sorry my mistake 20 volts.
I need to do some further reading on the usage of mosfets of source followers.
I need to do some further reading on the usage of mosfets of source followers.
MOSFETs have a very thin insulator layer between gate and source which can be perforated by applying too much voltage. This accounts for the sensitivity to static electricity damage.
It's customary to user a ~15V zener diode to clamp the Vgs. My personal rule is if there's nothing in the circuit to prevent Vgs from being exceeded, then the zener should be added. Many CCS circuits can get away without the clamp but for the source follower circuit I always use it.
Unlike vacuum tubes, MOSFETS have zero sense of humor when it comes to exceeding their voltage ratings. FWIW, I've not toasted a MOSFET in an audio circuit ... yet.
Cheers,
Michael
It's customary to user a ~15V zener diode to clamp the Vgs. My personal rule is if there's nothing in the circuit to prevent Vgs from being exceeded, then the zener should be added. Many CCS circuits can get away without the clamp but for the source follower circuit I always use it.
Unlike vacuum tubes, MOSFETS have zero sense of humor when it comes to exceeding their voltage ratings. FWIW, I've not toasted a MOSFET in an audio circuit ... yet.
Cheers,
Michael
It seems unlikely that you will run into issues with this but during turn-on it might be possible. A protection diode could sort that out easily enough.
MOSFETs often as not, used to have symmetrical internal construction.
Could exchange drain for source and realize no difference whatsoever.
But why is this trivia important enough to bother making a point???
Because you will see MOSFET capacitance from gate to source
(cathode) measures much higher value than gate to drain (plate).
But wait, didn't I just say they were symmetrical??? Yes, I did...
At very low voltage, both capacitors are similar and roughly equal.
But apply a bias voltage, and the shape of the channel changes.
Now one cap has wider dielectric than the other.... Its very very
important that drain cap becomes a smaller value.
Anyways, if you want Miller cap to be small and manageable,
you must keep drain voltage higher than gate, 10V at least.
Else this miller cap returns to its much larger resting value.
There is no follower action at the drain to hide this effect.
Any drain action conspires with Miller against you.
Anyways, a BJT can operate with only a few collector Volts
above full saturation. Capacitance can and will change with
base width, but this effect is still much less than MOSFET.
Don't be fooled the low RDS_ON of the MOSFET. May appear
to offer lower voltage drop than saturation of a BJT. But the
drive required to do so is extremely non-linear. Not an area
of operation you want to be in for audio purposes.
Give both 10V to play with, the MOSFET starts to look better.
These days, few MOSFETs are made symmetrical. But all still
need drain Voltage, and for the same reason.
Could exchange drain for source and realize no difference whatsoever.
But why is this trivia important enough to bother making a point???
Because you will see MOSFET capacitance from gate to source
(cathode) measures much higher value than gate to drain (plate).
But wait, didn't I just say they were symmetrical??? Yes, I did...
At very low voltage, both capacitors are similar and roughly equal.
But apply a bias voltage, and the shape of the channel changes.
Now one cap has wider dielectric than the other.... Its very very
important that drain cap becomes a smaller value.
Anyways, if you want Miller cap to be small and manageable,
you must keep drain voltage higher than gate, 10V at least.
Else this miller cap returns to its much larger resting value.
There is no follower action at the drain to hide this effect.
Any drain action conspires with Miller against you.
Anyways, a BJT can operate with only a few collector Volts
above full saturation. Capacitance can and will change with
base width, but this effect is still much less than MOSFET.
Don't be fooled the low RDS_ON of the MOSFET. May appear
to offer lower voltage drop than saturation of a BJT. But the
drive required to do so is extremely non-linear. Not an area
of operation you want to be in for audio purposes.
Give both 10V to play with, the MOSFET starts to look better.
These days, few MOSFETs are made symmetrical. But all still
need drain Voltage, and for the same reason.
Attachments
Last edited:
Thanks for the insights on mosfets Ken. The datasheet of the Phillips BF245A/B/C lists "interchangeability of drain and source connections" as one of the features of this device! Frequencies up to 700Mhz was the other feature, but it has a low max Vds of 30V, not much use in the tube application.
I have found very positive benefits from using mosfet source followers in a couple of small PP amps running in Class AB1, that is, an improvement in clarity and tighter/stronger bass. Subjective results of course. Anyone else found this? The only other known designs using mosfet source followers, that I know of, are the Baby Huey and the Tubelab SE (and Chris's 6L6 AB2 amp). The improvement in sound quality is like there was an impedance mismatch before the source followers were fitted. But before fitting the source followers, I did not think there was any impedance mismatch at all. Interesting stuff...
I have found very positive benefits from using mosfet source followers in a couple of small PP amps running in Class AB1, that is, an improvement in clarity and tighter/stronger bass. Subjective results of course. Anyone else found this? The only other known designs using mosfet source followers, that I know of, are the Baby Huey and the Tubelab SE (and Chris's 6L6 AB2 amp). The improvement in sound quality is like there was an impedance mismatch before the source followers were fitted. But before fitting the source followers, I did not think there was any impedance mismatch at all. Interesting stuff...
Me, with triode output stages. The reduction in Miller effect, increased headroom (AB2 whether you like it or not!), and freedom from blocking make a very big audible difference. I went to MOSFET drive for my triode amps back in the early '80s and never looked back. For screen drive, it's nearly mandatory.
Sgregory's Opus PP KT88 is another project with mosfet source followers.
http://www.diyaudio.com/forums/tubes-valves/161702-opus-5-0-modern-mullard.html
Nice design. Had something similar in mind but with a small pentode (EF86 maybe) as the sink on a LTP.
Also to use IRFU310.
Also to use IRFU310.
One important bit - do chose your MOSFETs propperly, not just because you 'had them in the bin'. I.e., don't use IRF840 if the current rating of an IRF810 will do (and it normally will as you'd be hard pressed to find a tube able to consistently pass a few amps of current!). The increased gate-source capacitance of the higher current parts is generally not a problem as it's matched with increased transconductance, and thus well bootstrapped. But, this does not happen with the drain to source capacitance, so you don't want to have it any larger than necessary. High current high voltage parts will easily amount to several hundred pF here, you might well need a tube follower just to drive that!
- Status
- Not open for further replies.