FX Loop for Express; Two Rock?

[Mr. Lime]

On my TW clone I would like to take use of the left over triode.
I already tried to set it up like a Komet K-50 with cathode follower but I liked the amp without it somewhat better.
Kevin O'Connor wrote on his TUT 1 about a transparent recovery stage of a tube fx loop which incorporates a local feedback loop.
Merlin Blencowe has written about such a gain stage on his website and in his book as well. Dumble and Two Rock had similar approaches but seem not as transparent as O'Connor has intended.
If I had to do something from scratch I would pad the signal down to 1/10 with a 100k/10k divider after the cold clipping stage.
On the return side I'd like to have a gain recovery which equals the reduction from the send output.
I wouldn't place the MV behind the recovery stage as it might work better as PPIMV. Doesn't the Two Rock miss a ground reference on the send? Also the recovery stage has a lot of more gain with a feedback R of 22M.


Any thoughts on my loop or the Two Rock one?
Has anyone tried something similar with success?

Ruby Schematic Two Rock Trainwreck Express FX Loop.jpg

FX Loop TW.png

Thanks a lot!

[Mr. Lime]

Well, as the local feedback circuit's divider made out of Ri and Rf the closed loop gain can be controlled by Rf made lager.
Aiken does have great info about LNFB stages and shows us the math behind them: http://www.aikenamps.com/index.php/desi ... amplifiers

Adjusting Rf with a slightly lager value a transparent gain boost over a large bandwidth should be archived.

The Two Rock Ruby has a very large Rf which results in very little feedback and lots of gain which is controlled with the MV behind it. Some users reported that the Ruby oscillates when the MV is turned way up. Generally I would guess that the circuit of the Ruby isn't that transparent at all but colors the distortion quite a lot.

Probably the previous gain stage with the FX send output should be buffered to keep the output impedance low and therefore the input impedance of the LNFB stage controlled.
A dry path could or should be added as the LNFB stage serves us a good summing amp and so the sound could be even more transparent.
I will try a LND150 mosfet as source follower dc-coupled to the cold clipping stage when all the parts have arrived.

Some thoughts?

[10thTx]

This is probably what I would try. Not sure of the value for the "send pot". I've tried using the CF as the FX "send" on 3 different amps and have enjoyed how it sounded using delay or reverb pedals.

With respect, 10thtx

[Mr. Lime]

Well the low output impedance of the MOSFET would be gone if a 250k pot is used.
A 25k pot might be the better choice or simply a fixed divider instead so the return stage can be set up for the needed boost.

The right feedback R value maybe has to be found with a pot or trim pot to archive the most transparent setting. Since nearly all pedals and rack effects have a volume control I don't think a send or return level control might be necessary.

[pdf64]

If the output impedance is already low enough to drive a regular guitar cable with negligible HF loss (eg <25k) then there's no real benefit in adding a buffer.
Hence the 2nd schematic in post #1 would work fine; make the 10k variable or a pot if a send level control is desired.

[Mr. Lime]

If the output impedance is already low enough to drive a regular guitar cable with negligible HF loss (eg <25k) then there's no real benefit in adding a buffer.
Hence the 2nd schematic in post #1 would work fine; make the 10k variable or a pot if a send level control is desired.

Since we have a cold unbypassed clipping stage here the output impedance is roughly 41k. This probably means a significant highs roll off.
If I did my math right the 100k/10k divider gives -20dB. A 10k pot instead of the resistor means even more attenuation which isn't useful here imo.

The Silver Jubilee has such a loop turned around so the NFB stage acts as a send stage giving low impedance due to the feedback.
As reported by many TW cloners the circuit is delicate about effects loops so a change of the cold stage would probably ruin the sound.

10thTx MOSFET recommendation is still the best bet for most transparency when you ask me.
Also the grid stopper (R29) between anode and gate could be increased a lot without HF loss which would help against blocking distortion, doesn't it?

If a send pot is desired I would try a 25k pot and increase the feedback R (R24) to 1M5 or 2M. This gives us a 23 - 26dB boost. So we have 3-6 dB over the attenuation of the send divider when the pot is turned all the way up. Just in case the triode doesn't operate as perfect like in theory.

transparent FX loop.png

[pdf64]

...Since we have a cold unbypassed clipping stage here the output impedance is roughly 41k...

No, the 'send' output impedance can't be higher than R22 = 10k (with regard to the schematic in post #1 of this thread). Assuming your 41k 3rd stage output impedance, the 'send' output impedance will = 10k//(100k+41k).

...If I did my math right the 100k/10k divider gives -20dB. A 10k pot instead of the resistor means even more attenuation which isn't useful here imo...

Yes, that divider will give -20dB. Consider the desired 'send' signal level; most pedal fx may not tolerate >500mV without being overdriven. Rack fx may cope with a bit more, perhaps up to 2Vrms. As the Trainwreck 3rd stage will likely be pushed to overdrive, I guess the signal level at its plate may be ~40Vrms? I don't know those tubes or your intended operating points, so maybe I'm in the wrong ballpark?

...The Silver Jubilee has such a loop turned around so the NFB stage acts as a send stage giving low impedance due to the feedback...

If you've got a design in mind, it would be best to link to it; the several silver jubilee models I've found seem to use the same arrangement as your post #1 schematic, eg https://el34world.com/charts/Schematics ... w_2555.pdf

[jjman]

All this loop talk got me thinking about mine and how it has been a little too weak in its ver 2 setup. All have been parallel and Ver 1 had a “blend” control going into the PI (with a phase flipping trans on the wet to preserve phase.) Although that allowed for very wet settings I did not want to get wetter by reducing so much dry.

Ver 2 has each “channel” feeding opposite sides of the PI to preserve phase, instead of the flipping tranny. The divider/send tap was between a 2.2meg and 18k dropper string to ground. The “wet” input to the PI was 50% the strength of the dry side. Ver 2 had been calibrated to avoid overdriving a particular electronic reverb unit. Time to beef up the wet potential.

I just changed the 18k for a 50k pot (bottom-left of scheme) and can now dial in 0-150% wet strength compared to the dry side. The wet “recovery” stage is not clipping. I will do some testing to see where various pedals clip via this “send” control when the preamp is maxed. I’m thinking that reverb unit (ART MR1) is too sensitive even on the “line” setting. Based on some notes I think it was clipping around 250-300MVpp. The send is likely maxing at ~1Vpp now based on the new dry-wet ratio to the PI.

[Mr. Lime]

How does your loop turn out? Are you happy with it?

I tried the one in the first post but I get some kind of blocking distortion. Maybe the approach with the second PI input delivers better results?
Any loop suggestions are very welcome!

[jjman]

I went to assess the pedals’ input capacity and realized I should not assess by feeding them a clipped signal even if clipped only on one half. I injected a sign wave into each pedal directly from my Heathkit generator with a total load on the pedals' outputs of about 500k.

Model.................Output clipped with input at
Boss GE-7 EQ:........4.4Vpp
Boss CE-2 Chorus:...3.4Vpp
Boss DM-3 Delay:....3.8Vpp
Digiverb:.............3.8Vpp
ART MR-1 Reverb:...1Vpp engaged
........................2Vpp bypass mode

I won’t be using the ART with this amp any more. I didn't test my Lexicon LXP-1 yet. It has some quirks I don't like despite it's internet status.

With a ~250MVpp input and the preamp on the amp maxed (cold clipping) the loop gets 1.8Vpp at max send. The recovery stage brings the return to 105Vpp into the wet PI side at that setting. The dry side of the PI gets 80Vpp there. Assuming the recovery is center biased, I could presumably have 2.6Vpp max coming from the loop w/o recovery clipping. Since these pedals can handle that I may try a 75k or 100k send. The 50k I’m using has a detent at 25k which is nice.

The 2 sides look almost the same shape and the phase is out a nice 180°. I guess my goal is max potential wet along with the dry in place. Or I could switch to series and forget the dry PI input. Still need to really play it instead of measuring.

[Mr. Lime]

Having an almost identical wave shape on both PI inputs seem to be a good start.
Usually I use a TC G Major as rack unit.
What I do miss is the global negative feedback applied to the PI. On clean settings I miss it most so I'm not sure if that would be the solution for that particular amp..
Some amps like the Fender 165AB have the feedback applied to the same input where the preamp is. A low output impedance of the previous stage is therefore needed.
IMO the presence control isn't needed but I would keep a portion of feedback.

I might try something like this:

loop LND150 Express.PNG