Lumba,
You know we have totally mad thoughts on amp design sometimes. And not just mad I mean weird. But you have to push the boundaries sometimes.
Mad thought, amp using laterals and Hexfets. Could you arrange for the Hexfets to take over from the laterals at higher outputs, different vgs and so on.
You know we have totally mad thoughts on amp design sometimes. And not just mad I mean weird. But you have to push the boundaries sometimes.
Mad thought, amp using laterals and Hexfets. Could you arrange for the Hexfets to take over from the laterals at higher outputs, different vgs and so on.
Workhorse said:
Cross conduction at high frequencies is due to not having low enough turn-off driving impedance and too slow a time constant. Also, the VAS diff transistors have different voltages. One has 27V-Vgs and the other has 54V-Vgs. A cascode at gnd potential is needed for the bottom half's driving VAS transistor so they will have same voltage and same Pd, otherwise DC output may drift a bit.
Hmmm
Totem pole gate drivers.....what a novel idea.
Mooly,
NO, the boundaries should be pushed in right direction, I would just use verticals in CFP configuration. You know, they are not very linear devices, but the CFP can do miracle, due to the way it works. In your amp, the driver`s tiny die mainly set the sound of the output stage, amazing isn`t it? So I would concentrate on linearizing the driver, the heart of the output stage. The CFP is a simple arrangement, which is essential in audio.
NO, the boundaries should be pushed in right direction, I would just use verticals in CFP configuration. You know, they are not very linear devices, but the CFP can do miracle, due to the way it works. In your amp, the driver`s tiny die mainly set the sound of the output stage, amazing isn`t it? So I would concentrate on linearizing the driver, the heart of the output stage. The CFP is a simple arrangement, which is essential in audio.
Even with Cascode it wont help.......!!!!CBS240 said:
Totem pole gate drivers.....what a novel idea.
Yesss, for Effective & Active gate Turn-OFF these are required, but regulars designers wont think outside of the box as usual , they think driving a vertical mosfet is just like driving a bipolar.............
Lumba, the only reason i never used mosfets is subjective, I listen to mostly rock music and found mosfets to not have bass slam compared to a good Bjt output. The best mos output I could find was stochinos fast amp, but driving laterals with diamond buffers although making amp complex gives me the same slam I can get out of bjts. This way, using the same front end I find it very hard to tell the two apart.
Lumba which devices are limited, bjts??? or laterals???
If you know the stochino, why does that amp suffer from crossconduction?? I had to take some measures to minimise this as it caused the fets to go into oscilation and heat up. Cure??? I have limited experience with mosfet outputstages but seems to me they suffer just as much from crossconduction.
Today while googling for some info, I came accross another guys design in another forum doing the same diamond buffer drive of laterals. I thought mine would be unique, but he designed it back in 2006. He also claims this method is the best he has found to drive mosfets, except for the frontend our outputs are near close copies. He s the co designer of some circuits at amb labs and has designs at headfi too.
If you know the stochino, why does that amp suffer from crossconduction?? I had to take some measures to minimise this as it caused the fets to go into oscilation and heat up. Cure??? I have limited experience with mosfet outputstages but seems to me they suffer just as much from crossconduction.
Today while googling for some info, I came accross another guys design in another forum doing the same diamond buffer drive of laterals.
I thought mine would be unique, but he designed it back in 2006. He also claims this method is the best he has found to drive mosfets, except for the frontend our outputs are near close copies. He s the co designer of some circuits at amb labs and has designs at headfi too.
Hi
IM
, the CFP has its stability problems, but there are advantages in that it has local feedback (or error correction) and voltage gain all in one stage. If you’re using verticals, Vth may be 3 or 4 volts but with CFP, this doesn't matter. To use verticals in complementary source follower arrangement, higher voltage is needed, higher than the drain voltage, to drive the gates. A lot of people don't want to deal with separate power supply rails. If linearity is also desired from verticals, (laterals tend to be much more linear and easier to drive) an additional wide bandwidth error amplifier is needed as well. This amplifier has to be able to drive a little bit of current due to the high order frequencies it generates to correct the high order non-linearity and being hampered by the input capacitance. IMHO, the error correcting ability of the CFP, while there, is limited by the GBW of the CFP amplifying stage. As for a source follower, the bandwidth of the error amp can be made much higher due to having a unity voltage gain for the input signal....but of course this is all at the cost of more complexity.
BJT's have no limit to the amount of current that can flow when fully saturated. They have virtually zero impedance, but a mosfet has a finite maximum conduction that is dependent on things like device, temperature, Vds. This might be why BJT’s seem to have more "slam". In order to get the same kind of damping and "slam" out of mosfets, they have to be driven with a low impedance source, preferably one that measures and corrects for the errors at the output. Vgs is a function of drain current, not input voltage. The drive required for linear operation across a reactive load is interesting. This may also be what happens to all your "slam". Any mosfet will turn on more than enough current way fast enough to get plenty of "slam" if it is driven properly.
I used to be a BJT guy, and criticized mosfets....until I learned how to use them. Now, I'm a FET guy.
IM
, the CFP has its stability problems, but there are advantages in that it has local feedback (or error correction) and voltage gain all in one stage. If you’re using verticals, Vth may be 3 or 4 volts but with CFP, this doesn't matter. To use verticals in complementary source follower arrangement, higher voltage is needed, higher than the drain voltage, to drive the gates. A lot of people don't want to deal with separate power supply rails. If linearity is also desired from verticals, (laterals tend to be much more linear and easier to drive) an additional wide bandwidth error amplifier is needed as well. This amplifier has to be able to drive a little bit of current due to the high order frequencies it generates to correct the high order non-linearity and being hampered by the input capacitance. IMHO, the error correcting ability of the CFP, while there, is limited by the GBW of the CFP amplifying stage. As for a source follower, the bandwidth of the error amp can be made much higher due to having a unity voltage gain for the input signal....but of course this is all at the cost of more complexity. BJT's have no limit to the amount of current that can flow when fully saturated. They have virtually zero impedance, but a mosfet has a finite maximum conduction that is dependent on things like device, temperature, Vds. This might be why BJT’s seem to have more "slam". In order to get the same kind of damping and "slam"
out of mosfets, they have to be driven with a low impedance source, preferably one that measures and corrects for the errors at the output. Vgs is a function of drain current, not input voltage. The drive required for linear operation across a reactive load is interesting. This may also be what happens to all your "slam". Any mosfet will turn on more than enough current way fast enough to get plenty of "slam" if it is driven properly.I used to be a BJT guy, and criticized mosfets....until I learned how to use them. Now, I'm a FET guy.
homemodder,
bipolars win the Grand Slam Prize, MOSFETs the High Frequency Response Reward. Low bandwidth means distortion, effecting a large part of the audible range, to get a nice sweet sound you need high bandwidth.
There are divergent opinions on how to drive MOSFETs in audio. When MOSFETs are used as switches, must be driven from low impedance sources for fast insertion and extraction of controlling charge. Workhorse`s beautifully composed totem-pole circuit would surely do its utmost driving clamped inductive switching stages.
Regarding their linear operating mode, MOSFETs are true transconductance devices, where the input voltage and output current are directly related (the current is determined by the gate to source voltage), so for highest accuracy in conversion, should be driven from voltage amplification stages.
bipolars win the Grand Slam Prize, MOSFETs the High Frequency Response Reward. Low bandwidth means distortion, effecting a large part of the audible range, to get a nice sweet sound you need high bandwidth.
There are divergent opinions on how to drive MOSFETs in audio. When MOSFETs are used as switches, must be driven from low impedance sources for fast insertion and extraction of controlling charge. Workhorse`s beautifully composed totem-pole circuit would surely do its utmost driving clamped inductive switching stages.
Regarding their linear operating mode, MOSFETs are true transconductance devices, where the input voltage and output current are directly related (the current is determined by the gate to source voltage), so for highest accuracy in conversion, should be driven from voltage amplification stages.
Lumba Ogir said:
The drain current of Vertical mosfets is not only proportional to gate drive voltage, but also to Vds Drain to Source Voltage. The Ciss changes non-linearly when Vds approaches less than 5volts that is near clipping condition, and at this condition the normal input capacitance at the gate is multiplied to huge value and only totem pole driver can then help in extracting large amount of charge starage and bring the amp out of Clip Sticking, otherwise you will end up with large cross conduction spikes which will lead to failure, so a passive resistor turn-oFF is no where beneficial in N-channel output stage. The pushpull drivers i use are baised in claas-A providing linear operation to some extent, but do great job on charge suck-up during hard clipping and reactive loading.
But things are different when it comes to complementary mosfet output stage in source follower topology.
To some extent, BJT's require charge suck-out too. Good luck building a high bandwidth amp without actively turning off the output stage devices, whichever they are. That's why most amps have a small cap across the bases/gates of the outputs. As for mosfets and linear operation, because Vgs is determined by Id, if the load is reactive, then the voltage source gate drive must reflect this change in phase of Id (and Vgs) wrt the input signal and load impedance. This is in addition to the change in Vgs wrt Vds as Workhorse pointed out.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.