6as7 Pse

[croccodillo]

Hi all,

I got a good deal, and bought 10 russian 6N6P + 8 6AS7G at about 30USD.
Now I'm waiting for the tubes, they should arrive next week.
I'm now thinking to realise a parallel SE amplifier.

THIS is the preliminary schematic.
It uses a single 6N6P connected as LTP as driver, and two 6AS7G as output tubes.
I have few questions for you:

1. Do the 6N6P have enough voltage swing to drive the 6AS7G tube (I would have about 120Vpp at 6AS7 grid)?

2. What about feedback: can I remove C4 and have no capacitor on feedback path?

3. I read on the forum and on the internet that 6AS7G tubes as usually unbalanced, so do I need any sort of servo circuit to keep the current in the paralleled sections the same?
I could use a current mirror in the cathode of each 6AS7G halves, both on top and on bottom, but I don't know how much would this solution interfere with sound.

4. What about fixed bias? Both the power supply (+300V) and the bias rail (-120V) are mosfet regulated, so cold I use fixed bias and expect to do not have bvoltage/current drift on the output valve?

5. Any oteher observation/suggestion on the schematic?

The CCS in the output stage is an Active Current Source, the circuit Nelson Pass uses in his amplifiers, adapted to tubes (in the previous prototype, built with KT88, it worked fine).

The calculated output power, with ACS, is about 16Wrms per channel.

What do you think?

Ciao,
Giovanni

[tubetvr]

Hi,

Some answers

1: You probably need more than 300V for th driver in order to get 120V Pk-Pk. Also as the gain of the 6N6P in differential mode is very low, something like 70% of µ/2 or about 7 you would need a very high input level, probably exceeding the grid bias, so no it doesn't work, you need one more gain stage and a driver circuit being able to give higher voltage. I checked some more, the anode voltage is 140V and the grid bias is therefore about 4.5V, so with a gain of 7 you can get maximum 63V Pk-Pk at clipping

2: Yes, you can remove C4 and C2, (and R1 and R3), they are not necessary as the grids are on ground potential anyway, I get the general feeling that the circuit is designed by someone with experience of SS but no experience of tubes, (he has assumed that the grid has potential towards ground which they have not as no grid current flow).

3: paralled tubes always need some mean of balancing, best is individual resistors in series with the cathode, drop at least 10V over these.

4: There is fixed bias on the lower tubes so they will drift without compensation but as there is a CCS involved the current will not drift, the result is that the voltage at the anodes of the lower tubes will drift but it will probably be alright.

5: The grid bias control for the output tubes doesn't work as no grid current flow nothing will happen when the pot is adjusted.

Find a better circuit that has been tested and is well designed, there are some obvious strange things in this one so I doubt that it exist other than on paper. Why do you want to have a Beta Follower output stage??? why not an ordinary PSE or Push-Pull?

Regards Hans

[Klimon]

3: paralled tubes always need some mean of balancing, best is individual resistors in series with the cathode, drop at least 10V over these.

Is it worthwile to implement this in an output stage with parallel output tubes sharing one cathode resistor; e.g. in my case 3.3k shared cathode resistor + separate 500ohms resistors in series with each cathode? Naturally this would also facilitate measuring current draw of each of the tube-pair. I also assume that the 500ohms resistors will not need cap bypass, considering their low resistance compared to the large (and bypassed) 3.3k shared resistor?

Some more info on parallel output tubes and balancing problems (and consequences):

http://www.stereophile.com/interviews/196rankin/ (subjective, no info if he used shared cathode resistors or any kind of balancing device)

http://digilander.libero.it/paeng/on...E_Mismatch.htm

Thanks - Simon

[tubetvr]

Is it worthwile to implement this in an output stage with parallel output tubes sharing one cathode resistor; e.g. in my case 3.3k shared cathode resistor + separate 500ohms resistors in series with each cathode?

The main reason for introducing siome mean of balancing is to avoid current hogging, i.e. one tube draw much more current than the other ultimately destroying itself as current or anode dissipation will be exceeded. Separate cathode resistors is one simple way to avoid this as any tube that have a tendency to draw more current will get more negative grid bias and this will counteract the increased current.

The larger the cathode resistors are the better but iot also have the effect of increasing output impedance and reduce gain so some of compromise is needed, a simple rule could be that you should drop at least 10% of the grid bias voltage over the unbypassed resistors.

Regards Hans

[Klimon]

Thanks Hans!

Very fun website btw (finally I can stop going to Ikea and make my own Swedish meatballs )

Regards -- Simon