current sharing) resistors, and on some of the bigger models the parallel them with the resistors.
Why would this be?
With no source resistors matching has to be 100% spot on is it not?
I mean how much power could be lost across a 0,1 ohm resistor?

[MarkusBass]

I have been fixing a lot of powered speakers by an Italian company called Montarbo.
What's interesting is that on some of the models they parallel mosfet output devices with no source (current sharing) resistors, and on some of the bigger models the parallel them with the resistors.
Why would this be?
With no source resistors matching has to be 100% spot on is it not?
I mean how much power could be lost across a 0,1 ohm resistor?

You can easily calculate it on your own; with 5A current you get 5*5*0.1= 2.5W. But the power loss on this resistor is not a way to balance transistors. They are balanced because the voltage drop on this resistor causes that the voltage on the gate changes and this causes in turn change in the current flowing through the transistor.
I don't think that this is interesting problem. With old amps (from e.g. the seventies), which used vertical MOSFETs (like e.g. 2SK135/2SJ50 from Hitachi) it was quite common that there were no ballast resistors. This was caused by high resistance of the drain-source channel and by positive coefficient of the channel resistance compared to the temperature of the structure. So at that time there were many amps without ballast resistors. Sometimes also the bias sensing transistor was missing. And the amps worked very well despite those "missing" parts.
Is it possible that the amp that you are talking about is old and it uses vertical MOSFETs like 2SK135, or BUZ901?

Mark

[J M Fahey]

You are right, of course, couple volts drop across ballast resistors swamps the less than 100mV Vgs difference in Mosfets same brand and batch.

Now I have seen a few cases (early VOX, Montarbo, Dean Markley) where some "designer" pulled some VFet schematic and saved a few bucks by building it with Switching types ... unmodified.

Those are commercial products and obviously don't have 90% returns of burnt amps, which would have forced them to review the circuits, so I think they survive because they are underbiased, period.

Crossover distortion is not easy to detect for many people, who just perceive "something is not right" but can't put a finger on it.

And/or if played loud or with certain material it's not noticeable.

A complex and pure sound instrument such as a piano sounds horrible with crossover distortion which is easily noticed, doubt a full blast DJ or the powered PA cabinets in a country fair will be blamed for it.

Of course the proper Mosfet type conversion involves ballast resistors, even if low value, and more important, heatsink mounted tracking thermal compensation.

[diydidi]

These Montarbo units don't have the thermal tracking element anywhere near the heatsinks. Outputs are IRFP240/ IRFP9240.

[J M Fahey]

Yes, I have a couple Montarbo schematics and they seem to love 240/9240.

Clearly then

I think they survive because they are underbiased, period.

They probably run one real hot at the factory for a couple hours, *then* adjusted bias, and used same value for the production line.

Very much doubt they adjust all hot, terrible waste of time.

FWIW I sold somewhat underbiased amps for years (decades) , less than a dozen people complained (all of them piano players) , those I biased properly.

Why not all others?
It gave me a slight but measurable safety advantage.

Musicians tend to add more cabinets (total impedance lower than minimum) , put amps against the wall semi covering heatsinks, dropping a leather jacket over the amp with same result, etc.

Mine have a widely perceive aura of robustness, and that's part of it.

Mind you, many Peavey SS amps are also customarily underbiased ... yet nobody complains.

Of course, that works fine with bipolars or VFets, switching type have an abrupt on-off below some 3.7V Vgs which no NFB can compensate.