Parasitic stopper: grid or anode?

[James Duncan]

It would be interesting to view the circuit that you used.
You should not be experiancing these problems as there are so many well proven valve circuits and some give as good and at often better results than the solid state circuit
So much information is available on tube circuits and I do not see the need for "circuit simulation" which very often does not provide a working result.
No need to reinvent the wheel here, got a good well proven circuit and go for it.
all the best, great to see a person actually constructing a valve radio.
MM0HDW

[G0HZU_JMR]

If it helps, my initial model of a basic E88CC circuit contained a 47R metal film resistor in the cathode and 6800R at the grid and 3900R at the anode and a 10nF shunt at the anode. The simulator predicted that it requires an anode stopper resistance of 30R to achieve unconditional stability when the Gm is 12.5mA/V. It would probably need a bit more than that when used in your detector circuit because of the way the device is driven by the LO.

My overall model and circuit included inductance for the wires inside the valve, the package inductance of the components and some extra inductance for the wire connections for each. It also had a basic model for the valve base. The most critical connection was the 10nF cap in the anode and this external connection (starting from the valve base pin) had a total of 20nH including the package inductance for the capacitor itself.

The model also predicted that an alternative grid stopper resistance of about 25R would also achieve a K factor >1.

[Alan_G3XAQ]

Thanks, Jeremy. The model does tie in with observation then, given that a small type 43 or 3B1 ferrite bead gives rather more than 50R loss resistance above 50MHz. Certainly, the circuit seems stable now.

73, Alan

[G0HZU_JMR]

Thanks.

I had a bit of fun this evening with the little 12AU7 test jig. I decided to do a slightly more controlled experiment. The idea was to place 100nH of inductance at the anode and terminate this is a shunt 1nF ATC porcelain capacitor. I used a tight tolerance Coilcraft inductor for the 100nH inductance. I also used a 3900R anode resistor to connect the HT. The 10nF cap at the anode is now gone.

The cathode was a 47R metal film resistor with a very tight connection to ground. So little more than the package inductance of the resistor here.

At the input I retained the 6800R shunt grid resistor and modelled some inductance for the wire connections. I then put another 100nH Coilcraft shunt inductor across the 6800R resistor.

I then simulated this as a 1port device looking into the ground end of the 100nH inductor at the grid using the datasheet model for the 12AU7 in Genesys. This predicted about -6R negative resistance and oscillation very close to 151MHz.

I also measured the real circuit using a VNA looking into the ground end of the 100nH inductor at the grid.

The Genesys plot below shows the VNA measurement vs the prediction using the datasheet based model and some basic models of the resistors and inductors that I have connected to it. Up to about 200MHz they both agree quite closely.

The VNA showed -8R and the simulation predicted about -6R negative resistance so a stopper at the grid would have to be at least 10 ohms. The simulation predicts that a 15R anode stopper would also (just) prevent oscillation.

You can see by the spectrum analyser plot that the circuit did indeed oscillate at about 158MHz. Above about 200MHz the model and the VNA started to disagree on the real part of the impedance and I'm not sure if this is an issue to do with the way I have modelled the valve base. But the very simple datasheet model of the valve did a fairly good job of predicting where the circuit would oscillate and what the stopper resistance would need to be to prevent oscillation.

To do the same with your E88CC valve may prove a bit trickier as I suspect I'd have to model the valve and the base with a more detailed model.