Mosfet Class A drive: Low voltage sharp attack Vgs :VS: Smooth gentle Vgs.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I want to build an audiophile, or better, grade amplifier. The theme is a minimalist class A design. A true 50 watt RMS output is the target power, mono.

Basically, a huge source follower drive stage. Which type of mosfets should be used?

The really low Vgs on, with a steep current curve.
Eg: N-Chan IRF3808
http://www.irf.com/product-info/datasheets/data/irf3808.pdf

:VS:

A soft progressive large voltage Vgs.
Eg: P-Chan IRF5210
http://www.irf.com/product-info/datasheets/data/irf5210.pdf


How would the choice of mosfets differ in a full feedback design :VS: low feedback design :VS: a 0 feedback design?

Has anyone designed a true balanced output full Class A amp? This design would probably be a power supply guzzler & a nice room heater…

_____
Brian.
 
I expect that in a no-feedback design a mosfet with a steep transconductance curve would be better because in a follower cct if the source gets a little behind in following the gate then won't take much until the fet digs in and pulls the source harder. Of course, the steepness of the transconductance curve has to be looked at *where you are actually using it*. No use choosing a fet that is crash hot linear and near vertical way up at 30 amps if you are actually using it at 3 amps.

"50 watts would be ridiculous, the power supply would need to draw around 500 watts continuous."

Naaah... My (almost) 50 watts rms per channel class A (source follower with choke load) has a 28v 3.5 amp drain per channel. It swings (almost) 56v p/p into 8 ohms. IIRC it measured about 47 watts or something.
 
Ive been thinking of using a linearized mosfet class a amp with a gentle gate slope. It adds 1 transistor at the gate, but, changing the value of the resistor between the colector and source of the mosfet will allow me to tune the Vgs slope any way I like.

It will also with ease make the output impedance run down into the mohms & help cut the gate capacitance.


Would it be wise to choose a mosfet where my bias current matches the temp balance point?
 
I agree with Circlotron. It all depends...a little risky to generalize in these matters. But I'll do so anyway. In general, minimizing capacitances is good and minimizing voltage changes is good. So a high transconductance FET with low capacitances is likely to be better than the opposite. Same goes for the BJT. In any case, in a SE design what you are trying to achieve is high linearity. In semiconductor devices capacitances and voltages are non-linear wrt output current, so minimize them or their effects. You can minimize the non-linear current due to capacitance by minimizing the voltage change across the capacitance.
 
Ok, for a P-Channel design, I think I found the ultimate component for a 4 amp biased Class A design.

IRF9510
http://www.irf.com/product-info/datasheets/data/irf9510.pdf

Notice that beautiful soft Vgs curve, no bumps, or clumping.

Each device:
43 watt
200 pf Ciss with vgs at 0v.
94 pf Coss.
At 1 amp drain current, Vgs stays at 4.6v from 25 degC thru 175 degC.
0.90$ Cdn, or, 0.58$ Us each.

For a target bias design current of 4 amps, use 4 devices in parallel.

Totals:
172 watt
800 pf Ciss with vgs at 0v
376 pf Coss.
3.60$ Cdn, or, 2.34$ Us.

Every other P-channel mosfet I looked at, both for singles & paralleling, for a nice curve with a bias at 4 amps, the Ciss start at around 1300pf, and they are much slower, & they would be less than 172 watts. Also, the fact that you can spread the 4 across a heatsink helps alleviate a very small hot spot, though, it will probably require a huge one to cool these guys drawing 1 amp from each.

If someone has seen better, post!

I will now scan for N-channel solutions, however, I've noticed that the majority of n-channel devices have steep, or bumpy Vgs curves.
 
Verbal said:
Before you start reinventing the wheel (which can be fun, too, of course), I suppose you are familiar with Nelson Pass' Zen and Aleph designs? If not: www.passdiy.com, Pass Labs forum


I have,

The exercise of re-inventing the wheel is to go through the history of amp development by one's self & finding / looking in directions which may have been dumped in the past which open's up a new point of view on an old dismissed design concept. Also, today's more advanced circuits may patch such a dismissed design offering new potential in expanding such a design.

Just going the passidy/Pass labs may lead to nothing more than pumped up clone of someone else work. There is too little chance of anything really new/old being discovered.

I prefer kicking / screaming / making mistakes over agin insted of the lazy going with the flow.

E.g. I have just recently designed a cheap, fully passive Vactrol Balanced In / Balanced out 2-thru 16 channel remote attenuator which goes from –0.1db to – infinity db. It seems to outperform mechanical pots & most of the relay / switched resistor passive attenuators, without aging / channel-channel gain drifting and it can attenuate 100Vrms signals. Because of the Vactrol attenuation flaws, (aging / channel-channel gain drifting, never achieving a true –infinity db), they have been abandoned as a feasible attenuator solution for Hi-Fi (more like Audio Pure-ist) audio pre-amp uses. I’ve been able to nail these problems down to such a degree that I feel that some die-hard mechanical switched fixed resistor attenuator lovers may switch. It’s taken me around a year to get it just right, worthy to take notice of.

Remember, today, we have much better regulated means of light source, microprocessors size, price & processing capabilities, 24bit ad/da converters in the 2$ range which we never dreamed of even just a few years ago & Vactrols have pretty much dropped their real potential place in audio over 20 years ago.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.