Hello, I have a question.
I am designing a push pull audio amplifier with 2SK1058 and 2SJ162 mosfets.
Normally they are used with a 0.22Ω source resistors.
What about to put these 0.22Ω resistors connected to the drain?
Thanks
I am designing a push pull audio amplifier with 2SK1058 and 2SJ162 mosfets.
Normally they are used with a 0.22Ω source resistors.
What about to put these 0.22Ω resistors connected to the drain?
Thanks
The lateral FET's which these types are do not normally need the source resistors for stability and control of quiescent current. Even with parallel pairs of FET's (only applies to laterals) and its questionable whether to use them. You can happily omit them for a single pair.
I also would not add them to the drain in that configuration, hazy thinking but they could compromise stability adding impedance at HF to the drain terminal and isolating the drain from the low impedance supply rail.
I also would not add them to the drain in that configuration, hazy thinking but they could compromise stability adding impedance at HF to the drain terminal and isolating the drain from the low impedance supply rail.
Than you very much for your replay.
If I would use them, is it better to use them on the source also in the following configuration?
If I would use them, is it better to use them on the source also in the following configuration?
I would recommend to use source resistors if you put devices in parallel, it will help to equalize the currents in despite of the mismatching characteristics of the FET's
Well I used resistors in a CFP configuration in my own amp but I suspect if I were designing this again I would not use them. It can be argued they also reduce the overall transconductance of the FET which is already low enough and we don't want to make it worse.
Just seen @anibal reply 🙂
My understanding is that the laterals are pretty much 'self balancing' in parallel. As one FET in a pair conducts more it heats more and that automatically tends to reduce the current in that FET. That only applies to laterals though as I mentioned before. I've a feeling I have some info on that somewhere in an old Profusion/Exicon application note.
Just seen @anibal reply 🙂
My understanding is that the laterals are pretty much 'self balancing' in parallel. As one FET in a pair conducts more it heats more and that automatically tends to reduce the current in that FET. That only applies to laterals though as I mentioned before. I've a feeling I have some info on that somewhere in an old Profusion/Exicon application note.
Noted thank, but these resistors are always to be put on the source, independently from the mosfet configuration (see the above pictures)?
I'm going to have to say I disagree on that one at this point 🙂 Laterals can be paralled with ease and with no special precautions.
I found this from nearly 20 years ago but I'm sure I've seen similar info direct from manufacturer application notes. Post #31
I found this from nearly 20 years ago but I'm sure I've seen similar info direct from manufacturer application notes. Post #31
There is a thing called "Thermal Runaway".
Semiconductor conducts more as the temperature rises. Then, more current leads more heat. You need something to break the vicious circle. The source resistor creates a local negetive feed back. It clamps down the gate voltage as the current goes higher.
Semiconductor conducts more as the temperature rises. Then, more current leads more heat. You need something to break the vicious circle. The source resistor creates a local negetive feed back. It clamps down the gate voltage as the current goes higher.
Latfets have negative (or at least non-positive) temperature coeeficient. They differ in this aspect from BJTs and vertical mosfets.
Thermal runaway doesn't happen to them (in 'normal' temperature ranges).
Thermal runaway doesn't happen to them (in 'normal' temperature ranges).
Hitachi 2SJ162 and 2SK1058 do not need any source or drain resistors but 220-330 series resistor on gates to prevent oscillation during amplifier turn on sequence.
Run100-150mA through each of them.
Built many Amps using these Mosfets and they sound great.
Run100-150mA through each of them.
Built many Amps using these Mosfets and they sound great.
Actually its a bit more complicated than this, most MOSFETs start at low current with positive tempco, then transition to negative tempco at higher currents. For lateral FETs this transistion is around 0.1A or similar(*), for VFETs, more like 20A, which is why VFETs require stabilization and laterals don't. BJTs are unremittingly positive tempco and require emitter resistors.
With all MOSFETs the gate voltages can drift slowly over time (ion migration effect in the gate oxide) and require re-biasing - modern devices are more stable than older ones due to manufacturing improvements.
(*) So in theory a lateral biased at 10mA could in theory run-away to 100mA and then stay there as its a stable equilibrium.
Source / drain resistors not needed for single pair.
Current sharing and cross conduction more a problem with parallel devices.
Easy solution is the only application suitable is single pair amplifiers.
Current sharing and cross conduction more a problem with parallel devices.
Easy solution is the only application suitable is single pair amplifiers.
Biggest I have built was 12 complementary pairs without global feedback. 100mA bias for each pair and no resistors on source nor drain.
Drivers output impedance should be no more than 1,5kohm and 1,5mA through the VAS stage for each complementary pairs. So in the case above I used a VAS with 1,2kohm impedance and 20mA of bias current.
Drivers output impedance should be no more than 1,5kohm and 1,5mA through the VAS stage for each complementary pairs. So in the case above I used a VAS with 1,2kohm impedance and 20mA of bias current.
I am glad to have sparked such an interesting discussion, but I have to highlight one point.
On the Hitachi Mosfet data sheet (Hitachi was one of the first, maybe the first, company to have introduced on the market this kind of components) they use a 0.2Ω resistor on the source of the MOS, so there must be a reason.
Anyway my first question was: if I would use these resistors on the CFP configuration (two couples of MOS) should I use them on the source or on the drain?
On the Hitachi Mosfet data sheet (Hitachi was one of the first, maybe the first, company to have introduced on the market this kind of components) they use a 0.2Ω resistor on the source of the MOS, so there must be a reason.
Anyway my first question was: if I would use these resistors on the CFP configuration (two couples of MOS) should I use them on the source or on the drain?
There may or may not have been a valid proven reason. This was new technology at the time and balancing resistors in parallel pairs of transistors had always been used. Maybe they just played it safe thing that to include them removes any doubt....
That there must be a valid reason for a design being the way it is doesn't always compute... think of all the vbe multipliers in commercial designs with the preset in the 'wrong place' such that failure of the wiper connection causes destruction of the output stage whereas if it were in the 'leg' of the resistive divider it would always fail safe to no current. They must have designed it that way for a valid reason... or maybe not.
The Hafler DH500 is a highly regarded lateral FET and that uses three pairs of outputs with no resistors.
I used 0.22 ohms on the Drain of my amp as mentioned earlier. Would I do the same today. Quite possibly not.
Because lateral are so so perfect and amazing.
each device so amazingly matched. They all current share perfectly.
And no matter how unstable DC is. Soon as you connect a lateral
That DC is perfectly stabilized. It fixes every circuit at every temperature.
We got so confused when parallel devices didn't all act the same and blew up.
How come 1 or 2 failed and the others didnt. Why just why?
We left resistors out because they are perfect. So that cant be it
Hitachi just invented them, those resistors with parallels devices are just probably a strange mistake.
Why worry about DC, current sharing is also fine.
Every one is exactly the same.
each device so amazingly matched. They all current share perfectly.
And no matter how unstable DC is. Soon as you connect a lateral
That DC is perfectly stabilized. It fixes every circuit at every temperature.
We got so confused when parallel devices didn't all act the same and blew up.
How come 1 or 2 failed and the others didnt. Why just why?
We left resistors out because they are perfect. So that cant be it
Hitachi just invented them, those resistors with parallels devices are just probably a strange mistake.
Why worry about DC, current sharing is also fine.
Every one is exactly the same.
No, they are not exactly the same and perfect. They just tend to be self-ballasting.
If someone had parallel “LatFets” that one hogged current and fried the first thing I’d suspect is FAKES. A standard switching hexfet stuck in a package and marked with a very desirable (usually unobtainium) part number.
Also with a poor enough match one could hog enough current to be a problem before the tempco became negative enough to cause negative thermal feedback. If you are going to parallel, buy them at the same time. They will be uniform “enough”.
If someone had parallel “LatFets” that one hogged current and fried the first thing I’d suspect is FAKES. A standard switching hexfet stuck in a package and marked with a very desirable (usually unobtainium) part number.
Also with a poor enough match one could hog enough current to be a problem before the tempco became negative enough to cause negative thermal feedback. If you are going to parallel, buy them at the same time. They will be uniform “enough”.