Anybody know how to slow down MOSFET's switching time? Will adding small capacitance between Gate and Drain do this?
If it's only for a switching purpose then increasing the gate resistor until you are satisfied would do...
It is for audio, for matching between positive and negative signals to be precise. I guess it is the same concept right?
I've heard people used that approach (the gate resistor) but I cannot understand the theory. I mean having extra capacitance to drive is more make sense to me. The capacitance will just delay the gate from reaching voltage above Vgs threshold.
The resistor may also slow down the rate at which the MOSFET turn on and off but I'm not sure if it is the same thing as what gate capacitance does to signal pulse at the gate. And I'm not sure either whether it is appropriate for matching purpose...
I think I have ever seen somewhere writing (from Nelson Pass?) regarding simulation of input, output and reverse transfer capacitance of a MOSFET but couldn't find it...
I've heard people used that approach (the gate resistor) but I cannot understand the theory. I mean having extra capacitance to drive is more make sense to me. The capacitance will just delay the gate from reaching voltage above Vgs threshold.
The resistor may also slow down the rate at which the MOSFET turn on and off but I'm not sure if it is the same thing as what gate capacitance does to signal pulse at the gate. And I'm not sure either whether it is appropriate for matching purpose...
I think I have ever seen somewhere writing (from Nelson Pass?) regarding simulation of input, output and reverse transfer capacitance of a MOSFET but couldn't find it...
Dear Pavel, I mean both with the same amount of reduction. I wonder if the "zobel" in your DPA-330 might also do the trick?
Jay said:
I would think for audio you would want the fastest response time, even if it is not symmetrical. If the fast response time is (much) higher than the audio max rate of change (which occurs at full power 20kHz zero crossing, for instance), that would make sure that at least in the output stage you have no slew-rate limiting.
If you mean that the driver stage cannot source/sink enough current to turn the FETs on/off fast enough, than I would look for a solution in a modified driver stage rather than slowing down the fastest FET to be as slow as the slowest one. It may be symmetrical, but you would still have slew rate limiting due to the limited driver stage.
Jan Didden
Increasing gate resistor value or adding extra gate capacitance will have virtually the same effect, since they both increase the time needed to charge the gate capacitance. I would suggest that increasing gate resistor value would be preferable to avoid any possible oscillation problems.
Jay,
The most simply and efficiency way is add different resistor to N/P MOST's Gate....
Cheers
Wenye
The most simply and efficiency way is add different resistor to N/P MOST's Gate....
Cheers
Wenye
Nonlinear Gate Capacitance
I do not recall in which thread, but it was mentioned about higher distorsions if the gate resistor is too high, due to nonlinear gate capacitance. Therefore, isn't adding a cap the better solution?
BTW, Charles Hansen posted a schematic of an Ayre V-3 prototype in another thread some months ago, and there are different gate resistors between the N- and the complementary P-channel Mosfets. I believe they are in for the same reason you are looking for remedial.
Regards
Tino
I do not recall in which thread, but it was mentioned about higher distorsions if the gate resistor is too high, due to nonlinear gate capacitance. Therefore, isn't adding a cap the better solution?
BTW, Charles Hansen posted a schematic of an Ayre V-3 prototype in another thread some months ago, and there are different gate resistors between the N- and the complementary P-channel Mosfets. I believe they are in for the same reason you are looking for remedial.
Regards
Tino
To slow down rise and fall times in switching applications and get flatter rise and fall slopes I recommend both adding external drain-gate capacitance and increasing gate resistance [gate clamping ie:zeners is also advidsed to prevent gate overvoltage]
Rise and fall times depend on gate resistance and on the relationship between Cd-g and Cg-s
Mosfets by nature show extremely non-linear capacitances so the rise and fall waveforms at constant current tend to be more exponential than linear
Class G amplifiers with switched supply rails are an example where controlled slew-rate is desirable for the rail-switching MOSFETs
Rise and fall times depend on gate resistance and on the relationship between Cd-g and Cg-s
Mosfets by nature show extremely non-linear capacitances so the rise and fall waveforms at constant current tend to be more exponential than linear
Class G amplifiers with switched supply rails are an example where controlled slew-rate is desirable for the rail-switching MOSFETs
Per a "whitepaper" on the Exicon website, the drain-gate capacitance and the drain impedance form an RC filter. Adding a gate resistor lowers the pole of the filter. It's nornally part of a feedback loop so this action reduces the bandwidth or "speed" of the MOSFET.
If I'm not explaining this right (I'm not crystal clear on how it works), someone please correct me.
The power MOS contributes to the second pole. Adding a cap across G-S without enhancing the driver stage and lowering the driving resistance will move the second pole closer to the first pole, and thus causing instability.
I would think the two resistor connecting to the gate can be balanced roughly in the way such that both rc time contanting.
Adding cap to G-S is a good idea to make the input cap less nonlinear. (But how about Cgd?) Doing so the resistance has to be lowered and the driving stage has to be enhanced. Otherwise stability and slew rate issues can be evident.
I would think the two resistor connecting to the gate can be balanced roughly in the way such that both rc time contanting.
Adding cap to G-S is a good idea to make the input cap less nonlinear. (But how about Cgd?) Doing so the resistance has to be lowered and the driving stage has to be enhanced. Otherwise stability and slew rate issues can be evident.
Umm...you should not try to slow the mosfet down. The output mosfet will only follow what your driver stage is doing.
If you have a problem controlling the mosfet then your drive impedance may be too high. This can be improved by increasing the current through this stage, you will have to lower the value of resistors here to maintain the same bias drive voltage.
If your problem is oscillation then you need to fix this at either the input stage or at the driver stage.
If your driver stage uses transistors then try a small (100-1000pf) capacitor between the base and emitter on both driver transistors.
If the input stage is a differential pair then a small capacitor (100pf-1000pf) between the pair may help.
If you have a problem controlling the mosfet then your drive impedance may be too high. This can be improved by increasing the current through this stage, you will have to lower the value of resistors here to maintain the same bias drive voltage.
If your problem is oscillation then you need to fix this at either the input stage or at the driver stage.
If your driver stage uses transistors then try a small (100-1000pf) capacitor between the base and emitter on both driver transistors.
If the input stage is a differential pair then a small capacitor (100pf-1000pf) between the pair may help.
In the Soundcraftmen amps, they used 2 different driver transistors. 2N3440 & MM4003
Normally the pair match to a 2N3440 is a 2N5416 Both are 1 amp devices, but the SC amps use a MM4003 as a pair to the 3440 which is only a 500ma device instead.
I can only make assumptions as to why they did this. My first thought however, was that they were trying to match the charge times of the 2SK135 & 2SJ50 mosfets. (which have a lot of gate capacitance) It my be more linear as such as well???
The Halfer amps use 2N3440 & 2N3415 drivers i believe.
Zero 😎
Normally the pair match to a 2N3440 is a 2N5416 Both are 1 amp devices, but the SC amps use a MM4003 as a pair to the 3440 which is only a 500ma device instead.
I can only make assumptions as to why they did this. My first thought however, was that they were trying to match the charge times of the 2SK135 & 2SJ50 mosfets. (which have a lot of gate capacitance) It my be more linear as such as well???
The Halfer amps use 2N3440 & 2N3415 drivers i believe.
Zero 😎
I've never seen that done. Is it between collectors or between emitters? I would whink emitters, but if I knew for sure I wouldn't ask.
Place 100pf (starting point) between the collectors of the input differential pair.
This has the effect of shunting any high frequency. In some circuits I have seen a resistor capacitor combination, but I just use a capacitor.
This has the effect of shunting any high frequency. In some circuits I have seen a resistor capacitor combination, but I just use a capacitor.
Transistor Sub.
I agree the 2N5416 seems a better match.
Closer specs than the MM003.
Charging / discharging the gate capacitance of these FETS would not be a big deal though provided the SOAR of the MM003 was ok.
I agree the 2N5416 seems a better match.
Closer specs than the MM003.
Charging / discharging the gate capacitance of these FETS would not be a big deal though provided the SOAR of the MM003 was ok.
- Status
- Not open for further replies.