'62 Princeton Phase Inverter Problem

[PlumBob]

Hi:
So I took my brown Fender Princeton to my tech for a general looking over and to see if he could do anything about a kind of harsh fuzzy nastiness I was hearing when the amp distorts. He identified the problem. The signal is clipping much earlier and much more sharply on one side of the phase inverter than on the other side. Does anyone know what would cause this? He tried a bunch of remedies and tested all the caps for leakage and such but so far hasn't been able to fix the problem. Any ideas of what to check?
Thanks.
-Matt

[Blue Strat]

When you say that this guy's "your tech", what exactly are his qualifications?

Has the amp every worked correctly for you? If so, what did you change if anything?

I assume he's tried replacing all tubes? How does he know one side is clipping more? Has he isolated the PI from the power section to see if the problem persists? Has he checked all voltages against the schematic? Subbed in new caps around the PI? Checked all resistor values around the PI?

[PlumBob]

The tech is a local guy who has done some good work for me. This problem with the Princeton has been there since I bought the amp. This is the first I've had someone look at it.

As far as I know, my guy has not isolated the PI from the power section, so that's one thing. Here's what he says about what he has tried so far:



The phase inverter of the Princeton is distorting the signal on one
side of the signal way before the other. The one side clips soon and
relatively abruptly as the amplitude of the signal is increased. The
other side squashes gradually with increased amplitude.

This is what I did to try and fix the problem:

I checked the coupling caps for leakage with a voltmeter, and also
temporarily replaced them . No change.

Replaced the 56K ohm plate and cathode resistors with 100k ohm
resistors to see if that would 'center' the waveform for symmetrical
clipping. That helped a little, but there was a loss of gain.

Used a variable resistor in place of the 1K ohm cathode resistor to see
if I could change the bias point. Change occurred, but in the wrong
direction.

All B+ voltages are around 25-30% higher then listed on the schematic
(filament voltage is only 5% higher). Temporarily put the amp on a
variac to bring the voltages down. This didn't help.

[Blue Strat]

I don't see anything there about trying different tubes, the most obvious thing to try.

[PlumBob]

He did try changing out the tubes. So what about your idea of isolating the PI from the power section and testing it like that? What would be involved in that? And what would it show or rule out?

[ChickenLover]

That amp uses a cathodyne splitter. If you search the net you'll find some mods for BF Princetons that would work here.

Being a vintage piece...I will guess that you want simple, reversible mods but here's a suggestion: tell your tech to set up the cathodyne in 'fixed-bias' and then maybe play with the bias a bit. All you do is move the connection of the existing 1M grid resistor straight to ground and then add a high-value resistor from B+ to the grid (try 2.2M). So you've basically put a voltage divider from B+ to ground and connected the grid to it...that's your fixed-bias. Now you can remove that 1K cathode resistor and leave both 56K resistors...that's pretty much it. Then you can play with the bias a bit to see if you can get it sounding better. This is a common BF Princeton mod (there's an internet name for it but I can't remember it...I did mine well before I had heard the 'name' for it). In fact...this might be called the 'Paul C' mod...Paul C being the guy from Tim pedal fame...or maybe it's the Stokes mod. One of these mods just increases the voltage in a BF Princeton a bit but that doesn't look like an option in your amp. HTH.

[Wakarusa]

I'm with Mike -- it sounds like the problem is most likely in the power section. If one side of the PI clips hard and early it could be a sign of grid conduction in the associated power tube. I'd check/replace the power tubes and also look for bad components/joints in the 220K bias feed and the 1K5 grid stopper. Could also be caused by DC leakage in the .1uF PI output coupling cap on the affected side.

[Old Tele man]

...I'd pull the power tubes and "o'scope" the PI signals to see if they were "clean & symmetrical" or "dirty & asymmetical" going into the power tube control grids...if yes, then the power tubes are the most likely the problem.

[Wakarusa]

...I'd pull the power tubes and "o'scope" the PI signals to see if they were "clean & symmetrical" or "dirty & asymmetical" going into the power tube control grids...if yes, then the power tubes are the most likely the problem.

And while you're in there, it wouldn't hurt to check DC levels on power tube grid pins for balanced and correct bias voltages.

[Old Tele man]

And while you're in there, it wouldn't hurt to check DC levels on power tube grid pins for balanced and correct bias voltages.+1, ie: "next step" if PI signals were cruddy...

[Wakarusa]

+1, ie: "next step" if PI signals were cruddy...

Just to pick nits I think I'd respectfully disagree and say the check is required in any event (particularly if the PI signals were not cruddy). Assuming we don't have a total failure of the output coupling cap, but instead just enough DC leakage to throw one power tube out of bias the PI operation would appear (at least close to) normal since the leakage would draw almost no current with the power tube removed. Only with the power tube installed and DC leakage throwing the bias off to the point where you get grid conduction/clamping/etc. in the power tube would you see the distortion in the PI from trying to drive the very low impedance forward conduction of the power tube grid.

'least that's how I see it.. :)

[Old Tele man]

oops! ...should've been: "next step" if PI signals WEREN'T cruddy.

...mea culpa!

[Luke Duke]

Wasn't a big problem with the Cathodyne PI that it always assymetrically clipped?

The thing one would have to consider is what would exacerbate the clipping? So I'd go with what someone above said about checking the coupling caps.

Luke

[Wakarusa]

Wasn't a big problem with the Cathodyne PI that it always assymetrically clipped?

The thing one would have to consider is what would exacerbate the clipping? So I'd go with what someone above said about checking the coupling caps.

Luke

Whether it's a problem or not is a matter of taste and degree ;)

From what I've read the big reason for developing first the paraphase and then the long-tailed pair phase inverters was the cathodyne's limited output voltage swing -- basically half that of the other two designs all else being equal. The goals, of course, were to minimize distortion so the objective was to never have the PI clip at all.

[Old Tele man]

...the longtail "Schmidt" PI provides GAIN, whereas the split-load "Cathodyne" PI does NOT provide gain, because it's just a combined anode-follower/cathode-follower...with less than unity gain on both sides.

[PlumBob]

Hey, thanks for all the thoughts. I forwarded them along to my tech, and he checked some more stuff and then did the Paul C. mod. He still couldn't stop the one side of the PI from clipping kind of hard, but he did improve the sound of the amp quite a bit. I just got it back. The clean sound is now stronger, and the distortion is more pleasing. Still a little harsh and buzzy though. I wonder if a Brownface Princeton just doesn't love to be pushed (especially with humbuckers). Anyway, that's my report. Thanks again.

[Luke Duke]

Whether it's a problem or not is a matter of taste and degree ;)

From what I've read the big reason for developing first the paraphase and then the long-tailed pair phase inverters was the cathodyne's limited output voltage swing -- basically half that of the other two designs all else being equal. The goals, of course, were to minimize distortion so the objective was to never have the PI clip at all.

Wait a sec....I'm confused. Everything I've read said the Paraphase came first, then the split load, or Cathodyne, after that the Schmidt or long tailed pair.

From what I understood the Cathodyne addressed the power handling problem that the Paraphase had, as well has clipping problems.

The LTP addresses the assymetrical clipping issue, and as someone else mentioned it provides gain.

That all said, each PI design has uses that people love. Look at a late tweed bassman or plexi, or BF Fender for LTPs in use. For Cathodyne look at the lower wattage tweeds up to the lower power twin. Also Orange used that same PI. The earliest Fenders used Paraphase inverters (again IIRC)

I'm a novice at best with tube amp design, so feel free to set me straight if need be.

Luke

[Wakarusa]

Wait a sec....I'm confused. Everything I've read said the Paraphase came first, then the split load, or Cathodyne, after that the Schmidt or long tailed pair.

From what I understood the Cathodyne addressed the power handling problem that the Paraphase had, as well has clipping problems.

The LTP addresses the assymetrical clipping issue, and as someone else mentioned it provides gain.

That all said, each PI design has uses that people love. Look at a late tweed bassman or plexi, or BF Fender for LTPs in use. For Cathodyne look at the lower wattage tweeds up to the lower power twin. Also Orange used that same PI. The earliest Fenders used Paraphase inverters (again IIRC)

I'm a novice at best with tube amp design, so feel free to set me straight if need be.

Luke

I could easily have my ordering askew regarding the paraphase and the cathodyne and admit to being too lazy at the moment to go look it up :(

One huge advantage of the cathodyne is that it uses only a single triode, but that gets a bit murky too because of the lack of gain -- so you end up seeing a lot of pentode/triode tubes (7199, 6AN8, etc.) doing PI duty with the pentode for gain and the triode for phase inversion. I think the cathodyne balance problems are also mitigated when run within modest specs -- but that doesn't really apply to a cranked guitar amp ;)

I'm personally a huge fan of paraphase in guitar amps (a 5D4 Super is a thing of absolute and utter beauty) because of the way they unbalance under different operating conditions. Part of the magic that's in some of the old Fender tweeds that disappears with the LTP amps is that the PI could be pushed into distortion adding to the overall distortion characteristics of the amp. This is also one of the reasons I think the hi-fi crowd walked away from them. Next, it's a pain to design a decent voltage divider to drive the non-inverting tube that stays balanced down the line (either due to signal conditions or all of the usual suspects like age and drift). You can put in all sorts of tweakables to adjust it, but a decent LTP design eliminates the hassles. Too, the part count and some of the design considerations get a bit stickier when you try to use a paraphase with a fixed bias power section.

And two final points -- First, the asymmetry is typically present in all of these to varying degrees whether the stage is clipping or not (and yeah, I get that asymmetry before clipping definitely implies asymmetry at clipping). However, the wunderkindern who thought 'em up assumed that nobody would use 'em clipped :)

Second, to the original poster -- I'm beginning to wonder just how severely different the clipping is in your PI. There's a distinct possibility that it's operating normally and you just don't much care for a brownface princeton :)

[Luke Duke]

I could easily have my ordering askew regarding the paraphase and the cathodyne and admit to being too lazy at the moment to go look it up :(

One huge advantage of the cathodyne is that it uses only a single triode, but that gets a bit murky too because of the lack of gain -- so you end up seeing a lot of pentode/triode tubes (7199, 6AN8, etc.) doing PI duty with the pentode for gain and the triode for phase inversion. I think the cathodyne balance problems are also mitigated when run within modest specs -- but that doesn't really apply to a cranked guitar amp ;)

I'm personally a huge fan of paraphase in guitar amps (a 5D4 Super is a thing of absolute and utter beauty) because of the way they unbalance under different operating conditions. Part of the magic that's in some of the old Fender tweeds that disappears with the LTP amps is that the PI could be pushed into distortion adding to the overall distortion characteristics of the amp. This is also one of the reasons I think the hi-fi crowd walked away from them. Next, it's a pain to design a decent voltage divider to drive the non-inverting tube that stays balanced down the line (either due to signal conditions or all of the usual suspects like age and drift). You can put in all sorts of tweakables to adjust it, but a decent LTP design eliminates the hassles. Too, the part count and some of the design considerations get a bit stickier when you try to use a paraphase with a fixed bias power section.

And two final points -- First, the asymmetry is typically present in all of these to varying degrees whether the stage is clipping or not (and yeah, I get that asymmetry before clipping definitely implies asymmetry at clipping). However, the wunderkindern who thought 'em up assumed that nobody would use 'em clipped :)

Second, to the original poster -- I'm beginning to wonder just how severely different the clipping is in your PI. There's a distinct possibility that it's operating normally and you just don't much care for a brownface princeton :)

Dang! Thanks Todd!

I'm trying to recite from memory the history in the Hunter Tube Amp book, so I could very well be wrong, and it's out on loan right now.

I do remember that it was here on TGP that I saw the explanations on why matched PI tubes weren't necessary due to the fact that guitar amp circuits are never symmetrical. So you figure that unles you push the Cathodynes pretty hard they don't perform as well as they could?

I also agree that the clipping is a problem only for someone wanting clean clean clean like Leo Fender. I've begun to wonder if that's part of the magic of a Princeton Reverb.....no LTP. I don't exactly have one to do experiments on though....nor am I sure I'd want to! :p

My buddy has a 52 Super Clone with the Paraphase Inverter (IIRC) and it makes for some SMOKIN interaction between the pre and power amps. You know there's just nothing to those circuits so all you get is guitar and amp goodness.

Here's a question for you Todd how much do you think the octal pres with the grid leak bias contribute to that tone?

That Super's tone made me more than a little curious about octal pres.

Luke

[Wakarusa]

So you figure that unles you push the Cathodynes pretty hard they don't perform as well as they could?


It's all about balance and goals. The cathodyne PI is a big part of why a Princeton behaves differently than a Deluxe (and a HUGE part of the classic 5E3 Deluxe). So I'm going to chicken out here and ask what you mean by "as well as they could"?

Keep in mind too that a cathodyne PI run in a well designed/balanced circuit can fit nicely into a "clean" tonal design. By balance I'm talking more about the balance between the different stages of the preamp and how the PI fits into that. As just typed in another thread, keep in mind that all kinds of big clean Ampegs use a cathodyne PI.


Here's a question for you Todd how much do you think the octal pres with the grid leak bias contribute to that tone?

That Super's tone made me more than a little curious about octal pres.

Luke


I'm gonna treat that like two questions ;)

I'm not a fan of grid-leak bias on a preamp stage regardless of the tube used. The grid-leak setup requires an input capacitor to isolate the grid leak resistor. I find that, for my tastes anyhow, the cap knocks the life out of things, particularly when you're trying for clean 'n sparkly. Note too that there's nothing stopping anyone from cathode biasing an octal twin triode.

Octal pres are, well, different. Leo liked a 6SC7 and later shifted over to 12AY7 and they just don't sound the same. Other octals have pretty close 9-pin equivalents (like 6SN7 vs 6CG7) both in terms of electrical characteristics and how they sound. The good is that an octal like a 6SC7 or 6SL7 sound a bit different than the stuff you usually hear. The bad is that, particularly in a combo, they can suffer from microphonics. Also, some of them are getting pricey. At last check NOS RCA 6SC7 were going for around $30/ea. So the answer is that it's really a matter of taste. For me, personally, the ultimate bad-ass tweed tone is a 5E7 bandmaster preamp with some tweaks feeding a 5D4 (or 5E9, or 5D8... you get the idea) style power amp.

[Luke Duke]

Once again, thanks for sharing your knowlege with a newb Todd.

It's all about balance and goals. The cathodyne PI is a big part of why a Princeton behaves differently than a Deluxe (and a HUGE part of the classic 5E3 Deluxe). So I'm going to chicken out here and ask what you mean by "as well as they could"?


All I meant was that they don't perform optimally (for guitar) unless driven kind of hard.

Keep in mind too that a cathodyne PI run in a well designed/balanced circuit can fit nicely into a "clean" tonal design. By balance I'm talking more about the balance between the different stages of the preamp and how the PI fits into that. As just typed in another thread, keep in mind that all kinds of big clean Ampegs use a cathodyne PI.

Honestly I haven't looked at the schematics or layouts on too many ampegs....but I would have guessed (incorrectly) that Hull used the LTP. That's very interesting.



I'm gonna treat that like two questions ;)

I'm not a fan of grid-leak bias on a preamp stage regardless of the tube used. The grid-leak setup requires an input capacitor to isolate the grid leak resistor. I find that, for my tastes anyhow, the cap knocks the life out of things, particularly when you're trying for clean 'n sparkly. Note too that there's nothing stopping anyone from cathode biasing an octal twin triode.

I'd never thought of it like that...I guess there is no reason to use grid-leak unless the characteristics that it lends are part of your tonal goals.

Octal pres are, well, different. Leo liked a 6SC7 and later shifted over to 12AY7 and they just don't sound the same. Other octals have pretty close 9-pin equivalents (like 6SN7 vs 6CG7) both in terms of electrical characteristics and how they sound. The good is that an octal like a 6SC7 or 6SL7 sound a bit different than the stuff you usually hear. The bad is that, particularly in a combo, they can suffer from microphonics. Also, some of them are getting pricey. At last check NOS RCA 6SC7 were going for around $30/ea. So the answer is that it's really a matter of taste. For me, personally, the ultimate bad-ass tweed tone is a 5E7 bandmaster preamp with some tweaks feeding a 5D4 (or 5E9, or 5D8... you get the idea) style power amp.

Yeah Bruce Collins told my buddy that if it gets to the point where he can't find any glass envelope 6SC7s he should just put the amp up till he can find some because the metal ones have terrible tone and microphonics. Honestly that's why I'd be scared to use octal pres in a build.

It's funny you mention a bandmaster because I've been toying with the idea of building a modded bandmaster on and off for almost a year now. This is not helping my GAS! :rotflmao

Luke

[Wakarusa]

More on PIs:
It's worth a peek at some Ampeg and Univox (and hi-fi) schematics since they dovetail nicely into what's been discussed here. A common setup is to use a pentode-triode (6AN8, 7199, etc.) in the PI slot. The triode is set up as a cathodyne PI with the pentode to provide gain. You'll also see lots of creative ways to balance the two signals from a cathodyne PI if you poke around long enough. Anything from different values for the plate and cathode resistors to using a following driving stage with a pair of triodes set for different gain (one vote for "most unusual" is a hi-fi design that followed the cathodyne PI with a 12DW7). And while we're talkin' PI balance vs gain it's worth a poke through some LTP designs paying attention to the selected values for the tail and plate resistors. A bigger Rt makes it behave more like a current source and yields better balance but at the cost of reduced gain (because effective plate voltage is reduced by the elevated cathode voltage from the tail resistor). Smaller Rt values allow for more gain at the cost of imbalance -- which is why your typical Fender (and Fender-based designs) have different Rp values for the two halves of the circuit. The mantra continues to be that it's all a matter of taste and design goals.

On bias:
I just don't have a good explanation for why early amp designs use grid-leak. Could have just been a habit. The input coupling cap and the requirement for very large valued Rg just makes me stay away from this design.

On roll-yer-own tweeds:
Heh. Don't get me started.... ;)

[hasserl]

I'm thinking of experimenting with a 6AN8 on my 5E5A. The stock gain level is just a tad less than I'd like, especially when using my Tele. Of course the Bass pot is key in this circuit, the gain loss thru the EQ is highly dependent on where you set this control. Turning it down reduces the gain a lot, but turning it up too much gives a blatty & muddy tone. A little mid scoop here would probably help. But I'm thinking that a little more gain from the stage just before the PI may give me what I'm looking for. Anyway, I've got a few on order headed my way.

[Wakarusa]

I'm thinking of experimenting with a 6AN8 on my 5E5A. The stock gain level is just a tad less than I'd like, especially when using my Tele. Of course the Bass pot is key in this circuit, the gain loss thru the EQ is highly dependent on where you set this control. Turning it down reduces the gain a lot, but turning it up too much gives a blatty & muddy tone. A little mid scoop here would probably help. But I'm thinking that a little more gain from the stage just before the PI may give me what I'm looking for. Anyway, I've got a few on order headed my way.

Interesting solution to your problem. You've got your work cut out for you. The triode in a 6AN8 is close to 1/2 of a 12AU7, but with lower plate resistance and a touch higher amp factor. It should do fine in the cathode follower slot, but you'll need to tweak Rk.
The pentode's gonna be a pain. If you run it in pentode mode you need to rig up a screen supply from somewhere. I'd have to sit down and do the math, but I wonder how the much higher plate resistance of the pentode will affect things when DC coupled to the cathode follower. If you're planning to run it triode connected then...

Wouldn't it be simpler to substitute a 12AX7 in the tone stack driver slot? If that doesn't get it done, another 12AX7 in the first preamp slot will boost things again. A couple of custom builds using basically this preamp and a cathode-biased tweed-inspired power section performed really well with Russian Tung Sol 12AX7 in all three slots.

[hasserl]

I've got 12AX7's throughout the preamp now. This is not a high gain amp, my Super Reverb has quite a bit more ooomph. I've got a nice Alamo here with a 7199 in pentode mode driving the triode as a cathodyne PI, it works very well. (I actually prefer that amp to the 5E5A, more grind & better tone) I can draw from that circuit for inspiration. Experimenting is not work, it's fun.

[ChickenLover]

Question: How would a 12DW7 work as a cathodyne twin triode with the 12AU7 half for the CF and the 12AX7 half for the AF?

I know a 12DW7 works great for FX loop Send/Return and for the GS/CF pairs you see in Marshalls (and many other amps) when you arrange it such that the 12AU7 half is for the CF. I always wire V2 (or any GS/CF pair) on my Marshalls 'backwards' so that I can use a 12DW7 there...and my FX loop Send/Return stages are usually wired that way anyway.

[ChickenLover]

Question: How would a 12DW7 work as a cathodyne twin triode with the 12AU7 half for the CF and the 12AX7 half for the AF?I usually don't quote myself but my browser locks up if I try to edit a post but anyway:

Maybe I have the bolded part backwards...in the case of the GS/CF pair and FX loop Send/Return we NEED the gain on one of the triodes to be up near a 12AX7 but here we don't (correct?)...so since the CF is already lower impedance than the AF (that is the case isn't it?) then maybe we should use the 12AU7 half for the AF since the 12AU7 half has lower plate impedance than the 12AX7 half.

[Wakarusa]

Once you arrange things with a dual triode I'm not sure it's fair to call it a cathodyne any more. I'm also trying to imagine how you'd wire this thing up... like a cascode circuit, but pulling a signal off the cathode of the driving triode? Or just as two separate stages, one inverting, the other non-inverting?

Not sure either way would work. In the former case, just guessing here, but I'd think you'd have problems keeping everything biased correctly (and this really isn't what cascodes are good for anyhow...). In the latter case you'd have pretty punishing imbalance since the gain of a cathode follower is always < 1 and the gain of your anode follower is always > 1. Unless I'm missing something I just don't see an advantage here over established twin-triode solutions (like a LTP ferinstance).

As far as driving tone stacks and effects loops and just general application of a 12AU7 (or a triode with 12AU7 characteristics) in cathode followers that all makes sense... the relatively high dissipation allowed and lower plate resistance make a 12AU7 more proficient as a current source for driving low impedance loads -- particularly when compared to a 12AX7 (which is, natch, more proficient as a voltage source).

The down side in all of this is that so many of the clever tubes that were commonly available 30 years ago are no longer in wide production (or, in some cases, in production at all). So while it's nifty to prototype interesting things that take advantage of the odd tubes you have laying around its nothing that can go to widespread use because there just aren't tubes to feed the demand (and I certainly don't have the hubris to assume our new designs will be so compelling that the tube manufacturers will ramp up production to allow them to spread ;))

[ChickenLover]

Todd, Sorry...total brain fart. You can basically just forget most of what I wrote above regarding using a 12DW7 in a cathodyne except use the 12AU7 half for the cathodyne and use the 12AX7 half for the gain stage preceding the cathodyne. Would using the 12AU7 half for the cathodyne be a possibility?

I might try this in my early SF Princeton when I return from vacation in a couple of weeks. I just got a 'scope and want to scope that PI anyway.

Ei makes current production 12DW7 don't they? In any case, I see your point but all my stuff is one-off 'just for me' stuff. I'm just a hobbyist/enthusiast and don't make stuff or even do any work for others (I won't even bias my friend's amps!).

[Wakarusa]

Todd, Sorry...total brain fart. You can basically just forget most of what I wrote above regarding using a 12DW7 in a cathodyne except use the 12AU7 half for the cathodyne and use the 12AX7 half for the gain stage preceding the cathodyne. Would using the 12AU7 half for the cathodyne be a possibility?


Oh sure, you can do that all day :) Essentially substituting the hi-mu triode in the 12DW7 for the pentode in stuff like a 6AN8 or 7199. You'll want to calculate new values for Rk and Rp since I doubt 56K (the Fender standard for a 12AX7) is what you want. Probably also useful to do the math (or spice simulation) to make sure the setup will swing the voltage you want, keep current draw reasonable, etc.

[epluribus]

Keep in mind too that a cathodyne PI run in a well designed/balanced circuit can fit nicely into a "clean" tonal design. By balance I'm talking more about the balance between the different stages of the preamp and how the PI fits into that...

Hey Todd.

Color me tuned in here, this whole issue of PI architecture and tuning is pretty fascinating. I hear ya about cathodyne-equipped amps (the Univox you just worked over for me) being capable of some pretty impressive cleans.

If I'm getting this, what we're really talking about (above quote) is the gain structure of the preamp stages and the PI, and how you can (and indeed did) use it to sculpt saturation in various parts of a circuit.

--Ray