Value of Rk on a AC-coupled Cathode Follower?

[SnickSound]

I just installed an FX loop in one of my amps, works like a charm! The buffer stage is fixed-bias AC-coupled cathode follower, very similar to what Mesa Boogie does, or many MOSFET kits.
Recovery stage is just a typical gain stage: 100k Rp, 2.2k Rk with 33uF bypass cap, 47nF coupling to Master Volume.

Now, I was looking at the Valve Wizard article on AC cathode followers: http://www.valvewizard.co.uk/accf.html
In there he says to use the smallest Rk that you can get away with to keep the output impedance low, in his example he's using an 18k. But 3 things:
- I'm only using pedals in this loop, which all have an input impedance of at least 470k
- I'm using two resistors in series for Rk and taking the output at the junction: 56k to put, then 1.5k to ground (approx 34dB of attenuation), so combined Rk is 57.5k
- He's using an ECC82, I'm using an ECC83 (to maximize recovery gain)

In this scenario, does it matters that I'm using a significantly higher value than he is? 57.5k is still much lower than 470k (the idea being that it has to be significantly lower than the input impedance it will be driving) and everything works fine. Would I gain anything by going lower? The ECC82 in his design can handle 2.25W while an ECC83 maxes at 1W (but it idles at 0.27W with the 56k+1.5k). Since I'm taking the output at the other end of the 56k, I'm unsure how that affects output impedance, I'm guessing it's higher already.

To make things more confusing, Mesa and other big companies with a similar loop seem to be using a 100k Rk, however they take the output from the cathode and then use a voltage divider (Send control) after the coupling cap to attenuate it. This does give the option of feeding line level devices, but I couldn't care less about that, my fixed 34dB attenuation is right on the mark. Turning "Rk" into a voltage divider seems to be common on MOSFET loops.

Maybe I'm overthinking this: it works, it sounds good, tube is running within specs.

[sluckey]

Sounds like you are right on the mark.

[2deaf]

In this scenario, does it matters that I'm using a significantly higher value than he is? 57.5k is still much lower than 470k (the idea being that it has to be significantly lower than the input impedance it will be driving) and everything works fine. Would I gain anything by going lower?

No and no.

What Merlin is doing and what you are doing are two different things.  Your cathode follower is swinging a large signal that you are only using a little bit of at the bottom of a voltage divider.  You would want to swing the largest signal possible for maximum headroom, which means using a large value R_k.  But you also want to keep the smaller resistor in the voltage divider as small as possible for a low output impedance, so that puts a limit on the larger resistor.

You want a low output impedance to drive the capacitive load presented by the cable used that is going from the effects send to the device.  The input impedance of your effect device isn't a design consideration with your cathode follower.  No matter what you connect to the effects send, the cathode follower sees 56K to 57.5K at the extremes.   

[tubeswell]

Any load between 47k to 100k would be fine for a 12AX7 CF stage

[Merlin]

What Merlin is doing and what you are doing are two different things. 

Exactly. Do you even need a cathode follower? Maybe the tube could be put to more interesting use..?

[SnickSound]

What Merlin is doing and what you are doing are two different things. 

Exactly. Do you even need a cathode follower? Maybe the tube could be put to more interesting use..?

I tried to do without (it's pretty much a Plexi and the loop is between the Treble and Master Volume, so technically already driven by a cathode follower), but it wasn't a success. Unpredictable input impedance of the following circuit due to non-buffered pedals made the tone stack change behaviour depending on which pedal was on, plus a loss of highs due to the long cable runs to and from the pedalboard. Also, when I tried to attenuate the signal to not overload pedals it got fuzzy for some reason. I figured putting a buffer on my pedalboard would help, but then why not just put a tube buffer right inside the amp, with a recovery stage while we're at it, so the FX loop was re-born.

I had an unused socket in the amp anyway, else I would have went the MOSFET route.