Question about guitar amp phase inverters

[ebartlet]

I have been researching guitar amp circuits and noticed vey few amps use a cathodyne type circuit vs the long tailed pair circuit for the phase inverter.

aside from the gain parameter, it seems to me (with my limited understanding) that the cathodyne PI would be better at driving the grids of the output tubes especialy when you get into AB2 land). Is this correct? And why is the long tailed pair so popular?

 

[PRR]

> I have been researching guitar amp circuits and noticed vey few amps use a cathodyne type circuit

Longtail is "newer", late 1950s. Go further into the past with a wider net. Many many cathodynes. Before that, many "paraphase" inverters (a natural for modding an SE model to P-P, especially with a common-cathode twin triode).

Just in the Fender DeLuxes:
Paraphase: http://schematicheaven.com/fenderamps/deluxe_5a3_schem.pdf
Semi-floating paraphase: http://schematicheaven.com/fenderamps/deluxe_5d3_schem.pdf
Cathodyne: http://schematicheaven.com/fenderamps/deluxe_5e3_schem.pdf
Longtail: http://schematicheaven.com/fenderamps/deluxe_6g3_schem.pdf

The 5E3 cathodyne works for many musicians. Some plucker named Neil owns dozens of them. His specific feature is the wonked Vol/Tone network, but he clearly knows how to hold a cathodyne on the very edge of grid blocking.

Fender/CBS felt some ownership of the longtail, so you see other amps using or reverting-to cathodyne well into the 1970s. My Ampeg VT-40 is cathodyne, but I think an earlier model was longtail. Early Sunns were all taken from Dynaco: pentode-cathodyne.

> it seems to me... that the cathodyne PI would be better at driving the grids of the output tubes especialy when you get into AB2 land...

No capacitor-coupled grid can go AB2. Don't matter what kind of driver, until you get to absurd dreams like a driver and caps bigger than the power stage. Geetar amps "touch" AB2, and it affects the way they are used, but they can't "work" AB2, the cap charges-down and puts the grid more negative.

The exceptions prove the rule. Ampeg SVT uses a 12BH7 buffer direct to the 6146/6550 grids, though a grid resistor limits Ig to a few mA. Fender 300PS uses a 6V6 (yes, a Champ) and an interstage transformer to slap grids. Before these beasts, we have to go to the late 1930s to find a guitar amp with significant grid current, and then only because before 6L6 there was no cheaper way to get stage power.

For the short moment when the cathodyne cathode tries to pull the lower P-P grid positive, the signal at the cathodyne plate slams negative. The longtail's plate loadings have little effect on each other.

The longtail is easy to cripple so that it can put plenty but not too-much drive on the power tubes' grids. For common power bottles, it tends to happen naturally. OTOH, for high-gain EL84, a stock-proportion longtail can slap the grids so hard into cutoff that it don't recover till halfway through the next verse.

The longtail has two hi-Z inputs, which allows using a pot and nice-size cap in a NFB loop. Cathodyne will invariably have a volt-amp, usually a simple tri/pentode, and a low-Z input at the cathode. Fine for simple resistive tapping of speaker winding, but not for fancy stuff.

The fancy stuff is the 5F6-A Presence control. Note also that it injects to the common cathode as well as one grid. This changes P-P balance. With the presence cap, P-P balance changes mid-band. This is apparently interesting to the ear.

> why is the long tailed pair so popular?

MANY more gitar geeks at Hoffman's.

 

[ebartlet]

[quote author="PRR"]

No capacitor-coupled grid can go AB2.... [/quote]

Thanks PRR. Ok I get it, I misread an article about blocking distortion -DC coupled cathode follower to OP tube grids will allow AB2, not cathodyne. And the long tailed pair allows for a presence control.

So I'm guessing that the cathodyne would still be less prone to grid blocking?

Yeah I've heard of that neil guy, I just ordered a 5C3 clone kit. I was gonna go 5E3, but an all octal, grid leak bias amp looked like fun...

[quote author="PRR"] His specific feature is the wonked Vol/Tone network[/quote]

Are you refering to the bypassing of the controls on his guitar or amp modifications?

 

[PRR]

> Are you refering to the bypassing of the controls on his guitar or amp modifications?

AFAIK, he likes to use one particular 5E3 he got for $50 in 1967 which does not sound quite like any other. The amp has been worked-in and runs 6L6 but isn't what you call "modded". (It does have a preset scheme but it is external.)

http://www.thrasherswheat.org/friends/amps.htm
http://www.thrasherswheat.org/ptma/equip.htm

"...a four input 1959 tweed Fender Deluxe. ...when you have the volume and tone all the way up, by turning that second volume knob up to about 10, it starts to fade away. But right before it starts to fade away, something else happens, and that has to be in exactly the right position. You can't breathe on it, or it ***** it all up."

To a tech-head like me, the 5F3 mixer is just funky and useless for precision work. Neil is not about precision, even though he can tell 115V from 120V in one twang. (I been playing with electric near as long as Neil, and it took me a while to notice my office was 109V....)

> an all octal, grid leak bias amp looked like fun...

Go hang at Hoffman's, some of them are crazy like you.

I will say that, within limits, I don't think specific topologies "matter". That's like asking if you compose a hit starting with the lyric, the melody, the beat, the bridge, the hook, or the cover art. I can take the greatest hook and make it dull. I suspect a good amp mechanic can make almost any amp topology play well.

 

[VacuumVoodoo]

I was taught many years ago that cathodyne is the circuit below i.e. a cathode follower directly coupled to a grounded grid amplifier stage. It makes a pretty nice wide band amplifier.
What many nowadays call a cathodyne phase inverter is a split load PI also known as "concertina PI". Wonder who/where/when started this confusion.

Cathodyne:

 

[PRR]

> I was taught many years ago that cathodyne is the circuit below i.e. a cathode follower directly coupled to a grounded grid amplifier stage. ...Wonder who/where/when started this confusion.

Perhaps whoever taught you.

Your picture is a Cathode-Coupled amp, and has been called many other things. It often caries Schmitt's name, O.H. Schmitt, “Cathode Phase Inversion,” J. Sci. Instrum., 15, March 1938.

http://www.aikenamps.com/cathodyne.pdf from Audio 1960 calls the split-load / concertina a Cathodyne without any hint that the name is new or novel.

Dynaco Mk II manual http://www.curcioaudio.com/st7_mnl.pdf calls spit-load a cathodyne, but there is no date on this copy.

 

[Hangman]

most magnatones I have seen are cathodyne.

 

[Joechris]

to hold a cathodyne on the very edge of grid blocking

Is there a way to control grid blocking? Could be useful in a guitar
distortion applikation.

jo

 

[pstamler]

If Fender is acting like they own the long-tailed pair, they've got another think coming. I believe they were used in the Acrosound Ultralinear II well before Leo put them into anything. And the biography of Blumlein I'm reading, although bollixed up on technical stuff, suggests that he was putting long-tailed pairs into radar equipment during World War II and perhaps earlier. They show up in early oscilloscope circuits, too.

A long-tailed pair without a negative supply -- now that might be a Fender innovation. Or is it?

Peace,
Paul

 

[VacuumVoodoo]

[quote author="pstamler"]

A long-tailed pair without a negative supply -- now that might be a Fender innovation. Or is it?

Peace,
Paul[/quote]

It is not. Connecting negative feedback to the common tail resistor instead of tube grid is probably his contribution, legend says he did by mistake.

 

[CJ]

whats up with the 82K vs 100 k plate resistors?

 

[dale116dot7]

That's my sound, man.

 

[dabo]

A phase inverter that has signal applied to only one of the inputs is a bit unbalanced, so adjusting the gain on the triodes to different values helps solve that.

I believe it goes something like this...

The signal on the side that has a cap from grid to ground is applied by the cathode voltage fluctuation driven by the other side (the one with signal applied at the grid). It's essentially the local negative feedback used as input. There are losses that need to be adjusted for.

If I'm wrong I'll take the beer punishment.

 

[Guest]

Take a punishment. :green:
Feedback on the 2'nd grid (like in modern opamps' usage) would cause compensation of even harmonics that Fender did not like.

 

[PRR]

> If Fender is acting like they own the long-tailed pair,

It is ancient history.

CBS paid a lot for Fender, then neglected it, then was shocked to notice it wasn't gushing cash. This would be mid-1970s. Among the minor-league plays which resulted was a general witch-hunt for other companies "stealing Fender intellectual property".

Leo's designs were sometimes unique, yet all were "obvious to one versed in the art". Likewise many modifications are "obvious".

The tone-stack everybody took from Fender suddenly got mis-wired (VOX did it wrong from the start). Traynor introduced a very whacked stack. Ampeg -may- have switched from longtail to cathodyne in that period.

What may have happened is a lot of lawyer billable time to send nasty-grams, quite modest design changes at the Other Guys, for a net loss to CBS.

In the last 20 years, everybody copies Old Fender, exact or modded. The current Fender organization seems to stick to hard-IP: tradename, trade dress. They apparently don't hassle small boutique builders even if a Fender logo winds up on a few non-Fender products.

The long-tail is old. Interestingly when good DC amps were needed in WWII, they sometimes used one 6V6 with G1 pulling current and G2 used as "input". The G1-K junction stabilized it against heater and aging drift. Sorta like we might use a BJT against a diode, with the advantage of having the exact same cathode for both devices. Oxide cathodes vary a LOT, even between two units in one bottle.

Self-bias longtail probably was not novel, though certainly not common. Longtails crop up in studio and hi-fi designs, but not self-bias, and none like Fender 5F6.

But the 5F6 puts speaker NFB into the bottom of the tone stack. It must have seemed reasonable at the time. That time may have been short...

It all comes together in 5F6-A. NFB hits the offside grid, but also goes common-mode up the longtail. The Presence pot adjusts, not just top-rise, not just HF THD, but the mix of odd and even partials in the top of the range. As a product, the Bassman 5F6-A was a yawn. The Fender Bass was and is a standard of the genre, but Ampeg generally had much better bass amps. But the 5F6-A is a fine wide-range amp, and one reason it was a so-so bass amp is speakers more suited for >82Hz abuse than deep rumble. As bassists moved to better amps, guitarists found the Bassman. Even then it might have been a momentary fad, except Jim Marshall thought UK import taxes on Fender gave him an opportunity. He had the 5F6-A cloned with UK parts and a tighter less-deep tuning (in part to suit closed-box speakers; I wonder if British conservatism disdained reckless open-back at high power). He sold a few to some punks, who sold a few concerts and records. Their real genius was their angst, but listeners thought the Marshall Amps played a part. Jim's daughter still runs the company, with some products as Leo-based as anything Fender sells now.

What other non-trivial 1950s circuit is still in use? The 5-tube radio? The Bell 500 telephone? The headlight dimmer on the floorboard? The latest AKG 414 has a little computer inside. I'm shocked to discover that the Marantz Model 9 (another longtail) has been "reissued". And there is an old-school Mac MC275 with XLR and IEC connectors. But nowhere near the numbers that 5F6-A descendants are turned out.

 

[walter]

[quote author="CJ"]whats up with the 82K vs 100 k plate resistors?[/quote] As I understand it, having matched plate resistors gives an even output, using an 82k, and a 100k (about 5v difference) knocks it out of ballance enough for a meaner tone :grin: Just one of the differences between silver and blackfaced. :sam: :sam: :sam: I am so sad that schematic heaven is not happening. :guinness: :guinness: :guinness: drown me sorrows.

 

[CJ]

schematic heaven is weird, sometimes i get in, sometimes not.
kind of like tan women. :twisted:


the last time, he siad his bandwidth was compromised by 100x traffic, so he was trying to up grade...

try this:

http://www.freeinfosociety.com/electronics/schempage.php?cat=1

 

[PRR]

> matched plate resistors gives an even output

No. The too-small longtail impedance sucks ~~10% away from the drive to the off-side cathode. Tube impedance stays about the same, and DC resistance is similar to plate resistance, so you need ~~20% different plate resistor to give "balanced" drive.

The NFB into the longtail appears to futher reduce even-order nonlinearity. But there are so many interactions happening that I don't trust either my cocktail napkin or my simulator.

 

[CJ]

how much pwr do the pwr tubes draw from the inverter to drive the 6L6GC grids ?

Must vary depending on the stomp being used...

how to calculate gain of inverter?

is it not supposed to be gain = 1 ?

 

[PRR]

> how much pwr do the pwr tubes draw from the inverter to drive the 6L6GC grids?

Simple computation. In guitar-amp work the peak swing will be around 40V per grid. Tube input is ~~200Megs + ~~50pFd. Grid-leak resistors run 33K-1Meg, though 220K gets used a LOT.

So 220K up to 14KHz, then falling.

About 4mW per grid at full undistorted sine. 8mW for square at edge of clipping. Maybe over 50mW each grid when driven 300% overdrive.


> is it not supposed to be gain = 1?

WHY?

There are great inverters which are unity gain. But isn't it sweet to get straing, inverted, and gain all in one lump?

The gain of this stage is ~~25 from one input to one grid. One plain bypassed 12AX7 will do 50, but we have both cathodes in series from one grid, so the gain to each output is half. The gain to both outputs is still near 50.

Gain depends on tube used. But low-Mu tubes give poorer balance.

Gain depends on load. 220K is out of spec for most power bottles run fix-bias, but they got away with it. Occasionally you see an over-honest design with 47K for two grids per side; 12AX7 won't drive that well. 12AT7 is also popular in this affair.