post PI gain stage?

[shooter]

found this schematic in Dougs library, Stromberg-Carlson ap-80.  Is V2 a post PI gain stage?   

http://el34world.com/charts/Schematics/files/stromberg-carlson/stromberg_carlson_ap80.pdf

[sluckey]

Yes. You see that a lot in quad big bottle amps.

[PRR]

It is a Williamson. An amp-classic worth looking up.

Well, THE Williamson used 12AU7-like tubes in all small holes, had triode or triode-strapped output section, and a very elaborate OT. "The Williamson" known from millions of 1950s kits often used Ultralinear output stage; because very-good UL OTs were readily available out of Philadelphia and the power output number on the same tubes got a bit higher.

The Williamson used "post PI" stages because it needed BIG grid swings and aimed at very low THD. The AP-80 needs like 40V swings into 56K grid resistors, which is pretty hefty for a cathodyne. It is maybe generous to add a booster, but this is a *150 Watt* beast, musta cost a lot, so the added frills were accepted.

Much later Bogen got the same output with a AX/AU driver and 8417 output bottles (smaller grid swings).

[HotBluePlates]

And typically the post-phase-inverter gain stage would be a differential (almost long-tail-looking) kind of gain stage.

However, that Stromberg-Carlson has negative feedback from the plates of the output tubes to the cathodes of that driver gain stage. A differential amp shares a cathode resistor; if you had feedback as in the AP-80, the 2 feedback signals would tend to cancel each other, defeating the point of the feedback.

[shooter]

it needed BIG grid swings

That was my guess.  Just the 1st schematic I came across with a P-PI gain stage.
So is Williamson an engineer or company?

the 2 feedback signals would tend to cancel each other

couldn't I " look at" the NFB as * stand-alone* for each side, sorta like a self biased PP amp

Thanks again for your time n trouble

dave
 

[jazbo8]

Williamson Amplifier (1947)

[shooter]

thank you

[HotBluePlates]

the 2 feedback signals would tend to cancel each other

couldn't I " look at" the NFB as * stand-alone* for each side ...

Look at it as feedback from the output tube to the gain stage driving that output tube. The point is to linearize the driver-output stage, and lose some of the excess gain not needed to drive the output tubes.

My bigger point in that post is that typically a differential gain stage is used as a driver after a phase inverter because it has push-pull input and push-pull output. Additionally, a good enough differential amp would tend to enforce equal drive signals to the halves of the output stage.

[jjasilli]

I'm intrigued by the negative feedback loops in this amp.  Also, I've been reading-up on the old electronics magazines from the mid to late '50's.  They state the purposes of inverse feedback (as it was sometimes called then) are to stabilize the amp and damp the speaker (I guess to reduce resonance and ringing); and to substantially increase the frequency response of a gain stage, though with a consequent loss in gain; or both.


There are two feedback loops in this amp.  One goes from a dedicated OT secondary winding to the first stage.  I'm guessing the primary purpose of this loop is to increase the frequency response of the 1st stage, but at the cost of much if its gain.  This in turn drives a split PI, presumably with unity gain.  On the plus side such a PI has great frequency response with little distortion.  The voltage needed to drive the power tubes comes from the next dual gain stages.  (Also as PRR pointed-out, those gain stages are heavily loaded by the small  grid leak resistors shared by each pair of power tubes; another reason to build-up a large driving voltage.)


A second negative feedback loop feeds these gain stages from the power tube plates.  I'm guessing the primary purpose of this loop is for speaker damping / amp stabilization.

Comments are welcome!

[PRR]

I am thinking the 100K:1.2K 6550-12AU7 loop mostly damps the supersonic resonances of that huge OT.

The tertiary winding to first stage cathode stabilizes and damps the *whole* amplifier, same as our speaker to first cathode NFB connections. Difference is that in a PA amp it may be desired to let the load winding "float", or be grounded at some arbitrary point (like the center). So overall NFB can't easily be taken from the load winding. The tertiary winding does the same thing, except it does not compensate for the small winding resistance of the load winding. (Damping Factor can't be super-low by hi-fi standards, but it sure can be low enough for PA work.)

[shooter]

Comments are welcome

I made it to page 12 of jazbo's link and most of it was along those veins.  My head started swimming, so I got a green beer and guitar to stop the spinnin!   

Thx HBP, learning ?, is a para-phase PI considered a deferential gain arrangement? 

back to fun playing badly 

[PRR]

> I made it to page 12

That's the meat of it.

This article spurred an enormous response, and the other 20+ pages are details and preamp suggestions.

[jjasilli]

I didn't read Jazbo's link.  However, as shooter points out, much of this stuff is for hi-fi (or mid-fi, anticipating a correction from PRR!!).  I.e., not directly related to guitar amps.

[HotBluePlates]

I am thinking the 100K:1.2K 6550-12AU7 loop mostly damps the supersonic resonances of that huge OT.

Thanks, I should have seen 100k:1.2k is a big enough ratio that there's only a touch of feedback for the stages in use.

... is a para-phase PI considered a deferential gain arrangement?  ...

No, a paraphase is a paraphase. 

The article points out a paraphase could replace a phase splitter (the no-gain split-load in this case) and push-pull driver stage (which is the shared cathode-resistor push-pull stage in the Williamson) because you can get a paraphase to effect phase inversion and provide gain. Then Williamson goes on to not use such a stage.

... much of this stuff is for hi-fi ...  I.e., not directly related to guitar amps.

It does pertain to identifying requirements and selecting a solution to meet your requirements. On that standpoint, it's educates the reader about amp design, regardless of whether the design is guitar or hi-fi related.

[shooter]

it's educates the reader about amp design

+1
And ya PRR, when I started seeing Greek symbols, I knew it was time for a pause   

my "view" of differential is something that takes a signal, and makes 2 out, sometimes changing dc levels, sometimes with gain, like line drivers (think rs232).  Or in Rf taking the incoming signal, split, encode, re-combine.  That's why I was thinking PI *fit the bill*, going 1 further, the OT *acts* to re-combine the differential signal
Thanks all for "the skoolin" 

[PRR]

"Differential" can mean about any dang thing.

It more often is used for Differential Input. The Fender Long-tail is a differential amplifier. One input takes signal, and the other side takes (a fraction of) the speaker voltage. The amplifier works to make them equal, which lowers distortion. (Fender's actual implementation may not do this.) There are other amps which take two input channels, or an input and a reverb return, to the two inputs.

Differential output is a tricky thing and needs much qualification. Cathodyne, Paraphase, and LongTail make two outputs, nominally equal and opposite, but the *difference* between them is not well controlled. Loading or shorting one output does not make the other output rise to keep the *difference* the same. There are such amplifiers; they are often quirky.

[HotBluePlates]

I, too, focus on the input when talking about a differential amp.

The principle characteristic of a diff-amp is that it responds to the difference of 2 input signals; more specifically, if the 2 input signal are identical (polarity and amplitude) then there is zero output. This is why I was saying think of the diff-amp as push-pull in, push-pull out. Even more directly, if you applied the exact same input signal to both inputs (a common-mode input signal), there's zero output. A shared constant current source in the tubes' cathodes (faked to varying degrees of success by a large resistance to ground or to a large negative voltage supply) acts to force equal and opposite tube current, and so equal and opposite output voltages.

The long-tail is something of a half-hobbled diff-amp. It responds to the difference of input signals, but we typically only drive one side instead of both (ignore the feedback for a moment). The 2 input signals are still "differential" because one is tied to a.c. ground. The drawback of this arrangement is that the gain to a single output is half what that tube would normally manage; however, it does a great job of taking a single input and generating push-pull outputs (though the balance isn't perfect without jiggering). The guitar-amp long-tail then uses the un-driven input for injecting feedback, for the purpose PRR noted.

[shooter]

Thanks, for the info, so far I've only built 2 PP amps and I wimped out and used inter-stage PI trannies, ala Gibson, hence my knowledge void on tube based PIs.