Best effects loop circuit?

[Ten Over]

For this reason I like the idea of using the master volume to determine drive, though the master volume would equally work well after the effects loop.

I would like to be able to switch it in and out of circuit.

One way to do the bypass switch:

[TUBEDUDE]

In this Mesa design the benefit of the cathode followers low output impedance is nullified by the 100k resistor in line.
The output of the loop send CF should be taken before the 100k resistor, from the junction of Rk and Rg, above the 100k resistor.
This would also be the place to install a send control for maximum versatility.
Also if you use a 12DW7, the AU side would be used here for a better FX send, making the loop able to drive any load impedance effortlessly.

[Ten Over]

In this Mesa design the benefit of the cathode followers low output impedance is nullified by the 100k resistor in line.
The output of the loop send CF should be taken before the 100k resistor, from the junction of Rk and Rg, above the 100k resistor.
This would also be the place to install a send control for maximum versatility.

There is no quote included, but I'm going to guess that you are talking about my Mesa loop above.

There are various ways to get a low output impedance at the Send jack. Using the low output impedance at the cathode is one method, but it has some problems associated with it. The coupling cap at the cathode is discharged when the amp is warm and in Standby mode. When the Standby switch is flicked, the DC voltage on that coupling cap suddenly shoots up which puts an alarming DC voltage on the Send jack. Some effects devices don't appreciate this. My Mesa loop has a greatly reduced DC spike when the Standby switch is closed.

The signal needs to be attenuated before the cathode follower when the signal is taken from the cathode in order to be suitable for effects devices. The greatly reduced signal level combined with the constant noise level means that the signal-to-noise ratio is degraded. My Mesa loop has a large signal at the cathode which is then attenuated along with the noise for the Send signal.

The advantage of taking the signal from the cathode is easily nullified by using a Send Level control that is too large. There is an example on the previous page that has a 100k Send pot. The worst case output impedance is 25.3k which completely wastes the 426 Ohm output impedance at the cathode. Let's say that we want to drive 15.24m of 100pF/m cable with a corner frequency of 15kHz. The maximum output impedance would be:
1/(2 *pi * 1.524nF * 15kHz) = 6.96k
This makes the maximum Send Level pot resistance 25k.
My Mesa loop has an output impedance of 6.3k, so it would have no problem driving 15.24m of patch cable.

[Ten Over]

Also if you use a 12DW7, the AU side would be used here for a better FX send, making the loop able to drive any load impedance effortlessly.

My Mesa loop can easily drive a very low impedance load with a 12AX7 and my driver is only using 1.1mA as compared to 10.6mA with the AU side on the previous page (10.6mA!?! Somebody better check that. That's more than the entire pre-amp combined.). Even shorting the Send jack won't cause the cathode follower to clip in my Mesa loop.

The output impedance at the cathode of the 12AX7 in my Mesa loop is something like 591 Ohms as compared to 426 Ohms for the AU as configured on the previous page. That's not really very much difference. How well a loop driver works depends mainly on the circuit design rather than on the choice of a 12AX7 or a 12AU7.

[TUBEDUDE]

Also if you use a 12DW7, the AU side would be used here for a better FX send, making the loop able to drive any load impedance effortlessly.

My Mesa loop can easily drive a very low impedance load with a 12AX7 and my driver is only using 1.1mA as compared to 10.6mA with the AU side on the previous page (10.6mA!?! Somebody better check that. That's more than the entire pre-amp combined.). Even shorting the Send jack won't cause the cathode follower to clip in my Mesa loop.

The output impedance at the cathode of the 12AX7 in my Mesa loop is something like 591 Ohms as compared to 426 Ohms for the AU as configured on the previous page. That's not really very much difference. How well a loop driver works depends mainly on the circuit design rather than on the choice of a 12AX7 or a 12AU7.

Ooh, so much! Good basic info. A couple points seem to be reaching though.
If you short your send jack, you have no output, so you're signal qualities don't matter. If you accept a more guitar relevant frequency response, you can drive an even longer cable. The longest I've needed was 25 feet, and it performs flawlessly.
This is the circuit used in London Power amplifiers. I wonder if they are aware of these shortcomings.
I used it in many amplifiers. It is quiet and clean. I did later modify them with a diode/resistor from grid to cathode as recommended by Blencowe. After your post I checked one of my amps, and there is no D.C. at the output, turn on or running. Possibly an issue with amps that still use a standby switch.
A smaller send pot is a great idea, and I'll bench it, thanks.

[Ten Over]

If you short your send jack, you have no output, so you're signal qualities don't matter.

It's just an example of finding the conditions at the extreme. This sort of exercise is often helpful in clarifying how things are changing.

FX Loop KOC Best All-Tube Loop.gif

The London Power loop has a 1k load on the CF with the Send jack shorted and the Send pot at max. This reduces the signal at the cathode to about 3Vp before distortion and clipping sets in. With no load at the Send jack, the CF can swing around 36Vp of clean signal with 300V on the plate. So that is the entire range for that setup.

If you accept a more guitar relevant frequency response, you can drive an even longer cable. The longest I've needed was 25 feet, and it performs flawlessly.

With 100pF/m cable, the worst case corner frequency would be 8.3kHz. This is too low for me. Of course at Send pot settings outside of the central region the performance improves, but somewhere, somehow, somebody is going to set that thing for the worst case. I feel that it would be better to have a design that has suitable performance at any setting.

[Ten Over]

After your post I checked one of my amps, and there is no D.C. at the output, turn on or running. Possibly an issue with amps that still use a standby switch.

The filter capacitor on the fixed bias voltage divider doesn't allow the grid voltage to suddenly shoot up when the Standby switch is closed. There are plenty of designs out there without that capacitor (particularly SS designs) that experience DC spikes at the Send jack.

A smaller send pot is a great idea, and I'll bench it,

Just set the attenuation before the CF so that you can't overdrive the CF. This might change the overall gain of the whole setup, but it has a Return pot to compensate for this.