Princeton Reverb - PI Distortion?

[Fresh_Start]

Conventional wisdom has it that a flaw in the Princeton Reverb is that the PI distorts before the power tubes do.  I spent some time this afternoon with my oscilloscope but am not sure how to interpret the results.

Here's a Gallery of 7 o'scope shots which you can look at either individually or as a slideshow. [edit 10/4/10 - added 4 shots of the PI discussed below]  The first four shots have the phase inverter grid (V4 pin 7) on CH 1 and a power tube grid (V5 pin 5) on CH 2. At first I thought "this proves it," but I'm wondering if the flat top on CH 2 is PI distortion or the power tube shutting off as the amp transitions to Class AB.  There is something funky happening on the bottom of the CH wave as volume reaches a bit above 4 and even more at vol = 6.

What surprised me is that a lot of the distortion in the 1st 4 pics disappears when you look at the PI grid and the speaker jack output at the same time (last 3 shots).  Shot #4 and #7 both have the Volume at 6, but the speaker output looks a little smoother and not as kinky as the 6V6 grid does.  I don't know how much of that has to do with NFB and/or the push-pull nature of the amp.

So, does this demonstrate that proponents of the "Stokes Mod" to get more clean headroom out of the phase inverter have a point?  HBP asked for data a while ago, so here it is... I think.

It may be relevant that the cathode current indicates combined plate & screen dissipation of about 10.95 watts or 78% of the 14 watt maximum plate dissipation for a 6V6.

Cheers,

Chip

P.S. Please let me know if this thread belongs in the "Tools" section.

[darryl]

Merlin addresses these issues with the cathodyne PI, both in his book, and also on his website: http://www.freewebs.com/valvewizard/cathodyne.html

[Fresh_Start]

Merlin addresses these issues with the cathodyne PI, both in his book, and also on his website: http://www.freewebs.com/valvewizard/cathodyne.html

OK, so we can see Merlin's "nipple distortion".   :book1:

After re-reading Merlin's essay, I've got a couple more questions relating to the load line and the bias point.  Would increasing the plate supply B+ increase the splitter's clean headroom? 

Also, Leo Fender used a 1K cathode resistor, not 3.7K to 4.7K, so Leo's bias point is in a diffent place on the load line.  What's the impact of that?

I didn't plot the anode & cathode of the PI on top of each other, only the anode signal.  Will do so tomorrow.

Chip

[Fresh_Start]

Added 4 more pics to the gallery link in 1st post.  They show both the anode & cathode of the PI, going from clean to full distortion.  I see the "nipple" again but not the frequency doubling/mirror image Merlin talks about.

The triode is biased with +1.1 VDC on the cathode WRT the 1 meg grid return resistor.  This looks like just what Leo designed, although the cathode voltage is just under 60 VDC instead of 50 VDC as shown on the attached schematic fragment.  B+ is high throughout the amp - I assume due to higher AC supply voltage.

I'm not sure if any of this is meaningful, so please give me a hand if there's anything useful to get out of this data.

Cheers,

Chip

[PRR]

Remove NFB. Drive the power-amp gain-stage (may be V3A) grid directly (a decent signal generator can overwhelm the 3.3M+470K network). Monitor the OUTPUT. Bring it to just-clipping.

Are the 6V6 grid signals clipped?

Is the gain-stage plate clipped?

I suspect the cathodyne clips just a little -after- the 6V6es are slammed. You get all the power you paid for, but when you hit "MORE!", the driver doesn't over-do things to the point of grid-blocking.

_With_ NFB applied, the internal signals all interact and when clipped they should go wild. When an amp "will work" without NFB, it is much simpler to study core functions that way.

> Would increasing the plate supply B+ increase the splitter's clean headroom?

A properly biased cathodyne is a very clean thing. It is more likely to go soft in the gain stage than the cathodyne.

You can drive audio power pentodes "easy" if your driver rail is similar to your screen voltage. What do we have here? 400V screens and 240V to driver. That would be fine for a hot Class A amp (though you can't usefully load 6V6 in Class A at 400V). Since this is a relatively "cool" AB bias we need big grid swing and 240V (60% of screens) is kinda low. I think Leo knew what he wanted. Works for many users too.

I do think it would be quicker to move the wire than to sit around talking about it.

[Fresh_Start]

PRR - many thanks.  As always, it will take me a while to digest your post because of its density.

The last sentence is very clear

EDIT 10/5:  I've been doing these experiments on an original BF Princeton Reverb.  Once again, PRR has pointed out what should be an obvious part of the the circuit as a whole - negative feedback.  Since the PI is inside the NFB loop, there are indeed some strange things going on.  I'm not going to disconnect NFB in this old amp but do have my PR clone/derivative to play with later.

Besides, Doug's lightning delivery with the parts I need for the old amp's rehab got here yesterday.  I placed my order on Thursday afternoon.  He's sooooo slow... NOT!

Chip

[PRR]

> I'm not going to disconnect NFB in this old amp

In this case you can just short the 47 ohm resistor with a clip-lead.

[jjasilli]

I'd been thinking that cathodyne is fine, but gives a "ratty" overdrive tone, which is fine if that's your goal.  LTP gives a smoother more modern overdrive.  I tried the Stokes mod in my SF Princeton.  It gave more headroom but made the overdrive tone way too ratty, though I think the result can vary from amp to amp.  More NFB might help, but it would work against the Stokes mod. 

I loaned the Princeton to a friend (Danny Kalb formerly of the Blues Project).  He liked it so much he insisted on buying it and I couldn't say no. 

Anyway, if you're looking for creamier overdrive you might really want LTP.  One member here suggests an LTP conversion for the Princeton -- Bluesbear maybe??? 

[Fresh_Start]

I'd been thinking that cathodyne is fine, but gives a "ratty" overdrive tone, which is fine if that's your goal.  LTP gives a smoother more modern overdrive.  I tried the Stokes mod in my SF Princeton.  It gave more headroom but made the overdrive tone way too ratty, though I think the result can vary from amp to amp.  More NFB might help, but it would work against the Stokes mod. 

I loaned the Princeton to a friend (Danny Kalb formerly of the Blues Project).  He liked it so much he insisted on buying it and I couldn't say no. 

Anyway, if you're looking for creamier overdrive you might really want LTP.  One member here suggests an LTP conversion for the Princeton -- Bluesbear maybe??? 

Not sure what "ratty" overdrive sounds like

The cathodyne PI is part of the Princeton Reverb's sound.  Add a LTPI and a choke and you basically have a one-channel Deluxe Reverb, right?

Cheers,

Chip

[jjasilli]

"ratty" overdrive: compare Tweed overdrive to later, but pre-Metal, Mesa Boogie (sculpted cascading gain for smooth overdrive) for stark difference in overdrive tone.

cathodyne PI is part of the Princeton Reverb's sound.  Yes, change that to LTP & you've changed the soul of the amp. But my point is:  if you don't like the tone of the overdriven cathodyne's distorted waveform, then you can't fix it within the cathodyne circuit; you need a different PI circuit, most likely LTP.  

[tubeswell]

I have heard that adding a 470k-1M grid stopper to the grid pin of the cathodyne stage can smooth out the overdriven sound quite a bit, tho' I have yet to try this with my BFPR clone.

[jjasilli]

OK, back to the first sentence in this thread:  Conventional wisdom has it that a flaw in the Princeton Reverb is that the PI distorts before the power tubes do. I'd say this is a feature & not a flaw.  My overall take, subject to technical refutation:  These amps with simple topology were not designed for overdrive tone, which was then considered something to be avoided.  The topology is very simple; stage1 > tonestack > tone recovery stage > Power Amp, i.e. PI > Power tubes.  Just about the only point to get distortion is for the tone recovery stage to overdrive the PI.  

The cathodyne PI has only one gain boost stage (the second stage being only unity gain).  Because of tube curves and the load line, overdriving only one stage will make the signal asymmetrical.  The unity following stage can only duplicate that.  And then the power tubes are being fed the asymmetric signal (even if they are made to add their own distortion).  Asymmetrical distortion sounds rattier than symmetrical distortion, which sounds smoother.  

Solutions:  1.  If the PI is the only stage to overdrive, then use an LTP.  That way both haves of the signal curve wind up with symmetrical distortion, for a smooth overdrive tone; 2.  Per KOC use the power amp (PI & Power tubes) for a clean boost only.  For overdrive, this requires a more complex preamp with cascading gain stages.  From gain stage to gain stage the signal reverses phase, so both halves of the signal can be distorted symmetrically.  Absolute symmetry is not required; some asymmetry may add synergy or mojo.  But at some point too much asymmetry will sound displeasing.  This is the inherent problem with the cathodyne.  (If I’m beating a dead horse somebody please shut me up!   )

[SoundmasterG]

In a Bogen project build that I have been working on, I initially tried the cathodyne before switching to the LTP. I also have used the floating paraphase in a Vox project. I think the cathodyne sounds like it amplifies with a wider frequency response or emphasis. The highs seem to be crisper when the amp is clean. This is with all other things being equal and just changing the phase inverter. I could be wrong...but it was what my ears were telling me. In the Bogen I was able to bring back some of those highs in other ways, but I definitly liked the LTP better in it. That said, my 1956 5E3 sounds great with the completely stock cathodyne too. :)

[phsyconoodler]

Ratty overdrive? Hmm....I build a few Princeton circuit amps and they have the best cranked amp tone I've heard to date.I crank them and use the guitar volume.The overdrive is killer good.
  Mine use 6L6's but the basic circuit is the princeton reverb.
I have heard some PR's that don't sound that great overdriven.In my experience they need more mids.I haven't looked at the waveform on the scope yet but I sure like what I hear with just the use of a Raw control.
  I'll have to try Merlins use of 100k grid stoppers and see what it sounds like.If it's better my eyeballs will pop out,because the tone I get now is pretty damn good.
  I also built one with cathode bias and the overdrive is outstanding.
Maybe 6L6's vs 6V6's?

[tommytornado]

FWIW, I tried a 100K grid stopper to pin 7 of V3 in a 5F4 I finshed a few weeks ago.  I've read that most use a much greater value 470k-1M.  In that circuit, the grid stopper, to my ears brings a little more headroom and definition to the tone party as the volume is cranked up (when compared to the 3 5E7's I've built). 

[HotBluePlates]

Conventional wisdom has it that a flaw in the Princeton Reverb is that the PI distorts before the power tubes do.  I spent some time this afternoon with my oscilloscope but am not sure how to interpret the results.

Here's a Gallery of 7 o'scope shots which you can look at either individually or as a slideshow. [edit 10/4/10 - added 4 shots of the PI discussed below]  The first four shots have the phase inverter grid (V4 pin 7) on CH 1 and a power tube grid (V5 pin 5) on CH 2. At first I thought "this proves it," but I'm wondering if the flat top on CH 2 is PI distortion or the power tube shutting off as the amp transitions to Class AB.  ...

Maybe I'm missing something... Isn't the power tube grid on the lower trace? I see it distort first.

So my next question is what is the peak voltage when the bottom trace begins to show distortion, and how does that peak voltage compare to the bias voltage. *If* the peak voltage equals the absolute value of the bias voltage (e.g., 34v peak, -34v bias), then you are seeing the 6V6 grid being driven positive and conducting rather than being a super-huge input impedance.

EDIT: The point behind the above paragraph is that if the grid is driven positive and starts to conduct, then distortion viewed at that grid could be the result of grid current, and not actual distortion in the phase inverter. Grid current doesn't necessarily happen all at once either, but once the peak input equals the bias, it becomes very severe very fast. And in a case of the serpent eating its tail, if the output tube draws grid current, the load represented by the grid becomes very small, and the phase inverter could be pushed to distort by trying to drive a load smaller than it can handle.

You could measure what is happening with the output tube current by inserting a 1 ohm cathode resistor and measure millivolts across this resistor to watch the current through the output tube (or use 10 ohms, remembering to multiply voltage readings by 10).

[Fresh_Start]

HBP - good question and perfect timing.  I was experimenting with a new set of power tubes tonight, so my Hoffman bias checker is connected to the amp right now.

Don't know if I'll have time this week to hook up the oscilloscope though...

Cheers,

Chip