diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Tubes / Valves (http://www.diyaudio.com/forums/tubes-valves/)
-   -   6N6P cascode + 6AS7 Cathode follower (http://www.diyaudio.com/forums/tubes-valves/105888-6n6p-cascode-6as7-cathode-follower.html)

croccodillo 26th July 2007 11:01 AM

6N6P cascode + 6AS7 Cathode follower
 
Hi all,

I really like the cascode idea, and I also like the Cathode follower topology.

So, I'd like to try a cascode driver and a CF output stage: THIS is the schematic.

I already tried the 6N6P "normal" cascode, and it sound really fine.

The cascode can deliver a maximum output voltage (peak-peak) of almost 200V (driven with a signal of 12Vpp); unfortunately this is not enough to drive the 6AS7G tubes connected as CF (I need about 300Vpp to have the desidered 200Vpp output to the trafo).

So, I'd like to improve the cascode driver, in manner to increase its output voltage.

So, my idea is as follows:
In the "traditional" cascode the upper triode grid is held at a constant voltage, and thus its plate voltage is only defined by the cathode voltage, set by the lower vale.

What if we change the grid voltage of the upper valve using part of the lower signal?

The upper valve will now have a plate voltage defined not only by its cathode voltage, but also by the grid voltage.

The proposed cascode circuit would (if it ever works) give an output voltage swing much higher than a traditional cascode.

In fact, in a traditional cascode the whole output voltage swing is given by the upper valve, and thus limited by its curve; with my idea, instead, the output swing is the sum of the upper+lower valve swing, so theoretically you can have double the swing.

The gain is however less, about 60 with the given resistors value.

With the given values, the output swing would be not more then 320Vpp, before to reach the limit of the upper valve.

In idle both the valves work at about 160V, with a grid voltage (bias) of about -6V.

What do you think?
Would it work?

Ciao,
Giovanni

316a 26th July 2007 12:29 PM

Has this circuit been built ? It looks like the cathode bias network on each cathode of the 6AS7 is redundant . I don't think a CCS on top of a cascode is a good idea , also for driving a cathode follower you won't need to use the low impedence output of the CCS . For high gain out of a cascode what is required is high gm , a pair of 6S45 would probably do the job a lot better than 6N6

cheers

316a

croccodillo 26th July 2007 12:50 PM

Quote:

Has this circuit been built ?
Not completely.

I built the "traditional" cascode with a CCS on top, it works great.

The output stage has been tried (as common cathode with CCS on top) without any cathode bias network (fixed bias), but a lot of people told me to avoid fixed bias because the 6AS7 valve are not matched at all (even in the same glass), so it is recommended to add some cathode bias to balance the current in the two halves of the valve.

This is what the bias network are intended for: balance the current between paralleled valves.
Actually I don't know if the solution will work, I'll have to try when I will build the prototype.

Quote:

For high gain out of a cascode what is required is high gm , a pair of 6S45 would probably do the job a lot better than 6N6
You are right, but i have a lot of 6N6P available and thus I'd like to use them.

Thanks,
Giovanni

croccodillo 26th July 2007 01:39 PM

Few mistakes found in the schematic.
No decoupling capacitor between driver and output stage, no fixed bias network for the outpout stage.
The parafeed capacitor has been moved down to ground, and substituted with an electrolytic one.
THIS is the revised circuit.

Ciao,
Giovanni

316a 26th July 2007 02:12 PM

Quote:

Originally posted by croccodillo

i have a lot of 6N6P available and thus I'd like to use them.

In that case parallel the 6N6 sections to obtain twice the gm . Why not go up a valve and use 6C41S instead of paralleled 6AS7 sections ?

cheers

Miles Prower 26th July 2007 10:45 PM

Quote:

Originally posted by 316a
I don't think a CCS on top of a cascode is a good idea
Don't see why not. That's how VT o'scopes attained multi-MHz bandwidths for the vertical deflection amps. If there's a problem here, that would be it since you don't really need that much BW in an audio amp, and it's just asking for supersonic oscillation if he decides to include gNFB.

revintage 27th July 2007 09:17 AM

6AS7s are cheap. Why not use a second one as current sink for the first.

You could then remake the output stage to a White-follower to get even lower output resistance and double the output current capacity.

If you do this, it might be good idea to skip to overall feedback and only have local feedback around the cascode.

Shoog 27th July 2007 09:27 AM

Quote:

6AS7s are cheap. Why not use a second one as current sink for the first.
Wouldn't the Mu of 2 make the 6AS7 a very poor current sink ??

Shoog

revintage 27th July 2007 10:01 AM

Shoog:
Yes indeed, but maybe I do not consider the 6AS7 to be the optimal tube overall. Also for CF you would want a tube with a higher gm.

As I considered Crocodillos amp more like a concept-idea I just had some private brain-storming on this theme.

Anyway, I had the White-follower concept in mind. With the 6AS7 we will then half the output Z and double the current capacity. If Ri=280 ohm and mu=2 of each half of the tube we would get Zout=35ohm with the halves in parallell. My guess about the top R of the White is 200-500 ohm depending on load.

Consider this as a theoretical discussion. I do not know how well it will work IRL.

Crocodillo: IS the CCS with low mu output the same as a mu-follower?

croccodillo 27th July 2007 02:13 PM

Quote:

6AS7s are cheap. Why not use a second one as current sink for the first.
Indeed, this was my intention from the beginning.
THIS is the schematic I have in mind.
The CCS "logic" will control the current and driver the 6AS7G grid down (theoretically) to -200V.
The feedback from R19 (load current) is used to vary the CCS current when needed: this is not a my idea, but a nelson Pass Patent (even if I completely redrawn the electronic circuit doing it).

Best regards,
Giovanni Albergoni


All times are GMT. The time now is 07:49 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2