Adding reverb to Bassman/Plexi (can anyone help double check my schematic?)

[passaloutre]

Not pleased with the options available for adding reverb to the Bassman/Plexi circuit, I decided to go about it myself. I didn't really like the compromises of the one-tube reverb or the Hoffman option which parallels the first stage (I want the reverb to come after the compression and tone shaping of the first stages and tone stack), and I really wanted Blackface reverb. So, I've essentially taken a Bassman with diode rectifier, and pasted the AB763 reverb in between the tonestack and the phase inverter, using the second half of V1 for reverb recovery rather than having two channels. Does anyone mind taking a look at my schematic and making sure I haven't made any major errors or oversights (particularly where the reverb is pasted in)?

Also, does anyone know of any commercial (boutique or otherwise) amps using a similar idea that I may borrow for ideas? Seems a no-brainer to marry this circuit to a reverb.

Thanks in advance

[passaloutre]

For the record, this will be using JTM45 style transformers from Mojotone (specifically the lower voltage PT with the 300-0-300 secondary) so I can use JJ6V6S or 6L6 tubes.

[sluckey]

I'm thinking you may want another triode just before the PI.

[passaloutre]

I have no triodes left!

People seem to have good luck putting the one-tube reverb in the same location without adding any more stages. Why not the blackface reverb?

[sluckey]

But the blackface reverb has the additional triode.

[passaloutre]

That's an interesting point. Presumably the extra triode makes up for any losses in the mixing stage (3.3M is an awful lot of resistance to put in series with your signal), but I wouldn't necessarily call it part of the reverb circuit. How does the one-tube reverb get away without this but the blackface can't? Also, the amps with the blackface reverb don't have the cathode follower buffering the tone stack. Would the CF see the mixing stage as just an extension of the impedance presented by the tone stack?

[sluckey]

That extra triode isn't exactly part of the reverb but it's necessary to boost the weak dry signal and weak reverb signal to a level needed to drive the PI.

The one tube reverb gets away with it because the mixing resistor is only about 100K. The one tube reverb is also very weak when compared to the fender reverb.

[PRR]

I figure it needs another gain of three to play like a 5F6a. (Which suggests an "active" mixer, slugged with NFB.)

Not having a triode is not a musically-valid reason to skip it.

Yes, some players will be fine with less gain.

Yes, resistive mixing is always lossy and 3Meg:0.5Meg is a brutal loss which must be made-up to get similar performance.

[HotBluePlates]

I gave my opinion of where to add the reverb (assuming Hoffman's plan, also linked, is unacceptable) on another forum here.

I also brought up the question of why add reverb inside the preamp at all, since the output tubes distort first.

I figure it needs another gain of three to play like a 5F6a. (Which suggests an "active" mixer, slugged with NFB.)

... 3Meg:0.5Meg is a brutal loss which must be made-up to get similar performance.

The 3M could be 1M to make-up some of the deficiency, though it's still about 0.5v short.

[passaloutre]

I figure it needs another gain of three to play like a 5F6a. (Which suggests an "active" mixer, slugged with NFB.)

Not having a triode is not a musically-valid reason to skip it.

Yes, some players will be fine with less gain.

Yes, resistive mixing is always lossy and 3Meg:0.5Meg is a brutal loss which must be made-up to get similar performance.

To what extent can this be made up with different mixing resistors? Or perhaps increasing the PI gain? Or both. Let's just say I'm not some bassman purist (or blackface reverb purist for that matter), so I don't mind making up the losses elsewhere if it gets me in the neighborhood.

When I had a blackface amp, I rarely turned the reverb knob past 3. Still I liked the brightness and attack of that reverb circuit. If tuning the mixing resistors gets me close to that sound (even if it's at a different place on the knob), that's much preferable than buying a new chassis just to add another tube.

Regarding the diagram you shared, could you help me understand how you calculated those voltages? Also, I have 4 and 8 ohm taps on the OT, if that changes the NFB calculation.

Thanks for your input by the way

[pdf64]

I have no triodes left!...

You might consider using an LND150 then; one could serve in a reverb recovery or the mixer recovery stage.

[tubenit]

You can consider this.  I use this in my D-763 amp.  It is right before the LTPI.  It has way more reverb then I'll ever use.  In fact, I considered changing it to the one tube reverb because it has SO much reverb.

IF you try it, experiment with the values where you need to. You can try a 2.2M reverb pot instead of the 1M if you want to.

With respect, Tubenit

[HotBluePlates]

... Also, I have 4 and 8 ohm taps on the OT, if that changes the NFB calculation. ...

If you mean the "4v" over by the feedback loop & speaker, that's about the voltage at the speaker & not about your feedback loop.

... Regarding the diagram you shared, could you help me understand how you calculated those voltages? ...

Speaking of that "4v (should be 12v in 4Ω)" over by the speaker:  Power = Voltage²/Impedance = 12²/4Ω = 36 watts.  If you have 4v across the 4Ω speaker load, you get only 4w of output power from the amp.

__________________________________________________

Let's start back at the Input jacks:

It's not an absolute, but a lot of guitar amps seem to have a "sensitivity" of ~20mV.  Meaning they only require ~20mV to drive the output tubes to maximum clean output power when the Volume control outputs half the signal applied to it.

You can split the hair finer with a lot of calculation, but a happy 12AX7 gain stage will have a voltage gain of something between 50-60.  To be conservative, a gain of 50 was assumed.  20mV * 50 = 1,000mV (or 1v).

We just mentioned "full clean power output when the Volume control tosses half the signal" But your amp won't get close to that.  Volume control full-up, so 1v in, 1v out.

2nd Gain Stage: No bypass cap, so half-gain of 25.  1v * 25 = 25v

Cathode Follower: These have a gain slightly less than unity, but call it "gain of 1" and so output is 25v.

Tone Stack Output: If you plug in 5F6-A values into the Marshall tab of the Tone Stack Calculator, the loss in the midrange is ~14dB.  20 log (5v/25v) = -14dB, so the output is 5v.

Resistive Mixer Loss: The 3.3MΩ resistor forms a voltage divider with the 470kΩ resistor coming from the 100kΩ Reverb pot.  The worst-case (Reverb control all the way down) looks to the Dry signal like R2/(R1+R2) = 470kΩ/(3300kΩ + 470kΩ) = 0.125x.  5v * 0.125 = 0.625v (call it 0.6v).

Phase-Inverter Input to Speaker:  You can calculate all the different things happening inside this circuit-block, but it's an enormous load of work for not much payoff.  Or you note that there is a feedback loop from the speaker to the phase inverter, which makes the entire block an opamp.  Closed-loop Gain of an Opamp is 1+(Rseries/Rshunt).

     Rseries = 27kΩ
     Rshunt = 5kΩ (Presence pot)
     Gain = 1+(27k/5k) = 6.4x

     Speaker Output = 0.6v * 6.4 = 3.84v (rounded up to 4v to offset some earlier rounding-down).

If you want the full 36 watts clean, 12v / 6.4 = 1.875v at the phase inverter input.  Hmm, 1.875v / 0.6v = ~3x which is why PRR said:

I figure it needs another gain of three to play like a 5F6a. ...

My suggestion gets you part of the way there.  Turning up your guitar louder gets you part of the way there.  Turning up the Volume control on the amp gets you part of the way there.

But at the end of the day you might be surprised to find, "This amp sounds a whole lot cleaner than I expected.  It's really hard to make the amp distort."

I can tell you from playing an acquaintance's 1958 5F6-A Bassman in a small airplane hangar that the original amp is brutally loud/clean and doesn't break up until waaaaaaayyy up the Volume dial.  Extra uncompensated loss due to the reverb circuit may produce an outcome you weren't looking for.  A late tweed Bassman is one of the cleanest and least-midrangey tweed amps out there.

[passaloutre]

That was a really great reply, and super helpful for someone who doesn't think about the math and just copies existing circuits.

Your comment about treating the power amp as an opamp was particularly enlightening for me, as I've dealt with opamps in guitar pedals a lot more than I've dealt with tubes (I've only made a couple amps, but I've made a million pedals). I imagine the analogy must break down at some point, as the gain cannot keep increasing with increasing feedback resistance, or else amps with no negative feedback would have infinite gain! I suppose with an opamp this happens when the output signal approaches the voltage rails, but I'm not sure what would break down first in the tube amp (I guess whichever tube starts to cutoff first?).

Regarding this particular amp, no I'm not expecting much distortion at low volume levels, though I would like at least *some* compression from the preamp without completely cranking it. If the mixing stage acts a bit like a pre-phase-inverter master volume, I'll consider that a good thing. Additionally, I will be using 6V6s and with a B+ around 400V, so I'm expecting a power-amp response more like a Deluxe Reverb. If I want more power amp distortion I have an attenuator.

Anyways, thanks for all the help above. I see now that you can't get something from nothing, and any series resistance in the signal path is limiting the amp to some degree. I've received some good suggestions, and I've got a few different options to try.

[HotBluePlates]

That was a really great reply, and super helpful for someone who doesn't think about the math and just copies existing circuits.

I made an error, and corrected it above:  PRR assumed the Volume pot was turned full-up.

The unbiassed cathode on the 2nd gain stage cuts its gain by half.  No longer 50x, gain for that stage is now 25x.  So 1v in, 25v out.  Which means you will not be able to get full power output (with 20mV input).  You'll distort the cathode follower & 2nd gain stages first, and have weak/clean output power from the power tubes.

Typical pickups are more like 100mV output; the 20mV sensitivity for full clean-power is in part a noise reduction technique.  So what do you get with your 100mV pickup?  5v out of the 1st gain stage, so Volume control at half still gives 500-1000mV (depending on taper), which is mighty close to enough to make the 2nd stage & cathode follower distort.

... Regarding this particular amp, no I'm not expecting much distortion at low volume levels, though I would like at least *some* compression from the preamp without completely cranking it. ...

That's not really gonna happen in a stock 5F6-A Bassman: the output tubes are the first thing to distort.  Kuehnel's Bassman Book proves that for the original circuit.

For your amp, the cathode follower will distort, but your output tubes would be making ~4w at this point.  They will be stark clean.

I think you're neither fish nor fowl as far as hoping the mixing resistors will act like a master volume control.  I'm leaning towards "worst of both worlds."

I haven't tried analyzing Hoffman's circuit for adding reverb to a Bassman.  I suspect it's quite a bit better than what we're looking at here.

... Your comment about treating the power amp as an opamp was particularly enlightening for me, as I've dealt with opamps in guitar pedals a lot more than I've dealt with tubes... I imagine the analogy must break down at some point, as the gain cannot keep increasing with increasing feedback resistance, or else amps with no negative feedback would have infinite gain! ...

The tube amp's feedback loop is simply taking the output and applying it to the inverting input of an earlier stage, just like a solid-state opamp.

The difference is the tube amp doesn't have an open-loop gain 100x or 1,000x or 1,000,000x the closed-loop gain.  More often something like 2-4x the closed-loop gain.  So if you keep increasing the series feedback resistor, eventually you just get "no negative feedback" (and it often doesn't take much resistance-increase).

... any series resistance in the signal path is limiting the amp to some degree. ...

It's not just "series resistance."  It's voltage divider made up of the 3.3MΩ resistor and the 470kΩ resistor coming from the Reverb Recovery gain stage.

To the Dry signal, the 3.3MΩ is the same as the resistance between a pot's Input and it Wiper; the 470kΩ is the resistance from the pot's Wiper to Ground.  Dry signal is much-reduced.

To the Reverb signal, the 470kΩ is the resistance from the pot's Input to its Wiper; the 3.3MΩ is the resistance from the pot's Wiper to Ground.  Reverb signal is less-reduced (but it's weaker to start with).

If you change the 3.3MΩ to 1MΩ, Dry signal doesn't get turned-down so much.  It goes up from 12% to 32% its starting strength.  But the Reverb gets turned down from 88% starting strength to 68%.

A right-balance involves experimentation.  And that's the key to circuits, whether you're talking pedals or amps: Balance.  Lift a circuit from one place & it doesn't necessarily behave the same shoe-horned into something else unless the balance is restored.

[passaloutre]

That's not really gonna happen in a stock 5F6-A Bassman: the output tubes are the first thing to distort.  Kuehnel's Bassman Book proves that for the original circuit.

Is this also the case with the Marshalls derived from the Bassman? I'm looking for more of a Plexi sound, even though I may have used the Bassman schematic here for convenience. I'll probably add a toggle to switch the cathode on V1 between Bassman and Plexi values.

It's not just "series resistance."  It's voltage divider made up of the 3.3MΩ resistor and the 470kΩ resistor coming from the Reverb Recovery gain stage.

To the Dry signal, the 3.3MΩ is the same as the resistance between a pot's Input and it Wiper; the 470kΩ is the resistance from the pot's Wiper to Ground.  Dry signal is much-reduced.

To the Reverb signal, the 470kΩ is the resistance from the pot's Input to its Wiper; the 3.3MΩ is the resistance from the pot's Wiper to Ground.  Reverb signal is less-reduced (but it's weaker to start with).

This makes sense to me, though it still seems analogous to a master volume control, albeit one that won't go above "1" on the dial. And I suppose even if I drop the "dry" mixing resistor down to 470k, I now have a master volume that only goes up to "5". And if I decrease that resistor even further, then I probably won't ever hear the reverb that I wanted so badly to begin with?

So let's say I could magically come up with another triode, and I want to use it as a mixing stage. I could simply steal the recovery stage from the AB763, but that's got way more gain than the factor of ~10 loss I had in the mixing stage. Would that drive the power amp into distortion *too soon*? Would I be better off with something like this: http://www.valvewizard.co.uk/localfeedback.html?

[tubenit]

So let's say I could magically come up with another triode

No magic needed.  You can use a mosfet as the cathode follower and free up a triode.  As in the example schematic shown. My experience is the mosfet in the CF position sounds exactly the same as a tube triode in that position. YMMV

http://www.geofex.com/Article_Folders/mosfet_folly/mosfetfolly.htm

I honestly think the schematic in reply #11 would work well, but I haven't specifically done that.  I have used one tube reverbs in Marshall topology and liked it a lot.  You could try a one tube using 12DW7 where the 12AU7 section is the driver triode.

With respect, Tubenit

[passaloutre]

Thanks tubenit, I'm leaning most towards the schematic you posted in reply #11. The biggest difference between that and the AB763 seems to be the mixing ratio (and some caps you presumably used to tame the brightness). Do you find this sucks the gain out of the preamp and can't drive the phase inverter?

You said youve used the one-tube reverb with the Marshall circuit before. Do you ever use an additional stage to make up the lost gain?

If I were more cynical, I'd say the people in this thread are listening with their calculators more than their ears (I jest), but I respect the knowledge and experience these people have. By all accounts people like the results of adding reverb to this part of the circuit (e.g., https://el34world.com/Forum/index.php?topic=19196.0, https://el34world.com/Forum/index.php?topic=14690.0, etc.) *in spite* of the mixing divider (whatever its ratio may be).

[HotBluePlates]

... If I were more cynical, I'd say the people in this thread are listening with their calculators more than their ears (I jest) ...

 

Tubenit has provided 2 different and very good solutions:  MOSFET to replace the tube cathode follower (freeing up a triode for use elsewhere), and using the "1-tube reverb circuit" (perhaps with as 12DW7, also freeing up a triode).  He's built quite a few amps using both ideas, I've played a few of them, and can attest they sound & work very well!

I would offer the alternative view that perhaps those providing the warnings have a number of amp-builds under their belt, and know how to use the calculator to predict performance, saving time compared to trial-n-error.  I don't usually tell people how many bad ideas I had, tried, and learned from the hard way before coming to learn how to let the calculator spot the bad ideas. 

... Is this also the case with the Marshalls derived from the Bassman? I'm looking for more of a Plexi sound ...

Yes, same.  Also proven in Kuehnel's book.

... So let's say I could magically come up with another triode, and I want to use it as a mixing stage. I could simply steal the recovery stage from the AB763, but that's got way more gain than the factor of ~10 loss I had in the mixing stage. Would that drive the power amp into distortion *too soon*? ...

Just do the same thing Fender did in their reverb-amps that your circuit came from:  Deluxe Reverb AB763  There's a triode gain stage between the 3.3MΩ resistor and the phase inverter.  You wouldn't care whether it "adds too much gain, cause that's what your volume control is for (you can always remove the cathode bypass cap if you want to reduce the stage's gain).

This is what Sluckey suggested in Reply #2.

It's not just "series resistance."  It's voltage divider made up of the 3.3MΩ resistor and the 470kΩ resistor coming from the Reverb Recovery gain stage.

This makes sense to me, though it still seems analogous to a master volume control, albeit one that won't go above "1" on the dial. And I suppose even if I drop the "dry" mixing resistor down to 470k, I now have a master volume that only goes up to "5". And if I decrease that resistor even further, then I probably won't ever hear the reverb that I wanted so badly to begin with? ...

Something like that.

Or you can move the reverb circuit to where I suggested here, at the output of the Volume control.  Maybe add a cathode bypass cap to the 2nd gain stage to make up some of the loss.

At the end of the day, you're gonna have to build it & see what happens.  It's only the practical "doing" that will finally let you know if the amp works the way you envisioned.

[passaloutre]

What do you think of this one? I've not used a MOSFET like this before, so let me know if I've done something wrong. Would I be better off using the mosfet for the recovery stage? That is, does the tube cathode follower contribute something to the tone that the mosfet can't?

I do hope you read the sarcasm in my "calculator" comment. I really appreciate all the help.

[pdf64]

... does the tube cathode follower contribute something to the tone that the mosfet can't?

See http://www.valvewizard.co.uk/dccf.html

[sluckey]

I suggest to use separate 1.5K/25µF R/Cs for the cathodes of the reverb recovery and mixer triodes.

[ac427v]

I'm getting really close with mine. The reverb sounds good but still tweaking to get smooth distortion tone on 10. I find that the theoretical discussion and math explanations in your thread have really helped steer me to small changes that improved my build. Thanks for getting this going!

On a larger Marshall 800 build with fender reverb I used the 3.3 meg mixing resister and included a mixing triode immediately before the PI. That added way too much gain and hum. Then I copied Tubeswill who used a split load on that stage--helped with the excessive gain problem but still too much hum in my build. Someday I'll get back to that amp...

[tubenit]

THANKS for sharing the Sumatra Plexi schematic with reverb!  Great job drawing the schematic, BTW. 

I used the 3.3 meg mixing resister and included a mixing triode immediately before the PI. That added way too much gain and hum.

So much is a matter of personal taste, but I am with you on this.  I think it would add too much gain for what I like also. I've always had enough reverb without the added mixing triode even with a one tube reverb.

With respect, Tubenit

[passaloutre]

If I were to use the mosfet for a mixing stage rather than the follower, is designing that stage really as simple as straight subbing it for a triode (gate for grid, source for cathode, drain for plate)? The characteristics graphs on the datasheet (https://www.vishay.com/docs/91059/91059.pdf) are on a log scale, cover a much lower range of drain voltages, and seem to go in the wrong direction.

[J Fletcher]

 I have added reverb to a few amps using either a 6SN7 or a 6FQ7 . Use one triode section , which pretty much equals both sections of a 12AT7 in parallel , to drive the tank , and the other triode as a gain stage. When I have done this , I sometimes use the 2nd triode as the cathode follower , or a split load phase inverter , where the loss of gain , the 6SN7 has an amplification factor of 20 , is not important.
Done this in a number of different builds , and have always been pleased with the results.
Jerry

[pdf64]

If I were to use the mosfet for a mixing stage rather than the follower, is designing that stage really as simple as straight subbing it for a triode (gate for grid, source for cathode, drain for plate)? The characteristics graphs on the datasheet (https://www.vishay.com/docs/91059/91059.pdf) are on a log scale, cover a much lower range of drain voltages, and seem to go in the wrong direction.

IIRC it depends on whether they're enhancement or depletion mode devices, and so require forward or reverse biasing for linear use.
Whatever, the IRF820 is fairly beefy device, and its capacitances may make it somewhat unsuitable for use with the typically rather high impedance mixer circuits. Have a look at how the LND150 is used in fx loop circuits, several examples in this thread https://el34world.com/Forum/index.php?topic=10208.msg288263#new

[passaloutre]

Well I decided the only way I would learn was to build it and see for myself. Based on the testimony of Tubenit and others who've added reverb to this circuit without any makeup gain, I skipped the extra stage and built it as below:



Rather than attempt to makeup the gain required to drive the power amp, I chose to bring the headroom of the power amp down (I didn't want a 36 watt amp to begin with), though a combination of: lower voltage power transformer, sag resistor, 6V6 tubes, and cathode bias (with a bypass cap to keep some of the fixed-bias flavor. I've got 350V on the 6V6 plates and 22V on the cathodes, so I reckon my output power is somewhere around 5E3 territory.

 I don't know if this was enough to account for the factor of three I needed to lose, and it's certainly no purist's Plexi, but the amp sounds great. Nice and crunchy on the second half of the dial, and not completely deafening when it gets there. The reverb isn't quite blackface surfy (due to the changed mixing ratio) but it's definitely enough for the amount of distortion this thing has.

For curiosity, I did include a pull switch on the reverb pot to defeat the mixer.

Here's a few progress pics, and some sound clips:





[Clean](https://drive.google.com/file/d/1cJGkXNT_aLBfnGRzQ9poHl2gU8XrIW_b/view?usp=sharing)

[Breakup](https://drive.google.com/file/d/1GplEntZE7pdS0vzFW-aVmE0XkVXE5_wj/view?usp=sharing)

[Crunch](https://drive.google.com/file/d/1hS9MIqbNiv4v9Y6OBYAm--9Oj_S2T46H/view?usp=sharing)

Let me know if you have any trouble accessing the files.

[tubeswell]

I did this 10 years ago, and it worked pretty spectacularly if I may say so myself (cross between a 5F6A and an Allen Encore)


(Edit: I just had a listen to your soundbytes passaloutre - very nice!)

[tubenit]

Passaloutre,

THANK you for sharing your success!!  Congrats!  The sound clips are great.  Nice tone!  Enjoyed your playing!

I'll offer some mods for your consideration (indicated in red).  I find the 47p snubbing cap to typically be a tone killer in my experience.  It has always sounded to me like a thin blanket was placed over the speaker killing the higher frequencies.

In contrast, the "enhance cap" indicated by the 120p across the 82k plate resistor doesn't kill the highs at all in my experience but instead simply smooths them out eliminating any harshness to them.  It's sort of like a child's spinning top where when the top is wobblying there is "harshness" in the high frequencies and when the top is spinning smoothly then there is smoothness in the high frequencies. 

The midboost switch allows 250p with the caps in series and .0015 without the caps in series giving a nice midboost.  You can use a .001 & a 500p cap in parallel for that.

The other mods are pretty straight forward.

I am NOT saying you should do any of these as all of this is personal taste in tone.  IF you only try one of these ideas, I hope you'll try removing the 47p cap and see if you like the tone better and find it has better clarity.

You could also try a different value on the 6V6's cathode cap.  You can safely parallel (with insulated alligator clipped wires) a 25uf or another 47uf cap to see if you like the tone more with 72uf or 94uf values there.  It tends to tighten the bass in my experience.  OR if you want a "softer" bass tone, then try a 22uf instead of a 47uf.

With respect, Tubenit

[ac427v]

Wow! Great amp-Great playing. I really like your "Gently Weeps" version.

After a lot of trial and error I ended up with a 220k reverb mixing resistor too. I've been thinking about trying the tweed bassman tonestack values to get a little more midrange scoop on tap. Hearing yours convinced me it's worth opening it up one last time. Thanks for reporting out and the sound clips.

[shooter]

nice! 




do you make Kettle drums also 

[pdf64]

I happened across a schematic of the DrZ Maz, and that seems like it might be characterised as being a tweed bassman ish preamp into a blackface type reverb and then a Vox ish power amp https://i211.photobucket.com/albums/bb138/jaydawg76/Dr_Z-MAZ.jpg

The hand drawing has errors, but you get the gist 

[passaloutre]

Thanks for all the feedback guys, indeed thanks for all the phenomenal help throughout this process.

Tubenit, could you explain the "mid-boost" a little more? It looks like it's just changing that cap value by adding another in series? Is the 4M7 necessary? I guess that's for anti-pop...

I'm curious why the cap across the PI plate load sounds different than the cap across both PI plates. Either way should be cutting ultra-highs out of the signal to the power tubes.

Not sure I follow the kettle drum comment, but sure

[shooter]

 
I'm sure you're in the majority


your paperweights make good tuners

[tubenit]

Tubenit, could you explain the "mid-boost" a little more? It looks like it's just changing that cap value by adding another in series? Is the 4M7 necessary? I guess that's for anti-pop...

Yes, it changes the value of the cap increasing or decreasing the mids.  Yes, the 4.7M helps prevent popping.  This midboost idea is on Dumble amps frequently which is where I first saw it.

I'm curious why the cap across the PI plate load sounds different than the cap across both PI plates. Either way should be cutting ultra-highs out of the signal to the power tubes

I have NO idea why it sounds different, but it sure has on my amps. It's become something I do with every LTPI amp I have.
I have not gotten along at all with the snubbing caps whether 10p or 47p or higher. 

There may be other factors involve that have made it work on my amps like component choices, speaker, guitar ………. who knows?   But the difference to me is snubbing caps smother tone and kill highs and the "enhance" cap allows transparency and simply smooths the highs without squashing them.

It would be a very easy experiment for you to try. 

With respect, Tubenit

[pdf64]

...

I'm curious why the cap across the PI plate load sounds different than the cap across both PI plates. Either way should be cutting ultra-highs out of the signal to the power tubes

I have NO idea why it sounds different, but it sure has on my amps. It's become something I do with every LTPI amp I have.
I have not gotten along at all with the snubbing caps whether 10p or 47p or higher. 

There may be other factors involve that have made it work on my amps like component choices, speaker, guitar …

My take is that the main factors involved are particular characteristics of the OT used (specifically its phase shift at high frequencies), any inherent high freq phase shift in the design (eg due to grid stoppers) and parasitics of the lead dress used. Depending on how those play out, and the degree of NFB (included how it may vary according to the presence control setting, if fitted), some methods of rolling off the ultrasonic response (ie to improve stability) can actually make it worse, causing ringing and even oscillation (parasitic to a signal, or at worst, free running).
With regard to the OT, even the secondary tap used can affect this, as at those extreme freqs, the primary to secondary coupling characteristics may well differ at each tap (even when loaded as per the design intent).
Hence checking the output with a scope on any build seems prudent, when loaded in both resistive and a real cab scenarios.

[pdf64]

I did this 10 years ago, and it worked pretty spectacularly if I may say so myself (cross between a 5F6A and an Allen Encore)
...

FYI the HT fusing may be improved there with a F / quick blow type fuse of the same rating. As the slow smooth HT ramp up provided by the indirectly heated cathode of the GZ34, and the absence of a HT standby, may mean that a T / slow blow type is unnecessary, and a F type fuse should respond to actual faults faster.
The 'fast min max' mode of posher Flukes etc can help to reveal what the peak currents there are, at power up and under load.

[passaloutre]

Tubenit, could you explain the "mid-boost" a little more? It looks like it's just changing that cap value by adding another in series? Is the 4M7 necessary? I guess that's for anti-pop...

Yes, it changes the value of the cap increasing or decreasing the mids.  Yes, the 4.7M helps prevent popping.  This midboost idea is on Dumble amps frequently which is where I first saw it.

I'm curious why the cap across the PI plate load sounds different than the cap across both PI plates. Either way should be cutting ultra-highs out of the signal to the power tubes

I have NO idea why it sounds different, but it sure has on my amps. It's become something I do with every LTPI amp I have.
I have not gotten along at all with the snubbing caps whether 10p or 47p or higher. 

There may be other factors involve that have made it work on my amps like component choices, speaker, guitar ………. who knows?   But the difference to me is snubbing caps smother tone and kill highs and the "enhance" cap allows transparency and simply smooths the highs without squashing them.

It would be a very easy experiment for you to try. 

With respect, Tubenit

Tubenit, it occurred to me today while looking at the amp and my photos of the build, that I just plain forgot to put that snubber cap in the amp in the first place [facepalm]. It was on my schematic and layout, but somehow never made it into the amp, oops.

Therefore I can confirm that it is not necessary, the amp sounds great without it, and I have no inclination to add it.

Here is the current state of the amp, in a combo cabinet with one Celestion Neo Creamback. Still waiting on my tolex and grill cloth.