New GU50 SE: SRRP or Cascode for driver?

[Globulator]

Hi,

I'm looking at a new SE design (a Sweet Peach rebuild) with the GU50 output tube in pentode mode at 430V, with a 6N6p driver tube at B2+ = 400V and a 6N3p input tube at B3+ = 350V. The concept is feedback only around driver+output to present a low impedance to the OPT.

So far so simple. So I can use a whole tube or a half tube for each section. I could use a CCS on the anode but I would have to buy those - whereas I have tubes here right in my hands...

The input stage: Needs to be simple, linear, lowish gain.
The driver stage: Needs to be linearish, high gain, good voltage swing
The output stage: Pentode mode, Anode 430V, 55mA idle, Screen at 250V?

Feedback will be from the GU50 output tube (anode or cathode) to the driver tube cathode or grid - ideally without a capacitor but probably will need one.

The input and driver stage can be GC ('grounded' (unbypassed) cathode, SRPP or Cascode. The SRPP and Cascode can be SRPP+ or UL Cascode courtesy of TubeCad's papers.

My idea is to use 7mA on the input 6N3p and 15mA on the driver 6N6p to get them linear.

Plan1: Input CG, driver SRPP+, feedback from GU50 anode to driver cathode.

Er.. I only have one plan at the moment

So could people comment on the choice of SRPP vs Cascode please - I'm a bit lost as to the practical difference between them!

[Miles Prower]

Originally Posted by Globulator

So could people comment on the choice of SRPP vs Cascode please - I'm a bit lost as to the practical difference between them!

The SRPP has a disadvantage of being balanced for just three load conditions: dead short, unloaded, and the one specific impedance for which it was designed. For every other load it goes out of balance and distortion rises with that imbalance. It really doesn't make a very good grid driver since grids going positive don't represent a consistent load.

You're probably better off using a CCS plate load instead. If you need the buffering, just use the other half of the dual triode as a cathode follower.

Cascodes have two possible disadvantages: limited output swing as compared to the Vpp supply. The cascode also has an unusually high effective r(p) (more in the territory of a small signal pentode) that may not play well with a load that isn't Hi-Z and Lo-C.

I did a project that used 6BQ7s to implement a cascoded LTP, and that worked out great: enough gain from a single phase splitter / gain stage, and the sonics were excellent.

[Wavebourn]

Originally Posted by Globulator

Feedback will be from the GU50 output tube (anode or cathode) to the driver tube cathode or grid - ideally without a capacitor but probably will need one.

A huge difference Feedback by current (from cathode) would increase output impedance of your amp. Feedback by voltage (from anode) would decrease it.
Some people argue that speakers sound cleaner driven by high output resistance, though. But others argue that frequency response is better when driven by low resistance.

I would try both.

[Globulator]

OK so I'm not going to implement a CCS right away (because of having to buy the bits), but I suppose I could build first with a simple GC and anode resistor - then upgrade later with a CCS. That would be easy too

I definitely want a very low output impedance driving the OPT, not because I'm convinced it's better, but because I want to hear what the effect is

I'm still a bit confused on a few things though:

a) I thought all pentodes are easy to drive - is this not the case?
b) So what use is a SRPP ?
c) What is a Cascode useful for (besides driving a Schade topology tube!)

Thanks for all the help!

[Arnulf]

I'm following this thread with great anticipation - I have GU-50s ready, my plan was 6N6P as amplifier/driver (common cathode) and either 6N1P or 6N2P as voltage amplifier (common cathode), one section per channel, feedback routed from the secondary to first stage.

As for your questions:

A: they are way easier to drive than triodes due to negligible Cmiller. Where does one draw the line between "easy" and "not easy" ?

B: line driver (think: transoceanic telco cables - fixed Z load, power gain = good due to cable losses)

C: low Cin at high f (think: UHF circuits before conception of multi-grid tubes); first stage disposes of Cmiller, because it is working into fixed potential (current gain), second stage provides voltage gain with current shoved into its bottom.

[danzup]

As I respect all other opinion , for the last 3 tube amplifier I am using SRPP as a driver .
For Gu50 in SE I am using SRPP with 6n2p and it work exceptional well for me .
FYI : I do not use feedback at all in SE , NEVER .
Why do not make some experiment ( as in trying all the circuit you mentioned ) and decide what is best for you ?
I have done such test for driver :
1.simple triode
2.pentode
3.triode with choke on anode
4.SRPP
5.Cascode
6. cathode repeater with triode
and the result was that number one for my ears ( best ) : 4.SRPP !!!!!!!
second place was : 6. cathode repeater with triode
last and the most crappy one : 3.triode with choke on anode

[Globulator]

Wow danzup, that was a comprehensive test!!

Intuitively I think the SRPP should work the best - I was puzzled why so few used it. My reasons why I thought it should work are:

a) It suits swinging the voltage well. For an SE you need a lot of drive voltage, and I suspect SRPP would be trying less hard than anything else - no straining against anything at the extents.

b) Output drive. It's used for simple output stages - so it has a lot of power capability. I know the grid of the output tube can be a lot to drive - but so are OPTs with speakers on them. With a reasonable grid stopper on the GU50 the drive should be pretty constant impedance I think - plus a pentode is pretty easy to drive. Sp a 6N6p SRPP idling at 15mA should simply not notice the grid load of a GU50.

So now (as I've now heard the answer I wanted ) I need to decide if I just use the time GC input 6N3p tube in a degenerated anode resistor way to drive this SRPP, or something else.

How easy is an SRPP to drive?

[danzup]

Originally Posted by Globulator

Wow danzup, that was a comprehensive test!!

Intuitively I think the SRPP should work the best - I was puzzled why so few used it. My reasons why I thought it should work are:

a) It suits swinging the voltage well. For an SE you need a lot of drive voltage, and I suspect SRPP would be trying less hard than anything else - no straining against anything at the extents.

b) Output drive. It's used for simple output stages - so it has a lot of power capability. I know the grid of the output tube can be a lot to drive - but so are OPTs with speakers on them. With a reasonable grid stopper on the GU50 the drive should be pretty constant impedance I think - plus a pentode is pretty easy to drive. Sp a 6N6p SRPP idling at 15mA should simply not notice the grid load of a GU50.

So now (as I've now heard the answer I wanted ) I need to decide if I just use the time GC input 6N3p tube in a degenerated anode resistor way to drive this SRPP, or something else.

How easy is an SRPP to drive?

Easy , I use something like this :
CD Player -->pasive R2R Volume --> ( SRPP --> Gu50 ) -->Speakers
On my "home" Romanian forum we build SE with Gu50 in penthode mode and drives by SRPP with following tubes : 6n6p , 6n2p , 6n3p , 6n1p and 6h13s .
For me 6n6p ( 16mA ) and 6n2p ( 8mA ) was the best !

[Globulator]

Originally Posted by danzup

Easy , I use something like this :
CD Player -->pasive R2R Volume --> ( SRPP --> Gu50 ) -->Speakers
On my "home" Romanian forum we build SE with Gu50 in pentode mode and drives by SRPP with following tubes : 6n6p , 6n2p , 6n3p , 6n1p and 6h13s .
For me 6n6p ( 16mA ) and 6n2p ( 8mA ) was the best !

That's an interesting idea - avoids a stage completely too.
What voltage do you run the GU50 and the SRPP from?
What screen grid voltage do you use for the GU50?

[danzup]

Originally Posted by Globulator

That's an interesting idea - avoids a stage completely too.
What voltage do you run the GU50 and the SRPP from?
What screen grid voltage do you use for the GU50?

+B = 400V at the Gu50 then a 4K7/3W resistor from +B to SRPP anodic with 470uF/450V and 0.1uF Wima MKP in parallel
For G2 = 250V , with a resistor and 250V zenner to ground and MKP capacitor in parallel with zenner .
( the final GU50 stage is like this Verstärker mit GU 50 in Klasse A–Betrieb )