diytube.com

I made some mods to the Poseidon circuit for running specifically PPP output stages. The mod requires a 20mA current source which is available from Mouser. It is part number IXCP20M45. The 6SN7's will run a little hot according to the simulation but the THD distortion figures came out to around .2%. Any comments?

Do you plan on using any NFB?

Dave

Hi,

**edit**
Ooops, nevermind the first post. It is N-channel.

Though I would suggest a totem CCS - MUCH higher dynamic impedance and you can adjust it.

20mA seems a little hot? I find 6mA/ea about right for SN7.

Cheers!

There will be feedback, a network where the 1k res bypassed with a cap is.

If you want to inject the NFB at the SRPP cathode, one method would be to replace R7 with an 820 ohm resistor, bypassed, at the cathode, and in series with an unbypassed 180 ohm resistor to ground, and then inject the NFB at the junction of the 180 and 820 ohm resistors. That preserves the original 1K total R7 value, and allows injection of an NFB signal.

NFB could also be injected at the right hand grid of the LTPI, replace the "short" to ground with a 100 ohm, or less, resistor, and feed it there, though this method could get a bit "squirrely". It does work well with simple PP amps, in which the audio input signal is fed into the grid of the left hand side LTPI triode, and there is no preceding VA gain stage.

FWIW, as an alternative to the fixed valye 20 mA CCS device, you could use an IXCP 10M45S, which is adjustable, and then tweak up the LTPI "tail" current to its optimum value, by changing the current set resistor value, or use a 500 0hm or less) trimpot, to trim it in.

Just a thought, I'm sure that there may be other, and better methods...

/ed B

Hello,

I rediscovered Orcad's PSPICE and just got done simulating my amplifier circuit. It is an SRPP followed by a Differential Phase Splitter. The output stage is a PPP with KT88 tubes. According to the simulator, I get:

Input sensitivity : 2V RMS.

Feedback : About 18dB

Power : Around 127 watts RMS.

THD : Around 0.3% at 1k at full power.

My goal was 120 watts or more of undistorted power. My output iron can handle up to 130W RMS. The KT88s are biased at around 70mA. Now to start building the thing. Parts are on order to prototype the power supply. What do you think about those numbers?

Is your power output number that which is available at the plates, or that which is developed at the secondary into an external load? Not knowing the B+ provided to the output stage under full power conditions, or the loading conditions you are using, it's hard to know how plausible your power output number is. It is certainly possible though, and otherwise, looks great!

Dave

The B+ is 500 volts for the driver and power output tubes. The OPT is 1900 ohms on the primary and the simulation is driving an 8 ohm resistive load.

Here are the schematics:

Driver:



Output:



Bias:

If your 500 volt B+ is actually maintained under full power conditions, and the OPT is capable, then you should get nicely over 150 watts RMS delivered into a secondary load. This also assumes that the screen taps are placed at no greater than 40% of the winding.

The 47 ohm plate resistors you show are unnecessarily large, and more than likely, not really even required at all, as long as the screen stability resistors and control grid stopper resistors you show are in place. If you do elect to keep them, they really should be no greater than 10 ohms to prevent losing any significant power across them.

Dave

150 watts my transformer will not handle, but it will handle 130. THE UL taps are 40%. I'll ditch the 47 ohm plate stoppers for 10's just to be safe. My transformer is a Heybor, they are much faster than Edcor and do not use plastic bobbins. The made them just like they were out of the 60s, without that planned obsolescence feeling to them!

Hey, neighbor, how did this amp work out?

Carl
Rochester

You near RIT by any chance? Perhaps we can meet up and I can show you what I am up to.

Yes, I'm on the south end of town. E-mail: cpultz at Earthlink.net