Re: Where is Graham's mixer design? (was Re: RAT Tube Mixer update?)

The Phantom wrote:
> 
> On Tue, 04 Mar 2008 19:41:35 GMT, John Byrns <byrnsj@[EMAIL PROTECTED]
>
wrote:
> 
> >In article <fqf4uh$tam$1@[EMAIL PROTECTED]
>,
> > Ian Thompson-Bell <ruffrecords@[EMAIL PROTECTED]
> wrote:
> >
> >> Eeyore wrote:
> >> >
> >> > Ian Thompson-Bell wrote:
> >> >
> >> >> Has anyone thought of using a grounded grid amplifier as a mix
amp?
> >> >
> >> > No point as far as I can see.
> >> >
> >> > Graham
> >> >
> >>
> >> I can think of several.
> >>
> >> 1. Only needs one triode
> >> 2. Input/output transfer characteristic is very linear when used with
> >> current source inputs
> >> 3. No Miller effect.
> >> 4. Input impedance approx 1/gm
> >
> >Are you sure about that?  By my calculation that approximation only
> >holds as long as Ra is small compared with the load resistor.  Perhaps
I
> >made a mistake in my hand done calculations and should try to look the
> >result up in the RDH4th?
> 
> I calculate input impedance as:
> 
>       RL + ra
>      ---------
>      1 + gm*ra


???? maybe this is right.

But in a grounded grid amp, suppose you *want* to make the Rin maximal.

Then use a CCS for the cathode dc supply.

The anode load produces the output required, and the Rout = the load RL
because the Ra is effectively high if the cathode load is high.

The gain = µ x RL / ( Ra + RL )

Suppose you have RL = 32k and gain = 16, then when 16V of output
appears, there is
0.5mA of current change.
if there is 1V between grid and cathode to cause this change then 
Rin = 1V / 0.5mA = 2k.

If there was a cathode load RL then this would be in parallel to the
Rin.


To drive 2k, you need a low source resistance maybe say 500 ohms.

2k is a rather LOW value for Rin and it should be at least 10k.

> 
> which would indicate that ra should be large compared with the load
> resistor for the approximation 1/gm to be good.
> 
> >
> >> 5. Gain approx anode load resistor over input resistor

?????

> >
> >Doesn't that gain approximation assume that Ra is low compared with the
> >value of the load resistor?
> 
> I calculate gain as:
> 
>           (1 + gm*ra)*RL
>     -------------------------
>     (1 + gm*ra)*Rin + ra + RL
> 
> which means that for the given approximation to be good, ra + RL must be
> small compared with (1 + gm*ra)*Rin.
> 
> >
> >There seems to be a serious compromise here with respect to achieving a
> >low input resistance while maintaining the gain close to unity!

Adding series R btween source and cathode can tailor anode gain
downwards from 
max above of 16 to anything you want, but what's the advantage?

If gain < unity is desired, then is not a voltage divider feeding
cathode follower better?

IE, a volume pot with CF after it?

Then  you keep Rout low, while Rin is the volume pot high resistance.

AFAIK, the ONLY beneit of grounded grid and cathode input is for phono
where a low loading resistance is OK.



But classic circuits used a triode CF to drive a cathode of a second
gain triode
with its grid grounded and with CCS common cathode dc supply.

This gives very high Rin, and naturally buffers the source from the
following amp and output current.

Some who used this circuit also applied NFB from 2nd triode anode output
to the same tube's grid
instead of grounding it.

It works well but only for small signals.

Patrick Turner. 
 
> As one might expect, to get this result gm*ra should be large.
> 
> >
> >> Disadvantages are:
> >>
> >> 1.Heater cathode leakage can inject hum/noise into input path.
> >
> >Is this a serious problem assuming a DC heater supply is used, and the
> >heater bus is well filtered/isolated to prevent the entry of other
noise
> >sources?
> >
> >
> >Regards,
> >
> >John Byrns