primary impedance for two 12sn7s - two 6K6s

[andrew]

Hi all, I have a project for a push pull using two 12sn7 power tubes ( each tube is wired in parallel )
My power transformer is 340 volts on plates and my output transformer has a 6.6K primary.
I have one speaker jack at 4 ohm and one at 8 ohm. The speaker is 8 ohm.
I also have two sockets wired for two 6K6 power tubes and I want to compare the sound of the tube types.
What would you consider the best primary impedance for the 12sn7s and the 6K6s.
Thanks for any help, Andrew

[Stevem]

First off how are you planing to get the heater voltage up to 12 .6 volts for the 12sn7s?

[andrew]

I have a 3 amp 12 vt ac to dc power supply I got off eBay for about $8 seem work no problems. thanks for the reply.

[Tony Bones]

The load on triodes is not nearly as critical as with pentodes. By that I mean that there's no danger trying wildly different loads.

In general, with triodes as output tubes, higher resistance loads give lower distortion and lower power... and vice-versa, but don't be afraid to try every combination of speaker and OT connection that you can throw at it. Hear for yourself without fear of anything bursting into flames.

[gingertube]

6K6 want to see 12K Raa ideally. That would also suit 12SN7

Put an 8 Ohm speaker on your 4 Ohm speaker tap (or 16 Ohm speaker on the 8 Ohm tap).

Cheers,
Ian

[andrew]

Thanks guys. That is great advice from and all tubes shall be happy. Andrew

[Phil_S]

I had difficulty with a pair of 6SN7's as output tubes. They didn't seem to work very well when fed by a LTP or paraphase inverter. I think they did not like the voltage swing. The cathodyne seemed to be ideal. Plate voltage in that amp is around 440VDC.I don't remember much else. I built that amp about 10 years ago and haven't played it recently.

[tubeswell]

For centre-bias class A operation, the nominal reflected load can be ballparked by:

Zout = Vp/(Pmax/Vp)

Where:

Vp = plate idle voltage
Pmax = Max plate dissipation

For tubes in parallel, sum the Pmax first.

Using this formula I get

5k3 for 6K6 || 6K6 @ 300V (each 6K6 is 8.5W Pmax)

12k for 1 x 12SN7 @ 300V (both triodes in parallel have a Pmax of 7.5 W)
https://frank.pocnet.net/sheets/093/6/6SN7GTB.pdf

It’s a little bit more nuanced for the 6K6s because they are pentodes, (and you need to keep an eye on Pg2), but if you keep the screens about 50 V below the plates, they’ll be okay with the above reflected load.

[andrew]

Thanks for the screen information. My house voltage varies upon the time of day and I now get 360 on the plates before every AC unit in the city is running full time. Later in the day it will be around 340.
It's been 95 - 99 degrees where I live. I measured the plate voltage at 360 and the choke fed screens are at 356 with 1.5k screen resistors. If it still works I have a VVR I could use just for the screens but could I run a resistor from the screen node to feed both screen resistors ? If so is there a ballpark resistor value that would bring the voltage down 50v ? Thanks again, Andrew

[sluckey]

One fairly easy way to drop the screens by 50V is to use a 50V/5W zener rather than a resistor. Just connect the zener's cathode to the normal screen B+ node then connect the anode to the junction of the screen resistors that feed the tube. I would also put a 20µF filter cap on the anode of the zener but may not be necessary.

[andrew]

I just ordered some zeners. Thanks

[tubeswell]

I measured the plate voltage at 360

6K6 || 6K6 Zout = 360V/(17W/360V) = 7,624R (~8K 10W SE OT is close enough)

12SN7 Zout = 360V/(7.5W/360V) = 17,280R. A little Fender Reverb transformer (17k - 25k) is probably ideal. Or a small 4k:2R SE OT with an 8R speaker, or 8k:4R OT with a 16R speaker.

the choke fed screens are at 356 with 1.5k screen resistors. If it still works I have a VVR I could use just for the screens but could I run a resistor from the screen node to feed both screen resistors ? If so is there a ballpark resistor value that would bring the voltage down 50v ? Thanks again, Andrew

You could probably slug a 5k-10k screen dropper resistor in there. The point is that with a classic centre-biased class A operation for pentodes, an optimum load line often ends up hitting the Vg=0 load line below the knee, and in this region of operation the screen current will tend to increase sharply (when the plate voltage swing is at the nadir part of the signal cycle - pentodes being 'constant-current' devices'n'all). So if you keep the screen current lower - by about 50V or so (compared to the plates) - it will bring the Vg=0 grid curve down to the point where the load line intersects at or above the knee, reducing the risk of over dissipation under heavy signal conditions.

Or you could keep the tube bias cooler - that's another way to do it.