Biasing a vertical mosfet such as the irfp240/9240 series using resistor is unstable due to thermal issues. You may try a Vgs multiplier similar to the Pass A75 amp thermally connected to the output stage heat sink and 0.47 ohm or higher source resistors.
The irfp240/9240 pair have more than twice Crss compared to the 2sk1058/2sj162 pair, hence distortion rises higher towards high frequency.
Better cascode the input jfet for a higher pre rails than sorry. The resulting circuit becomes more complex though.
To me, the circuit in post #1 and #66 are comparable in simplicity. I don't think there is any missing component. However, resistor values have to be adapted to the actual jfet used. I think idss matching between left and right channel jfets is required to avoid channel imbalance. The missing detail is how to do the adaptation correctly.
The added complexity in post #101 and #118 is due to the use of bjt as input devices that requires current injection at emitter side to maintain bias and the use of current sources as collector load to help stabilize thermal drift.
Quoting Juma, this topology is infinitely tweakable, enough permutations with unique sonic signature to keep interested parties busy for quite awhile.
would it make sense to mix SMD transistors with through hole resistors and caps
If the PCB come with the SMD attached, it greatly simplifies the building process.
In general, however, there is no issue mixing the parts.
tech issue
would it make sense to mix SMD transistors with through hole resistors and caps
Yes, but BF862 is only available in SMD package and that's the reason why it's appeared on PCB. If we can go here with something else in other package I can easily change existing layout.
think I have seen Per Anders do both SMD and through hole options for some selected components, on the same board, placing SMD pads between the legs holes (please do not comment on my use of words
)
This looks practical, thanks Dobrivoje. I'll add pins for such solution to the PCB layout.
ryssen,
if i m not wrong,
yes but not very loud without significant distortion.
you would have to boost the ccs and parallel more bf862.
what i would personally try is to parallel a few complementary followers using k170/j74
or even maybe use that Indra1 schematic and scale down a bit
if i m not wrong,
yes but not very loud without significant distortion.
you would have to boost the ccs and parallel more bf862.
what i would personally try is to parallel a few complementary followers using k170/j74
or even maybe use that Indra1 schematic and scale down a bit
Last edited:
juma,
can you elaborate a bit on how you came about with buffer bf862 Vds value of around 5? i usually seen more than that...
If you mean buffer from post #66 the Vds is not 5V. It's 1.7V and it's been held pretty much constant by means of cascoding.
I chose that value (1.7V @ 10mA) after some experimenting that brought me to conclusion that it is the best sounding working point for BF862 as cascoded source follower.
Another reason for such a low Vds is extremely low gate leakage current which enables the use of Mohm input resistor values without causing the input DC offset.
Circlotron buffer based on BF862 will nicely drive 300 Ohms headphones up to 2V_RMS per phase (about 100mW_peak). For more drive use more JFETs in parallel.
Small multi-turn pot (10R or so) in place of R8 or R9 will adjust the DC offset.
Attachments
