I think the Class B / Class AB thing is a red herring here. It depends on whose definition of Class B you adhere to. Personally I go with Doug Self's definition of Class B as output stage biased just on at idle, at some optimum level, usually a few tens of mA.
Back to the point. I think the OP wants an amp design with all JFETs for input and VAS stages and MOSFETs for outputs. Is that correct Indir ?
I don't recall ever seeing a design with small MOSFETs in input and VAS stages. I can't point the OP to anything specific but perhaps a search through Mr Pass and Mr Curl's contributions might show up something ?
Simon
Back to the point. I think the OP wants an amp design with all JFETs for input and VAS stages and MOSFETs for outputs. Is that correct Indir ?
I don't recall ever seeing a design with small MOSFETs in input and VAS stages. I can't point the OP to anything specific but perhaps a search through Mr Pass and Mr Curl's contributions might show up something ?
Simon
It's like wanting to use valves without an output transformer. The Gm characteristic of mosfets is not suitable for small bias. This is why mosfets amps are invariably class AB and then some. For low bias, tens of mA, bipolars are suitable.
Otherwise you would have to design a special local circuit around each mosfet. Something that linearizes the FETs...
Otherwise you would have to design a special local circuit around each mosfet. Something that linearizes the FETs...
May be you need to build more amps to conclude that (bipolar) transistors are not your cup of tea. But I agree that mosfets have special charm, especially when built properly.
If your experience with mosfets is in class-A, it is the most suitable implementation for the device. For class-B output stage, IRF240 (TO-240) has been used and favored by some, but in general we need low capacitance mosfets like those in TO-220, and laterals.
For VAS, we have not many options. ZVN3110, ZVN2110, IRF610/510 are the most common. For input, IRFD110 (see the Death Of Gainclone, DOGH).
Fetzilla amplifier uses JFET-->ZVN-->LATFET.
There is an old amplifier by JLH (schema hosted by Paul Kemble) using all FET: VP0808(LTP)-->ZVN2110-->LATFET.
I run irfp240/9240 class ab amps with a very small bias current.
Just enough to get rid of crossover distortion.
They seem to work fine like that and sound good.
I believe Peavey also tend to use small bias currents.
An all-FET design is certainly possible, even more so if the various 'house-keeping' functions are relegated to BJTs, such as cascodes and current sources, and especially current mirrors.
That being said, there are some points where the designs will differ. For instance, a regular input differential built with BJTs will have a MUCH smaller tail current than something similarly useful made with JFETs and especially MOSFETs, as you want the best transconductance, but doing that by increasing tail current also increases bias current and also noise current so you might end up needing another follower stage in front.
JFETs have are generally depletion mode so the tail current is limited by Idss (Id at Vgs=0) - and if one operates the JFET very close to Idss, the input capacitances are really non-linear.
MOSFETs don't have this problem and you can go as high as you can while being able to cool them (as you want small ones to keep capacitances low) but then they are much more noisy to begin with, though in a power amp this might not be a major issue.
Going from that, extra bits have to be added as the regular idea behind a LTP at input followed by a VAS is current drive to the VAS, which is essentially NOT the way a FET will work. Also, while any semiconductor has non-linear capacitances, FETs can also make them quite large in magnitude, so things like Miller compensation must be re-thought to a degree. Cascoding is your friend here.
Finally, there are plenty of MOSFETs to use in the output stages but the choice is not trivial because what might look like complementary is usually not (and that's physics).
That being said, there are some points where the designs will differ. For instance, a regular input differential built with BJTs will have a MUCH smaller tail current than something similarly useful made with JFETs and especially MOSFETs, as you want the best transconductance, but doing that by increasing tail current also increases bias current and also noise current so you might end up needing another follower stage in front.
JFETs have are generally depletion mode so the tail current is limited by Idss (Id at Vgs=0) - and if one operates the JFET very close to Idss, the input capacitances are really non-linear.
MOSFETs don't have this problem and you can go as high as you can while being able to cool them (as you want small ones to keep capacitances low) but then they are much more noisy to begin with, though in a power amp this might not be a major issue.
Going from that, extra bits have to be added as the regular idea behind a LTP at input followed by a VAS is current drive to the VAS, which is essentially NOT the way a FET will work. Also, while any semiconductor has non-linear capacitances, FETs can also make them quite large in magnitude, so things like Miller compensation must be re-thought to a degree. Cascoding is your friend here.
Finally, there are plenty of MOSFETs to use in the output stages but the choice is not trivial because what might look like complementary is usually not (and that's physics).
Peavey makes guitar amps.
Don’t get me started on the complete rape and pillage caused by roadies and their crap equipment of highly talented acoustic musicians and singers I saw at the Brecon Jazz Festival recently.
Peavey do PA as well.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.