Tube mic design... need some explanation!

[KrIVIUM2323]

I everybody.

I m trying to better understand how tube mic works especially the venerable U47.

I did some search on the forum and read a post which i do not understand:

http://www.groupdiy.com/index.php?topic=9993&highlight=u47

Dale116dot7 wrote:

"On a U47, for example, the input resistors are somewhere around 160 meg. With an 80pF capsule, that sounds like it should really hurt the low frequencies. But on the U47, the very high gain of the tube and the tube's Miller capacitance make this resistance value a non-issue - the equivalent source capacitance is the capsule (around 80pF) in parallel with the Miller capacitance (somewhere between 300 and 2000pF). That capacitance is actually an effective feedback path in a condensor microphone amplifier circuit."

As i did understand Miller capacitance is the inter-capacitance between cathode/plate; grid/cathode; grid/plate.

So i did not understand how this capacitance are an effective feedback path (for the audio signal). Is this acting as the feedback in U67 and U87?

An other question too about the U47: in PSW, Oliver Archut said that one important factor with the vf14 is the cathode voltage (1,1V), which as i did understand too is defined by cathode resistor. I do believe this resistor act for the grid bias so why this voltage is so important?

Maybe something to do with the metal from which the cathode is made of?
And how determine an optimum cathode voltage for other kind of tube ( grid bias is determined by capsule capacitance/voltage applied to it isn't it ?)

Maybe i did'nt understand things right... :?

I've got an other question about capacitors in tube circuit:
are cathode capacitors in G7 and other common grounded cathode design part of the audio signal path?

I've got this question because i'm curently trying to design my first tube mic circuit and plan to use oil/paper cap (i do like the sound of this ones!) for the output of the tube(plate) to the trannie and as Jacob wrote somewhere in the forum, no need for this pricey kind of capacitors if they are associated with low cost parts(electrolytic cap).

When i look at Slowblow mic pramp design it seems that the unbalanced version is running only the output of the tube, not ground as audio output like designs with transformer output (maybe the cathode follower design?)

thank you in advance.

 

[Gus]

look carefully at a u47 schematic there is no seperate fil power supply. The fil is supplied from the same B+ via a dropping resistor to the fil to the cathode.

The cathode is a 29 ohm cathode bias with 29ohms?

Think ohms law current = voltage /resistance

current = 1.1V/29ohm

a lot more current going thur the resistor not from the tube plate current

what does this mean? it is more a fixed bias gain stage that does not shift with signal level.

Very clever design IMO.

 

[sismofyt]

have you ever pondered about that Neumann might have chosen the VF14 for it's high heater voltage and not sound as such? it makes the whole deal a lot chaper to make. wasn't the VF14 even scarce back the in the 40's?

 

[dale116dot7]

The whole Miller thing can be confusing at first. The first thing to remember is that the signal source is a capacitor. You can (provided you provide a different way to DC-stabilize the circuit) use capacitors in place of resistors in a feedback circuit. They have some sort of AC impedance (1/(2*pi*f*C)) and this actually works, provided that the amplifiers can drive the capacitors, and that the frequencies are kept high enough. Fortunately, since the signal source is a capacitor, and the impedances are very high, it all works out ok. But you can't do this with, say, a mic preamp.

You can look at a non-inverting op-amp circuit, replace the two resistors with capacitors, and you basically have that configuration in a tube mic. The source capacitance is whatever the capsule capacitance is (in the case of a K47 Neumann capsule, that capacitance is about 80pF). The feedback capacitance is whatever you have as a capacitance from the plate to the grid. It is typically going to be between 0.5 and 10pF. Remember that the actual resistance values at a given frequency will be inversely proportional to the capacitances.

One other thing - a single tube or transistor does not have an infinite gain (as an op-amp is often approximated as), so if you want to analyze this you need to use the non-ideal op-amp gain formulas.

I ran a simulation using a BF245C JFET (which will behave similarly to a vacuum tube as far as this exercise goes), and here are the results:

Signal source is 10mV, load resistor was 47k, no external loads, bias somewhere around 10V on the drain with a 24 volt supply.

0.1uF source capacitance, no feedback capacitance, output = 1200mV
80pF source capacitance, no added feedback capacitance, output = 600mV
80pF source capacitance, 1pF added feedback C = 360mV
2pF added feedback = 235mV
5pF added feedback = 125mV
10pF added feedback = 75mV

Note that the frequence response of the output level on each run was pretty much flat (within a fraction of a dB) from about 50 Hz to over 1 MHz. You can also see that the SPICE model of the BF245C includes its capacitance, which is somewhere around 1pF.

Basically what the Miller effect states it that the effective input capacitance includes the capacitance from grid to plate multiplied by the gain of the tube. You can look at it either as an input capacitance or as a feedback circuit, whichever is easier to understand. Given that adding only 10pF of feedback capacitance cut the signal down by almost an order of magnitude, you can see that this is a very important parameter. In a tube mic, I find it easier to look at this as a feedback charge amplifier, whereas in a tube mic preamp (with a resistive source), looking at it as an effective capacitance (where it only affects the frequency response) is easier for me.

It also shows that no matter how much people insist they have no negative feedback in a tube mic, or how they hate the sound of negative feedback, or whatever - in a tube mic you basically cannot get rid of it. Heck, in a solid-state mic it's still there. Well, you can get rid of it but the circuitry is not typically what you see in a tube mic.

I like the U47 bias circuit. No big electrolytic needed, no electrolytic or ceramic capacitors in the signal path is good, right? Especially if we're talking about 1930's to 1950's electrolytics. I've used that system with a 6AH6 tube which I like the sound of in a tube mic. The only thing is the filament current is something like 450mA so that resistor ends up being somewhere around 3 ohms.

Hopefully that helps a bit.

-Dale