I have been building and testing a single stage amplifier for a condenser microphone. Everything in my Qspice file has been matching my (real world) results I have been getting with the multimeter almost perfectly. However, there is one exception. It has to do with the grid leak resistor. Here is a photo of my circuit…
When I change R1 (in real life) from 100Meg to 250Meg, my anode voltage drops from 60V to 41V. Also, cathode voltage goes down to 1 volt. If I change R1 to 250Meg in Qspice, it does not have any affect on the DC bias points at all.
Is this just a product of the model being incomplete? Or, is this a limitation of spice coding in general?
Here is a version that can probe into its model subcircuit.
Well… I have never used a tube and only based on general knowledge for this one.
In DC analysis, V(G) in theory is 0V regardless of the value of R1 if there is no other leakage path. With this simulation model, assuming leakage resistance is added between A and G, V(G) will have voltage and can affect the bias condition.
But with this leakage added, in this model, R1 needs to go down to 1Meg to get the V(A) ~ 60V back. Anyway, maybe the real device with its other nonlinearities and parasitic has to be counted. Let’s see if this set of simulations can give you more insight.
Thank you very much for the work you did here. It’s gonna take me some time to apply this to my situation. For anyone else in the future that may stumble upon this problem. I found a post on another forum from an engineer that sheds some light on this. They say that as long as the tube is running at really low current (less than 1mA), the value of the grid leak resistor doesn’t really affect bias operation. However, they do say the more the current, the more change is made by the grid leak resistor.
Here is the actual quote –
There is a maximum value that you should not go above, because it can lead to unreliable or drifting bias. It is not always stated for preamp valves, which is why we default to the safe, comfy value of 1Meg. But up to 10Meg is usually OK for valves running at low currents (<5mA).
Blockquote
From this I can only assume that the spice models for valves are rather incomplete. Because the people that code these, aren’t taking this into account.
