Princeton Reverb AA1164 Modified

[Ed_Chambley]

I have a Silver Princeton Reverb (not original) from someone who wants me to convert it to one like I built.  Only thing, he wants to add the stokes mod and trim bass.  I made some revisions to the schematic.  The only revisions are in red.  Best I can understand is the stokes mod requires a upgraded PT and OT which I have.  And simply moving the the PI plate up one node.  I have never done this, but I think the change in caps will trim the bass.  I don't know if anything else is necessary for the stokes.

Please review and let me know what I am going to blow up.

Thanks

Add a jpg so everyone can see it without download.

[Willabe]

Hi Ed, where you been?    

V4b plate 56K PSU connection is wrong. It's a boot straped split load, you moved the plate B+ feed but left it's plate R back at the old (?) B+ node and it's not connected to V4b plate. So there's no signal feeding V5 coupling cap.

It's hooked up like a CF the way it's drawn.


             Brad    

[Ed_Chambley]

Working Brad.  Too much to do right now.  I'll be back more.  I'll make a change and see if I understand.  I read the about the stokes and what I read it bypasses the 56k.

Here is another attempt.  I was also wondering about the change in cap values.  Do not know what will happen here.

[Willabe]

I need to add a wire from the 18K to the 56K?

Yes and hook it up to the plate.

I don't see how you can bypass the plate load 56R? It's a PI you need both halves of the signal to feed the power tubes coupling cap, 1 for each side of the push/pull output stage.

                
                      Brad          

[Willabe]

That looks right to me.

Working Brad.  Too much to do right now.

That's good. A lot of people wish they were in your shoes right now.
            
                
                        Brad      

[plexi50]

I used a Deluxe reverb output tranny in my princeton reverb. Even though i gained a little more headroom i lost that sweet compression that the Princeton reverb is known for. I put back the princeton OT and im happier knowing that the amp is what it is and sounds great as it was built in them old days. JMO

[HotBluePlates]

If you want to trim bass, the most organic way to do it is to drop the 22uF bypass caps to 1-2.2uF. You won't need to do it on every stage. Try a few and taste-test.

[Ed_Chambley]

I used a Deluxe reverb output tranny in my princeton reverb. Even though i gained a little more headroom i lost that sweet compression that the Princeton reverb is known for. I put back the princeton OT and im happier knowing that the amp is what it is and sounds great as it was built in them old days. JMO

I agree, he will lose the Princeton in this almost complete rebuild.  I offered to simply make him a Blackface Deluxe, but noooooooooooooo.  It will be fun to see where this goes.

[Ed_Chambley]

If you want to trim bass, the most organic way to do it is to drop the 22uF bypass caps to 1-2.2uF. You won't need to do it on every stage. Try a few and taste-test.

1-2.2uf, that much?  I trust you experience.  I will give it a try.  I was thinking 4.7uf at the first stage, but it is simple enough to try smaller values.  Thanks for the input.

[Ed_Chambley]

That looks right to me.

Working Brad.  Too much to do right now.

That's good. A lot of people wish they were in your shoes right now.
            
                
                        Brad      

Did not say I was making money.  I have had a period of time where everything that can go wrong has.  Well not everything.  The most important things are fine.  Mine and my wife's health is good so counting blessings.

[sluckey]

The PaulC mod was often done in conjunction with the Stokes mod. It changes the PI from bootstrapped self bias to fixed bias operation. Easy to do or undo. I've never done this mod. See page 6 of this pdf for more info...

http://home.comcast.net/~seluckey/amps/5e3/5e3.pdf

[Willabe]

The Paul C mod was often done in conjunction with the Stokes mod. It changes the PI from bootstrapped self bias to fixed bias operation.

There's your bypass (sort of) it gets ride of the 1K5 R in the PI cathode.

I offered to simply make him a Blackface Deluxe, but noooooooooooooo.

Yeah well. You can lead a horse to water but you can't make him drink.    

Sorry to hear your going through a tough time, you'll pull out of it. Glad to hear you and your wife's health is good. I like the way you count.    


                    Brad      

[tubeswell]

Try 4M7 instead of 2M2 for the upper voltage divider resistor in the 'Paul C' mod. (Raises the input impedance slightly (820k instead of 688k) and makes the bias cooler (for more headroom).

[Ed_Chambley]

I see the reason for the Paul C mod.  Question. 

The stokes raises the plate voltage therefore adds headroom.  What is the max plate voltage for a 12Ax7?  The reason I ask is I checked the voltage of the previous node in my Princeton (which has more voltage from the PT) and do not think the tube could stand the increased voltage without a larger plate resistor.

Datasheet states 100 volts in series.  Any idea what might be optimum voltage?

[Ed_Chambley]

The PaulC mod was often done in conjunction with the Stokes mod. It changes the PI from bootstrapped self bias to fixed bias operation. Easy to do or undo. I've never done this mod. See page 6 of this pdf for more info...

http://home.comcast.net/~seluckey/amps/5e3/5e3.pdf

Thanks for the nice layout on the 5e3 as well.  I am building one for someone and I was just getting ready to make the board.  I made a nice pine cab with burled walnut inlay.  Not really my type of amp, but a lot of people really seem to like them.

[HotBluePlates]

1-2.2uf, that much?  I trust you experience.  I will give it a try.  I was thinking 4.7uf at the first stage, but it is simple enough to try smaller values.  Thanks for the input.

I used to own a great-sounding '67 Princeton Reverb; after I measured the bypass cap values, most were around 1-4uF.

1uF and 1.5kΩ results in a -3dB point of 106Hz, right near the bottom of the guitar's range.

Actual total resistance from cathode to ground will be a bit smaller, and your cap's value will vary within tolerance. Regardless, you'll find ~1uF will take the mud out of the bass without being very obviously bright.

[Ed_Chambley]

Try 4M7 instead of 2M2 for the upper voltage divider resistor in the 'Paul C' mod. (Raises the input impedance slightly (820k instead of 688k) and makes the bias cooler (for more headroom).

Here is one of my limitations, so I have researched and came up with nothing so far.  I get that the larger value will raise the impedance in this instance.  Since I cannot find (formula) to calculate it, my assumption is simply a 4m7 produces 820k.

While I am on this, the first bypass cap on V1 when it is 4.7uf, why is it common to raise the resistor value.  Normally when using 22uf you have a 1k5, but lowering the cap value to 4.7uf I see many times the schematic show a 2k7 or such resistor.  I understand the reason in general, but how do you calculate it?  Knowing this would help me save a lot of tinkering time.

[HotBluePlates]

... Question.  

The stokes raises the plate voltage therefore adds headroom.  What is the max plate voltage for a 12Ax7?  ...

According to G.E., 330v.

Note that "plate voltage" is the voltage from plate-to-cathode. I have occasionally seen supply voltage for such a 12AX7 stage in excess of 450vdc, and have also seen 12AX7 plate-to-cathode voltage over 350v in some vintage gear (not guitar amps) which I would deem "well-designed". I think we can find many examples of 6V6's being run well over their maximum plate voltage rating. My preference, though, is to stay within data sheet ratings.

Datasheet states 100 volts in series.  Any idea what might be optimum voltage?

I'm not sure what you mean by this.

Try 4M7 instead of 2M2 for the upper voltage divider resistor in the 'Paul C' mod. (Raises the input impedance slightly (820k instead of 688k) and makes the bias cooler (for more headroom).

Here is one of my limitations, so I have researched and came up with nothing so far.  I get that the larger value will raise the impedance in this instance.  Since I cannot find (formula) to calculate it, my assumption is simply a 4m7 produces 820k.

Look at the schematic Sluckey linked.

It's not obvious from that drawing that the 2.2MΩ connects from the B+ node to the grid of the phase inverter. There's also a 1MΩ from the grid to ground.

Now, pretend you're an electron that's part of the a.c. signal at the previous stage's plate. After you pass through the coupling cap, there are 2 directions you can move. You can either go through the 1MΩ to ground, or you can go through the 2.2MΩ to the filter cap, which is effectively the same as ground to alternating current. Remember, we assume the cap looks like a short-circuit to a.c., so it looks like a wire from the B+ node to ground.

Since as an a.c. electron you have 2 paths to move, those paths are effectively in parallel. I'll call "parallel resistance" Rp.
Rp = R1*R2/(R1+R2) = 2.2M*1M/(2.2M+1M) = 687.5kΩ

Repeat with 4.7M and 1M, and you get 824.6kΩ

I don't know if I'd arbitrarily raise/lower the resistors connecting to the grid of the inverter. Doing so, by definition, changes the bias of the phase inverter. I'd want to know the B+ voltage supplying that inverter to attempt to calculate a suitable value, or I'd closely check the value of voltage drop across the plate and cathode resistors to ensure they're same/similar. I do trust Tubeswell's judgement, though.

[HotBluePlates]

While I am on this, the first bypass cap on V1 when it is 4.7uf, why is it common to raise the resistor value.  Normally when using 22uf you have a 1k5, but lowering the cap value to 4.7uf I see many times the schematic show a 2k7 or such resistor.  I understand the reason in general, but how do you calculate it?  Knowing this would help me save a lot of tinkering time.

V1 of what amp? The AA1164 Princeton Reverb uses 25uF caps everywhere there is a bypass cap.

But let's assume your amp does have a 4.7uF bypass cap. If that's the case, you're thinking about the question backwards.

The cathode resistor sets the bias of the preamp stage, and that is an overriding decision. In other words, the resistor value is chosen for some very good reason.

But an unbypassed cathode resistor introduces local negative feedback, reducing stage gain. A "fully bypassed" resistor results in full stage gain as determined by the interaction between the plate load resistor and internal plate resistance of the tube.

"Fully bypassed" means that for any a.c. signal frequency of interest, the cap present such a low impedance compared to the resistor such that to the a.c. signal, the resistor appears to not be there. This gives an opportunity to tailor the frequency response of the stage. Because a cap's impedance rises and frequency goes down, you can select a cap such that higher frequencies aren't affected, while at some low frequency the gain of the stage drops to the value it has with an unbypassed cathode resistor.

We normally approach this in terms of the -3dB point; when the bypass cap's impedance equals the cathode resistor's resistance, the gain of the stage will drop 3dB from its fully-bypassed value.

So rearrange the formula for capacitive reactance, and set resistance, R, equal to capacitive reactance:
X_c = 1/(2*Π*f*C)

X_c -> Capacitive Reactance
f -> frequency
C-> capacitance, in Farads

to:
f = 1/(2*Π*R*C)

Example:
Assume a 2.7kΩ cathode resistor and a 4.7uF bypass cap.

f = 1/(2*Π*2700*0.0000047F) = 12.54Hz

So the stage gain will be down 3dB at 12.54Hz, and will continue to drop until it reaches the value of unbypassed gain at some lower frequency. Since the low E of the guitar is ~82Hz, you could argue this stage has more than adequate bass response, and could stand a smaller bypass cap especially if less bass is desired.

There is a simplifying assumption that makes this formula easier to use, but somewhat less accurate. However, it gets close enough given standard part values and tolerances, and the likelihood that you can know the exact number of Hz you want a change to occur at. The completely accurate method takes several more steps and more knowledge of the stage's operating characteristics.

[Ed_Chambley]

While I am on this, the first bypass cap on V1 when it is 4.7uf, why is it common to raise the resistor value.  Normally when using 22uf you have a 1k5, but lowering the cap value to 4.7uf I see many times the schematic show a 2k7 or such resistor.  I understand the reason in general, but how do you calculate it?  Knowing this would help me save a lot of tinkering time.

V1 of what amp? The AA1164 Princeton Reverb uses 25uF caps everywhere there is a bypass cap.

But let's assume your amp does have a 4.7uF bypass cap. If that's the case, you're thinking about the question backwards.

The cathode resistor sets the bias of the preamp stage, and that is an overriding decision. In other words, the resistor value is chosen for some very good reason.

But an unbypassed cathode resistor introduces local negative feedback, reducing stage gain. A "fully bypassed" resistor results in full stage gain as determined by the interaction between the plate load resistor and internal plate resistance of the tube.

"Fully bypassed" means that for any a.c. signal frequency of interest, the cap present such a low impedance compared to the resistor such that to the a.c. signal, the resistor appears to not be there. This gives an opportunity to tailor the frequency response of the stage. Because a cap's impedance rises and frequency goes down, you can select a cap such that higher frequencies aren't affected, while at some low frequency the gain of the stage drops to the value it has with an unbypassed cathode resistor.

We normally approach this in terms of the -3dB point; when the bypass cap's impedance equals the cathode resistor's resistance, the gain of the stage will drop 3dB from its fully-bypassed value.

So rearrange the formula for capacitive reactance, and set resistance, R, equal to capacitive reactance:
X_c = 1/(2*Π*f*C)

X_c -> Capacitive Reactance
f -> frequency
C-> capacitance, in Farads

to:
f = 1/(2*Π*R*C)

Example:
Assume a 2.7kΩ cathode resistor and a 4.7uF bypass cap.

f = 1/(2*Π*2700*0.0000047F) = 12.54Hz

So the stage gain will be down 3dB at 12.54Hz, and will continue to drop until it reaches the value of unbypassed gain at some lower frequency. Since the low E of the guitar is ~82Hz, you could argue this stage has more than adequate bass response, and could stand a smaller bypass cap especially if less bass is desired.

There is a simplifying assumption that makes this formula easier to use, but somewhat less accurate. However, it gets close enough given standard part values and tolerances, and the likelihood that you can know the exact number of Hz you want a change to occur at. The completely accurate method takes several more steps and more knowledge of the stage's operating characteristics.

Got the first formula, no problem.  Having trouble with f=.  In the formula you have Π.  Does Π=3.14? if so I do not get the same solution as you.

[sluckey]

Got the first formula, no problem.  Having trouble with f=.  In the formula you have Π.  Does Π=3.14? if so I do not get the same solution as you.

You will if you multiply 2 times 3.14 times 2700 times 0.0000047, press equal, THEN get the reciprocal (1/x). Or, if using a scientific calculator, don't overlook the importance of the parentheses.

[Willabe]

THEN get the reciprocal (1/x).

So that's what that buttons for! HaHa, and it works!       

Thanks!


                 Brad     

[Ed_Chambley]

Got it.  I guess sometimes my ignorance helps.  Another one for my "formula book".  Thanks HBP and Sluckey.

@ Brad, that is what that thing is for.  I simply missed it in the formula.

Wow, that is so cool to know. Going back to modding the princeton in an organic fashion.

so resistor 1k5 and cap 2.2=48.25Hz?  If this is correct, and applied to V1 of the princeton would it be a cut off?  Where would you get the guitar bass range?  This is why the suggestion of 1uf as you get 106.873...Hz + or - the caps tolerance.  Then further up the circuit additional adjustments can be made.

Sorry for the questions, but a lot of things are coming together at the same time in my mind.

[Ed_Chambley]

The lowest c note on piano is 16.35hz.  If I am calculating correctly, the stock princeton first bypass is ~4.25Hz.  Why would this be the case?  Is is just because the low watt amp was never intended to be dimed?

[HotBluePlates]

so resistor 1k5 and cap 2.2=48.25Hz?  If this is correct, and applied to V1 of the princeton would it be a cut off? 

Yes for your calculation of frequency (the calculator on the computer has a "pi" button, so I get a little different answer than when you used 3.14). No on that frequency being "cut off."

We could calculate all this, but take it on faith that with typical 100k/1.5k resistors a 12AX7 has a gain when bypassed of, say, 60. If the cathode resistor was unbypassed, gain will be cut to around half, or 30.

So for any frequency that sees the bypass cap as a small impedance compared to the cathode resistor will have the full amount of gain, while a frequency that sees the cap as a large impedance compared the cathode resistor will have its gain reduced to 30.

What's a "large impedance?" The usual rule of thumb is 10 times, so when the cap looks like 15k, we can confidently say gain of the stage is 30. Similarly, when the cap impedance is 1/10th the resistor (150Ω), we can be confident the gain is 60. Between these extremes is the case where the impedance of the cap equals the resistance of the resistor, which is what the earlier formula gave you.

A 2.2uF cap looks like 15kΩ at 4.82Hz, and 150Ω at 482Hz. Use/rearrange the formula for Xc to find/verify these.

So the gain at 48.25Hz in your example should be something between 30 and 60, and will be 3dB below the maximum gain of 60. -3dB correlates to half-power, but if we were talking voltage levels then "half voltage" would be -6dB. So talking voltage, -3dB is about 70.7%.

We have a max gain of 60, so -3dB is 60 * 0.707 = 42.4.

So with your example, we'll have a gain of 60 at 482Hz falling to 42.4 at 48.2Hz, ultimately leveling off at a gain of 30 at 4.82Hz.

[HotBluePlates]

Where would you get the guitar bass range?  This is why the suggestion of 1uf as you get 106.873...Hz + or - the caps tolerance. 

See this musical note to frequency conversion chart. The guitar Low E is E₁.

So you balance the selection of available parts with where you'd like to have an effect.

I should also point out that the assumption that an unbypassed cathode resistor results in half-gain is just a simplification. Unbypassed gain may be more than half the bypassed gain, or very much less, and depends on tube type, operating point and cathode resistor value.

But what I wanted you to see was that the "cut-off frequency" is not a brick wall below which nothing passes. Really, the whole things is a smooth slope from a higher gain to a lower gain (or in a hi-pass filter from full output to no output) which occurs over a broad range of frequencies. The -3dB point is called the cut-off frequency out of convenience, and to give an indication of where the slope is occurring.

I called this approach "organic" before because it seems if you want to reduce bass, why not just amplify it less? That seems to be more sensible than amplifying bass a lot, then trying to strip it out again with coupling caps.

The lowest c note on piano is 16.35hz.  If I am calculating correctly, the stock princeton first bypass is ~4.25Hz.  Why would this be the case?  Is is just because the low watt amp was never intended to be dimed? 

Can of worms. A lot of things might factor into the choice of a bypass cap.

Reactive components (caps and chokes) cause phase shift. Where their impedance is small, phase shift is near 0 degrees, is 45 degrees at the -3dB point, and approaches a limit of 90 degrees as their impedance becomes very large. Caps and chokes shift phase in opposite directions. So one reason to have such a low cutoff frequency may be to keep phase shifts well outside the range of frequencies that will be amplified. This can be particularly important inside a feedback loop. That's a whole other book...

If you have multiple filter networks (bypass caps, coupling caps and/or tone circuits) that share ranges where they're acting on the signal, the phase shift increases and the steepness of the filter slope increases. What that means to a bypass is instead of having such a wide range of frequencies where it goes from full-gain to minimum-gain, that range is narrowed and the cut seems more extreme. Within a feedback loop, this means more phase shift and more risk of turning negative feedback into positive feedback. To many shifts (roll-offs) too near each other will make feedback loops unstable and turn your amp into an oscillator.

Tubes might have hum due to heater voltage being electrostatically coupled to the cathode, or heater current leaking through the insulation to the cathode. This causes hum, but it can be eliminated by using a large bypass cap to shunt these voltages/currents to ground. I believe this is one of the big reasons the 5F6-A Bassman used a 250uF bypass cap on the first stage (as well as to ensure no cutting of gain in the bass region, since it was intended for Fender's electric bass).

Anyway, the staggering of cutoff points is part of the reason many amps use big coupling caps early in the amp, with progressively smaller caps as you work towards the output tubes.

[Ed_Chambley]

The cathode resistor sets the bias of the preamp stage, and that is an overriding decision. In other words, the resistor value is chosen for some very good reason.

[/quote]
The overriding decision to use 1k5 for a 12ax7 in V1 on all the fender amps I am familiar with is based in what?  In other words, what would a different value change?  Is this based on the voltage from the data sheet whereas 250v the datasheet states 1500 ohms and 200v states 1800?  So this is a given based on voltage?

Where in the princeton, 240v is published on the schematic so 1k5 is used where voltage is increased the cathode resistance drops.  Max bias -55 from the Philips sheet.  How do you measure bias on preamp tubes?

[sluckey]

How do you measure bias on preamp tubes?

Same as on a power tube. Bias is defined as the voltage difference between the grid and cathode. So, if you have zero volts on a grid and +2 volts on the cathode, the bias is -2v. If you have 50 volts on a grid and 52 volts on the cathode, the bias is still -2v.

[Ed_Chambley]

How do you measure bias on preamp tubes?

Same as on a power tube. Bias is defined as the voltage difference between the grid and cathode. So, if you have zero volts on a grid and +2 volts on the cathode, the bias is -2v. If you have 50 volts on a grid and 52 volts on the cathode, the bias is still -2v.

That is what I thought, but the reason I asked is information from the philips data sheet stating max bias -55 where I have never seen anything close to this in a preamp.  Just wondering what application would require a negative bias anywhere near -55?

Sorry for all the questions, but things are aligning in my mind as I study Merlins preamp book.

[Willabe]

Cathode followers have elevated voltage on it's cathode.

Concertina splitters have elevated cathode voltage.

So much sometimes new manufactor tubes can't take it and die unless you put a dc stand off voltage on the heaters. I think that's why the data sheet listed -55 max on the cathode?


              Brad      

[Ed_Chambley]

Cathode followers have elevated cathode voltage.

So much sometimes new manufactor tubes can't take it and die unless you put a dc stand off voltage on the heaters.


              Brad      

Got ya, thinking in the wrong area.

[Willabe]

I was wrong on that. They do have an elavated cathode but it's grid is elavated too. At least on some/most (?) CF.

Look at tweed bassman. It's cathode is 180vdc but so is it's grid.


                  Brad      

[Willabe]

It changes the PI from bootstrapped self bias to fixed bias operation.  See page 6 of this pdf for more info...

http://home.comcast.net/~seluckey/amps/5e3/5e3.pdf

The bootstrap PI's grid is elevated like the cathode but the fixed bias PI's grid is not.

(I think that's correct.)


                    Brad    

[Ed_Chambley]

I was wrong on that. They do have an elavated cathode but it's grid is elavated too. At least on some/most (?) CF.

Look at tweed bassman. It's cathode is 180vdc but so is it's grid.


                  Brad      

Look at this. This is confusing to me, but it does show a cathode follower with a high bias.

[Willabe]

Yes but a 300B is a power triode not a preamp/small bottle tube.

I'm sure that Sluckey, HBP, PRR or tubeswell will know an exaple off the top of their head. It might be for a radio circuit?

 
                 Brad    

[Willabe]

Oh, wait a minuite. I just saw the 6SN7 cathode.    

They have a high -dcv on it. -384dcv on it's cathode.


                    Brad    


(Edit; Sorry Ed but look in reply #29, I added something.)

[Ed_Chambley]

Oh, wait a minuite. I just saw the 6SN7 cathode.    

They have a high -dcv on it. -384dcv on it's cathode.


                    Brad    


(Edit; Sorry Ed but look in reply #29, I added something.)

I have been reading tubecad.com.  That is where the attachments come from.  Very difficult to follow as I have never seen anything like it, but as usual enough is sinking in to generate curiosity.

[sluckey]

Just wondering what application would require a negative bias anywhere near -55?

Many radar timing circuits used high bias in circuits that need to be absolutely cutoff and then switched on to saturation, ie, pulse generators, pulse amplifiers, multivibrators, switching circuits, etc. Not much use to audio devices though.

[HotBluePlates]

Is this based on the voltage from the data sheet whereas 250v the datasheet states 1500 ohms and 200v states 1800?  So this is a given based on voltage?

Bias is used to set an operating point, and to give desired operation to meet a particular goal or accommodate existing conditions.

Don't overlook the fact that resistance-coupled charts on some data sheets give conditions with a given supply voltage, and what happens to Eo (size of the output voltage swing) and Gain with various plate and cathode resistors. These charts are there to give a builder quick reference info to find suitable parts to use when they have a given supply voltage and either need maximum gain (or a particular amount of gain less than maximum), or a specific size of voltage output.

Why is 90v and 180v so common on data sheets? Because at one time all supply voltages were derived from batteries, or equipment had to be designed to operate the same way with either batteries and a.c. (or d.c.) wall voltage. 90v is ten 9v batteries, although high voltages used to be supplied by a dedicated single battery (maybe a 250v battery for your output tubes, then feeding the rest of the supply).

Anyway, short answer is note the value of bias for V1, given on the schematic as the 1.3v at the cathode. That's big enough that no pickup of the day would drive the grid appreciably close to 0v at any time, and typical guitar levels are small enough not to result in significant distortion in the output of that stage.

Where in the princeton, 240v is published on the schematic so 1k5 is used where voltage is increased the cathode resistance drops.  ...

So let's draw the loadline and operating point.

See the curves below. There is a 100k loadline drawn with a 240v supply voltage; the other end of the line lands at 0v and 2.4mA (240v/100kΩ).

There are 3 cathode resistor lines drawn, for 820Ω, 1.5kΩ and 2.7kΩ. These values were chosen because they are typical 10% resistor values that were used in lots of guitar amps.

The lines are drawn by assuming a bias voltage to the left/below the loadline (0.5v for instance), and dividing the chosen voltage by the resistor value to determine the current at that point. Then a second bias point is chosen that will likely be above/right of the loadline, and the process repeated to find a second point for the cathode line.

Where the cathode resistor line crosses the loadline is the idle plate voltage and idle plate current of the stage. The location of that point in relation to the grid voltage curves also indicates the resulting bias voltage.

With the 1.5k line I drew, I get a plate voltage of ~155v and a bias voltage between -1.25 to -1.3v (a little hard for me to guess). I note that the portion of the loadline from the 0v gridline down to 0mA and 240v is nearly-bisected by the cathode line. That means the tube is biased to an operating point where the plate voltage is sitting about mid-way of the usable plate voltage swing and thereby maximizes the output signal for that tube.

Note the 820Ω line crosses the loadline further towards the 0v gridline, and the 2.7kΩ line crosses closer to cutoff at the plate voltage axis.

The 2.7k line will likely result in less output voltage swing as it will be limited during negative-going inputs by approaching cutoff at the bottom of the graph. The 2.7k cathode resistor will likely result in more distortion that the 1.5k resistor, as the internal plate resistance changes significantly during signal swing, and the spacing of the grid lines is very different during positive and negative portions of the input signal.

The 820Ω line will also result in less output voltage, as the signal swing is limited on the positive-going portion of the input signal. However, it might actually have less distortion for small signals than the 1.5kΩ line, because the spacing of grid curves is more consistent, as is the angle of the  grid lines where they cross the loadline (this indicates internal plate resistance).

So 1.5kΩ was likely chosen as the best compromise of biggest output voltage while reducing distortion to an acceptable amount. To be fair, the input signal is small enough under normal circumstances that distortion will be minimal.

[HotBluePlates]

Max bias -55 from the Philips sheet.  ...

I can't find such a number on data sheets I've looked at.

But it doesn't really matter. Look at the plate curves I posted. The -5v gridline is down near 0mA of plate current for just about any plate voltage (not even supply voltage) up to near 400vdc. So for all practical purposes, -5v of bias will turn a 12AX7 off.

If you look hard enough at any stage using a 12AX7, you can be confident it will be biased somewhere between 1-3v between the grid and cathode. The voltage difference may not be obvious with some schematics.

[HotBluePlates]

Let's say you have your ideal tone with a 1.5kΩ resistor and a 22uF cap. The -3dB point is 4.8Hz.

But you've just acquired the most perfect of all caps: the Ultima Extremo with gold foil plates and teflon film dielectric immersed in extra virgin olive oil made from olives grown on Mount Olympus, hand-assembled by the delicate fingers of fairy strippers. No other cap will do in your amp! The only catch is they're only available in 2.2uF, and you're dead-set on using it. You absolutely must use this cap, so you're gonna adjust the cathode resistor value to maintain the same low frequency response.

Quick cheat: if you move the decimal one place to the left on the cap, you need to move it to the right on the resistor to arrive at the same -3dB point. So your new cathode resistor is 15kΩ.

This resistor and cap are going back into V1 of your Princeton.

Problem is, the large cathode resistance drops plate current, raises the bias voltage and causes the idle plate voltage to rise. The new bias lands at about -2.6v, plate voltage is now 222v, with a 240v supply! The operating point is very near cutoff and distortion is guaranteed with even a small input signal. Any input about 0.6v peak (or bigger) will result in hard-clipping.

This is an extreme case, but shows that if the cathode resistor is changed while everything else is kept constant (except your new bypass cap value), the operation of the tube can change significantly. That said, moderate changes of the cathode resistor are fine, often with corresponding changes in the plate load resistor. Or rather, a different plate load is selected, with a new value of cathode resistor to land on a desired operating point.

[Ed_Chambley]

Let's say you have your ideal tone with a 1.5kΩ resistor and a 22uF cap. The -3dB point is 4.8Hz.

But you've just acquired the most perfect of all caps: the Ultima Extremo with gold foil plates and teflon film dielectric immersed in extra virgin olive oil made from olives grown on Mount Olympus, hand-assembled by the delicate fingers of fairy strippers. No other cap will do in your amp! The only catch is they're only available in 2.2uF, and you're dead-set on using it. You absolutely must use this cap, so you're gonna adjust the cathode resistor value to maintain the same low frequency response.

Quick cheat: if you move the decimal one place to the left on the cap, you need to move it to the right on the resistor to arrive at the same -3dB point. So your new cathode resistor is 15kΩ.

This resistor and cap are going back into V1 of your Princeton.

Problem is, the large cathode resistance drops plate current, raises the bias voltage and causes the idle plate voltage to rise. The new bias lands at about -2.6v, plate voltage is now 222v, with a 240v supply! The operating point is very near cutoff and distortion is guaranteed with even a small input signal. Any input about 0.6v peak (or bigger) will result in hard-clipping.

This is an extreme case, but shows that if the cathode resistor is changed while everything else is kept constant (except your new bypass cap value), the operation of the tube can change significantly. That said, moderate changes of the cathode resistor are fine, often with corresponding changes in the plate load resistor. Or rather, a different plate load is selected, with a new value of cathode resistor to land on a desired operating point.

Thank you!  I understand.  What a great explanation.  BTW, I have never been dead set on any cap.  I actually harvest them from consumer grade electronics which have little use.  I will say once I get what I am looking for from an amp I will replace them with something of good quality for durability.  I used Sozo blues in my last build, but they were free.  I cannot hear any significant difference, but I can readily hear a difference in tubes.

Don't want to open the worm can on caps, but I do dip all my caps in extra virgin olive oil. :)

[Willabe]

I actually harvest them from consumer grade electronics which have little use.  
Don't want to open the worm can on caps, but I do dip all my caps in extra virgin olive oil. :)

So that's how you get that smooth tone.


                  Brad       

[Ed_Chambley]

I actually harvest them from consumer grade electronics which have little use.  
Don't want to open the worm can on caps, but I do dip all my caps in extra virgin olive oil. :)

So that's how you get that smooth tone.


                  Brad       

I want to see a Fairy Stripper.

[Willabe]

I want to see a Fairy Stripper.

Now I get it. That's what Ozzi saw.

"Fairies where boots and you gotta believe me. I saw it , I saw it with my own two eyes."


               

[PRR]

> I have been reading tubecad.com.

He's fond of tricky-stuff.

The plan with -384dcv on a cathode, the plate is directly-coupled to the grid of a *cathode follower output stage*. Whereas in plate-loaded power stage the plate swings 300V either way with only 30V of grid swing, a *cathode follower* making good use of a 400V supply will need *330V* of both-ways swing on its grid. That 6SN7 sits within 800V of total supply voltage, plate in the middle swinging up and down >300V both ways.

It works on paper. It damps a speaker without overall NFB. IIRC it was done once on one low-production commercial product. The disadvantages (heroic voltages, added tube, added heater windings) make it maybe not such a dandy idea except for experimenters who have run out of novel ideas.

> The overriding decision to use 1k5 for a 12ax7 in V1 on all the fender amps I am familiar with is based in what?

A 1950s Sylvania data-sheet. 300V supply, 220K next-stage load, Sylvania recommended 100K and 1.5K.

It works. Gain could be slightly higher with 1K, but hardly enuff to matter.

I think the main thing is: the oldest Fenders used much higher resistors, giving a treble limit near 15KHz (or KC to old guys). When Leo copied these 100K+1.5K values he got 30KC, which in a several stage amp means less loss at 5KC, a brighter tone and stand-out sound. He basically never questioned that result. History shows it is a good working area: most of Leo's imitators over 60 years have used the same values, pretty much under "all" 12AX7 stages.

> maybe a 250v battery for your output tubes

Consumer battery power never went above 135V.

90V is both a popular battery, and a near-enuff for 110V AC/DC radios' filtered B+.

250V comes from the maybe-450V limit of electrolytics, 100V drop in rectifier, 100V drop in field-coil. While the second filter cap lands near 250V, the first runs 350V, and at turn-on both rise near 450V for a few seconds until the power-output bottle gets warm. Conservative design leads to 250V on the amplifiers. Hence all the power-bottles with 250V suggested conditions, the RF IF tubes with 250V characteristics, and R-C tables which sometimes give a 250V B+ column.

[Ed_Chambley]

> I have been reading tubecad.com.

He's fond of tricky-stuff.

He is fond of tricky-stuff, for sure.  It is interesting.  The autobias circuit especially.  I will attach it.  The history of the origin of this is interesting and the idea seems, in theory, the best I have ever read.

Thank you for your input.  I know many things have been tried in tube amps outside the norm for guitar and it seems proven designs are best.  But understanding what you are doing first, before tweaking, can make life much easier.

Let me know your thought son the autobias.

[catnine]

When I look at fenders schmatics the 1.5K cathode and 25 uf  and 100K plate was common on the 12ax7 no matter what the preamp rail voltage was. They even used the same values with a 12ay7 .

 The main difference I see just based on P-P princetons was the rect tube. The 6G2 was a bit different because of the tone stack . Yet it had 280VDC on the preamp rail and a higher 12ax7 plate . Same thing on the AA964 but here there is a GZ34  290 vdc rail and even a higher 12ax7 plate then the AA1164 a different rect and lower rail of 240 VDC and lower 12ax7 plates then the added gain stage after the tone stack .