Placement of a parallel triode gain stage in a Princeton Reverb

[Hoodnight]

Hey folks, cobbling together a Princeton Reverb without tremolo and decided to use the spare triode left over to convert an existing triode gain stage to a parallel triode stage - I'm not a gain-lover so didn't want to do an extra stage as a boost or anything.

I'm wondering - in terms of lowering the overall noise floor of the amp - would I be better served using the parallel stage as the first gain stage and have less noise downstream, or use it for the reverb recovery stage and have less reverb hum coming through?

[BrainDeadAmps]

Input stage, if noise reduction is your goal.

[pdf64]

… or use it for the reverb recovery stage and have less reverb hum coming through?

I can’t see how parallel triodes could reduce hum?

[Hoodnight]

@pdf64 - I've found a heap of hum comes from the reverb recovery triode in my DRRI. Realising that the noise reduction of a parallel stage probably won't do anything for hum though!

Looks like V1 is the ticket.

[pdf64]

@pdf64 - I've found a heap of hum comes from the reverb recovery triode in my DRRI…

It may just be amplifying hum that’s being fed into its grid, from whatever source.

[tubenit]

Kevin O'Connor states in one of his books that paralleled triode results in 30% increase in gain and no increase in floor noise.  I don't remember him saying that it ever reduces noise in contrast to his statement it doesn't increase noise.


Based on numerous experiences I've had with paralleled triodes (12A_7), I would venture to say it does slightly increase gain with no floor noise increase.



With respect, Tubenit

[dwinstonwood]

Merlin Blencowe states that because less anode current is needed in parallel triodes to create the same gain as a single triode, the parallel tubes have less "flicker" noise. I don't think he says anything about hum.
With my "playing" would I even ever hear flicker?

[jordan86]

Not to derail but have you considered doing a long tail pair phase inverter instead? Would be more of a deluxe at that point but I think that’s give you more bang for your buck.

[BrainDeadAmps]

From what I’ve gathered it doesn’t reduce noise, it just increases the signal/noise ratio.

[tubeswell]

If you use a parallel triode on V1 with a halved plate resistor (47k) and a halved cathode resistor 820R and a 47uF cathode bypass cap, it will reduce* the noise floor and preserve  the gain.

(*An improved S:N ratio amounts to the same thing as ‘less noise’ in the signal).

[Hoodnight]

@tubeswell - Absolute legend, much appreciated! Will try designing a layout around that setup, might make another thread once I've got something going.

@tubenit - Yeah I guess they kind of end up being interchangeable if circuit is modified so that gain is a constant?

@jordan86 - Yeah had thought about it, but I think I'd like to keep the cathodyne - I've got a Deluxe Reverb anyway, so may as well have some variety!

[tubeswell]

@tubeswell - Absolute legend, much appreciated! Will try designing a layout around that setup, might make another thread once I've got something

I should add that the parallel triode I referred to above will also have lower output impedance than a single triode stage with a 100k plate resistor - all the better to drive the following tone stack load.

[HotBluePlates]

Kevin O'Connor states in one of his books that paralleled triode results in 30% increase in gain and no increase in floor noise.  I don't remember him saying that it ever reduces noise in contrast to his statement it doesn't increase noise. ...

When seeking to reduce noise, the Old Books first blamed resistors, noting that there is an amount of noise that is unavoidable that is proportional to resistance, and due to temperature being above Absolute Zero.

     -  So plate-load resistors and grid-reference resistors are noise generators.
     -  Few point is out, but faulty resistors contribute more noise when there is large DC voltage across them:  plate load resistors.
     -  The grid-reference resistors tend to be high-resistance, and unfortunately have gain afterwards.  So they are critical.
     -  This leads to noise being "referred to an equivalent grid resistance" to describe the amount of noise.

RDH4 page 783:
"The equivalent resistance for shot-effect valve noise is inversely proportional to the [transconductance (Gm)]" [so maximizing Gm is helpful in reducing tube noise].

     -  Paralleling tubes gives a single "composite tube" with half the internal plate resistance and twice the Gm.
     -  Whether said explicitly or implicitly, authors lean on this "Gm-doubling" if/when they claim parallel triodes reduce noise.

NOTE:  You probably need to halve the plate & cathode resistor values to allow the double-current that results in the double-Gm if these tubes share components.  Tube Gm is proportional to plate current, and Gm falls with lower current.


The noise-reduction is likely small, and a material reduction of noise usually requires attacking from several directions.

However, we're talking about random hiss noise, not the hum Hoodnight mentions, which is likely the result of some other cause.

[Hoodnight]

Great info, thanks HBP.

Also, just to clarify - I'm an absolute hack whose amp-education is sporatic ingestion of RobRob, Uncle Doug, Merlin, and Lyle Caldwell, so anything I say should be taken with a metric tonne of salt.

I'm building a Princeton clone and don't care for tremolo, so thought I'd do something with the spare triode. I mostly play squeaky clean+reverb, and have also been futzing around with my DRRI that has excess hiss and lots of hum from the reverb recovery that I've lost the will to keep tracking down. A parallel triode stage seemed like a low-risk option in terms of adding complexity to my first build that would apparently make a minor improvement to tone, and the advertised noise reduction was a bonus.

The hum in the DRRI that I've got is coming from V4 somewhere, and I'll get back in there and figure it out when I develop some more patience - lifting the main PCB on that thing makes me want to go and buy a drumkit.

[PRR]

...Tube Gm is proportional to plate current, and Gm falls with lower current.

Tube Gm is a function of plate current, often the square-root.

Twice the current is often 1.4 times Gm.

But as Merlin says, in the audio range, in commercial tubes, shot noise may be exceeded by flicker noise over most of the audio band. (The "high current" approach is more valid above a MHz.) Frequently you want the "least" current that will still pull your load. (Remember that very low current truncates your treble and "hisses less".)

And yes, none of this is about HUM.

[HotBluePlates]

... (Remember that very low current truncates your treble and "hisses less".)

Restating for myself & others:

    lower current ---> higher internal plate resistance ---> higher output resistance/impedance ---> reduced treble bandwidth (when interacting with output & wiring capacitance)

... Frequently you want the "least" current that will still pull your load. ...

Good point!  Lots of reasons to be "just good enough."

[Hoodnight]

Right, so in terms of noise reduction it's diminishing returns when you're using increased current to increase Gm?

Have been playing around with a layout and decided to make up a schematic for it. My changes are highlighted. Everything in the power section is the same as the Hoffman Reverb No-Trem circuit in the library, so not included. Schematic is attached.

Any glaring issues poking out to anyone? I'm gonna keep playing around and see whether it makes more sense to keep the 2nd gain stage in V3 or whether I should move the reverb driver itself down to V3.

[tubeswell]

Your cathodyne stage isn’t balanced. The plate load and the cathode load need to be the same resistance (or close to the same)

[Hoodnight]

Ah yep, copy paste error. Probably shouldn't be working on this so late at night! Thanks for the catch.

Updated schem attached.