> my dealer dont have this tube..
He probably has something better... the Germans and neighbors were making good tubes after the US lost interest.
I've advocated the Television Vertical Sweep Deflection tubes. The oldest ones were twin medium/low-Mu triodes like a heavy-duty 12AU7, but the last big tube TV sets used a high-Mu triode with a small power triode or pentode. 6EM7 is the biggest of the US breed with triode output. I know there is a smaller Euro tube that has become faddish for small hi-fi speaker amps. That's pentode output but it can be strapped triode. And the TV V-sweep tubes are often very cheap: huge numbers were made, replacement market was once strong, and all those TV sets are now gone.
> is there some better, perhaps with transformer models out there?
I'm an old fart and I think SPICE is a waste of time unless you can sketch-out and estimate your design on the back of a cocktail napkin. SPICE can do the math to high precision, but it will never tell you that R16 R9 are a waste of several pennies, and it won't tell you the noise disadvantage of R12 unless you ask very carefully.
I suspect SPICE itself does not know what a transformer is. There are several ways to make one, and some vendors may have pre-made transformer models.
What I've always done in Pspice is:
Put in a coil. I usually know the impedance, and the lowest frequency it needs to keep impedance. From that I know what inductance I need. For 200Ω source and 20Hz, about 2 Henries. If you are not modeling a specific transformer, an approximate value is OK. Don't get silly, like 1,000H, because there may be places like C9 where an unrealistic inductance will give unrealistic results.
Put in another coil for the secondary. Assuming good coupling (always true in audio) its inductance is higher than the primary inductance by the impedance ratio or the square of the turns ratio. For 2H primary and 1:4 turns, the secondary will be 32H.
Then you need to couple the two coils together. In Pspice, we have a K_linear. This takes the name of 2 to 4 inductors, and a "COUPLING" value. In audio terms, L*(1-Coupling) is your series leakage inductance, and 1/(1-Coupling) is in many cases your relative bandwidth (Coupling=0.999 on 2H gives 2mH leakage inductance and probably roughly 1:1,000 or 20Hz-20KHz bandwidth). The coupling of an ideal audio transformer is 1.0, but that is physically impossible and SPICE will barf. Realistic values for coupling run from 0.99 for split-bobbin power transformer to over 0.9999 for a highly interwound audio transformer. If not modeling a specific part, use 0.9999 and remember that your real iron probably won't be as good.
This model neglects winding resistance, core loss, and parasitic capacitance. Resistances can be added, in fact classic SPICE may barf on a zero-ohm coil. But in most audio the resistance losses are small. Not negligible, but small. I often neglect them, and round-down the results. For most audio iron-core inductors giving less than 1K impedance, parasitic capacitance is negligible over the range of human hearing, but over 10K it dominates a design just trying to hit 20KHz, and even at 1K impedance you may need to add capacitance to show supersonic effects that can be heard in the audio band.
Actually, my copy of Pspice has a part "XFRM_LINEAR" that is two coils and a Coupling value, plus a pre-made graphic. So it is equivalent to the above process except limited to two coils.
There is also "XFRM_NONLINEAR", which takes the same parameters but must be different. Since I don't see where to specify the nonlinearity, and hope my audio iron is always linear, I don't use it. Obviously a switching regulator simulator must be rich with non-linear iron, since switchers usually run near to or past the knee of the iron or ferrite.
