Complementary diff input with JFET and BJT cascode

[Bensen]

Please, give me some remarqs off this circuit.

I've always used BJT's but many off you made me curious about using JFET's.

Jfet's should be cascoded to minimize the voltage differences between drain and source to keep the input capacitance relative low.
But I've seen many circuit's where the Jfet is cascoded with another Jfet. This way you'll have about 0.7V as VDS for the Jfet. When I see in the datasheets, I will have then pretty hich input capacitance. So I wanted to cascode the Jfet with a BJT, this way I could have a VDS for the Jfet off about 3.5 to 4 volts when using 2 LED's.

What do you guys think off this differential input circuit? Or should I just use two Jfet's instead of Jfet + BJT.

Greetz
Ben

[Bensen]

I've got the idea off the SERVO 50 amplifier off Borbely.
And Mr. Curl has me convinced to use complemantary diff input instead of single diff, while he is using this in his Halo JC1 amp.

http://www.borbelyaudio.com/special_articles.asp

[LineSource]

Hi Benson,

There are several good threads on cascodes and folded cascodes. The thread below cites several good papers and provides some data comparing FET-FET, FET-Bipolar, and Bipolar-Bipolar cascodes. There are also threads on "modulated cascodes" using FET-FET topologies that you might find interesting. I've been using a JFET-Bipolar folded cascode topology with good results in my power amps.

Cascodes- the truth is out there...

[Bensen]

LineSource,

Yes, I had read that thread already, but for a non EE is it pretty hard to understand sometimes. But in post#18, Forr is explaining a circuit that for me looks like the one I've drawn.

Is in my circuit the CCS for the diff pair exually needed, I use a regulated power supply for the diff pair. And can I than connect the four resistors with each other? ...Oh wait, than I have the Borbely amp
This should make things a bit more easier.

I used two LED's in series for biasing the cascode BJT, is a simple resistor or zener maybe better? Noise? Osscilation?

Thanks
Ben

[amplifierguru]

Hi Benson,

One of the good things about the FET/BJT cascode as in your pic is that it doesn't need the CCs's for biassing and a simple R tail between the two halves can set the bias.

The best CCS is no CCS. You still need to bias the cascode though.


Cheers,
Greg

[AndrewT]

Hi Bensen,
I too have read Borbely and agree that the logic for connecting the cascode to the LTP tail is well founded.
I think you have made the correct decision to cascode this way rather than using ground for the cascode reference.

The two current sources (& two sinks) do seem a little complicated and maybe Guru is right to remove the tail sink/source and just keep the cascode source. But with a regulated PSU even a resistor here should be adequate to feed the two LED string.

However, keep the PCB as you have drawn and compare the four versions;- all ccs, all resistor, ccs to cascode and resistor to LTP, ccs to LTP and resistor to cascode. Then you can publish the sound quality effect of your experiments. I will be waiting.

ps I will send back the latest version of SOAR.

[amplifierguru]

Hi Benson,

There could be an opportunity to bias the cascode from a modulated driver as in my latest design which not only eliminates the need for the CCs but benefits from a 3 times reduction in THD - why wouldn't you?

Cheers,
Greg

[Bensen]

Hi amplifierguru ,

I've did yesterday a search for "modulated cascode" but found nothing interesting. Have you maybe a link to some sort of circuit where did is used.
I remerber that you' ve writin some where that you used the current throught the VAS also for biasing the diff pair.

Thanks for the reply.
Ben

[Bensen]

Hi AndrewT,

You mean by referencing to ground, something like in the attached pic I presume.
But, this gives me one very big advantage. Namely, this way I can supply the diff input pair with a regulated PS +-75V , without destroying the JFET's due to overvoltage, and have a large enouph output swing for the second stage. Otherwise I have to use some kind of circuit posted in the next post.

But this cascode give me more voltage variations over the Jfet --> input capacitance raises.

[Bensen]

This is the topology I should use when I would supply the diff pair with only 15volts. Don't mind the resistor values....

I think always that: the less stages there are in an amp the better.