Is this OK?

[lumanauw]

I'm building an input Jfet differential pair using K30 Jfet.
But the voltage rail are +/-48V, while the rating for 2SK30 is only 50V. The value will be too close to the max Jfet voltage rating, I'm afraid.

Usually people cascode Jfet (with small bipolars) to work only about 15V.

To reduce the voltage seen by the Jfet, is it OK to put 15V zener on the drains of the Jfets before the current mirror? This way the Jfet will only stressed by voltage about 30V (far enough from max=50).

What is the pro and cons using this method?

[sparkybg]

The zeners are noisy. You might bypas them with capacitors, say 100nF. Using two resistors, bypased with capacitors will cause nearly the same result. You must calculate the values for the resistors for half the current if differential stage in order to obtain the preffered voltage drop.

It might work, but cascode with two bipolars or another two jfets is preferable for me.

But I may be wrong. Let others say their opinions furst.

[darkfenriz]

I think it is OK.
Another member (mr. Evil ?) used several leds paralleled with small caps as far as I remember to get the same result.

[AndrewT]

Hi,
do you realise that you can reduce the Tail Voltage considerably?

using +48V and -10V will work just as well.
The tail voltage can be reduced by resistor, zener//cap feeding the CCS.

As for reducing the drain voltage, why do you want to avoid the cascode? It works well with either BJT or FET cascodes.

You can cascode high using 30V Zener reference to source or to Ground.

Or

Low voltage reference using two series connected LEDs giving about 3.5 to 4V. Again to source or to ground.

I do not know how the drain Zeners will perform in your proposal, but I suspect the Zener noise will be an issue. Forward biased diodes might be quieter.

[darkfenriz]

Andrew
why -10V?
Noise? NO, not here.

[AndrewT]

Hi Darko,
why -10V, just a number a lot less than -48V. Leaves lots of room for really big RC Time Constant to remove all -ve rail modulation effects.

Are you saying that Zeners in the drain supplies do not cause a noise problem for the LTP amp stage?

[darkfenriz]

Ah, I see, make CCS as clean as possible, yes? This is a good practice.

Concerning noise I think it isn't an issue on drain, because output impedance charcteristic of jfet drain. A change of drain voltage does very little change in drain current. Yes, worse than bjt collector output impedance, but still not very bad I think.

regards
Adam

[lumanauw]

Hi, Sparky,

The zeners are noisy. You might bypas them with capacitors, say 100nF. Using two resistors, bypased with capacitors will cause nearly the same result. You must calculate the values for the resistors for half the current if differential stage in order to obtain the preffered voltage drop.

Will the zener noise enter the differential system?

I try not to use resistor (in the place of Z15V), because resistor there will modulate drain voltage of the K30 (next=Vds of the K30 will be modulated), following V=IxR, I need something fixed voltage drop, and I think zener is suitable. But I don't know about that noise characteristic for zener in that position.

It might work, but cascode with two bipolars or another two jfets is preferable for me.

Cascoding with Jfet might be OK. I might be over-concerned here, but if the cascoding(s) are bipolars, I think there are additional current that is entering/leaving the differential system (yes, Ib is very small, but it still isn't music signal ).
At the attachment below, there will be current leaving base Q3-Q4, and it is not music signal.

At first, I tought that if I use only 2 legs device (like zener), all currents will be there between Anode-Cathode, none is leaving or entering the differential system, not disturbing the music itself.

Hi, Adam,

Yes, I remember Mr. Evil do that. Thanks for the hint

Hi, AndrewT,

It is the Vd-s of the K30 that I want to reduce. The rail is +/-48V, it makes Vd-s of the K30 will be about 48V, too near to the max voltage rating of the K30.
The tail ccs is formed with bipolars, I can use high voltage bipolars, so no problem there. It is the K30 that makes problem for me now

Ah, I see, make CCS as clean as possible, yes? This is a good practice.
Concerning noise I think it isn't an issue on drain, because output impedance charcteristic of jfet drain. A change of drain voltage does very little change in drain current. Yes, worse than bjt collector output impedance, but still not very bad I think

Really? I shouldn't worry about somekind of "Early Effect" (modulated Vd-s modulate gm of the Jfet)? Can I use resistor in the place of Z15?

[myhrrhleine]

Personally, I use cascoding for the linarising effect.
reduced Voltage is just a convenient, added benefit.

[darkfenriz]

Originally posted by lumanauw

Really? I shouldn't worry about somekind of "Early Effect" (modulated Vd-s modulate gm of the Jfet)? Can I use resistor in the place of Z15?

You SHOULD worry, but I meant zener noise (of low amplitude) won't introduce significant Early effect. Reststors in place of zener MIGHT introduce significant Early effect, stay with zeners.

regards