This was a big debate back in my formative days.
Most use(d) a small value (~0.1u) to transcend the
speed of the (usually) Vbe multiplier. Some swore
the larger the better. I use whatever I use somewhere
else in the circuit for economy.
I would use a resistor instead of the LED if the JFET
would act like the current source it should be. BUT
with a resistor I get xover distortion with a heavy
load that somewhat collapses the rail.
Is this un-ideal current sourceness some manifestation
of the "leakiness" ?
It is the clearest amp I have come up with (for my ears anyways)
THD ~ 0.06 (FWIW) quasi-monotonic (distortion doesn't go down
with level, but not up either) I attribute it to lack of global feedback.
The biggest issue with the lateral MOSFETs in this application is the high and variable gate capacitance. It takes a lot of drive at high frequencies.
If you bootstrap from the output to the drain resistor you would reduce the effect of voltage swing. The cost being an extra cap.
(This feels like working with a China OEM on reducing parts count because a resistor has a cost above zero)
If you bootstrap from the output to the drain resistor you would reduce the effect of voltage swing. The cost being an extra cap.
(This feels like working with a China OEM on reducing parts count because a resistor has a cost above zero)
Edde, the Idss of the cascode fet just has to be equal or greater than the input Idss. Whatever the input device can do, the cascode device has to do. IF you use mosfets for the cascode part, like the CTC Blowtorch, then I found that operating at HIGHER Id than the input device is better, because it gives slightly lower distortion.
This simple circuit might become a 'revelation' for you, hitsware. It reminds me of the extremely simple and 'crude' amplifier that Richard Heyser designed at JPL and took home one night in the '1950's'. This is where the idea that global negative feedback is a problem, rather than a solution, came from first. Long before Charles Hansen, or me, even.
> The biggest issue with the lateral MOSFETs in
> this application is the high and variable gate capacitance.
> It takes a lot of drive at high frequencies.
Not nearly as bad as IRF types. Way lower capacitance.
I can get a nice 10kHz square wave which is my criteria.
The only drawback (for me) with the laterals is the
relatively high channel resistance (~ 1 Ohm/device)
> If you bootstrap from the output to the drain resistor
> you would reduce the effect of voltage swing.
> The cost being an extra cap.
And extra resistor.
If I bootstrap it ups the voltage gain which I don't want
because the ratio of the source and drain resistors set
a predictable gain. (since the circuit has no global feedback
as is the usual gain setting mechanism)
> This feels like working with a China OEM on reducing parts
> count because a resistor has a cost above zero
Low parts count is an aestetic for me .....
__________________
> this application is the high and variable gate capacitance.
> It takes a lot of drive at high frequencies.
Not nearly as bad as IRF types. Way lower capacitance.
I can get a nice 10kHz square wave which is my criteria.
The only drawback (for me) with the laterals is the
relatively high channel resistance (~ 1 Ohm/device)
> If you bootstrap from the output to the drain resistor
> you would reduce the effect of voltage swing.
> The cost being an extra cap.
And extra resistor.
If I bootstrap it ups the voltage gain which I don't want
because the ratio of the source and drain resistors set
a predictable gain. (since the circuit has no global feedback
as is the usual gain setting mechanism)
> This feels like working with a China OEM on reducing parts
> count because a resistor has a cost above zero
Low parts count is an aestetic for me .....
__________________
The laterals really don't need it and I don't like the awkwardness
of making the thermal contact.
( Thats the whole reason I prefer lateral mosfets. Otherwise I'd
use darlingtons )
I have. Measures and sounds the same. For some reason it's habitual.
Based on input from gurus in my past.
If I replace the LED with a resistor, why does the bias change
if the rail changes ? Shouldn't the JFET act as a constant current
source at idle ?
That's fine, but Vth will be all over the place without very careful matching.
For 1.65V, you'd do better with red LEDs- the HLMP6000 would be perfect, very low impedance.
I found these empiracly...........(after trying every diode and LED
that RadioShack stocks)
For some reason big green is different than small green.
- Status
- Not open for further replies.